International Reference Life Cycle Data System Ilcd-Gp - Eudor - WEB - LBNA24571ENC - 002

Input products
Resources
ILCD handbook Emissions
International Reference Life Cycle Data System

(Co-)products
Tier-2
supplier Aa
Tier-2
supplier Ab
Tier-1
supplier A
Tier-1
supplier B
Manufacturer Use
Emissions
Waste
service
provider 1
Waste
service
provider 11
Waste
service
provider 12
Extraction 1
Material
production 1
Assembly 1 Use 1 Dis
Re
Tier-2 Tier-1
Input products
supplier Ac supplier C
Wastes
(Co-)products
Material
Extraction 2 Assembly 2 Use 2 Dis
production 2
Resources
Material
Extraction Assembly Use Disposal
production
Wastes
Assembly 1 Use 1 Disposal 1
Emissions
Recycling
Material
Emissions Extraction 1 Recycling
production 1
Input products
Material
tion 1 Assembly 1 Use 1 Disposal 1
production 1
Input products Assembly 2 Use 2 Disposal 2
(Co-)products Resources
Recycling Resources
Wastes
Material
production
Wastes
Material
tion 2 Assembly 2 Use 2 Disposal 2
production 2
Emissions
Recycling
Input products
Tier-2 Tier-1
Manufacturer Use
(Co-)products supplier Aa supplier A
Material p
Extraction 1 Recycling
production 1
Resources
Emissions Tier-2 Tier-1
supplier Ab supplier B
Wastes
Tier-2 Tier-1
supplier Ac supplier C
(Co-)products
Emissions
Material
production
Wastes Material
Input products Extraction 1 Assembly 1 Use 1 Disposal 1
production 1
(Co-)products Recycling
Resources
Material Recycling
production
Wastes
Waste Waste Material

Tier-2 Tier-1 Extraction 2 Assembly 2 Use 2 Disposal 2
Manufacturer Use service service Recycling production 2
supplier Aa supplier A
provider 1 provider 11
Waste
Tier-2 Tier-1 service Assembly 1 Use 1 Disposal 1
supplier Ab supplier B provider 12 Emissions
Material
Tier-2 Tier-1 Extraction 1 Recycling
production 1
supplier Ac supplier C Assembly 1 Use 1 Disposal 1
Input products
(Co-)products
Material
Extraction 1 Recycling Resources Assembly 2 Use 2 Disposal 2
production 1
Emissions
Wastes
EUR 24571 EN - 2011
Input products
Material
Extraction 1 Assembly 1 Use 1 Disposal 1
production 1 (Co-)products
Recommendations for Life Cycle Impact Recycling

Resources
Wastes
Assessment in the European context Extraction 2

Material
production 2
- based on existing environmental impact assessment models and factors

First edition
ILCD Handbook: Recommendations for Life Cycle Impact Assessment in the European context First edition
The mission of the JRC-IES is to provide scientific-technical support to the European Union’s Policies for the
protection and sustainable development of the European and global environment.
European Commission-Joint Research Centre - Institute for Environment and Sustainability: International
Reference Life Cycle Data System (ILCD) Handbook- Recommendations for Life Cycle Impact Assessment in the
European context. First edition November 2011. EUR 24571 EN. Luxemburg. Publications Office of the European
Union; 2011
European Commission
Joint Research Centre
Institute for Environment and Sustainability
Contact information
Address: Via E. Fermi, 2749– 21027 Ispra (VA) - Italy

E-mail: [email protected]
Fax: +39-0332-786645
http://lct.jrc.ec.europa.eu/
http://ies.jrc.ec.europa.eu/
http://www.jrc.europa.eu/
Legal Notice
Neither the European Commission nor any person acting on behalf of the Commission is responsible for the use
which might be made of this publication. References made to specific information, data, databases, or tools do
not imply endorsement by the European Commission and do not necessarily represent official views of the
European Commission.
Europe Direct is a service to help you find answers

to your questions about the European Union
Freephone number (*):
00 800 6 7 8 9 10 11
(*) Certain mobile telephone operators do not allow access to 00 800 numbers or these calls may be billed.
A great deal of additional information on the European Union is available on the Internet. It can be accessed
through the Europa server http://europa.eu/
JRC 61049
EUR 24571 EN
ISBN 978-92-79-17451-3
ISSN 1018-5593
doi:10.2788/33030
Luxemburg: Publication Office of the European Union, 2011
© European Union, 2011

Reproduction is authorised provided the source is acknowledged
Printed in Italy
i
Preface
To achieve more sustainable production and consumption patterns, we must consider the
environmental implications of the whole supply-chain of products, both goods and services,
their use, and waste management, i.e. their entire life cycle from “cradle to grave”.
In the Communication on Integrated Product Policy (IPP), (EC, 2003), the European
Commission committed to produce a handbook on best practice in Life Cycle Assessment
(LCA). The Sustainable Consumption and Production (SCP) Action Plan (EC, 2008)
confirmed that “(…) consistent and reliable data and methods are required to assess the
overall environmental performance of products (…)”. The International Reference Life Cycle
Data System (ILCD) Handbook, based on the existing international standards on LCA, ISO
14040/44, provides governments and businesses with a basis for assuring quality and
consistency of life cycle data, methods and assessments.
This guidance document provides recommendations on models and characterisation
factors that should be used for impact assessment in applications such as Life Cycle
Assessment (LCA). This supports the analyse of emissions into air, water and soil, as well as
the natural resources consumed in a single integrated framework in terms of their
contributions to different impacts on human health, natural environment, and availability of
resources. It supports the calculation of indicators for different impacts such as climate
change, ozone depletion, photochemical ozone formation, respiratory inorganics, ionising
radiation, acidification, eutrophication, human toxicity, ecotoxicity, land use and resource
depletion.
ii
Executive Summary
Overview
Life Cycle Thinking (LCT) and Life Cycle Assessment (LCA) are scientific
approaches behind a growing number of environmental policies and decision support
in business in the context of Sustainable Consumption and Production (SCP). The
International Reference Life Cycle Data System (ILCD) provides a common basis for
consistent, robust and quality-assured life cycle data, methods and assessments.
These support coherent and reliable business and policy instruments related to
products, natural resources, and waste management and their implementation, such
as eco-labelling, eco-design, and green procurement.
This guidance document provides recommendations on the methods to apply for
modelling of the most common impact categories, linking emissions and resources
consumed over the life cycle to the impact indicators.
About Life Cycle Impact Assessment (LCIA)

In a Life Cycle Assessment, the emissions and resources consumed linked to a
specific product are compiled and documented in a Life Cycle Inventory (LCI). An
impact assessment is then performed, generally considering three areas of
protection: human health, natural environment, and issues related to natural resource
use.
Impact categories considered in the so-called Life Cycle Impact Assessment
(LCIA) include climate change, ozone depletion, eutrophication, acidification, human
toxicity (cancer and non-cancer related), respiratory inorganics, ionizing radiation,
ecotoxicity, photochemical ozone formation, land use, and resource depletion. The
emissions and resources derived from LCI are assigned to each of these impact
categories. They are then converted into indicators using factors calculated by
impact assessment models. These factors reflect pressures per unit emission or
resource consumed in the context of each impact category. Emissions and resources
consumed, as well as different product options, can then be cross-compared in terms
of the indicators.
About the International Reference Life Cycle Data System (ILCD) Handbook
The ILCD Handbook is a series of detailed technical documents, providing
guidance for good practice in Life Cycle Assessment in business and government.
The ILCD Handbook can serve as “parent” document for developing sector- and
product-specific guidance documents, criteria and simplified tools. The ILCD
Handbook is based on the existing international standards on LCA, ISO 14040/44,
that provide the indispensable framework for LCA. This framework, however, leaves
the individual practitioner with a range of choices that can change the results and
conclusions of an assessment. Further guidance is therefore needed to support
iv
consistency and quality assurance. The ILCD Handbook has been set up to provide
this guidance.
Role of this Guidance Document within the ILCD Handbook

This guidance document presents
recommendations on the methods to
apply for modelling of the most
common impact categories, linking
emissions and resources consumed
to the indicators. It builds on two
other LCIA related ILCD documents,
the “Analysis of existing
Environmental Impact Assessment
methodologies for use in LCA” (EC-
JRC, 2010a) and the “Framework
and Requirements for LCIA models
and indicators” (EC-JRC, 2010b).
The recommendations are based on existing models assessed in the overall
framework of the Areas of Protection “Human Health”, “Natural Environment”, and
“Natural Resources”.
Approach and key issues addressed in this document

Several methodologies have been developed for LCIA and some efforts have
been made towards harmonisation. Starting from the first pre-selection of existing
methods and the definition of criteria, this report describes the recommended
methods for each impact category at both midpoint and endpoint.
Recommendations are given for the impact categories of climate change, ozone
depletion, human toxicity, particulate matter/respiratory inorganics, photochemical
ozone formation, ionising radiation impacts, acidification, eutrophication, ecotoxicity,
land use and resource depletion. Research needs are identified for each impact
category and differentiated according to their priority.
No method development has taken place in the development of this document.
The intention was to identify and promote current best practise. This document does
not provide recommendations for weighting across impact categories, nor for
normalisation within a given category relative to e.g. impacts in a given region.
v
CONTENTS
EXECUTIVE SUMMARY .................................................................................................................. IV

1 INTRODUCTION ......................................................................................................... 1
1.1 Summary of Recommended Methods ................................................................ 3
TABLE 2 RECOMMENDED MODELS AND THEIR CLASSIFICATION AT ENDPOINT.................................... 6
2 RECOMMENDED METHODS FOR IMPACT CATEGORIES .................................................. 7
2.1 Procedure for analysis and classification of methods ........................................... 7
2.1.1 Application of criteria and sub criteria .................................................................. 8
2.2 Recommendation levels ........................................................................................... 9
2.3 Research needs ...................................................................................................... 11
2.4 Geographic scope................................................................................................... 12
3 BACKGROUND INFORMATION ON THE EVALUATION OF EXISTING LCIA METHODS ........ 13
3.1 Climate change ....................................................................................................... 13
3.1.1 Introduction ....................................................................................................... 13
3.1.2 Pre-selection of methods for further evaluation ................................................. 13
3.1.3 Method evaluation ............................................................................................. 15
3.1.4 Discussion on method evaluation ...................................................................... 16
3.1.5 Discussion on uncertainties and the importance of spatial differentiation .......... 17
3.1.6 Recommended default method at midpoint level ............................................... 18
3.1.7 Recommended default method at endpoint level ............................................... 18
3.1.8 Consistency between midpoint and endpoint methods ...................................... 18
3.1.9 Classification of the recommended methods ..................................................... 19
3.1.10 Calculation principles ........................................................................................ 19
3.2 Ozone depletion ...................................................................................................... 19
3.2.4 Discussion of uncertainties ................................................................................ 22
3.2.8 Classification of the recommended default methods.......................................... 24
3.3 Human toxicity ........................................................................................................ 25
3.3.2 Environmental mechanism for human health effects ......................................... 25
3.3.5 Discussion of uncertainties and the importance of spatial- temporal
differentiation..................................................................................................... 29
3.4 Particulate matter/Respiratory inorganics ............................................................ 33
3.4.4 Recommended default method at midpoint (fate and exposure level) ............... 41
3.4.5 Recommended default method at endpoint (effect and severity level) ............... 41
vi

3.5 Ionizing radiation .................................................................................................... 42
3.5.3 Discussion on method evaluation: uncertainties and limitations......................... 45
3.5.4 Recommended default method for human health impact of ionizing
radiation ............................................................................................................ 46
3.5.5 Recommended default method for the ecosystem impacts of ionizing
radiation ............................................................................................................ 46
3.6 Photochemical ozone formation ............................................................................ 47
3.7 Acidification ............................................................................................................ 55
3.7.4 Discussion of uncertainties and the importance of spatial differentiation ........... 62
3.8 Eutrophication......................................................................................................... 65
3.8.4 Discussion of uncertainties and the importance of spatial differentiation ........... 74
3.9 Ecotoxicity............................................................................................................... 78
3.10 Land use .................................................................................................................. 84
3.10.4 Recommended default method at the midpoint level ......................................... 91
3.10.5 Recommended default method at the endpoint level ......................................... 91
vii

3.11 Resource depletion ................................................................................................. 92
3.11.2 Pre-selection of midpoint methods .................................................................... 94
3.11.3 Pre-selection of endpoint methods .................................................................... 95
3.11.6 Recommended default method for category 1 ................................................... 99
3.11.7 Recommended default method for category 2 ................................................. 100
3.11.8 Recommended default method for category 3 for water .................................. 100
3.11.9 Recommended default method for category 4 ................................................. 100
3.11.10 Consistency between midpoint and endpoint methods .................................... 101
3.11.11 Classification of the recommended default methods........................................ 101
3.11.12 Calculation principles ...................................................................................... 101
3.12 Other impacts ........................................................................................................ 102
3.12.1 State of the field .............................................................................................. 102
3.12.2 Framework ...................................................................................................... 102
3.12.3 Cause-effect chains......................................................................................... 107
3.12.4 Method selection ............................................................................................. 107
3.12.5 Method evaluation ........................................................................................... 107
3.12.6 Conclusion ...................................................................................................... 107
4 REFERENCES ....................................................................................................... 108
5 ANNEX 1 - CONSISTENCY ACROSS MIDPOINT AND ENDPOINT INDICATORS ................. 123
5.1 Handling of cultural perspectives and time perspective .................................... 124
6 ANNEX 2 - RESEARCH NEEDS ................................................................................ 125
6.1 Climate change ..................................................................................................... 125
6.2 Ozone depletion .................................................................................................... 126
6.3 Human toxicological effects................................................................................. 126
6.4 Particulate matter/Respiratory Inorganic ............................................................ 127
6.5 Ionizing radiation .................................................................................................. 127
6.6 Photochemical ozone formation .......................................................................... 128
6.7 Acidification .......................................................................................................... 129
6.8 Eutrophication....................................................................................................... 130
6.9 Ecotoxicological effects ....................................................................................... 131
6.10 Land use ................................................................................................................ 132
6.11 Resource depletion ............................................................................................... 133
6.12 Other impacts ........................................................................................................ 133
6.13 Ecological footprint .............................................................................................. 134
7 ANNEX 3 – DEVELOPMENT OF THIS DOCUMENT ....................................................... 135
viii
FIGURES
Figure 1 Environmental mechanism for climate change and associated LCIA methods.............15
Figure 2 Environmental impact pathways of ozone depletion ....................................................20
Figure 3 Position of LCIA methods in the impact pathways of human toxicity. ...........................26
Figure 4 Environmental mechanism for respiratory inorganics (derived from Humbert 2009) ....34
Figure 5 Overview of impact pathway stages of radioactive releases for human health
(adapted from Frischknecht et al. 2000).............................................................................43
Figure 6 Overview of impact pathway on ecosystem for radioactive releases to freshwater.
Plain lines refer to physical transfers of radioactive substances, whereas dotted lines
correspond to exposures to radioactive radiation. ..............................................................43
Figure 7 Flow diagram for photochemical ozone formation and position of analysed LCIA
methods along the cause-effect chain................................................................................49
Figure 8 Flow diagram for acidification. The LCIA methods are positioned along the cause-
effect chain. .......................................................................................................................58
Figure 9 Flow diagram of the cause-effect chain for eutrophication Method evaluation .............68
Figure 10 Flow diagram for ecotoxicity ......................................................................................79
Figure 11 Flow diagram of the cause-effect chain of land use (adapted from Weidema and
Lindeijer (2001)). ...............................................................................................................88
Figure 12 Overview of methods classification for resource depletion .........................................93
ix
TABLES
Table 1 Recommended methods and their classification at midpoint .......................................... 5

Table 2 Recommended models and their classification at endpoint ............................................ 6
Table 3 Summary of the evaluation results of 5 models that assess climate change in LCA
context. ..............................................................................................................................16
Table 4 Overview of the environmental pathways and indicators modelled in the selected
endpoint methods. .............................................................................................................20
Table 5 Summary of the evaluation results of 5 models that assess Ozone depletion in LCA
context. ..............................................................................................................................21
Table 6 Selected methods and underlying models for human toxicity effects for analysisa ........25
Table 7 Summary of the analysis of the available characterisation methods against the
adapted criteria for human toxicity (in two sub- tables) ......................................................27
Table 8 Selected methods and models for respiratory inorganics. .............................................33
Table 9 Comparison of intake fractions between different models .............................................35
Table 10 Treatment of effect and severity in the different models and methods evaluated
(modified from Humbert 2009). ..........................................................................................35
Table 11 Summary of the analysis of the available methods against the criteria for respiratory
inorganics. (divided in two sub tables) ...............................................................................37
Table 12 Selected methods and underlying models ..................................................................43
Table 13 Summary table on the analysis of the characterisation methods for ionizing
radiation.............................................................................................................................44
Table 14 Summary of the evaluation results of the midpoint characterisation methods against
the criteria for photochemical ozone formation...................................................................50
Table 15 Summary of the evaluation results of the midpoint and endpoint characterisation
methods against the criteria for photochemical ozone formation. .......................................51
Table 16 Summary table documenting the analysis of the midpoint characterisation methods
against the adapted criteria for acidification. (split into two tables) .....................................59
Table 17 Summary table documenting the analysis of the endpoint characterisation methods
against the adapted criteria for acidification. ......................................................................61
Table 18 Summary of the evaluation results of the midpoint characterisation methods against
the criteria for aquatic eutrophication. ( at mid and endpoint) .............................................69
Table 19 Summary of the evaluation results of the endpoint characterisation methods
against the criteria for terrestrial eutrophication. ................................................................72
Table 20 Selected midpoint methods and underlying models for ecotoxicity..............................78
Table 21 Summary table documenting the analysis of five midpoint characterisation methods
against the adapted criteria for ecotoxicity. ........................................................................80
Table 22 Summary table documenting the analysis of the two remaining midpoint models
and three endpoint models for ecotoxicity. .........................................................................81
Table 23 Selected midpoint methods and underlying models for land use. ................................85
Table 24 Summary of the evaluation results of three midpoint models that assess land use in
an LCA context. .................................................................................................................89
Table 25 Summary of the evaluation results of five endpoint models that assess land use in
an LCA context. .................................................................................................................90
Table 26 Selected methods and underlying models for resource depletion. (see description
of each method below) ......................................................................................................93
Table 27 Summary of the analysis of six midpoint characterisation methods against the
adapted criteria for resources. ...........................................................................................97
Table 28 Summary of the analysis of four endpoint characterisation methods against the
adapted criteria for resources. ...........................................................................................98
Table 29 Research needs for impacts covered under “other impacts” .....................................134
x
ACRONYMS
AG Advisory Group of the European Platform on LCA

AoP Area of Protection
CF Characterisation Factors
DALY Disability Adjusted Life Year
ELCD European Reference Life Cycle Data system
EPLCA European Platform on Life Cycle Assessment
GWP Global Warming Potential
ICRP International Commission on Radiological Protection
ILCD International Life Cycle Data system
IPCC Intergovernmental Panel on Climate Change
JRC Joint Research Centre
LCI Life Cycle Inventory
LCIA Life Cycle Impact Assessment
MIR Maximum Incremental Reactivity
NMVOC Non-Methane Volatile Organic Compounds
NPP Net Primary Production
PAF Potentially Affected Fraction of species
PDF Potentially Disappeared Fraction of species
SETAC Society of Environmental Toxicology and Chemistry
UNEP United Nations Environment Program
UNECE United Nations Economic Commission for Europe
US EPA United States Environment Protection Agency
WHO World Health Organisation
WMO World Meteorological Organisation
In the text
LCIA – Analysis document: ILCD - Handbook Analysis of existing Environmental
Impact Assessment methodologies for use in Life Cycle Assessment (EC-JRC, 2010a)
LCIA – Framework and requirements document: ILCD – Handbook Framework
and Requirements for LCIA models and indicators (EC-JRC, 2010b)
xi
GLOSSARY
Definiendum Definition
Area of A cluster of category endpoints of recognisable value to society, viz.
protection (AOP) human health, natural resources, natural environment and sometimes
man-made environment (Guinée et al., 2002)
Cause-effect or environmental mechanism.System of physical, chemical and
chain biological processes for a given impact category, linking the life cycle
inventory analysis result to the common unit of the category indicator
(ISO 14040) by means of a characterisation model.
Characterisation A step of the Impact assessment, in which the environmental
interventions assigned qualitatively to a particular impact category (in
classification) are quantified in terms of a common unit for that
category, allowing aggregation into one figure of the indicator result
(Guinée et al., 2002)
Characterisation Factor derived from a characterisation model which is applied to
factor convert an assigned life cycle inventory analysis result to the common
unit of the impact category indicator (ISO 14040)
Characterisation Throughout this document an “LCIA methodology” refers to a

methodology, collection of individual characterisation “methods” or characterisation
methods, “models”, which together address the different impact categories,
models and which are covered by the methodology. “Method” is thus the individual
factors characterisation model while “methodology” is the collection of
methods. The characterisation factor is, thus, the factor derived from
characterisation model which is applied to convert an assigned life
cycle inventory result to the common unit of the category indicator.
xii
Definiendum Definition
Classification A step of Impact assessment, in which environmental interventions are

assigned to predefined impact categories on a purely qualitative basis
(Guinee et al 2002)
Elementary Material or energy entering the system being studied has drawn from
flow the environment without previous human transformation (e.g. timber,
water, iron ore, coal) , or material or energy leaving the system being
studied that is released into the environment without subsequent
human transformation (e.g. CO2 or noise emissions, wastes discarded
in nature) (ISO 14040)
The category endpoint is an attribute or aspect of natural environment,
Endpoint
human health, or resources, identifying an environmental issue giving
method/model
cause for concern (ISO 14040). Hence, endpoint method (or damage
approach)/model is a characterisation method/model that provides
indicators at the level of Areas of Protection (natural environment's
ecosystems, human health, resource availability) or at a level close to
the Areas of Protection level.
Environmental A consequence of an environmental intervention in the environment
impact system (Guinee et al 2002)
Environmental A human intervention in the environment, either physical, chemical or
intervention biological; in particular resource extraction, emissions (incl. noise and
heat) and land use; the term is thus broader than “elementary flow”
(Guinee et al 2002)
Environmental The result of the characterisation step showing the indicator results for
profile all the predefined impact categories, supplemented by any other
relevant information (Guinee et al 2002)
Impact Class representing environmental issue of concern (ISO 14040). E.g.
category Climate change, Acidification, Ecotoxicity etc.
Impact Quantifiable representation of an impact category (ISO 14040). Eg Kg
category CO2-equivalents for climate change
indicator
Life cycle "Phase of life cycle assessment involving the compilation and
impact quantification of inputs and outputs for a given product system
assessment throughout its life cycle." (ISO 14040) The third phase of an LCA,
(LCIA) concerned with understanding and evaluating the magnitude and
significance of the potential environmental impacts of the product
system(s) under study
The midpoint method is a characterisation method that provides
Midpoint
indicators for comparison of environmental interventions at a level of
method
cause-effect chain between emissions/resource consumption and the
endpoint level.
Sensitivity A systematic procedure for estimating the effects of choices made
analysis regarding methods and data on the outcome of the study (ISO 14044)
xiii
1 Introduction
Life Cycle Thinking (LCT) is a core concept in Sustainable Consumption and Production
(SCP) for business and policy. The environmental pillar of LCT is supported by Life Cycle
Assessment (LCA), an internationally standardised tool (ISO14040 and ISO14044) for the
integrated environmental assessment of products (goods and services). Upstream and
downstream consequences of decisions must be taken into account to help avoid the shifting
of burdens from one impact category to another, from one country to another, or from one
stage to another in a product’s life cycle from the cradle to the grave.
A Life Cycle Assessment consists of four phases (ISO 14040). In the Goal and scope
definition phase, the aim of the LCA is defined and the central assumptions and system
choices in the assessment are described. In the Life Cycle Inventory (LCI) phase, the
emissions and resources are quantified for the chosen products. In the Life Cycle Impact
Assessment (LCIA) phase, these emissions and resource data are translated into indicators
that reflect environment and health pressures as well as resource scarcity. This calculation is
based on factors which represent the predicted contribution to an impact per unit emission or
resource consumption. These factors are generally calculated using models. In each phase,
in the Interpretation phase, the outcome is interpreted in accordance with the aim defined in
the goal and scope of the study.
Since the early 1990’ies numerous LCIA methodologies1 have been developed. The use
of several different LCIA methods makes it difficult to compare LCA results and interpret
them. To some extent the differences represent different LCIA approaches that may be of
interest in certain applications. But a default/baseline method is needed and a single method
may be needed in some applications.
The ISO 14042 standard on impact assessment published in 1999, and the later update in
the ISO 14044 standard in 2006, brought some standardization on basic
principles/framework. This addresses the choice of models2 in very general terms, and most
of the existing LCIA methodologies can be seen as ISO compatible.
As the ISO guidelines on LCA provide a framework rather than technically detailed
standardisation, the SETAC working groups, later followed by task forces under the UNEP-
SETAC Life Cycle Initiative, started voluntarily work on scientific consensus and
development of a recommended best practice. These have been complemented by activities
of many other organisations such as JEMAI, US EPA and the European Commission. As
result of these activities, recommendations on the best approaches and the underlying
principles were developed; see for example Udo de Haes et al, 2002. Achievements include:
1
See the glossary: Throughout this document an “LCIA methodology” refers to a collection of individual
characterisation “models” or characterisation “methods”, which together address the different impact categories,
which are covered by the methodology. “Method” is thus the individual characterisation model while
“methodology” is the collection of methods.
See the glossary
2
1
1 Introduction
 Consensus on the need to integrate midpoint and endpoint3 models in a consistent

framework to combine the advantages of both concepts (Bare et al., 1999, Bare et
al., 2000).
 A generic set of quality criteria for assessing different methods, and the application
of these criteria on the most widely used impact assessment methods (Udo de
Haes et al., 2002, Margni et al., 2008).
 A growing global consensus among model developers based on best practice for
e.g. toxicological effects (fate, exposure and effect). (Hauschild et al., 2008,
Rosenbaum et al., 2008).
This is the setting of the International Reference Life Cycle Data System (ILCD), to
develop a coherent and consistent LCIA methodology (framework, characterisation models,
and characterisation factors) based on an analysis of existing characterisation models,
factors and insights.
The International Reference Life Cycle Data System (ILCD) Handbook is a series of
technical guidance documents for LCA that complement the International Standards to
provide the basis for greater consistency and quality of life cycle data, methods and
assessments. This is also the basis for the ILCD Data Network, an open network of inventory
data sets provided by government and private organisations that help guarantee quality and
consistency.
Reflecting the global nature of product life cycles and the necessity of having globally
agreed methods and data, the ILCD is developed in close collaboration with UNEP and with
participating national authorities developing LCA databases. This is facilitated by the
European Commission, including interactions with representatives of its 27 Member States
and Advisory Groups (AG) from business associations, software and database developers,
as well as life cycle impact assessment methodology developers4. Recognising that most
product systems include activities at global level, the recommendations aim for a global
validity.
This document is a part of a series of documents developed to give recommendations on
the framework and methods for LCIA:
1. ILCD Handbook: Analysis of existing Environmental Impact Assessment
methodologies for use in Life Cycle Assessment [LCIA – Analysis document- EC-JRC
(2010a)]
2. ILCD Handbook Framework and requirements for LCIA models and indicators [LCIA –
Framework and Requirements document EC-JRC (2010b)] focused on definition of
evaluation criteria for recommended LCIA methods and general recommendations for
characterisation models and Areas of Protection
3. Guidance on recommended LCIA characterisation methods (models and factors)
See the glossary.

3
see http://lct.jrc.ec.europa.eu/assessment/partners for details

4
2
1 Introduction
4. Characterisation factors of the recommended methods, accessible as spreadsheet via

the website of the European Platform on LCA5
This report describes, the recommended methods for each impact category at both
midpoint and endpoint level starting from the first pre-selection of existing methods (LCIA –
Analysis document, EC-JRC, 2010b) and the definition of criteria (LCIA – Framework and
Requirements document, EC-JRC, 2010b).
1.1 Summary of Recommended Methods

The tables below present the summary of recommended methods (models and
associated characterisation factors) and their classification both at midpoint and at endpoint.
The recommended characterisation models and associated characterisation factors are
classified according to their quality into three levels: “I” (recommended and satisfactory),
level “II” (recommended but in need of some improvements) or level “III” (recommended,
but to be applied with caution). A mixed classification sometimes is related to the application
of the classified method to different types of substances.
Out of the methods that were listed, other methods were included in the analysis but not
recommended because they were not mature for recommendation.
In the summary table, the classification “interim” indicates that a method was considered
the best among the analysed methods for the impact category, but still immature to be
recommended. This does not indicate that the impact category would not be relevant but
further efforts are needed before a recommendation for use can be given. The evaluation of
the methods is reported in the description of each impact category (Chapter 3).
For more clarification, the reader is referred to the section of Chapter 2 where details on
the assessment procedure, based on both scientific and stakeholder’s acceptance criteria
are presented.
The recommendations in this document take into account models that have been
available and sufficiently documented for an in depth evaluation in mid 2008. Models
developed after this date, have not been taken into account.
If a study intends to claim to be in compliance with the ILCD Handbook, and uses
midpoint indicators for the assessment, the models and factors at midpoint that have a
level I, level II or level III recommendation shall be used. Any geographical differentiation,
addition of factors for individual flows and addition of impact methods for not yet covered
impacts or improvements on methods have to be explicitly justified and reflected in both the
goal and scope definition and in the results interpretation (please also refer to the “ILCD
Handbook – General guide – Detailed guidance”, chapter 6.7.2 to 6.7.5 and the related
“Provisions 6.7 Preparing the basis for the impact assessment.”)
Analogously, if a study intends to claim to be in compliance with the ILCD Handbook, and
wants to use endpoint indicators for the assessment, the recommended models and
5
3
1 Introduction
factors at endpoint, that have a level I, level II or level III recommendation, are to be used.
Any geographical differentiation, addition of factors for individual interventions and addition of
impact methods for not yet covered impacts or improvements on methods have to be
explicitly justified and reflected in both the goal and scope definition and in the results
interpretation as stated in the “ILCD Handbook – General guide – Detailed guidance”,
chapter 6.7.2 to 6.7.5 and the related “Provisions 6.7 Preparing the basis for the impact
assessment.” Due to the unavailability of sufficiently mature models for most endpoint
categories, recommendations currently can be provided only for a few categories. In order to
meet ISO and ILCD requirements to include all relevant environmental impacts, the study
needs to provide and use endpoint models and factors for all relevant environmental impacts
also for those where currently no recommended ILCD method exists. This applies, unless an
explicit restriction to a limited set of categories is stated in the study goal as defined in the
“ILCD Handbook – General guide – Detailed guidance”, chapter 5.2.2 and the related
“Provisions: 5.2 Six aspects of goal definition”.
For those impacts where no models have been recommended, the methods classified as
“interim” can be considered as the best among the analysed methods but not mature for
recommendation.
Please note that the use of a reduced set of impact categories shall be explicitly
considered in the results interpretation and be explicitly communicated to the target
audience. See the “ILCD Handbook – General guide – Detailed guidance”, Provisions: 6.10
Comparisons between systems”, provision VII.”
As stated in the ILCD Handbook, the selection of the impact categories must be
consistent with the goal of the study and the intended applications of the results, and it must
be comprehensive in the sense that it covers all the main environmental issues related to the
system.
4
1 Introduction
Table 1 Recommended methods and their classification at midpoint
Recommendation at midpoint
Impact category Recommended default LCIA method Indicator Classific

ation
Climate change Baseline model of 100 years of the IPCC Radiative forcing as Global I
Warming Potential (GWP100)
Ozone depletion Steady-state ODPs 1999 as in WMO Ozone Depletion Potential I

assessment (ODP)
Human toxicity, cancer USEtox model (Rosenbaum et al, 2008) Comparative Toxic Unit for II/III
effects humans (CTUh)
Human toxicity, non- USEtox model (Rosenbaum et al, 2008) Comparative Toxic Unit for II/III
cancer effects humans (CTUh)
Particulate RiskPoll model (Rabl and Spadaro, Intake fraction for fine particles I
matter/Respiratory 2004) and Greco et al 2007 (kg PM2.5-eq/kg)
inorganics
Ionising radiation, Human health effect model as developed Human exposure efficiency II
235
human health by Dreicer et al. 1995 (Frischknecht et al, relative to U
2000)
Ionising radiation, No methods recommended Interim

ecosystems
Photochemical ozone LOTOS-EUROS (Van Zelm et al, 2008) Tropospheric ozone II

formation as applied in ReCiPe concentration increase
Acidification Accumulated Exceedance (Seppälä et al. Accumulated Exceedance II

2006, Posch et al, 2008) (AE)
Eutrophication, Accumulated Exceedance (Seppälä et al. Accumulated Exceedance II

terrestrial 2006, Posch et al, 2008) (AE)
Eutrophication, aquatic EUTREND model (Struijs et al, 2009b) Fraction of nutrients reaching II
as implemented in ReCiPe freshwater end compartment
(P) or marine end
compartment (N)
Ecotoxicity USEtox model, (Rosenbaum et al, 2008) Comparative Toxic Unit for II/III
(freshwater) ecosystems (CTUe)
Ecotoxicity (terrestrial No methods recommended

and marine)
Land use Model based on Soil Organic Matter Soil Organic Matter III
(SOM) (Milà i Canals et al, 2007b)
Resource depletion, Model for water consumption as in Swiss Water use related to local III
water Ecoscarcity (Frischknecht et al, 2008) scarcity of water
Resource depletion, CML 2002 (Guinée et al., 2002) Scarcity II

mineral, fossil and
6
renewable
6
Depletion of renewable resources is included in the analysis but none of the analysed methods is mature for
recommendation
5
1 Introduction
Table 2 Recommended models and their classification at endpoint
Recommendation from midpoint to endpoint
Impact category Recommended default LCIA method Indicator Classific

ation
Climate change No methods recommended interim
Ozone depletion No methods recommended interim
Human toxicity, DALY calculation applied to USEtox Disability Adjusted Life Years II/interim
cancer effects midpoint (Adapted from Huijbregts et al., (DALY)
2005a)
Human toxicity, non- No methods recommended interim

cancer effects
Particulate DALY calculation applied to midpoint Disability Adjusted Life Years I/II
matter/Respiratory (adapted from van Zelm et al, 2008, (DALY)
inorganics Pope et al, 2002)
Ionising radiation, No methods recommended interim

human health
Ionising radiation, No methods recommended

ecosystems
Photochemical ozone Model for damage to human health as Disability Adjusted Life Years II
formation developed for ReCiPe (Van Zelm et al, (DALY)
2008)
Acidification No methods recommended interim
Eutrophication, No methods recommended

terrestrial
Eutrophication, No methods recommended interim

aquatic
Ecotoxicity No methods recommended

(freshwater,
terrestrial and
marine)
Land use No methods recommended interim
Resource depletion, No methods recommended

water
Resource depletion, No methods recommended interim

mineral, fossil and
renewable
6
1 Introduction
2 Recommended methods for impact categories

The recommendations in this guidance document are made based on an analysis of a
wide range of existing methods used in LCIA, supplemented by a selection of environmental
models that cannot currently be found integrated into LCIA methodologies, but which have
interesting features to consider in the development of recommendations for LCIA.
First step of the analysis was a pre-selection of models based mainly on whether they
were in use in LCIA or contained elements which could be interesting for the straightforward
development of models for use in LCA. If a method is used in multiple LCIA methodologies,
only the most recent and up to date version of that method was considered (please refer to
the LCIA- Analysis document, EC-JRC, 2010a).
The second step was the development of general recommendations for each category
and the definition of assessment criteria to be used in the evaluation and comparison of the
pre-selected methods. The “LCIA- Framework and requirements” (EC-JRC, 2010b)
document describes the results in term of recommendations and criteria developed.
2.1 Procedure for analysis and classification of methods

It is the purpose of this document to recommend global default/baseline characterisation
models and characterisation factors for each impact category. In doing so, this Guidance
Document draws on what is available in existing LCIA methodologies7 supplemented by a
selection of environmental models that cannot currently be found as integrated into LCIA
methodologies, but nevertheless have features which may be interesting to consider in the
development of recommendations for LCIA. The first activity has thus been the identification
and pre-selection of characterisation models at midpoint and endpoint level (see “LCIA-
Analysis” document, EC-JRC, 2010a)).
In order to support the selection of the best methods, criteria for good characterisation
modelling practice have been developed in advance to be used in the evaluation and
comparison of the pre-selected methods.
Next to the evaluation criteria also expert judgement was applied in the decision making
process and helped select the LCIA methods that are recommended.
The development of criteria and their application in evaluation of methods from the
different impact categories have been decided in a consultation process involving domain
The following LCIA methodologies were scrutinized for characterisation models which would be potential
7
candidates for recommendation: CML 2002 (Guinée et al., 2002); Eco-Indicator 99 (Goedkoop and Spriensma,
2000); EDIP (1997-2003) (Wenzel et al.,1997, Hauschild and Wenzel, 1998a, Hauschild and Potting, 2005,
Potting and Hauschild, 2005); EPS2000 (Steen, 1999a,b); Impact 2002+ (Crettaz et al., 2002, Jolliet et al., 2004,
Payet, 2004, Pennington et al., 2005, Pennington et al., 2006, Rochat et al., 2006, Rosenbaum, 2006,
Rosenbaum et al., 2007); LIME (Itsubo et al., 2004, Hayashi et al., 2000, Hayashi et al., 2004, Hayashi et al.,
2006, Itsubo et al., 2008a-d); LUCAS (Toffoletto et al., 2007); ReCiPe (De Schryver et al., 2007, Huijbregts et al.,
2005a,b, Struijs et al., 2007, Van Zelm et al., 2007a-b, Wegener Sleeswijk et al., 2008); Swiss Ecoscarcity or
Ecological scarcity (Brand et al., 1998, Müller-Wenk, 1994, Ahbe et al., 1990, Frischknecht, 2008,2006a); TRACI
(Bare, 2002, Bare et al., 2003, Hertwich et al., 1997, Hertwich et al., 1998, Hertwich et al., 1999, Hertwich et al.,
2001, Norris, 2002); MEEuP methodology (Kemna et al., 2005); EcoSense (IER 2008)
7
experts for the respective impact categories, the European Commission and EU Member
States representatives, and international partners, as described in Annex 3. Finally, a public
stakeholder consultation has been carried out (partecipants also acknowledged in Annex 3).
2.1.1 Application of criteria and sub criteria

The criteria consist of general criteria based on fundamental requirements for LCIA
methods (both characterisation models and factors), which are the same for all impact
categories. These are complemented by minor groups of specific sub-criteria, which are
addressing the characteristic features of each individual impact category and are outlined in
LCIA- Framework and requirements document (EC-JRC, 2010b).
General criteria
The general criteria focus separately on scientific qualities and on stakeholder acceptance
and applicability to LCI data sets.
Scientific criteria
1. Completeness of scope
2. Environmental relevance
3. Scientific robustness and certainty
4. Documentation, transparency and reproducibility
5. Applicability
Stakeholder acceptance criterion

6. Degree of stakeholder acceptance and suitability for communication in a business
and policy contexts
Each criterion is specified through a number of sub criteria.
Specific criteria
Prior to developing the specific criteria, the environmental mechanism of the impact
category in question was described with a flow diagram with all relevant pathways and flows
which might be included in a characterisation model.
Based on the methods analysis and supported by the diagram, a limited number of
additional category-specific sub criteria were developed under the two criteria:
‘Environmental relevance’ and ‘Scientific robustness and certainty’ to complement the
general criteria and adapt them to the specificities of the impact category, capturing the
central characteristics of that category and the decisive points at which the analysed
characterisation methods differ and thus supporting discrimination between the different
methods.
8
A detailed description of criteria and sub criteria is given in the LCIA- Framework and
requirements document.
The detailed assessment of the characterisation methods for each impact category is
provided in separate spreadsheets8. The spreadsheets were used as supporting working
documents during the expert judgement processes
For each criterion and sub criterion a score was assigned to the characterisation models
reflecting the compliance of the model with the criterion or sub criterion requirements. The
used score are provided below:
A: Full compliance
B: Compliance in all essential aspects
C: Compliance in some aspects
D: Little compliance
E: No compliance
For the overall evaluation of the characterisation model, the importance of each criterion
and sub criterion needs to be assessed for the impact category in question. A differentiation
between normal (N) and high (H) importance is applied. Criteria of high importance are
criteria which address fundamental aspects of significance for the resulting characterisation
factors.
Some of the sub criteria are so important that an exclusion threshold is defined as a
required minimum performance below which the characterisation model will not be
considered any further in the analysis. Whenever a characterisation model fails to pass such
an exclusion threshold, the analysis of that characterisation model stops.
In order to support an overall evaluation, a score for each criterion is developed based on
an evaluation of the scoring of the sub-criteria. For the science based criteria, an overall
score is then developed based on the scoring of each science based criterion. The
compilation of the scores is based on “expert judgement” including consideration of the
importance of different criteria and sub criteria. The resulting statements on the science
based criteria and on the stakeholder acceptance criterion are the bases of the final method
recommendations.
The findings from the evaluation are summarized in recommendations on the
characterisation method for each impact category.
2.2 Recommendation levels

The recommended characterisation methods (models and associated characterisation
factors) are classified according to their quality into three levels: “I” (recommended and
satisfactory), “II” (recommended but in need of some improvements) or “III” (recommended,
but to be applied with caution). A detailed description of the levels is provided below:
Available at http://lct.jrc.ec.europa.eu
8
9
Level I: Recommended and satisfactory

Definition: These models and characterisation factors are recommended for all types of
life cycle based decision support. Although further research needs may have been identified,
these are not preventing the models/factors being seen as satisfactory given the current
state-of-the art. However, updating and improvement via established mechanisms, such as
e.g. the IPCC, should be followed also for these methods and factors.
Level II: Recommended, some improvements needed
Definition: The models and characterisation factors are recommended for all types of life
cycle based decision support. The uncertainty of models and the resulting characterisation
factors is to be more strongly highlighted. The impact on results and interpretation has to be
more carefully evaluated, especially in published comparisons. The need for dedicated
further research is identified for these methods/factors to further improve them in terms of
precision, differentiation, coverage of elementary flows etc.
Level III: Recommended, but to be applied with caution
Definition: These models and characterisation factors are recommended to be used but
only with caution given the considerable uncertainty, incompleteness and/or other
shortcomings of the models and factors. These models/factors are in need of further
research and development before they can be used without reservation for decision support
especially in comparative assertions. The recommendation is to calculate and present the
results of the LCIA with and without methods that are level III and to discuss the differences,
e.g. in the interpretation of the LCA. It is also recommended to conduct sensitivity analyses
applying – if available - other methods than the level III recommended ones and to discuss
differences in the results, e.g. in the interpretation of the LCA. However, the level III
recommended method should remain the baseline.
Interim: immature for recommendation but the most appropriate among the existing
approaches
Definition: The methods and characterisation factors defined as interim are to be used
only with extreme caution, and limited to in-house applications, given the considerable
uncertainty, incompleteness and/or other shortcomings of the methods and factors.
Note that for some impact categories there were no existing models and factors that met
the criteria for level III. For these impact categories no method is recommended in the ILCD
System, as the level of maturity and/or available documentation is considered too limited to
facilitate general use.
The fact that an impact category at midpoint or endpoint has no recommended methods
hence does not mean that it is not relevant to include in a study, but merely that at the
moment no existing method was found sufficiently mature for recommendation.
This should not be taken as a recommendation to exclude this specific impact category,
but to apply a method which has been identified by the practitioner as the current best
practise for the specific application. However, in the study the uncertainties and the
limitations have to be clearly stated, in particular for this impact category.
10
In specific situations, even if there is a recommendation to use a LCIA method, the use of
a different LCIA method could be accepted, provided that two conditions are met:
1. The LCIA method different from the recommended method is more suitable for the
circumstances of the specific case
2. The LCIA method is in compliance with ILCD requirements.
To 1: This can be relevant especially in case of specific geographical relevance of the
chosen model (models or factors developed for a specific country/ climate etc.). It has to be
justified that this will significantly reduce the uncertainty associated with the impact
assessment in the particular life cycle assessment, and it has to be justifiable in accordance
with the goal of the study.
To 2: In the ILCD Handbook “Review schemes for Life Cycle Assessment”, minimum
review requirements for LCIA are listed. An independent external review is requested for
LCIA factors, whereas an independent panel review is requested for the underlying LCIA
models.
Any deviation from the recommended LCIA method has to be justified and the
recommended LCIA method has to remain the baseline for comparison and it has to be
reflected in the interpretation.
Not necessarily all LCIA methods that are recommended within this document are fully
compliant with all ILCD requirements, especially related to the requirements for review of
LCIA models and factors. However the recommendation reflects that after expert and public
consultations these methods were seen as being of sufficient quality. Until the methods
comply with all ILCD requirements, they may be considered a preliminary recommendation.
2.3 Research needs

Research needs are identified for each of the impact categories and prioritised according
to their importance for the characterisation modelling for the impact category, in particular
where the recommended methods are classified as level II or level III, or where the methods
are classified as interim. The research needs are classified according to their priority (high-
medium-low) and the associated workload is estimated. The research needs are reported in
Annex 2.
In a cross-cutting activity it is analysed to which extent the impact pathways, which are
modelled by the recommended characterisation models, are complementary at midpoint level
and at endpoint level or whether they present overlap or insufficient coverage of the relevant
environmental mechanisms. Detected inconsistencies are corrected if possible, or the
selection of recommended characterisation models is modified with the aim of ensuring
complementarity between the impact categories to the extent possible. The check on
consistency across impact categories also helped identify future research needs in order to
ensure coverage of all relevant impacts.
11
2.4 Geographic scope

Life Cycle Assessment typically has a global scope as the supply chains behind products
tend to be global in nature, crossing national and geographic borders, particularly in terms of
raw material and energy supplies. In many cases the location of emission sources or
resource use may not be known. Hence, life cycle impact assessment models and factors
must firstly be globally applicable.
This guidance is intended to support Life Cycle Impact Assessment on a global level
recommending default characterisation models and associated factors for each impact
category. As far as available, global models were recommended. In some cases no
international consensus exists on globally representative characterisation models and
factors. In the absence of sufficiently sound global models, a choice had to be made for
models that represent large heterogeneous regions. These may be at continental or national
scale. It is assumed that the central-tendency estimate for these smaller regions will be a
sufficiently good estimate of the global default value.
These choices can be seen as reflecting a European perspective on models and factors
for use in Life Cycle Assessment.
Further distinctions in relation to e.g. emissions scenarios (e.g. from a high stack, in a
densely populated area) and geographic/political boundaries may be helpful if associated life
cycle inventory or unit process data are available (e.g. factors for China used in the context
of Chinese emissions). The additional collection and use of such specific inventory/unit
process data and impact assessment factors is justified when this will significantly reduce the
uncertainty associated with a particular life cycle assessment (see Section 2.3 for deviations
from the recommendations/baseline).
12
3 Background information on the evaluation of

existing LCIA methods
The following sub-sections present the results of the analysis of existing LCIA methods at
midpoint and endpoint level for each of the impact categories conducted in accordance with
the evaluation criteria as developed in the guidance document LCIA- Framework and
requirements (EC-JRC, 2010b). The recommended methods are classified according to the
classification system as reported in the Chapter 2. The detailed assessment of each method
is documented in a separate spreadsheet9 for each impact category. These spreadsheets
were used as supporting working documents for the expert judgement during the
assessment of the methods.
3.1 Climate change

3.1.1 Introduction
All LCIA methodologies have an impact category Climate Change (sometimes called
Global Warming), and they all use the Global Warming Potentials (GWPs) developed by the
Intergovernmental Panel on Climate Change (IPPC). However, there are some differences in
the use of GWP’s
 IPCC periodically publishes updates, and not all methodologies use the latest
factors (but could easily be updated)
 IPCC publishes GWP’s for different timeframes.
3.1.2 Pre-selection of methods for further evaluation

The pre-selection of characterisation models for the climate change impact category has
been explained in the LCIA - Analysis document (EC-JRC, 2010a) and is summarized below.
As there is a wide consensus on the use of IPCC’s GWP’s for characterisation at midpoint
level, only this method was selected as representative for all midpoint methods currently
used in LCA. At the endpoint level, four methods that are based on different models were
selected: Ecoindicator 99, EPS2000, Recipe and LIME.
Midpoint
The GWP’s published in IPCC’s Fourth Assessment Report (AR4, 2007) were taken as
this indicator is used as midpoint indicator in every characterisation model. All methods can
in principle be easily updated with these latest figures.
9
http://lct.jrc.ec.europa.eu/assessment/assessment/projects#consultation_impact
13
3 Background information on the evaluation of existing LCIA methods
IPCC has three versions of the method, indicationg three different timeframes. The impact
in terms of cumulative radiative forcing of greenhouse gas (GHG) emissions is either cut off
after 20, 100 or 500 years.
The 500-year perspective is considered sufficiently long to assess the majotity of the
damage caused by the substances with the long atmospheric residence times, while the 100-
and 20-year timeframes capture partially the impact of substances with a long lifetime. In
some circles, the 100-year timeframe is used as this is the basis adopted for the Kyoto
Protocol. I
It should be noted that GWP’s also are used as the basis in all the endpoint models, and
also here the time perspectives is an issue.
Endpoint
Eco-indicator 99 (Goedkoop and Spriensma, 2000) takes into account human health
damage from climate change. Health effects considered include heat stress, vector borne
diseases and flooding. The change in radiative forcing is determined using GWP’s from
IPCC’s Second Assessment Report (SAR, 1995) for three pilot GHGs (CO2, N2O and CH4),
each representing a group of GHGs with a certain lifetime. The characterisation factors are
expressed in Disability Adjusted Life Years (DALY). There is no model for the damage to
ecosystems. The climate model used is an unpublished version of the FUND model
developed by Tol (1999).
EPS2000 (Steen, 1999a,b) takes into account human health damage, loss of species and
effects on primary production. Health effects considered include thermal stress, flooding,
malaria and malnutrition. The change in radiative forcing is determined using GWP’s from
IPCC’s First Assessment Report (FAR, 1990). The characterisation factors are expressed in
Years Of Lost Life (YOLL), person-years of severe morbidity, person-years of morbidity,
Normalized EXtinction of species (NEX), kg of crop-productivity loss and kg of wood-
productivity loss. All characterisation factors are subsequently harmonized in an additional
step using monetarization.
ReCiPe (De Schryver and Goedkoop, 2009a) includes human health damage and loss of
species. The health effects considered include heat stress, malaria, malnutrition, diarrhoea
and flooding. The change in radiative forcing is determined using the GWP’s published in
IPCC’s Fourth Assessment Report (AR4, 2007). The characterisation factors of human
health damage are expressed in DALY, while loss of species is expressed in Potential
Disappeared Fraction of species (PDF). A meta-study was used (Thomas et al., 2004) for the
link to biodiversity. Two versions of ReCiPe have been considered for the human health
assessment: De Schryver and Goedkoop (2009a) and De Schryver et al. (2009). The
difference lies in the way the temperature factor is calculated, but in essence the approaches
lead to the same result. In the first approach, the damage is calculated for CO 2 only, and the
midpoint (using GWP’s published in IPCC’s Fourth Assessment Report (AR4, 2007)) is used
to cover other GHGs. In the second approach, GWP’s are not used, but the entire
environmental mechanism is calculated in a way that is compatible with the different time
perspectives used. In the first approach, there is no time cut off, but different time horizons in
the equivalency factors are used. In the second approach different time horizons are used.
14
LIME follows some of the principles of Eco-indicator 99, as it also develops mechanisms
for CO2, N2O and CH4 as pilot substances, and then applies GWP’s to incorporate additional
substances. The impacts on heat and cold stress, floods, malaria, disasters, crop, plant,
energy and dengue fever as well as malnutrition are covered. These are linked to human
health damage (in DALY/kg), social assets (in Yen/kg), plant production (in Dry-ton/kg) and
biodiversity10 (in EINES/kg). It uses a climate model (DICE Model, Nordhaus, 1994).
The figure below describes the environmental mechanism and the position of each LCIA
methods along the cause-effect chain.
Emissions into the atmosphere
Time integrated concentration
- Direct effect
- Indirect effects Radiative forcing
IPCC 2007 Midpoint
- temperature
level
- extreme weather Climate change
- precipitation
- droughts
Effects on Effects on
ecosystems humans
Net primary Changing Wild other Malnutr- Flooding Infectious Heat

production biomes fires impacts ition Diseases stress
Endpoint
level
Decreasing
bio diversity
EPS 2000
Ecoindicator 99
ReCiPe
LIME
Figure 1 Environmental mechanism for climate change and associated LCIA methods.
3.1.3 Method evaluation

The five models have been rated against the criteria defined in the LCIA- Framework and
Requirements document (EC-JRC, 2010b). The results are summarized in the table below11.
Background information for the assessments can be found in a separate Excel file (Climate
change.xls12)
EINES:Expected Increase in Number of Extinct Species

10
11
A: full compliance; B: compliance in all essential aspects; C: compliance in some aspects; D: little compliance;
E: no compliance
12
15
Table 3 Summary of the evaluation results of 5 models that assess climate change in LCA context.
IPCC EPS2000 ReCiPe Ecoindicator 99 LIME
Midpoint Endpoint Endpoint Endpoint Endpoint

Completeness A No specific A Considers A Considers C Only human B Considers
of scope endpoints. human health, Human health health is human
biodiversity and and considered. health
crop biodiversity. damage,
productivity. plant
productivity,
as well as
ecosystem
damage.
Environmental A B Complete A Complete C Complete A Complete
relevance model, model. human health model.
although model.
several
assumptions
are made.
Scientific A IPCC D Models contain B No uncertainty C Climate model B Human
robustness & combines several factors is not entirely health well
Certainty stakeholder estimations included, most transparent. modeled,
acceptance and up to date data, Uncertainty uncertainties
with best approximations scenarios factors included, not specified.
science. . Uncertainty included. scenarios Links to crop
factors included. loss
included. uncertain
due to limited
model.
Documentation A IPCC A Information is A Information is A Information is Information
,Transparency provides very easily easily easily available. E in non-
& detailed accessible. accessible. Japanese
Reproducibility background language
documentatio only partially
n. available.
Applicability B Good B Good B Good B Good B Good
applicability applicability applicability applicability applicability
Overall A Broadly C Rough model, B Up to date, well C Link to eco- C Good human
evaluation of accepted partially described systems health
science based scientific outdated. method. missing. model, old
criteria basis All climate
methodologie model, lack
s use this of
method at information.
midpoint .
Stakeholder A Generally D Not generally D Not generally D Not generally E Not generally
acceptance accepted. accepted. accepted. accepted. accepted.
3.1.4 Discussion on method evaluation

All the endpoint models have considerable uncertainties, as the link between cumulative
radiative forcing and damages to human health and ecosystems is difficult to establish.
There is an extensive amount of literature describing the link between emission scenarios,
temperature increase and associated damages to human health, ecosystems and economy
(e.g. crop losses). However, LCIA methods focus on the marginal effect of one kilogram of a
GHG emission. The different LCIA methods rely on the following assumptions:
1. EPS converts IPCC damage estimates to estimates that can be related to a
kilogram of CO2-equivalent emission. It applies an average (and not marginal)
approach. LIME has a similar approach, but deducts a marginal damage factor
(the additional damage of an additional kilo). It also uses anold climate model.
Eco-indicator 99 asked a known expert, Richard S. J. Tol, to make a specific
16
number of runs using the Fund model (Tol, 1999). That version of the model was
never described, and as the model is constantly developing, the model is not clear
or generally accepted. The original version of ReCiPe uses a large comparison
study by Meinshausen (2005), who compared many international authoritative
climate models, and deducted a marginal temperature curve per mass load of CO 2
equivalents. This study is well recognised and often referred to in other reviewed
literature (De Schryver and Goedkoop, 2009a). Recently a new version of ReCiPe
was developed using the IMAGE model to link emission flows to increase in CO 2
concentration, radiative forcing and resulting increase in temperature (De Schryver
et al., 2009).
2. The models that link temperature to human health damage assume different
scenarios reflecting the degree of adaption of humans to changes in the climate.
Questions related to whether malnutrition will be prevented with good policy, or to
whether a cure for malaria will be found, or to whether heat strokes are avoided
because people purchase air conditioners, are crucial to determining the damages,
as the latter are highly sensitive to the former. EPS2000 assumes little adaptation,
and LIME seems not to take this into account. In Eco-indicator 99 and in ReCiPe,
three versions are used that provide three adaptation scenarios, allowing the user
to choose a version. Most models heavily rely on consensus documents, such as
those published by the WHO (McMichael et al., 2003).
3. For damage to ecosystems, there are some important assumptions. For example,
the assumption on the speed with which species can migrate, or how fast species
adapt to a changed climate. The studies available disagree on the magnitude of
the damages (see e.g. De Schryver and Goedkoop, 2009a).
4. The links to crop losses have the problem that temperature change can be
beneficial for crop production at some latitudes, while there are damages in other
latitudes or regions. There is also a dispute on how pests and diseases will be
affected and on whether this is fully counterbalanced by the expected increase in
crop yields, if any.
The scientifically-robust link between radiative forcing, temperature and ecosystem
impacts makes ReCiPe the scientifically most robust endpoint method. There are three
different versions that are based on different assumptions regarding adaptation and time
perspectives. One of these versions is regarded as the default model; the other two versions
can be used for sensitivity analyses. LIME also has some promising models, but due to the
lack of information available in English it is difficult to interpret. EPS2000 has the benefit of
using a clear model, but the model relies on some assumptions and older models.
3.1.5 Discussion on uncertainties and the importance of spatial

differentiation
Spatial differentiation is not relevant for locating the emission origin, but is relevant for the
damage assessment. Therefore, most endpoint models do the assessment on a regional-
specific basis.
17
3.1.6 Recommended default method at midpoint level

At midpoint level, GWP’s from the Intergovernmental Panel on Climate Change (AR4,
2007) is recommended. It is based on the most up-to-date and scientifically-robust
consensus-based model available, which produces characterisation factors based on
radiative forcing and residence time of the GHG emitted.
All LCIA midpoint methodologies available apply characterisation factors based on
GWP’s, although these are generally not updated to the latest version. Presently the up-to-
date characterisation factors are from the IPCC Fourth Assessment Report published in
2007.
From a scientific/sustainability point of view, it seems best to use the 500-year time
horizon, as only in this perspective all relevant impacts of all relevant emissions are better
captured. On the other hand, it is clear that in almost all policy instruments, like for instance
the Kyoto Protocol, the 100 year perspective is used and that this time perspective has the
broadest acceptance.
Recommendation of the 100-year timeframe is proposed as default, but it is also
suggested to use the shorter (20-year) and longer (500-year) timeframes as a sensitivity
analysis. This check is especially relevant when assessing agricultural systems, as the N2O
often emitted in these systems has a long lifetime, and thus has a significantly higher
characterisation factor (factor 2) in the 500 year perspective compared to the 100 year
perspective. Methane has almost a factor 4 lower characterisation factor in the 500 years
perspective.
3.1.7 Recommended default method at endpoint level

At endpoint level, no method is considered here mature to be recommended.
As interim, the method developed by De Schryver et al. (2009) and implemented in
ReCiPe could be adopted as it considers damages on both ecosystems and human health.
The method used by EPS2000 stands at the second place as interim method. Like
ReCiPe, it considers damages on both man-made environment and human health, but
EPS2000 is based on relatively old and simple models often relying on estimates. Eco-
indicator 99 scores the same ranking, but it less up-to-date, and the models are not well
documented.
3.1.8 Consistency between midpoint and endpoint methods

There is a benefit in having a midpoint and endpoint method for this category, as
comparisons are scientifically robust at the midpoint level while the endpoint method
provides natural-science based estimates at the Area of Protection level. The interim
endpoint default method builds directly on the recommended midpoint default method, so
there is a fine consistency.
18
3.1.9 Classification of the recommended methods

At midpoint, the recommended IPCC (2007) method for (GWP100 years) is classified as a
level I method (recommended and satisfactory) for characterisation.
At endpoint, no method is recommended. If an endpoint method is required that
expresses the impact in terms of DALY and species loss, the method developed by De
Schryver et al. (2009) can be used, but this method is classified as an interim method
because it is not sufficiently mature to be recommended.
3.1.10 Calculation principles

Additional midpoint factors cannot be calculated by the LCA practitioner but are also not
foreseen to be needed as an essentially exhaustive list is provided by IPCC.
3.2 Ozone depletion

The pre-selection of characterisation models for the ozone depletion impact category has
been explained in the LCIA – Analysis document and it is summarized below. All LCIA
methodologies have an impact category Ozone Layer Depletion (sometimes called
Stratospheric Ozone Depletion) and use the Ozone Depletion Potentials (ODPs) published
by the World Meteorological Organisation (WMO) (www.wmo.ch).
As there is a wide consensus about the use of OPD’s for characterisation at midpoint, only
one representative was selected, in this case the EDIP method, which is based on the 1999
WMO assessment (WMO, 1999). It will be referred to as the WMO 1999 midpoint approach.
For the endpoint, the following four methods have been selected as they are based on
different models
Eco-indicator 99 (Goedkoop and Spriensma, 2000)
EPS2000 (Steen, 1999a,b)
LIME (Hayashi et al., 2006)
ReCiPe (Struijs et al. 2009a and Struijs et al. 2010)

At the endpoint level, the impact pathways between the midpoints and the endpoints have
been developed (Figure 2) and in table 4 is presented an overview of the environmental
pathways and indicators modelled in the selected endpoint methods.
19
Emissions into the atmosphere
Time integrated concentration of Halons
Decreased ozone concentration Midpoint (WMO)
Increased UV-B level
Effects on ecosystems Effects on

and productivity humans
Wood Crop Fish Immune Skin Cataract

production production production suppression cancer
Figure 2 Environmental impact pathways of ozone depletion
Table 4 Overview of the environmental pathways and indicators modelled in the selected endpoint
methods.
Method Human health Crops Plankton Wood
ReCiPe DALY
Eco-indicator 99 DALY
LIME DALY Yen loss Kg Loss Kg loss + Yen loss
EPS2000 YLL, Morbidity

13

The five models have been rated against the criteria, identical to the climate change ones.
The table below summarises the assessment14. Background information for the assessments
is in a separate Excel file (Ozone depletion.xls 15).
13
When reviewing the EPS method it seems that it also has links between CFC and crops, wood and also
biodiversity, but these links only include the climate impact of CFC11 and its equivalents and are therefore not
considered here.
14
E: no compliance
15
20
Table 5 Summary of the evaluation results of 5 models that assess Ozone depletion in LCA context.
WMO Midpoint Eco-indicator 99 EPS2000 LIME ReCiPe
Mid or endpoint Mid End End End End
Completeness of A No specific links are C Only the link to B Only human health A Links to human health, Net primary C Only the link to human health
scope made human health is production and manmade resources is modelled
modelled
Environmental A Midpoint method B Complete model for C Only skin cancer is A Human health well covered, partial B Human health well covered,
relevance also used in all human health included based on coverage of crop productivity effects no ecosystems or crop
endpoint methods WMO damage impacts
estimates
Scientific A Based on WMO C Fate and damage D The way the total B State of the art method for human A State of the art method,
robustness & consensus models relatively damage is allocated health, somewhat limited models for (most recent of all), using
Certainty old, and rough using WMO is rather wood productivity novel approaches and
coarse models
Documentation & A Detailed reports and A Detailed reports B The description of B Backgrounds are only available in A Detailed reports and models
Transparency & models are available and models are the pathway is very Japanese are available
Reproducibility available brief
Applicability B ODP substances B ODP substances B ODP substances are B ODP substances are widely reported B ODP substances are widely
widely reported are widely reported widely reported reported
Overall evaluation A WMO report is based C Somewhat D Rough, models, B State of the art model for human B Most recent state of the art
of science based on widely accepted outdated relies on some WMO health, unique attempt to model crop model for human health
criteria science estimates for future losses
damage
Stakeholder A CFC equivalents are C DALY not generally B Relatively easy to D Well accepted in Japan, limited B DALYs are not generally
acceptance criteria widely used in policy accepted, CFC understand model, availability on scientific backgrounds accepted in EU but widely
equivalents are indicators not widely accepted in WHO and other
taken from accepted institutes
alternative source
21

As stated before, all LCIA methodologies have an impact category Ozone Layer Depletion
and use the ODPs published by the WMO. Hence, the WMO characterisation model has high
scores in almost all criteria and sub criteria, the factors are widely accepted and therefore it
is the preferred choice as a default method for the calculation of midpoint characterisation
factors.
For the endpoint characterisation factors all the methods have developed factors for the
AoP Human Health while for crops, plankton and wood only the LIME methodology has
factors. Among the others, the method proposed by Struijs (Struijs et al. 2009a and 2010)
could be used as interim method, as it is based on state of the arts models, the method was
published in a peer review journal, and at the moment is the most recent developed. LIME
could be also considered as interim since it received high score in several of the central
criteria, (the coverage of AOPs human health, natural environment and natural resources;
peer reviewed; updated model) but much of the documentation is in Japanese which
prevents a wide diffusion of this background data. Full translation of the methods and their
background documentation is still awaited.
3.2.4 Discussion of uncertainties

3.2.4.1 Discussion of uncertainties in the WMO equivalencies as a
basis for midpoint characterisation
The uncertainties in the equivalency factors published by the WMO are widely
documented and discussed in extensive stakeholder debates (WMO 1999 and 2003); this
discussion is not repeated here.
3.2.4.2 Uncertainty in the modelling of human health impacts

In all human health models, except EPS2000, the same principles are used. The fate of a
marginal increase of emission of ODS’ and the resulting worldwide increase of UVB
exposure is calculated, taking into account population density, latitude and altitude etc. There
are several factors that contribute to the uncertainty in such models:
The sensitivity for skin cancer highly depends on skin colour and on individual behaviour.
UV radiation is latitude dependent, and so is skin colour distribution. Overlaying the maps
of predicted UV-B increase and maps of skin colour distribution is a problem, as most
data reflect the original distribution of skin colour patterns over the globe, not reflecting
the huge migration waves. Individual behaviour is also difficult to take into account. Sun
bathing behaviour is a very important factor. Eco-indicator 99, LIME and ReCiPe all
struggle with this problem. For ReCiPe a special GIS model was developed, to improve
this situation. In this model the increased UV-B levels, population density, original skin
colour and other factors were modelled per grid cell.
22
Cataract is often associated with UV-B, but there are significant problems in proving that
there is indeed a link, and the data that is used to model this link is uncertain. The main
reason is that it is difficult to trace back the occurrence of this disease to high or low UV
exposure. Eco-indicator 99 and LIME include this link. EPS2000 does not refer to it.
ReCiPe uses the link only for one of the three cultural perspectives, that it models
(egalitarian) but not in the default perspective (hierarchical), because of this difficulty.
In the EPS2000 method, the model has been simplified, using an overall damage
assessment report from WMO and dividing the predicted damage by the total expected
emissions for the next 100 years. Due to this simplification, this method is probably the
most uncertain method.
3.2.4.3 Uncertainties in damage to crops

The damage pathways to crops and wood in LIME are well defined, but there are some
problems due to lack of data. For wood production only data on a single species is available
(Pinus Taeda), and this species only occurs at latitude of 30 to 40 degrees; still the sensitivity
of this species is extrapolated over a global scale. For Plankton growth, a model is used for
which data are available, especially for latitudes above 50 degrees, which is expected to be
the latitude where major damages occur, so this choice can be justified to some extent.

The WMO steady state method is in some form applied in all LCIA methodologies and is
also selected as the midpoint method to be preferred. The recommendation is to use the
latest WMO published ODP equivalents (currently WMO, 1999). A point of attention is that
WMO publishes equivalents representing different timeframes. Different stakeholders may
prefer different timeframes, but as the default it is proposed to use the infinite time
perspective, as this is the most widely used version in policy. In practice, there is very little
difference with the 100 year perspective as most currently used ODP substances have a
lifetime shorter than 100 years. Following the WMO reasoning that after 2040 the
anthropogenic impact on ozone depletion will be negligible, a shorter timeframe can also be
used, but the greater policy acceptance for the 100 year perspective is taken as guidance.

At endpoint, no methods are recommended to be used because no method is sufficiently
mature to be recommended.
As interim, the model of Struijs et al. (2009a and 2010) as implemented in ReCiPe
methodology uses an up to date model (AMOUR model - den Outer et al., 2008; van Dijk et
al. 2008) to assess human health damages on endpoint level caused by ozone depletion.
The most important limitation of the method is that there are no links to ecosystem endpoints.
It is recommended to adapt this method as the best available for the endpoint level, although
it has a very limited stakeholder acceptance, as it is a quite new and not easily
understandable method.
The LIME method (Itsubo et al., 2008c) rely on really interesting and advanced models. It
is the only method that links to crop loss, wood production and plankton loss. The ozone
layer depletion model has been published in a peer reviewed scientific journal (Hayashi et al.
23
2006), but that article, refers to several publications in Japanese, and these publications are
needed to understand the details of the method, especially for crop and plankton losses. We
know that meanwhile work is ongoing, especially regarding documenting and reducing
uncertainties. In some presentations researchers have shown that characterisation factors
are also changing in this new work, but information in these developments is also limited to a
few conference proceedings and posters. Very recently a partial draft translation has been
made available. This lack of information makes it impossible to recommend this method at
this stage.
The EPS2000 has a relatively simple model. It simply divides the total expected future
damage predicted in a WMO report by the total expected releases over 100 years. An
important benefit of this approach is that it is easy to explain and does not differ too much
from a midpoint model in this respect. EPS2000 scores relatively low in the scientific criteria,
but because of its relative simplicity; it scores among the best in stakeholder acceptance
criteria, because of its simplicity. Due to the limited scientific quality, and limited scope (no
cataract) it is not recommended.
The Eco-indicator 99 has been the starting point for the LIME and ReCiPe method. It is
also the oldest, and relatively primitive, because of its age, compared to the two later
methods. As with the EPS2000, the simplicity can be seen as strength if stakeholders are to
understand it, but the ozone depletion problem is certainly more complicated. It is not
recommended.

The interim method developed for ReCiPe claims to have a real midpoint/endpoint
structure, but on a closer look it has not. ReCiPe has a midpoint indicator (using WMO 1999
as reference), but this midpoint is divided in 6 sub-groups that have a specific damage
characterisation factor each. This adds to precision, but is a little inconsistent with the overall
framework.
3.2.8 Classification of the recommended default methods

The recommended midpoint method, the WMO Ozone model for ODPs, is classified as a
Level I method (recommended and satisfactory) for characterisation at the midpoint level, as
it is widely accepted, and highly environmental relevant.
At endpoint, no method is recommended.
If an endpoint method is required, the model of Struijs et al. (2009a and 2010), as
implemented in ReCiPe methodology, and based on the AMOUR model (den Outer et al.,
2008; van Dijk et al. 2008) could be used as interim method to assess human health
damages caused by ozone depletion.

Additional midpoint factors cannot be calculated by the LCA practitioner but are also not
foreseen to be needed as an essentially exhaustive list is provided by WMO.
Additional endpoint factors are not foreseen to be needed as those presently available
cover all relevant types of substances from midpoint to endpoint.
24
3.3 Human toxicity

The pre-selection of characterisation methods for the human toxicity impact category is
provided in LCIA- Analysis document (EC-JRC, 2010a). Table 6 summarises the results.
Table 6 Selected methods and underlying models for human toxicity effects for analysisa
Methodology Underlying model Reference
USEtox USEtox 1.0: Model developed by a Rosenbaum et al. (2008)

task force within the UNEP-SETAC
Life Cycle Initiative
b
ReCiPe USES-LCA version 2.0 Huijbregts and van Zelm (2009)
c
IMPACT 2002+ IMPACT2002 Jolliet et al. (2003), Pennington et al.
(2005)
TRACI CalTOX 4.0 Bare et al. (2003), McKone et al. (2001)

d
EDIP2003 EDIP1997, combined with site Potting et al. (2005)
dependent factors
CML 2002 USES-LCA version 2.0 Huijbregts et al. (2000 )
MEEUP Based on emission limit values Kemna et al. (2005)
Endpoint only
method
EPS2000 Direct empirical relationship between Steen (1999a,b)

global emission and observed
exposures or health impact for a few
pollutants
a
Though the present study focuses on Life Cycle Impact Assessment methods, it must be emphasized that other
environmental tools such as risk assessment, substance flow analysis or environmental impact assessment
provide complementary information and are more appropriate to assess e.g. localized health impacts associated
with peak individual exposures, etc.
b
The most recent version of the model USES-LCA 2.0 is the underlying model for the calculations of
characterisation factors for human toxicity in ReCiPe. Previous versions of the model family USES-LCA and
EUSES, employed in CML2002 and Eco-indicator99, were not included in the evaluation.
c
The European version of the model IMPACT2002 is the underlying model for the calculations of characterisation
factors for toxicity in IMPACT2002+. LUCAS and LIME contain respectively Canadian and Japanese versions of
IMPACT2002 and were not included in the evaluation to avoid duplication.
d
The most recent version of the EDIP method is evaluated (2003 version). A previous version, EDIP1997, was
not included in the evaluation.
3.3.2 Environmental mechanism for human health effects

Figure 3 describes the position of LCIA methods in the environmental mechanism for
human health effects:
Recipe, IMPACT 2002+, and USEtox are based on similar models, representing a full
model-based description of chemical fate, exposure, effect and optionally severity. TRACI
and CML 2002 are also similar, but differ in the way that the effect and severity indicators are
calculated and their scope.
25
EPS2000 directly assesses impacts at the endpoint of a few contaminants based on

human response to emissions, but without disaggregating the various human exposure
pathways.
EDIP is a simplified approach that approximates some of these processes, without fully
describing them.
MEEUP is directly based on emission limits that reflect policy objectives as the basis of
the indicators, hence reflecting more a policy-based weighting than impact assessment.
MEEUP
ground-, fresh- or agricultural Outdoor Indoor air
marine water or natural soil air home or worplace
fate
Algae Vegetation crop Intake
fraction
crustacae
EDIP fish oral animal meat inhalation
exposure exposure
gastrointestinal tract blood lung, nose
USES-LCA
other target organs dose - 1.0 CML
response
CalTOX
cancer type
other non cancer type k,l respiratory disease m,n TRACI
i,j disease severity:
overall cancer overall other non cancer overall respiratory inorganics USEtox
Damage on human health overall ionic radiation

USES-LCA
EPS IMPACT 2002
2.0 Recipe
Figure 3 Position of LCIA methods in the impact pathways of human toxicity.

The pre-selected models have been rated against the criteria defined in the LCIA-
Framework and Requirements (EC-JRC, 2010b). The table below summarises the
assessment16. Background information for the assessments is in a separate Excel file 17
(Human toxicity.xls).
16
A: full compliance; B: compliance in all essential aspects; C: compliance in some aspects; D: little
compliance; E: no compliance
17
26
Table 7 Summary of the analysis of the available characterisation methods against the adapted criteria for human toxicity (in two sub- tables)
IMPACT2002+ (midpoint and
Criteria
USEtox (midpoint) ReCiPe (midpoint and endpoint) endpoint) TRACI (midpoint)
The scope of the model for the
The model scope is applicable to the The scope of the model for the The scope of the model is applicable
evaluation of toxic chemicals on the
A comparative evaluation of toxic chemicals on evaluation of toxic chemicals on the to the comparative evaluation of toxic B
Completeness of European scale is largely applicable,
/ global and European scale. No spatial B European scale is applicable. No A chemicals on the European scale. /
scope although the model is parameterised
B differentiation beyond continent and world parameterization for other continents Parameterization is available for all C
for the US. Some conservative
compartments so far continents
assumption for human health effects
Environmental relevance is high for all Environmental relevance is high, best
Environmental relevance is high. Not
environmental pathways, except dermal basis for direct application of Environmental relevance is good. Best
valid for dermal uptake nor direct
uptake. Best basis for TD50 calculations, pesticides on crop. Includes method for impact pathways including
Environmental application of pesticides on crop.
B cancer-negative chemicals and route-to route B B intermittent rain. Not valid for dermal C dermal uptake. Use of RfD's
relevance Best basis for estimating severity for
extrapolation. Not valid for direct application of uptake. Basic assumptions for embedding uncertainty factors is
non cancer. Some factors for indoor
pesticides on crop. Implicit equal severity. severity. Some factors for indoor problematic. Implicit equal severity
exposure available
Preliminary for metals exposure available
Chemical input data checked and model Chemical input data have been peer
Data mostly from reviewed
components extensively reviewed by a large Model components extensively reviewed at least for Toxic Release
datatabases. Model components
group of model developers, model uncertainty reviewed and uncertainty estimates Inventory. Model components
extensively reviewed and uncertainty
Scientific evaluated but no parameter uncertainty available, but chemical data not extensively reviewed and uncertainty
estimates available, but chemical
robustness & B available. Carry over rates are kept below 1. B B always reviewed. Carry over rates B estimates available. Carry over rates
data not always reviewed. Carry
Certainty Metal and pesticides treatment for human are kept below 1 in the latest may be above 1, unless latest version
over rates are kept below 1. Metal
toxicity require improvements. Severity factors developments. Metal treatment for of CALTOX is used. . Metal and
and pesticides treatment for human
can be taken from other method since CTU human toxicity require improvements pesticides treatment for human toxicity
toxicity require improvements
represent cases of cancer and non cancer require improvements
The model, documentation is The model, documentation and results
The model, documentation and results are
Documentation, A The model, documentation and A available, but details on processes are published, very well documented
published, available on line and the model can
Transparency & A / results will be published and the / are not readily available. Results are A and the excel spreadsheet relatively
be easily used free of charge to calculate new
Reproducibility: B model can be used free of charge B published and the model can be used transparent. The model can be used
chemicals
free of charge free of charge
Database with > 1250 human toxicological
Database with > 1000 human toxicity Database with > 800 human toxicity
characterisation factors (recommended A A Database with > 380 human toxicity
characterisation factors is available characterisation factors is available
Applicability: A /interim). Intake fractions compatible with / / B characterisation factors is available
that can be easily applied and that can be easily applied and
future indoor and work environment exposure B B that can be easily applied and updated
updated updated
factors
USEtox includes all vital model elements in a TRACI and CALTOX include all vital
ReCiPe addresses human toxicity IMPACT2002+ addresses human
scientifically sound way, except for metals and fate model elements in a scientifically
Science based and includes all vital model elements toxicity and includes all vital model B
direct impact of pesticides. It is sufficiently sound way, except for metals and
criteria overall B B in a scientifically sound way, except B elements in a scientifically sound /
documented and has the largest substance direct application of pesticides on
evaluation for metals and direct application of way, except for metals. It is well C
coverage. Uncertainty may require further crops. It is well documented. Use of
pesticides. It is well documented documented
attention uncertainty factors should be avoided
Principles of the model are
transparent and based on the
Stakeholders Principles of the model are transparent and
A EUSES-system applied in the EU to Principles of the model are Principles of the model are easy to
acceptance: the parsimonious nature of USEtox reinforces
/ B evaluate new and existing chemicals, C transparent, but the model is not yet B understand and endorsed by the US-
Overall transparency. The model is being endorsed by
B but the LCA version is not officially endorsed by an authorative body EPA and other state agencies
evaluation an international authorative body (UNEP)
endorsed by an international
authorative body
27
Criteria
EDIP2003 (midpoint) CML 2002 (midpoint) MEEuP (midpoint) EPS2000 (endpoint)
The EPS2000 framework has been
the precursor of endpoint methods.
The scope of the model for the Many pathways or mechanisms are
The scope of the model for the evaluation of evaluation of toxic chemicals on No human toxicological impact not covered. This is because
B A
Completeness of toxic chemicals on the European scale is fully the European scale is fully mechanisms included. Indicators EPS2000 only models probable
/ / E C
scope: applicable but the cause-effect chain is only applicable. Some derived from policy-based emission impacts from present, average
C B
partial parameterization for other limit values emissions of toxic substances, and
continents so far these are estimated to mostly occur in
trace amounts and result in impacts
that are considered insignificant
Some data may be used to evaluate
Environmental relevance is good No specific focus on human other models. Incomplete pathways
Environmental relevance is good for all D
Environmental to high. Not valid for dermal toxicological impacts, as emission and questions of consistency across
C environmental pathways but dermal uptake. B / D
relevance: uptake nor direct application of limit values are used as impact locations in the operational calculation
Cause-effect chain not fully described. E
pesticides on crop. indicator of factors to be addressed. The
monetarisation approach is of interest
Data mostly from reviewed
datatabases. Model components
Not further evaluated, because the
Data mostly from reviewed datatabases. extensively reviewed and Relative weakness in data consistency
Scientific thresholds within the categories
Model published in peer reviewed book. No uncertainty estimates available, between regions, etc. Willingness to
robustness & C B 'completeness of scope' and C
uncertainty or experimental verification but chemical data not always pay interesting in case of
Certainty: 'environmental relevance' were not
available reviewed. Metal and pesticides monetarization approaches
reached
treatment for human toxicity
require improvements
The model, documentation and
Documentation, The model, documentation and published in
results are published in detail Not further evaluated, because the The approach is relatively well
Transparency & A detail. Results are published and the model A B
and the model can be used free thresholds were not reached documented and explained
Reproducibility can be used free of charge
of charge
Database with > 850 human
Database with > 180 human toxicity A thresholds within the categories
toxicological characterisation Impacts of emissions not specifically
Applicability B characterisation factors is available that can / 'completeness of scope' and C
factors is available that can be mentioned are modelled as zero
be easily applied and updated B 'environmental relevance' were not
easily applied and updated
reached
EDIP addresses human toxicity and includes
The model addresses human Coverage limited in number of
the effect part in a scientifically sound way, No compliance with science-based
toxicity and includes all vital substances and impact pathways.
Science based except for metals and direct application of criteria for the evaluation of human
model elements in a scientifically Empirical data may contribute to the
criteria overall C pesticides. It is well documented. The fate C E toxicity impacts. Political emission C
sound way, except for metals empirical evaluation of other models.
evaluation assessment is, however, very simplified and targets are used in the indicator
and direct application of Willingness to pay data may be used
no information is available on the uncertainties development
pesticides. It is well documented for valuation purposes
involved in the model results
Principles of the model are easy
to understand and based on the
Stakeholders
Principles of the model are transparent, but EUSES-system applied in the EU Not further evaluated, because the Principles of the model are
acceptance:
C the model is not endorsed by an authorative B to evaluate new and existing thresholds within the science based C transparent, but the model is not
Overall
body chemicals, but the LCA version is criteria were not reached endorsed by an authoritative body
evaluation
not officially endorsed by an
international authoritative body
28

As stated in the LCIA- Framework and requirements document (EC-JRC, 2010b), LCA
characterisation models and factors for human toxicity effects must be based on models that
account for a chemical’s fate in the environment, human exposure, and differences in
toxicological response. Therefore, one of the midpoint methods (MEEuP) was not further
evaluated since the proposed approach didn’t have a human toxicity model behind the
calculations.
For the five remaining human toxicity methods at midpoint, one (USEtox) has an almost
full compliance with the science-based criteria, two (IMPACT2002+ and ReCiPe) show
compliance in all essential aspects, one (TRACI) has a good science-based criteria
compliance, while the remaining three (EDIP2003, CML2002 and EPS2000) show
compliance only in some aspects. For the evaluation of stakeholder’s acceptance criteria, the
USEtox model also stands out compared to the other models as the principles of the model
are easy to understand and UNEP encourages its use by businesses and governments.
Several features make USEtox the preferred choice as a default method for the
calculation of characterisation factors:
- straightforward multimedia models are widely used in LCIA for modelling chemical fate
and human exposure. USEtox reflects the latest consensus amongst such modellers and
their associated models. It also reflects the principles of the earlier OECD consensus model
(Klasmeier et al., 2006) that focused on fate and long range transport of contaminants.
Similar to the other multimedia model based approaches, USEtox includes a number of vital
model elements of toxicological effects assessment (Hauschild et al., 2008). Nevertheless it
has undergone limited testing and shows the same fundamental limitations as all simple
multimedia models
- it offers the largest substance coverage with more than 125018 human toxicological
characterisation factors and reflects more up to date knowledge and data on cancer effect
factors than other approaches.
- the model has been set up to model a global default continent, and it has a nested
multimedia model in which it is possible to consider global, continental and urban scale
differentiation.
3.3.5 Discussion of uncertainties and the importance of spatial-

temporal differentiation
USEtox has similar uncertainties when compared to many of the other fundamentally
similar models such as USES-LCA, Impact 2002, and CALTOX.
Degradation half-lives are in most cases the parameters driving uncertainty for the fate
part of the assessment (Hertwich et al, 1999). The low dose extrapolation and dose-
18
The number covered is relative compared to the number of classes of chemicals with similar behaviour, hence
similar factors; however, no guidance on this yet exists in order to have default factors per chemical group/class.
Such developments would extend the application of such models to a much broader range of chemicals
29
response modelling are responsible for the highest uncertainties in the effect part (Crettaz et
al.2002, Pennington et al.2002).
For the quantification of the uncertainty of human toxicity characterization factors, e.g.
Hofstetter (1998) provided expert based estimates yielding a 95% percent confidence limit of
a factor 2 to 80 assuming on a lognormal distribution. Based on comparisons among the
different models, e.g. Rosenbaum et al. (2008) suggested an additional model uncertainty of
a factor 10. This generally results in a factor 100 for the uncertainty of recommended
characterisation factors and a factor 1000 for the factors that are characterised as “interim” in
the USEtox context (similar to Level III here, i.e. recommended, but to be applied with
caution)
It is expected that the accuracy and overall reliability of the factors will lie at least in this
range. But an uncertainty of 2 to 3 orders of magnitude is significantly lower than the roughly
12 orders of magnitude variation between the characterisation factors of different chemicals.
Similar situations may exist for other impact categories.
As with all LCA results, best-estimates must be used for decision support, reflecting the
current state of scientific knowledge and often predictions to low concentrations at which
actual impacts may not be known. As discussed in Rosenbaum et al. (2008), characterisation
factors presented here must be used in a way that reflects the large variation of 12 orders of
magnitude between characterization factors for toxicological effects for different chemicals as
well as the 3 orders of magnitude uncertainty on the individual factors.
In practice, this means that for the LCA practitioner, these characterisation factors for
human toxicity can be useful to identify the 10 or 20 most important chemicals pertinent for
their application. The life-cycle human toxicity scores enable thus the identification of all
chemicals contributing more than e.g. one thousandth to the total score. In most applications,
this will allow the practitioner to identify 10 to 20 chemicals to look at in priority and perhaps
more importantly to disregard 400 other substances whose impact is not significant for the
considered application. In practice, this means that for the LCA practitioner these toxicity
factors are very useful to identify the priority contaminants pertinent to their application. The
factors for toxicological effects thus enable the identification of chemicals contributing more
than e.g. one thousandth to the total indicator result. In most applications where this is
important, this will allow the practitioner to identify the chemicals that contribute the most to
the indicator and, perhaps more importantly, to disregard 400+ other substances whose
impact is not significant for the considered application. This is important in the interpretation
phase, as well as where refinement of the study may be needed.
Furthermore, spatial differentiation may influence results, especially for chemicals with
short lifetimes: the population density around the point of emission in case of inhalation being
the dominant route , the agricultural production intensity in case of food dominant pathways,
the vicinity of the emission relative to a drinking water source, etc. No comprehensive
assessment or approach currently exists to account for these spatial, as well as temporal,
variations in LCA studies. These may be partially cancelled out by other factors, such as
having multiple sources of emissions or may be negligible relative to other sources of
uncertainty/variation for many contaminants. Nevertheless, at the time of writing, it is not
possible to provide general recommendations for differentiations in LCIA for toxicological
effects that will reduce uncertainty and justify the collection of additional emission-scenario
specific data.
30

The use of USEtox as multimedia model, combining chemical fate and exposure with
toxicological data, is recommended for midpoint indicators.
It results from a consensus building effort amongst related modellers and, hence, the
underlying principles reflect common and agreed recommendations from these experts. The
model accounts for all important parameters in the impact pathway as identified by a
systematic model comparison within the consensus process.
This type of multimedia model integrates all environmental media into one consistent
model, provides default estimates for use in applications such as LCA, and is widely
adopted. It is adopted e.g. for regulatory assessments in e.g. the European Union (EUSES,
see EC, 2004) and for persistence screening calculations as recommended by bodies such
as the OECD (Klasmeier et al., 2006). This type of model is already widely used in LCIA and
was recommended by SETAC (Udo de Haes et al, 2002) as well as by the working group of
the UNEP/SETAC Life Cycle Initiative (Jolliet et al. 2006).
In USEtox, a distinction is made between recommended and interim characterization
factors, reflecting the level of expected reliability of the calculations in a qualitative way
(Rosenbaum et al. 2008).
Some characterisation factors (e.g. for ‘metals’, ‘dissociating substances’ and
‘amphiphilics’ - detergents) are classified based on expert opinion as interim due to the
considered higher uncertainty of the factors for these substance groups relative to others in
current practice. For the remaining set of chemicals, factors are also classified as interim
when route-to route extrapolation of the effects data is particularly uncertain or when the
target site is linked to the considered exposure route (nasal, lung or gastrointestinal target
sites).
The calculation of separate midpoint factors for cancer and non cancer is recommended,
as at least this distinction of effects is generally feasible in current practice and likely
significant. Equally exposure to particulate matter or respiratory inorganics and ionizing
radiation are to be considered separately.

For the endpoint characterization, it is proposed as an initial basis to apply the most
recent values that are proposed by Huijbregts et al. (2005a) using Disability Adjusted Life
Years (DALYs) per case, as recommended in the section 3.1 of the requirement document
(LCIA - Framework and requirements document, EC-JRC, 2010b). Since present knowledge
does not enable to determine the exact effect endpoint for many chemicals, averages
severities are calculated separately for cancer (11.5 DALY/case) and non-cancer (2.7
DALY/case) based on individual illnesses and used by default.
For cancer effects, the additional uncertainty linked to the severity factor will be limited, as
these numbers are primarily based on statistical data for years of life lost and the variation
between effects is about one order of magnitude (see e.g. Crettaz et al. 2002).
31
For non-cancer effects, the variation can be several orders of magnitude. Therefore no
method is recommended and USEtox has to be considered as interim for non-cancer effects
at the endpoint level.
As highlighted by Pennington et al. 2002, extreme caution is advocated when comparing
the likelihood and potential consequence estimates across chemical emissions in an LCA
study, particularly between noncancer and cancer effect results.
These estimates provide preliminary or screening level, insights only due to high model
uncertainty. While the framework for the calculation of LCA characterization factors allows for
the consideration of nonlinear low-dose response curves, mechanistic thresholds, and
multiple background exposure concentrations, the availability of required data is limited in
practice. Assumption of a default linear low-dose-response relationship remains
pragmatically necessary. For truly nonlinear dose-response curves with mechanistic
thresholds, likelihood measures may only reflect an erosion of the margin of exposure—an
impact on the capacity of the world to accommodate such emissions. Acknowledging this
high model uncertainty is important when interpreting the results of an LCA study”.

Compatibility between midpoint and endpoints recommendations is ensured since the
midpoint indicator defined in USEtox as Comparative Toxic Units (CTUhuman) corresponds to
cases of cancer and non cancer, whereas the severity factor reflects the Disability Adjusted
Life Years per case. These can therefore be combined in a straightforward way.

The midpoint method is generally classified as “recommended but in need of some
improvements” (Level II out of III) for both cancer and non-cancer effects due to non polar
organics.
The recommended endpoint method is described as Level II for cancer effects except for
polar organics and as interim for non cancer effects, since the derivation of severity factors
for the latter is much more variable and uncertain than for carcinogens. For non-cancer
effects, no method is recommended and USEtox as to be considered as interim.
Note that, both for midpoint and endpoint, in the mixed classification in the summary table
(II/III and II/interim) the second level of classification refers to substances belonging to the
classes of metals and amphiphilics and dissociating chemicals, where the characterisation
factors are down scored.
The present version of the USEtox model is not applicable to account for the contribution
to population exposure of the direct application of pesticides on crops, or for direct human
exposures associated with e.g. the use stage.

In case a midpoint characterisation factor is missing for an important elementary flow in
the inventory, it can be determined using the model as documented in Rosenbaum et al.
(2008).
32
The latest version of the USEtox model may be downloaded at www.usetox.org to

calculate characterization factors for new substances.
The calculation to fill data gaps requires the availability of the needed substance
properties among which particularly the toxicity and degradability data can be uncertain and
difficult to find. These are normally the input parameters contributing most to the overall
uncertainty of the characterisation factor.
3.4 Particulate matter/Respiratory inorganics

The pre-selection of characterisation models for the particulate matter/respiratory
inorganics impact category has been described in LCIA - Analysis document (EC-JRC,
2010a) and is summarized in Table 8.
Table 8 Selected methods and models for respiratory inorganics.
Model/Method Underlying model
Generic models. Do not consider secondary aerosols.
USEtox (iF) (Rosenbaum et Model based on a thorough evaluation of a large set of existing human
al. 2008) toxicological and ecotoxicological models developed for LCA under the
auspices of the UNEP-SETAC Life Cycle Initiative.
IMPACT 2002 (iF) Steady-state model. Can easily be adapted to any spatial characteristics.
(Pennington et al. 2005)
Humbert (2009) (iF, uF, Fate and exposure based on the UPFM model (Humbert, 2009). Effect and
endpoint) severity based on epidemiologic studies. Humbert. (2009) evaluate intake
fractions, but also uptake fractions.
Simplified compilation of results.
Hofstetter (1998) (iF, Compilation of different results. Effects are based on epidemiologic studies.
endpoint) Underlying method for the LCIA methodologies Eco-indicator 99 (Goedkoop
and Spriensma, 2000) and IMPACT 2002+ (Jolliet et al., 2003).
Simplified spatial approaches.
Greco et al. (2007) Only fate and exposure evaluation.

Two outputs are made: 1) specific iF for the different US counties, based on
the S-R matrix, and 2) regressions.
The underlying model of Greco et al. (2007) is a Source to Receptor (S-R)
matrix. The S–R matrix is a regression-based derivation of output from the
Climatologic Regional Dispersion Model (CRDM) which uses assumptions
similar to the Industrial Source Complex Short Term model (ISCST3).
The considered parts for the present assessment are the regressions that are
derived from the results of the 3080 US counties. These regressions evaluate
iF as a function of population (density) at different radius.
RiskPoll (Rabl and Spadaro This simplified model has been calibrated with different projects to reflect
2004) main factors on influence on intake and subsequent damages. Effects are
based on epidemiologic studies.
33
Detailed spatial models underlying the results.
TRACI (Bare et al. 2003) Fate and exposure based on Wolff (2000), using the CALPUFF model.
Effect based on epidemiologic studies (Nishioka et al., 2002).
van Zelm et al. (2008) Fate and exposure based on EUTREND. Effect based on epidemiologic
studies. Underlying model for the LCIA methodology ReCiPe (Goedkoop et
al., 2009).
EcoSense (IER 2008) Fate and exposure using a source-receptor matrix (based on EMEP), WTM
dispersion model and ISC model. Local scale modelling using the ISC model.
Effect based on epidemiologic studies.
In the figure below the description of the environmental mechanism for respiratory
inorganics is provided.
Emission of primary PM Emission of precursor

(composition, PSD, stack height) (NOx, SOx, NH3, stack height)
Chemical transformation
Secondary PM
Evolution of PSD and composition of Evolution of PSD and composition of

primary PM secondary PM
Exposure to primary PM Exposure to secondary PM

(composition, PSD) (composition, PSD)
Concentration/dose – response Concentration/dose – response

(multiple endpoints) (multiple endpoints)
Severity Severity
Figure 4 Environmental mechanism for respiratory inorganics (derived from Humbert 2009)
3.4.1.1 Intake fraction evaluation

The intake fractions (iF) calculated by different models are compared in table 8.3. Urban
conditions represent a city of 10’000 km2 having 10’000’000 inhabitants (1’000 pers/km2).
These parameters are the one used to represent urban conditions in the USEtox model
(Rosenbaum et al., 2008). The values have been “corrected” for differences in breathing
rates among the articles. The adjusted breathing rate is 13.3 m3/pers·day (based on EPA,
1997). Only the adjusted intake fractions are presented in the table below.
34
Table 9 Comparison of intake fractions between different models
Model Urban Rural Average Continental

Greco et al. 2007’ counties 5.2 ppm 0.64 ppm 1.6 ppm 0.93 ppm
Greco et al. 2007’ regressions 6.5 ppm 0.56 ppm 0.84 ppm
USEtox (low advection (500 m2/s)) 16 ppm 0.79 ppm
USEtox (high advection (3200 m2/s)) 4.7 ppm 0.79 ppm
RiskPoll 6.7 ppm
van Zelm et al. (2008) 4.9 ppm (4.9 ppm)
Humbert et al. (2008) 18 ppm 0.18 ppm 5 ppm (5 ppm)
Marshall et al 2005’ regressions
37 ppm
(intra-urban only)
Heath 18 ppm 0.78 ppm
Heath (regressions) 21 ppm 0.89 ppm 2.2 ppm
Table 9 shows that the PM2.5 intake fraction varies more between low (rural median iF of
0.5ppm) and high (urban median iF of 15ppm) population densities with a factor 10 to 100
variation than between the model themselves with a factor 5 variation. Thus the ability to
differentiate between low and high population densities is a key characteristic before
considering the quality of the model itself.
3.4.1.2 Effect and severity evaluation

The treatment of effect and severity in the different models and methods evaluated is
presented in the table below.
Table 10 Treatment of effect and severity in the different models and methods evaluated (modified from
Humbert 2009).
Model/Method Type of endpoint considered Effect factor
USEtox Not considered. Only iF is valid at present stage. No effect factor, only intake fraction
IMPACT 2002 Chronic mortality; Respiratory admission; Chronic Based on Hofstetter (1998): 43
bronchitis incidence (adults); Bronchitis (children); DALY/kg PM10 inhaled (average
3
Restricted activity days; Asthma attacks (adults breathing rate of 20 m /pers·day),
and children) corresponds to 64 DALY/kg PM10
inhaled (average breathing rate of
3
13.3 m /pers·day)
Humbert et al. Chronic mortality; Respiratory admission; Chronic 67 DALY/kg PM10 inhaled (average
3
(2009) bronchitis incidence (adults); Bronchitis (children); breathing rate of 13.3 m /pers·day)
Restricted activity days; Asthma attacks (adults
and children)
Hofstetter Bronchodilator usage; Cough; Lower respiratory 64 DALY/kg PM10 inhaled (average
3
(1998) symptoms (wheeze); Chronic bronchitis; Chronic breathing rate of 13.3 m /pers·day)
cough; Restricted activity days (RAD); Respiratory
hospital admissions; Acute Mortality (AM); Chronic
mortality; Expiratory Reserve Volume (ERV) for
Chronic Obstructive Pulmonary Disease (COPD;
ERV for asthma; ERV for croup in preschool
children
35
Model/Method Type of endpoint considered Effect factor
Greco et al. No effect and severity modelling N/A (average breathing rate of 20
3
(2007) m /pers·day)
TRACI Premature mortality; chronic bronchitis; 25 DALY/kg PM2.5 inhaled. 10.9

cardiovascular hospital admissions; restricted DALY/case
activity days
van Zelm et al. Chronic mortality; acute mortality; acute respiratory 57.8 DALY/kg PM10 inhaled
(2008) morbidity; acute cardiovascular morbidity (average breathing rate of 13
3
m /pers·day)
Van Zelm et al. (2008) re-evaluated the effect and
severity factors using clear input parameters.
However, it does not consider effects caused by
chronic bronchitis (adults) that are identified
important by Hofstetter (1998) and Humbert et al.
(2009).
RiskPoll short-term mortality; long-term mortality; PM10 = 32 YOLL/kg inh (long-term

respiratory hospital admissions; cerebrovascular mortality)
hospital admissions; chronic bronchitis – adults;
restricted activity day – adults; asthmatics
(bronchodilator use) – adults; asthmatics (lower PM2.5 = 54 YOLL/kg inh (long-term
respiratory) – adults; asthmatics (coughing) – mortality)
adults; chronic cough – children; asthmatics
(bronchodilator use) – children; asthmatics (lower
respiratory) – children; asthmatics (coughing) – (nitrates are considered PM10;
children; congestive heart failure - elderly sulphates are considered PM2.5)
EcoSense Increased mortality risk (infants); new cases of 1E-4 DALY/kg PM10 emitted
chronic bronchitis; increased mortality risk –
YOLLacute; life expectancy reduction – YOLLchronic;
respiratory hospital admissions; cardiac hospital
admissions; work loss days; net restricted activity
days; minor restricted activity days; lower
respiratory symptoms; LRS excluding cough;
cough days; medication use/bronchodilator use.
Concentration Response Function are published,
and have been aggregated in “DALY due to
morbidity” and “YOLL due to mortality”, resulting in
DALY per emissionof primary particulate matter
and per precursor for nitrates and sulfates
In terms of severity, long-term mortality dominates most analyses of the effect factors. The
effect and severity factor varies by a factor 4 (between 20 and 80 DALY depending on the
size of PM considered and the model considered).

The nine models have been rated against the criteria. The tables below summarises the
assessment19. Background information for the assessments is in a separate Excel file 20
(Particulate matter.xls). Note that Greco et al. (2007) reports only the intake fraction.
19
E: no compliance
20
36
Table 11 Summary of the analysis of the available methods against the criteria for respiratory inorganics. (divided in two sub tables)
Humbert et al.2009 Hofstetter 1998 (iF,

USEtox (iF) IMPACT 2002 (iF) (iF, uF, endpoint) endpoint) Greco et al.2007 (iF)
Not complete since no factors for The scope of the model for
evaluation of respiratory effects
secondary/precursor at this stage. Not complete since no the evaluation of respiratory
Not complete since B from inorganics is applicable for
Completeness Useful for calibration (i.e., compare factors for effects from inorganics on
C C D only primary PM are C / fate and exposure (B). Indeed, it
of scope the results of other methods to the secondary/precursor at this the European scale is
considered. E only evaluates intake fraction. Does
one of USEtox for PM10) and ensure stage. applicable. It is not spatially
not evaluate the effect and severity
consistency with other methods. adaptable.
(E).
Good for primary PM.
Acceptable for first Good overall. Lack
Consider the Good overall. Regressions very
Acceptable for first approximation of approximation of CF for differentiation between
influence of the PSD. B useful. No effect is considered. Can
Environmental CF for primary PM10. No secondary primary PM10. No urban and rural. Lack
C C C Not complete since C / be used for fate and exposure, but
relevance PM treated. Urban emissions can be secondary PM treated. difference between low/high
no secondary PM. E not for effect and severity. Only low
explicitly modeled. Urban emissions can be stack (i.e., mobile/point
Urban emissions can (i.e. mobile) sources. Lack NH3.
explicitly modeled. sources).
be explicitly modeled.
Chemical input data and model
Science based but with
components extensively reviewed by Some limitations (dose-
some limitations (dose- The model is science based, but
a large group of model developers, response function only
response function only only until intake fraction. Does not
Scientific but no uncertainty estimates mass based and not B
Good basis for future mass based and not surface go further. Some limitations
robustness & C available. No secondary PM C surface or number, B C /
research. or number, influence of PSD (influence of PSD or composition
Certainty provided. Some limitations (dose- influence of PSD or E
not considered). Does not not considered). Treatment of
response function only mass based composition not
reflect the latest scientific secondary PM simplified.
and not surface or number, influence considered).
knowledge.
of PSD not considered).
Good documentation. Factors
Easy to reproduce results
Documentation, Extensively documented. Easy to At the time of the Acceptable. Published as a published. Background data and
and modify input
Transparency & A reproduce results and modify input B D evaluation not B book so not as easy to get B models referenced. Not easy to
parameters. But only partly
Reproducibility parameters. published access to. reproduce results and modify input
documented.
parameters.
Good for primary PM10. Very good A Good applicability but only for fate
Good for primary PM10. No
Applicability B B No secondary/precursor B applicability. But only A Very good applicability / and exposure. Lack factors for
secondary/precursor PM.
PM. for primary PM. E NH3. (E) for lacking effect/severity.
Good science. Lack of latest Good science behind. Lack of latest
Acceptable for first knowledge regarding knowledge regarding that the
approximation of CF for surface or number might be surface or number might be better
Acceptable for first approximation of
primary PM10 (lack of better than mass based and than mass and influence of PSD.
CF for primary PM10 (lack of latest
latest knowledge regarding Good basis for future influence of PSD, as well as Regressions easy to adapt to
Science based knowledge regarding surface or
C C surface or number might C research. C fate and damage modeling. B specific spatial conditions. Allow to
criteria number might be better than mass
be better than mass and Unpublished yet. CF specific to urban have CF for urban emissions. Risk
and influence of PSD). No secondary
influence of PSD). No emissions not considered. to underestimate iF caused by
PM treated. Can have an urban box.
secondary PM treated. No difference between emissions in densities higher than
Can have urban box. low/high stacks 1000 pers/km2. For mobile sources
(mobile/point sources). only. Lack values for NH3.
Principles of the model are easy to Not published at the
Method already B
Stakeholders’ understand and the model is time of the Good for fate and exposure(B).
B C extensively used. But only D B Good. /
acceptance endorsed by an international preliminary analysis Effect and severity missing (E)
primary PM is present. E
authoritative body (UNEP) of methods
37
RiskPoll (iF, endpoint & CBA) TRACI (iF, endpoint) van Zelm et al. 2008 (iF, endpoint) Ecosense (iF, endpoint & CBA)
Complete assessment of impacts and The scope of the model for the
The scope of the model for the The scope of the model for the
damage costs due to primary and evaluation of respiratory effects
evaluation of respiratory effects from evaluation of respiratory effects from
Completeness secondary PM, including model for creation from inorganics is applicable if
B C C inorganics on the European scale is A inorganics on the European scale is
of scope of secondary PM due to SO2 and NOx European and US conditions
applicable. But not easily spatially applicable. The user has to provide
emissions. Both local and regional impacts considered comparable. But not
adaptable. longitude and latitude
can be modeled. spatially adaptable.
Complete assessment of impacts and
Good overall. But valid for US. Good overall. Lack value for PM2.5.
damage costs due to primary and Good overall. But lack easiness of
Lack difference between urban Lack explicit difference between
Environmental secondary PM, including model for creation spatial adaptability. Lack explicit
B C and rural emissions. Lack easy C urban and rural, though underlying B
relevance of secondary PM due to SO2 and NOx difference between urban and rural.
adaptability to specific approach is spatially adaptable. Has
emissions. Both local and regional impacts Has low/high (mobile/point) difference.
population density. low/high (mobile/point) difference.
can be modeled.
The model is science based.
But with some limitations (dose-
Model is based on detailed and thorough response function only mass The approach is science based. But The model is science based. But with
Scientific review of epidemiological evidence. based and not surface or with some limitations (dose-response some limitations (dose-response
robustness & B Recently updated for the NEEDS project of B number, influence of PSD or B function only mass based and not B function only mass based and not
Certainty ExternE. Some limitations (influence of composition not considered). surface or number, influence of PSD surface or number, influence of PSD
PSD or composition not considered). Secondary PM treatment not or composition not considered). or composition not considered).
simplified, as in Greco et al.
(2007).
Documentatio Detailed documentation available in Good documentation. Factors
n& "Methodology Update 2005", at published. Background data and Good documentation Easy to
Documentation referenced. Not
Transparency www.externe.info. Easy to reproduce models referenced. Not easy to reproduce results with the simplified
B B easy to reproduce results and B B
& results with the simplified version. The reproduce results and modify input version. But not easy to modify input
modify input parameters.
Reproducibilit model can be used to re-run different cases parameters, though underlying parameters.
y with different inputs. approach is spatially adaptable.
Applicability B Applicable on all continents A Very good applicability A Very good applicability A Good applicability
Good science based. But valid
for US and not easily adaptable Good science. Lack of latest
for regionalization. Lack of knowledge regarding surface or
Model is based on detailed and thorough Good science based. Valid for EU and
latest knowledge regarding number that might be better than
review of epidemiological evidence, not easily adaptable for
Overall surface or number that might be mass based and influence of PSD.
recently updated for the NEEDS project of regionalization. Lack of latest
evaluation of better than mass and influence CF specific to urban emissions not
B ExternE. Lack of latest knowledge B B B knowledge regarding surface or
science based of PSD. CF specific to urban considered, though underlying
regarding that the surface or number might number that might be better than
criteria emissions not considered, approach is spatially adaptable. Lack
be better than mass and influence of PSD. mass and influence of PSD. CF
though underlying approach is value for PM2.5. Has different values
CF specific to urban emission considered. specific to urban emission considered.
spatially adaptable. CF for for different stack height (proxy for
mobile and point sources (proxy mobile and point sources).
for low and high stack).
Stakeholders
Principles of the model easy to understand, Good. There is an authoritative Good acceptance. There is an
acceptance C B B Good. Not spread out yet. B
and full documentation is readily available. body behind (US EPA). authoritative body behind (EU).
criteria
38

The intake fraction calculated for primary particle matter by the different models ranges
from 0.5 ppm to 40 ppm depending for example on the population density or the height of
emission points. It is essential to select a model that covers both primary and secondary
particulates and can be adapted to various population densities and heights of emission
points. Both RiskPoll and Greco et al. (2007) enable, in a simplified approach, to calculate
the intake fraction, adjusting the population density to be consistent with other human health
impacts (e.g. human toxicity, ionizing radiation, etc.). Any user can customize these two
models to calculate factors for specific situations without excess time investment.
Greco et al.(2007) represents an interesting alternative to RiskPoll: a) it covers both
primary and secondary aerosols (apart from NH3); b) population densities can be adapted to
match any landscape parameters (because of regressions). Weaknesses are: 1) does not
evaluate secondary PM from NH3, 2) only addresses mobile (i.e., low stack) sources, and 3)
cannot adapt to different wind speeds. The secondary PM from NH3 as well as the stack
height would need to be extrapolated from other studies21.
RiskPoll (Rabl and Spadaro, 2004) makes a complete assessment of impacts and
damage costs due to primary and secondary PM, including model for creation of secondary
PM due to SO2 and NOx emissions. It also parameterizes the dominant factors of influence
for generic landscape characteristic. It can be adapted to match various landscape
parameters, though not as easily as Greco et al. (2007). Calculations for new substances are
not straightforward and intake fractions have to be calculated separately since they are not
readily available in the present software. The advantage over Greco et al. (2007) is that
RiskPoll differentiates between low and high stacks.
a) Other methods for sensitivity study or for further research
Results of van Zelm et al.(2008) and TRACI can be used for sensitivity analysis. Average
and country specific values for different height of release, from EcoSense EU27, can be used
for hotspot and sensitivity analysis. Actually, EcoSense makes a complete assessment of
impacts and damage costs due to primary and secondary PM, including model for creation of
secondary PM due to NH3, SO2 and NOx emissions for two different emission scenarios (i.e.
2010 and 2020)
EcoSense: Being a regularly updated model of seemingly high scientific quality developed
for policy support in the European energy sector, EcoSense could be an interesting
candidate for future recommendations within the impact categories for respiratory inorganics
but also for photochemical ozone formation, acidification and eutrophication. Since it does
not provide characterisation factors, it cannot be considered for recommendations at this
point, but investigation of the possibilities to adapt EcoSense to characterisation modelling at
both midpoint level and endpoint level.
Furthermore, it has good data for Europe (EC, 2005). However it is less well adapted for
global calculations, and lacks ease of adaptation to match any landscape parameters. Part of
21
Note that if Greco et al. (2007) is used, regressions parameters have to be multiplied by 13.3/20 to account for
3
the fact that the regressions were derived for an average breathing rate of 20 m /pers·day, whereas the average
3
breathing rate of the US population (EPA 1997) is 13.3 m /pers·day.
39
it is possible with the online version that also provides overall correlations. This however
needs further research for application within the LCA framework since simplified data are
only available for overall costs without intermediary results such as intake fraction. The
environmental relevance of Ecosense is good but lack explicit difference between urban and
rural, though underlying approach is spatially adaptable even not easily. Has low/high
(mobile/point) difference.
TRACI and van Zelm et al. (2008) have the advantage to differentiate between low/high
(i.e., mobile/point) sources, but both are not easily adaptable to other landscape parameters,
though the underlying models can be modified. Van Zelm et al. (2008) don’t have values for
PM2.5. EcoSense makes the assessment of impacts and damage costs due to primary and
secondary PM, including model for creation of secondary PM due to NH3, SO2 and NOx
emissions for emission scenario 2010 and 2020. EcoSense provides values for release
height >100 m and average of all sources.
b) Models not selected as the recommended model but useful for consistency or further
research:
a) USEtox (lack of value for secondary particles but useful to calibrate the selected
model and ensure consistency with other impacts on human health, USEtox is better for
human toxicity impact than for respiratory inorganics because several mechanisms
influencing the fate of PM are not considered (e.g. coagulation). Apart from USEtox, all the
other method input date were not peer reviewed
b) Hofstetter (1998) (van Zelm et al. (2008) is a better update),
c) IMPACT 2002 (no value for secondary particles),
d) Humbert et al. (2009). Good method to differentiate between particle size and only
method that considers surface or number instead of mass as the proxy for health effects. It
has an excellent adaptation to specific spatial conditions. CF specific to urban emission
considered. However, it is only available for primary particles.
3.4.3.1Discussion of uncertainties and the importance of spatial

differentiation
Spatial differentiation between emissions in low and high population density is a key factor
for the fate and exposure for primary and secondary particulate matter (Humbert, 2009). On
the effect side, it has been suggested that Reactive Oxidant Stress is a major determinant in
the health effects (e.g., Donaldson et al., 2003). The exact mode of action is however still
undefined, especially whether impacts are due to a physical effect of PM or to other organic
or inorganic substances adsorbed to particulate (making in that case surface a better proxy
than mass to evaluate the adverse health effects). In addition, the use of epidemiological
data implies that PM attributed impacts can be due to other pollutants or physical factors
whose concentration could be correlated to PM. Care must especially be taken to avoid
possible double-counting between impacts of PM-attributed impacts and impact of other
correlated variables in case of common endpoint.
The uncertainty of the intake fraction is evaluated to approximately a 95th percentile
confidence limit of a factor 3 assuming a log-normal distribution (e.g., Marshall et al. 2005).
40
The uncertainty of the effect (i.e., slope) factor is judged to approximately a factor 2, and the
uncertainty of the severity factor to approximately another factor 2 (between approximately 6
and 12 DALY/case of long term mortality). By uncertainty is meant scientific knowledge of the
phenomena, and not the variability in function of the population density. Overall, the
characterisation factor therefore has an uncertainty of a factor 6 (5th and 95th percentile).
3.4.4 Recommended default method at midpoint (fate and

exposure level)
Following the discussion above, it is recommended to use RiskPoll (Rabl and Spadaro,
2004) as a basis for calculating intake fractions and to use Greco et al. (2007) and the other
models described in the discussion on method evaluation to check and possibly adjust the
order of magnitude of the intake fraction. According to WHO, it is recommended to consider
in a first step the fraction below 2.5 μm as a harmful fraction that reaches the target site.
Furthermore, if surface or number is considered a better proxy of impacts than mass, PM2.5 is
a good approximation.
3.4.5 Recommended default method at endpoint (effect and

severity level)
It is recommended to recalculate the effect and severity factors, starting from the work of
van Zelm et al. (2008) that provides a clear framework, but using the most recent version of
Pope et al. (2002) for chronic long term mortality and including effects from chronic bronchitis
as identified significant by Hofstetter (1998) and Humbert (2009).

Since the endpoint model is an effect and severity calculation applied to the exposure
estimate provided by the midpoint characterisation, there is good consistency between
midpoint and endpoint model.

At midpoint, intake fraction calculations based on RiskPoll (Rabl and Spadaro, 2004) and
Greco et al. (2007) are classified as Level I (recommended and satisfactory) since the
human health effects of PM has been extensively studied.
At endpoint, for the effect and severity factor, these effects are well demonstrated for
primary particles (Level I) but more uncertain for secondary particles (Level II out of III, i.e.
“recommended with some improvements needed”). The user must however be conscious
that the estimated effect of PM may be an indicator of the overall effect of the air pollution
rather than based on a proven cause-effect relationship for PM.

The number of flows contributing to this impact category is rather modest and largely
covered by existing factors. However, the location and the associated population density in
the deposition area is crucial to the impact at midpoint level (intake fraction), and a study
41
may require that characterisation factors be calculated to represent the population density
conditions in the actual area affected by important emissions from the product system. This
requires running the recommended midpoint model with the relevant population density and
possible emission height information. The endpoint factors (effect and severity) are
substance/agent specific and not influenced by local conditions.
3.5 Ionizing radiation

For damage to human health related to the routine releases of radioactive material to the
environment (see Figure 5 for the impact pathway), the method described in Frischknecht et
al. 2000 has been considered, since to our knowledge this is the only method that meets the
general requirements for a quantitative approach. The fate and exposure model has been
based on the ExternE work carried out by Dreicer et al., 1995, who described the routine 14
atmospheric and liquid discharges in the French nuclear fuel cycle. Data from UNSCEAR
(1993) were used for 3 additional radionuclides.
Frischknecht et al. 2000 list the DALYs for the same types of cancers which are used for
human carcinogens, with 0.05 fatal and 0.12 non-fatal cases per Man-Sv, as reported by Ron
and Muirhead (1998). Radiation induced cancer cases are assumed to occur at the same
age pattern as for other cancer causes. The number of severe hereditary effects is assumed
to be 0.01 case per Man-Sv [ICRP22, 1999], resulting in 61 DALYs per case without age
weighting.
For damage to ecosystem (see the impact pathway at Figure 6), the model developed by
Garnier-Laplace et al. (2008 and 2009) has been analysed, that uses " the Dose Rates
associated with a 50% Effect defined as the percent change in the (average) level of the
observed endpoint during a chronic external gamma irradiation exposure experiment, named
EDR50 expressed in µGy/h" (Garnier-Laplace et al., 2006). The ecotoxicological effect factor
is calculated by converting the dose rates into the corresponding medium concentration (i.e.
water and sediment for freshwaters) for nine commonly adopted reference organisms
covering different phyla. The final effect factor is calculated as 0.5 divided by the geometric
mean of the HC50r,o for the nine reference organisms and associated with the 95%
confidence interval. To model the endpoint indicator a PDF calculation is foreseen to be
applied to the midpoint indicator score (Based on Posthuma and de Zwart, 2006 as for
ecotoxicity).
The pre-selected models are reported in the table below.
22
ICRP is the International Commission on Radiological Protection
42
Table 12 Selected methods and underlying models
Human toxicity Underlying model Reference
Human health Dreicer et al. 1995 Frischknecht et al., 2000 also used in
damages due to Ecoindicator 99, IMPACT 2002, ReCiPe
ionizing radiation UNSCEAR, 1993 and Swiss Ecofactor
.
Ecotoxicity
Screening Level AMI – Payet et al., 2004 Garnier-Laplace, 2008, 2009 also based
Ecological Risk on Garnier-Laplace, 2006
Assessment for EDEN, v.1.5
radioactive releases
Figure 5 Overview of impact pathway stages of radioactive releases for human health (adapted from
Frischknecht et al. 2000).
radioactive releases
freshwater sediment
internal irradiation due
to bioaccumulation
Gy/day per Bq/kg
BCF L/kg
external irradiation Algae external irradiation

due to water due to sediments
Gy/day per Bq/kg Gy/day per Bq/kg
crustacae
fish
Damage on ecosystems
Figure 6 Overview of impact pathway on ecosystem for radioactive releases to freshwater. Plain lines
refer to physical transfers of radioactive substances, whereas dotted lines correspond to
exposures to radioactive radiation.
43

Since currently only a single method is presently considered relevant for each of the
ionizing radiation subcategories, no detailed criteria-based comparison were planned as with
the other impact categories (See LCIA- Framework and requirements document, EC-JRC,
2010b). Hence, no specific criteria have been developed for this impact category. Instead the
evaluation is focused on the level of quality reached by the available methods within each
main criterion. Table 13 summarizes both model performances23 for human toxicity and
ecotoxicity. The assessments are documented also in a separate Excel file24 (Ionising
radiation.xls).
Table 13 Summary table on the analysis of the characterisation methods for ionizing radiation.
Frischknecht et al. 2000 Garnier-Laplace et al. 2008 and 2009

Criteria
human health ecosystem health
The model scope is applicable to the

comparative evaluation of impact of
radioactive substances on human health. It
is valid on global and European scale and
compatible with human toxicity category,
For ecotoxicity of radioactive substances, the
also leading to DALYs at endpoint. The
model is fully applicable to the comparative
Completeness method is based on site-specific fate and
B evaluation of radionuclides on global and
of scope exposure models for French Nuclear
European scale. Its underlying framework is
facilities and data has been generalised for a
consistent with that used for ecotoxicity
site-independent assessment. No spatially
differentiated factors are presently available.
There is no independent midpoint with this
category, but intermediary data on fate and
exposure and on number of cases.
Environmental relevance is high for freshwater

ecotoxicity, but ecotoxic impacts in marine and
Environmental relevance is high. The terrestrial compartments are not considered. The
framework enables to provide separate model applied to convert radiative doses to
intermediary results on fate and exposure, corresponding concentration represents the
Environmental
B on dose-response and related number of state-of-the-art and accounts for organism
B
relevance cases and on DALY per case in a consistent specific absorption rates relative to their
way with the human toxicity framework. No preferred life medium (sediment, water). The
urban area since releases mostly occur model does not cover the endpoint, but here the
outside of large agglomerations method applied for ecotoxicity can be adapted
since chemical and ionizing radiation are directly
comparable in this respect.
The scientific quality of the model used is
good and uncertainty has been documented
at each assessment step. The endpoint
All factors are based on at least three Effect
Scientific approach based on DALY is sound and
B Dose 50% at 3 different trophic levels, therefore
robustness & consistent with other human health impact B
qualifying for the ecotoxicity requirement for
Certainty categories. Dose-response is directly based
recommended factors (rather than interim).
on human subject exposed in Nagasaki and
Hiroshima and is therefore more reliable
than for human toxicity.
The model, documentation and results are The underlying models and principles have been
Documentatio, published in peer reviewed journals and very published in peer reviewed journals. However
transparency & A well documented with detailed data and C documentation of the characterization factors
Reproducibility sources used for each radionuclide and applicable in LCA is only available as grey
model part, ensuring further reproducibility literature
23
E: no compliance
24
44
Frischknecht et al. 2000 Garnier-Laplace et al. 2008 and 2009

Criteria
human health ecosystem health
Frischknecht et al (2000) calculate the

characterization factors per emission (Bq) for
21 radionuclides to outdoor air, 13 to water Characterization factor for the 13 most frequent
B and 15 to ocean covering 31 nuclides as a releases from nuclear power plants to water.
Applicability B
whole, which are supposed to be the most Factors can be easily applied. Update requires
important in nuclear power plant operations. specific knowledge
Factors can be easily applied. Update
requires specific knowledge
The proposed framework includes all vital The model addresses the freshwater part of the
Science based
B model elements in a scientifically sound way. environment problem, includes all vital model
B
criteria There is a need to update some of the input elements in a scientifically sound way.
data and/or models Documentation needs to be improved.
Principles of the model are transparent and

Principles of the model are transparent The
compatible with the USEtox framework, but the
Stakeholders model has been widely used but not
B/C C LCA version has not been peer reviewed yet and
acceptance endorsed by an international authoritative
is not officially endorsed by an international
body
authoritative body
3.5.3 Discussion on method evaluation: uncertainties and

limitations
For human health impacts, the uncertainty in the fate analysis is approximately a factor 2
to 4 (Dreicer et al., 1995), but for globally dispersed radionuclides (i.e., Tritium (H-3), Carbon-
14, Krypton-85, and Iodine-129), for which the uncertainty is probably greater than an order
of magnitude. An important source of uncertainty is related to the very long half-lives of these
radioactive materials. The global exposure of Tritium, Carbon-14, Krypton-85, and Iodine-129
has been calculated for a time horizon of 100.000 and 100 years. The uncertainties for the
global exposure are considerable for these substances and a coefficient of variation of 10 to
50 has to be assumed (a coefficient of variation of 10 means that the confidence intervals
varies between the median divided by 10 and the median multiplied by 10). The coefficient of
variation typically amounts to a factor 3 for cancer endpoint and to a factor 5 for hereditary
diseases. The uncertainty on the DALY per case of hereditary is estimated to a factor 1.5.
Uncertainty for the ecosystem impacts has been characterized using bootstrap methods
and leading to 95% confidence intervals on the HC50 for each radionuclide.
It is important to note that the present framework involves significant simplifications that
need to be further refined long-term. The different biochemical reactions in the body tissues
are not fully defined and the multiple radiochemicals toxicity might lead to non additive
synergies. Low-dose ionizing radiation might induce trans-generational instability that differs
significantly from one species to another. Delayed genomic instability in human, animal and
vegetal cells would be difficult to quantitatively model in detail.
45
3.5.4 Recommended default method for human health impact

of ionizing radiation
For the midpoint and endpoint characterization, the method proposed by Frischknecht et
al., 2000 is recommended. It includes all vital model elements in a scientifically sound way
and the method is well documented. Intermediary results on fate and exposure and on
number of cases are available and should be kept separate. Anyway, at endpoint no method
is recommended, due to the high uncertainty on the DALYs per case of severe hereditary
effects. Frischknecht et al., 2000 at endpoint could be only used as interim.
3.5.5 Recommended default method for the ecosystem

impacts of ionizing radiation
At midpoint and at endpoint, no method is recommended for ecosystem impacts of
ionising radiation.
As interim, for the midpoint characterization, the approach developed by Garnier-Laplace
et al.2008 and 2009 could be used. The model addresses the freshwater part of the
environment problem, includes all vital model elements in a scientifically sound way. The
midpoint method for impacts of ionizing radiation is classified as not recommended since at
the moment there has been no peer review on the characterisation. The method can be
upgraded to level III, once characterization factors have been published in a peer reviewed
publication and properly documented based on an ERA- type approach recently published
(Garnier Laplace et al, 2009)
The model does not cover the endpoint. The method of Posthuma and De Zwart, 2006
could be applied as for ecotoxicity (since chemical and ionizing radiation are both based on
Hazardous Concentration affecting 50% of species (HC50) at their 50% effect (EC50) and are
directly comparable in this regard), but, due to the existing limitation of the methods, no
recommandation of using this endpoint is done.

If also interim methods are considered, consistency is good between midpoint and
endpoint both for human health impacts and ecosystem impacts.

For human health, at midpoint (e.g. incidences of cancer or other diseases) the method of
Frischknecht et al., 2000 is classified as being recommended for human toxicity impacts of
ionizing radiation at level II out of III (recommended but in need of some improvements).
At endpoint no method is recommended, due to the high uncertainty on the DALYs per
case of severe hereditary effects. The method of Frischknecht et al., 2000 can be used as
interim
No method is recommended for ecosystem impacts at midpoint and at endpoint.
At midpoint, as interim, the method developed by Garnier-Laplace et al. 2008 and 2009
could be used.
46

Most important radionuclides are covered by existing factors; the need for calculation of
additional factors is limited.
3.6 Photochemical ozone formation

The impact category appears under a number of different names in the various LCIA
methodologies: (tropospheric) ozone formation, photochemical ozone formation or creation,
photo oxidant formation, photo smog, or summer smog. There are minor differences in terms
of substances included and atmospheric and meteorological conditions assumed in the
modelling, but in essence they all address the impacts from ozone and other reactive oxygen
compounds formed as secondary contaminants in the troposphere by the oxidation of the
primary contaminants Volatile Organic Compounds (VOC) or carbon monoxide in the
presence of nitrogen oxides, NOx under the influence of light. In the “LCIA-Analysis
document” the pre-selection of characterisation methods for the ozone formation impact
category is described. The pre-selected methods are briefly introduced below:
Midpoint
 CML2002 and EDIP97: Photochemical Ozone Creation Potential (POCP) based
on UK AEA model using a simplified description of the atmospheric transport
(Lagrangian type model). CML2002 applies the latest version from Derwent et al,
1998, EDIP97 adopts a distinction between areas with high and low background
concentration of NOx and applies a Swedish modification (Andersson-Sköld et al,
1994) for estimating the ozone formation in low NO x regions – expressed as C2H4
equivalents. Both represented by CML2002.
 EDIP2003: Two sub categories, both site-dependent at the level of country in
Europe based on RAINS model (Hauschild et al, 2006), which applies the Eulerian
EMEP model (Heyes et al, 1996). Impact on humans modelled as number of
people exposed in excess of WHO guidance value for chronic effects times
duration (WHO, 1989) – expressed as pers·ppm·hours. Impacts on vegetation
modelled as area of ecosystem exposed above guiding threshold for chronic
effects times duration (WHO, 1987) – expressed as m2·ppm·hours. It has factors
for Non-Methane Volatile Organic Compounds (NMVOC), CH4, CO and NOx.
Substance-specific characterisation factors are possible through correction with
relative POCP-factor.
 LIME: Models ozone formation from 8 archetypes of VOCs in C2H4 equivalents at
midpoint using a Japanese modification of the Photochemical Box Model from US
EPA (Schere & Demerjian, 1984, modified 1992, as quoted in the Japanese LIME
documentation) to produce Ozone Conversion Equivalency Factors (OCEF) which
are geographically differentiated for seven Japanese regions (Itsubo et al., 2008d).
47
 MEEuP: No environmental science-based characterisation model – the emitted

masses of individual NMVOCs are simply added (Kemna et al., 2005). Not further
evaluated
 ReCiPe: Models marginal increase in ozone formation due to emissions of
NMVOC or NOx applying the LOTOS-EUROS spatially differentiated model
averaging over 14000 grid cells to calculate European factors. Substance-specific
CF possible through correction by the substance’s POCP factor relative to the
POCP-factor for an average NMVOC (van Zelm et al, 2008).
 TRACI: Maximum Incremental Reactivity (MIR) model from Carter, 2000 for
characterisation factors, average factors for US based on weighting according to
population density patterns, characterisation factor for NOx based on national
influence relative to NMVOCs (Norris, 2003).
Endpoint
 EcoSense: Fate and exposure modelling spatially differentiated at the level of

country in EU15 using a simplified description of the atmospheric transport
(Lagrangian type model). Linear effect model for human health damage based on
epidemiological data and for crop damage based on laboratory data for loss of
crops (Krewitt et al, 2001). The model has been updated into EcoSenseWeb 1.3
(Preiss & Klotz, 2008).
 EPS2000: Fate model based on a pre-1992 version of UK AEA’s POCP model,
linear effect model based on epidemiological data for human health impacts and
on empirical or lab data for loss of crops (Steen, 1999b).
 LIME: Damage to human health (expressed as DALY) is modelled based on
epidemiological data from ExternE taking regional differences in population density
into account, damage to crop and wood production (in economic figures) and
damage to primary production (loss of NPP) are modelled based on empirical data
from Japan.
 ReCiPe: Models human health damage by applying constant damage factor for
acute human health effects (expressed as DALY) based on European statistics.
Chronic effects and non-linearity due to existence of thresholds are disregarded
due to lack of evidence. Damage to vegetation is not considered.
For the evaluation the following methods are left out: EI99 (ReCiPe is a follow up),
IMPACT 2002+ (same as EI99), LUCAS (same as TRACI but applied in Canada) and Swiss
Ecoscarcity (in accordance with Swiss regulation, no distinction between NMVOCs, i.e.
weight-based summation as in MEEuP).
Figure 7 shows the cause-effect chain for photochemical ozone formation from airborne
emissions of VOCs, carbon monoxide or nitrogen oxides with the most important pathways
highlighted (bold arrows). The analysed LCIA methods are positioned along the cause-effect
chain.
48
Figure 7 Flow diagram for photochemical ozone formation and position of analysed LCIA methods along
the cause-effect chain.

The pre-selected models have been rated against the criteria defined in the LCIA-
Framework and Requirements (EC-JRC, 2010b). The results are summarized in the Table
14 and Table 1525. Background information for the assessments is in a separate Excel file 26
(Photochemical ozone.xls)
25
E: no compliance
26
49
Table 14 Summary of the evaluation results of the midpoint characterisation methods against the criteria for photochemical ozone formation
CML 2002 EDIP2003 LIME midpoint MEEuP ReCiPe midpoint
. Criteria
(Derwent et al., 1998) van Zelm et al., 2008
The scope of the model is fully
The scope of the model is fully The scope of the model is No characterisation
applicable for the evaluation of The scope of the model is fully
B applicable for the evaluation of fully applicable for the model, NMVOC A
Completenes photochemical ozone formation applicable for the evaluation of
/ B photochemical ozone formation B evaluation of photochemical C emissions are simply /
s of scope on the European scale - photochemical ozone formation,
C parameterised to European ozone, parameterised to summed on a weight B
estimated from realistic worst parameterised to European conditions
conditions Japanese conditions basis
case for North Western Europe
Environmental relevance is
Not further evaluated,
Environmental relevance is high, high and both AOPs are Environmental relevance is high, but
B A because a threshold
Environment Environmental relevance is high and both AOPs are addressed. addressed. Regional spatial damage to natural environment not
/ / B within the category C
al relevance and both AOPs are addressed. Spatial differentiation is supported differentiation within Japan represented. Spatial differentiation can
C B 'environmental relevance'
within Europe supported. Does not cover easily be developed for Europe
was not reached
NOx
Underlying fate model and Not further evaluated,
Underlying fate model heavily Midpoint model reviewed and bench
Scientific B adaptation to LCIA use reviewed, Underlying fate model hardly because a threshold B
reviewed and continuously marked against other models from the
robustness & / B uncertainty from spatial variability D state-of-the-art, no treatment within the category /
updated, no detailed treatment field, no treatment of uncertainty in
Certainty C quantified and several temporal of uncertainty in resulting CFs 'environmental relevance' C
of uncertainty in resulting CFs resulting CFs
scenarios investigated was not reached
Documentati The method is documented
The method and CFs The method and CFs documented Not further evaluated,
on & and accessible, the CFs only The method and CFs documented and
documented and accessible for B and accessible for reproducible B because a threshold
Transparenc available in Japanese, accessible for reproducible application
C reproducible application in LCA. / application in LCA. / within the category B
y& Characterization model and in LCA. Characterization model and
Characterization model and C Characterization model and input C 'environmental relevance'
Reproducibili input data not easily input data not easily accessible.
input data not easily accessible. data not easily accessible. was not reached
ty accessible.
Characterisation factors for Not further evaluated,
127 characterisation factors are Characterisation factors for 8 Characterisation factors for NMVOC
NMVOCs, CH4, CO and NOx are because a threshold A
available and can easily be archetype VOCs are and NOx are available and can easily
Applicability A A available and can easily be applied B within the category /
applied - update depends on available and can easily be be applied - update depends on
- update depends on developers of 'environmental relevance' B
developers of POCP model applied and updated developers of LOTOS-EUROS model
RAINS model was not reached
Addresses the European Parameterised to European
conditions and supports spatial conditions, only addresses human
Addressing the European scale Addresses Japanese
differentiation within Europe. health impacts. Spatial differentiation
based on realistic worst case for conditions and supports Not further evaluated,
Underlying fate model and can easily be developed for Europe.
B North Western Europe. Well B regional spatial differentiation because a threshold
Science adaptation to LCIA use reviewed. Midpoint model reviewed and bench
/ reviewed but no treatment of B / within Japan. Underlying fate within the category B
based criteria Uncertainty from spatial variability marked against other models from the
C uncertainty in CFs. Method and C model rather old. No 'environmental relevance'
quantified and temporal scenarios field, no treatment of uncertainty in
CFs documented and treatment of uncertainty in was not reached
investigated. Method and CFs resulting CFs. Method and CFs
accessible for 127 substances resulting CFs.
documented and accessible for documented and accessible for
NMVOCs, CH4, CO and NOx NMVOC and NOx.
Based on models and data used Moderate stakeholder
Moderate stakeholder acceptance, Good stakeholders Moderate stakeholder acceptance,
B in the evaluation of photo B B acceptance, Japanese B
Stakeholders Danish government behind the acceptance, but not Dutch government behind the method.
/ oxidants for the EC, but difficult / / govern behind the method. A /
acceptance: method. Expert knowledge is further evaluated, Expert knowledge is required to
C to understand without expert C C Expert knowledge is required C
required to understand the model threshold not reached understand the model
knowledge to understand the model
50
Table 15 Summary of the evaluation results of the midpoint and endpoint characterisation methods against the criteria for photochemical ozone formation.
TRACI (mid) EcoSense (end) EPS2000 (end) LIME (end) ReCiPe (end)
Criteria
Norris, 2003 Krewitt et al., 2001 van Zelm et al., 2008
The scope of the model is
The scope of the model is The scope of the model The scope of the model is
The scope of the model is fully applicable for the
fully applicable for the for the evaluation of fully applicable for the
A fully applicable for the evaluation of photochemical
Completeness of evaluation of photochemical ozone evaluation of photochemical
B / B B evaluation of photochemical B ozone formation,
scope photochemical ozone formation on the ozone formation
B ozone, parameterised to parameterised to European
formation, parameterised European scale is parameterised to European
Japanese conditions conditions, but addresses
to US conditions applicable. conditions (for POCP)
only human health impacts
High environmental
relevance for both AOPs.
Environmental relevance is
Environmental relevance High environmental Damage model for HH
high, but damage to natural
B is high although relevance for HH and simple and based on Environmental relevance is
Environmental environment not
/ characterisation is B partially for Natural B B European data, damage B high, but damage to natural
relevance represented, damage
C weighted towards human environment model for vegetation and environment not covered
models very simple based
health impacts (represented by crops) crops not available for
on empirical data
review. It does not cover
NOx
Midpoint model reviewed and

Midpoint model Fate model reviewed but not
benchmarked against other
extensively reviewed, Good science based updated version, effect Model components
models of the field. Effect
further components and reviewed. Valid for C model based on rough B reviewed but old and hardly B
Scientific robustness model developed based on
C derived from reviewed B EU. No treatment of / empirically based estimates. / represent state of the art /
& Certainty empirical data, chronic effects
information, no treatment uncertainty in resulting D Consistent uncertainty C globally. No uncertainty C
and threshold for effects not
of uncertainty in resulting CFs considerations, but not of all considerations.
considered due to lack of
CFs. aspects.
evidence
The method principles The method principles and The method is documented
Good documentation The method and CFs
and the CFs are the CFs are documented and accessible, the CFs
Easy to reproduce documented and accessible
Documentation & documented and and accessible for B only available in Japanese,
results with the for reproducible application in
Transparency & C accessible. B B reproducible application in / reproducibility not clear. B
simplified version. But LCA. Characterization model
Reproducibility Characterization model LCA. Characterization C Characterization model and
not easy to modify and input data not easily
and input data not easily model and input data not input data not easily
input parameters. accessible.
accessible. easily accessible. accessible.
65 characterisation factors
Characterisation
are available and can easily Characterisation factors for
App. 580 characterisation factors for NMVOC and
be applied - update of fate Characterisation factors for NMVOC and NOx are
factors are available and A NOx are available and A A
model depends on 8 archetype VOCs are available and can easily be
Applicability A can easily be applied - / can easily be applied - / B /
developers of POCP model, available and can easily be applied - update depends on
update depends on B update depends on B B
update of damage model applied and updated developers of LOTOS-
developers of MIR model developers of
easy based on new EUROS model
EcoSense model
empirical data
51
TRACI (mid) EcoSense (end) EPS2000 (end) LIME (end) ReCiPe (end)
Criteria
Norris, 2003 Krewitt et al., 2001 van Zelm et al., 2008
The fate model addresses Parameterised to Japanese
the European scale based conditions, addressing
Parameterised to US
on realistic worst case for human health impacts
conditions, weighted Parameterised to European
Parameterised to North Western Europe (based on European effect
towards human health conditions, addresses only
European situation (POCP, old version). data), vegetation and crops
impacts (?). Fate model human health impacts.
addressing human Damage to natural (Japanese effect data).
extensively reviewed, Midpoint model reviewed and
health and partially for environment partly Model components
B further components B B benchmarked against other
Science based criteria vegetation damage represented (through crops), reviewed but old and hardly
/ derived from reviewed B / / B models of the field. Effect
overall evaluation (represented by crops). damage models very simple represent state of the art
C information, no treatment C C model developed based on
Model is strongly based on empirical data. globally. No uncertainty
of uncertainty in resulting empirical data, chronic effects
reviewed. No treatment Consistent uncertainty considerations. The method
CFs. Method principles and threshold for effects not
of uncertainty in considerations, but not of all and CFs documented, CFs
and CFs documented and considered due to lack of
resulting CFs aspects. Method principles available for 8 archetype
accessible for app. 580 evidence.
and CFs documented and VOCs, but only in Japanese
substances.
accessible for 65 for the moment. Does not
substances. cover NOx
Moderate stakeholder Low stakeholder Moderate stakeholder Moderate stakeholder

acceptance, USEPA Good acceptance. acceptance, no authoritative acceptance, Japanese acceptance, Dutch
B B B
Stakeholders behind the method. There is an body behind the method. government behind the government behind the
B B / / /
acceptance Expert knowledge is authoritative body Pure scientific work, method. Expert knowledge method. Expert knowledge is
C C C
required to understand behind (EU). understood with expert is required to understand required to understand the
the model knowledge the model model
52

differentiation
Spatial differentiation has been found more important than differentiation between
substances for vegetation impacts and in particular for human health impacts, where, for
example, Hauschild et al, 2006 find variations up to several orders of magnitude between
countries within Europe (primarily due to variations in population densities on the
geographical scale which is relevant for photochemical ozone). It is thus important that the
recommended model is able to support spatial differentiation.

The LOTOS-EUROS model as applied in the ReCiPe method for photochemical ozone
formation (van Zelm et al, 2008), consist of a detailed fate and exposure model for human
health impacts and is developed in a form which makes it readily adaptable for calculation of
a set of consistent CFs for each continent if integrating continent-specific atmospheric fate
models. A global default factor can be found as a weighted average of the continent factors
or perhaps calculated using a global average atmospheric fate model. Furthermore, the
present version of the model can provide spatially differentiated factors, but only for Europe.
This method is recommended for characterisation at midpoint level of photochemical ozone
formation impacts on human health.
ReCiPe currently calculates the indicator value by summing impacts from grid cells in
which there is a resident human population. This gives the indicator a bias towards human
health impacts and makes it inappropriate to represent impacts on the AOP Natural
environment. The recommendation is therefore to calculate the area- and time integrated
ozone concentration increases by the LOTOS-EUROS model, aggregating over all of Europe
without giving priority to inhabited regions, before applying these as characterisation factors
at midpoint level for photochemical ozone formation. Factors should be provided for NMVOC,
CH4, CO and NOx, at present factors for CO and CH4 are missing (more long-lived than the
typical NMVOC and hence dispersed over a larger region).
EDIP 2003 is based on the RAINS model and meets the science based criteria. Respects
non-linearity of photochemical ozone formation and addresses both human health and
vegetation impacts. It provides spatially differentiated CFs as well as overall site-generic
factors for Europe. Adaptation to other continents is not straightforward.

The method developed in ReCiPe is a recommended for human health impacts at
endpoint. It meets the science based criteria well, is peer reviewed and benchmarked against
other models, provides factors for NOx as well as for NMVOCs, and its environmental
relevance is high for a European setting. It has a good link to the recommended midpoint and
applies of the same framework which makes it adaptable to other continents and to the
global situation if the relevant effect data can be incorporated. The recommendation of the
model in its present form depends on the justification of the assumption that only acute
effects of ozone exposure are important, and that there is no threshold for exposure below
53
which effects can be disregarded. Factors should be provided for NMVOC, CH4, CO and
NOx.
For impacts on vegetation at endpoint level, a model might rather easily be built on
EDIP2003 midpoint model, which already models the time and area-integrated exposure
above a critical level for vegetation.
The characterisation factors both at midpoint and endpoint level should be extended with
the possibility for substance differentiation applying additional factors based on the
substance POCP or MIR value. Derwent and co-workers find the two systems to generally
show a fine agreement over a wide range of reactivities (Derwent et al., 1998) with a
tendency for the POCP to give an increased resolution of substances with low POCP values
due to its focus on long-range transport. Since the MIR values are available for around 600
individual VOCs compared to less than 140 for POCP, this favours the use of MIR to
distinguish the individual NMVOCs if wanted. The differentiation thus obtained is most
important for distinction between pure hydrocarbons on one side, and halogenated VOCs on
the other.
EcoSense obtained the highest scores in many criteria, but it applies a rather simplified
atmospheric transport model the relevance of which has been put into question for the
complex photochemical formation of ozone. An updated version of the EcoSense model
(Ecosense web 1.3) applies a more realistic modelling for ozone formation and should be
considered for potential future recommendation. It considers both human health and
ecosystem quality (only considering damage on crops). Spatial and temporal explicit
evaluation was already done for Europe, South America and Asia (Preiss and Klotz, 2008).
Within the EcoSenseWorld model, the population data are based on SEDAC global gridded
population, background emission and meteorological data grid could be adjustable to any
region of the world in order to apply the WTM model.

For the recommended models for human health impacts there is a fine consistency
between midpoint and endpoint as they have been developed applying the same model in a
consistent framework.

At midpoint, the recommended default method is the LOTOS-EUROS model as applied in
the ReCiPe method (Van Zelm et al, 2008), is classified as Level II out of III (Recommended,
some improvements needed).
At endpoint, the recommended method is the method developed by Van Zelm et al., 2008
as implemented in ReCiPe. It is classified as Level II out of III (Recommended, some
improvements needed).

The models provide unspecific factors for NMVOC and NOx respectively, but the
characterisation factors both at midpoint and endpoint level should be extended with the
54
possibility for substance differentiation applying additional factors based on the substance
POCP or MIR value. The two systems generally show a fine agreement, and since the MIR
values are available for around 600 individual VOCs compared to less than 140 for POCP,
this favours the use of MIR to distinguish the individual NMVOCs if wanted. Calculation of
additional substance factors outside this selection is not possible for the LCA user, but the
difference between individual substances is generally modest, so the importance is limited.
3.7 Acidification
Current LCIA characterization models focus on terrestrial acidification as it tends to
precede aquatic acidification when inland water is acidified after the depletion of the acid
neutralization capacity of its watershed. The impact indicators of existing methods cover the
majority of impact mechanisms and relevant elementary flows for the Area of Protection
(AOP) Ecosystem Quality.
Only few methods, such as EDIP97 and the CML method also cover waterborne
emissions, but the methods are not sufficiently developed; besides, aquatic acidification may
be considered as a separate impact category.
Characterization factors for acidification are traditionally calculated at midpoint level, as it
is the case for the majority of the LCIA methods considered in this analysis (Potting et al.
1998a,b; Norris 2003; Kemna et al. 2005; Seppälä et al. 2006). Others are damage oriented
LCIA methods and relate emissions of acidifying substances to impacts on the endpoint
biodiversity (Goedkoop et al. 2000; van Zelm et al. 2007). The analysed midpoint and
endpoint approaches follow the same cause-effect chain up to the modelled changes in soil
parameters, but they differ in the effect factor.
In the LCIA- Analysis document (EC-JRC, 2010a) pre-selection of characterisation models
for the acidification impact category is reported. Here, these methods are briefly presented:
Midpoint
 TRACI acidification potentials are based on the model developed by (Norris,
2003). It provides generic and spatially differentiated characterization factors for
the US. A fate model, ASTRAP is used to link the emission to the deposition on
land area. TRACI considers acidification potentials due to the acidic deposition on
the entire land and inland water area whether soil and ecosystems are sensitive or
not. The dose-response curve implicitly equals 1.
 EDIP 2003 provides European Country-dependent Characterisation Factors
adopting the Unprotected Area (UA) method (Potting et al. 1998a), which is based
on a category indicator measuring changes in area of unprotected ecosystems due
to emission reductions at country level within Europe. UA also considers a fate
transport model linking the emission to the deposition to land and inland water
55
area using the European RAINS model (Amann et al. 1999). The indicator
measures the increase in area of ecosystem that becomes unprotected by
exposure over its critical load. The dose-response curve implicitly equals 1.
 MEEUP (Kemna et al. 2005). Its framework relates to European Community
legislation and strategies and the Gothenburg protocol. It considers acidification
potential in term of H+ releases without addressing chemical fate of chemicals in
air and in soil, i.e. all emissions and subsequent depositions generate an
acidification potential. The dose-response curve implicitly equals 1.
 The method of Accumulated Exceedance (AE) (Seppälä et al. 2006) provides
European Country-dependent Characterisation Factors for Acidification and
Terrestrial Eutrophication. The atmospheric transport and deposition model to land
area and major lakes\rivers is determined using the EMEP model combined with a
European critical load database. The acidification potential is expressed in
accumulated Exceedance. The dose-response curve implicitly equals 1. A more
recent publication (Posch et al. 2008) updated the factors of the AE method using
the newest 2006 version of the EMEP Eulerian atmospheric dispersion model
(Tarrason et al. 2006), which provides also depositions onto different land cover
categories, and the newest critical load data base (Hettelingh et al. 2007)
consisting of about 1.2 million different ecosystem such as forests, surface waters,
and semi-natural vegetation.
 CML 2002 uses the method of Hazard index (HI) (Huijbregts et al. 2001) and also
provides European spatially-specific characterization factors (CF) for acidifying
and eutrophying air pollutants. The CFs express the marginal change in the
hazard index of all ecosystems in Europe, comparing the actual load to the critical
load weighted over ecosystems and region. Atmospheric transport and deposition
is determined using the European RAINS model. The HI method assumes a dose-
response slope inversely proportional to the critical load itself.
 ReCiPe is a midpoint-endpoint method. The midpoint indicator adopts the Base
saturation method developed by Zelm and colleagues (2007a), which calculates
the atmospheric fate with the EUTREND model (Van Jaarsveld et al. 1997). It only
considers terrestrial ecosystems. It uses the simulation model for acidification’s
regional trends, SMART 2, (Kros 2002) to characterize soil sensitivity at midpoint
level as a change in soil base saturation. The change in base saturation per unit
deposition is presently only available for Europe. Dose-response is determined in
endpoint modelling.
 LIME (Hayashi et al. 2004) is a midpoint-endpoint method. The midpoint indicator
expresses SO2 equivalency of Atmospheric Deposition Factor (ADF), which
indicates an increase of H+ deposition per unit area to an additional emission
acidifying chemical. The fate (deposition) of the emissions is calculated with an
atmospheric transport model or with empirical data depending on the chemical. It
only considers terrestrial ecosystems. Dose-response is determined in endpoint
modelling.
56
Endpoint
 ReCiPe adopts the approach of van Zelm and colleagues (2007a), which further
model the cause-effect chain up to damages on biodiversity with a physiologically
based dose-response model. Based on Monte-Carlo simulations for 240 plant
species, it expresses the change in potentially not occurring fraction of plant
species per change in base saturation [dimensionless].
 Eco-indicator 99 (Goedkoop and Spriensma. 2000) uses a simplified fate
assumption to determine the fraction of an acidifying emission that is deposited on
Europe (equal for all the chemicals). The effect factor is determined applying the
Dutch Nature Planner model that focuses on the percentage of threatened species
in The Netherlands caused by acidifying emissions. The indicator expresses the
change in Potentially Disappeared Fraction of species (PDF)* m2 * yr per marginal
change in deposition.
 The damage factor of LIME (Hayashi et al. 2004) indicates the total Net primary
production (NPP) damage in all of Japan due to the additional emission of
causative substances. Net primary production (NPP) of existing vegetation was
adopted as an impact indicator of terrestrial ecosystems. The aluminium toxicity
was adopted as the major factor influencing the effect on terrestrial ecosystems
due to acidification.
Payet (2006) proposed in relationship with the European funded NOMIRACLE project
(and IMPACT 2002+ developments) a dose-effect relationship to assess a change in pH
concentration in a non-buffered water body in terms of fraction of affected, or disappeared
species. As this method is not readily operational it has not been further considered in this
evaluation. However, despite the fact that it has still not been validated with field
measurements and needs to be complemented by a fate model, it could set an interesting
basis for further developments in assessing the effect of acidifying chemicals on aquatic
ecosystems.
Figure 8 shows the cause-effect chain for airborne acidifying emissions with the most
important pathways highlighted (bold arrows). The analysed LCIA methods are reported
according to their position in the cause-effect chain. The MEEUP method is not included in
the figure as it doesn’t follow the cause effect chain, but merely represents a potency related
to an elementary flow.
57
Airborne emissions
Atm. fate& transport TRACI, LIME (midpoint)

and deposition on land
Change in sensitive Non sensitive area

area Sensitive area
EDIP 2003
Area with limited Area above Change in critical
buffer capacity critical load load exceedence
AE,
Diminishing buffer Change in soil CML 2002
capacity parameters
Leaching of Leaching of H+ Leaching of Al3+

nutrient cations and pH decrease
Reduced Acid stress Ecotoxicity

nutriments Al3+
Decrease in Decrease in
biodiversity bioproductivity
EI99, ReCiPe LIME (endpoint)
Figure 8 Flow diagram for acidification. The LCIA methods are positioned along the cause-effect chain.

The 10 models have been rated against the criteria reported in the “LCIA- Framework and
requirements document” (EC-JRC, 2010b). The results are summarized in the table below27.
Discussions of specific issues in the evaluation of the models against the criteria are
presented in the background documentation28 (Acidification.xls), where a table with the
detailed evaluation including the whole sets of sub-criteria is given.
27
E: no compliance
28
58
Table 16 Summary table documenting the analysis of the midpoint characterisation methods against the adapted criteria for acidification. (split into two tables)
TRACI EDIP 2003 MEEUP Accumulated Exceedence (Seppälä et al.,
(Norris, 2003) (Potting et al.,1998) (Methodology report) 2006)
The scope of the model for the The scope of the model for the
evaluation of acidifying chemicals is
evaluation of acidifying chemicals is evaluation of acidifying chemicals is
B largely applicable, although the model The scope of the model for the evaluation
Completeness not compatible with LCA the method not applicable, as lacking of an
/ is parameterized for US and D C B of acidifying chemicals on the European
of scope discounts effects of acidifying atmospheric fate model. CF are
C addressed terrestrial acidification only. scale is fully applicable
deposition occurring in area above therefore rather at the safe side in
CFs are rather at the safe side in soil
the critical load line with a regulatory context
fate
Limited environmental relevance. Lack of environmental relevance is
Limited environmental relevance. The High environmental relevance for
Consider a full atmospheric fate, the obvious, do not consider
Environmental method fully considers atmospheric biodiversity. Full atmospheric and soil
C D marginal change in biodiversity, but E atmospheric fate and soil sensitivity, A
relevance fate, but not the soil sensitivity to assessment considered. Sensitive to
discount deposition in area above thus it doesn't enable any regional
acidifying deposition emission scenario and current critical load
critical load differentiation
Not further evaluated, because the Not further evaluated, because the
Scientific thresholds within the categories Model components extensively reviewed
thresholds within the category thresholds within the category
robustness & B 'completeness of scope' and B and uncertainty estimates available in term
'environmental relevance' were not 'environmental relevance' were not
Certainty 'environmental relevance' were not of spatial variation and emission scenarios
reached reached
reached
The method and the CFs are well
documented and accessible.
Documentation Not further evaluated, because the Not further evaluated, because the Characterization model and input data not
thresholds within the categories
& Transparency thresholds within the category thresholds within the category easily accessible. Can potentially be
D C 'completeness of scope' and B
& 'environmental relevance' were not 'environmental relevance' were not adapted to generate CFs for different
'environmental relevance' were not
Reproducibility reached reached continents if complemented with a global
reached
atmospheric model and expert judgment on
sensitive areas
Not further evaluated, because the Not further evaluated, because the Not further evaluated, because the
thresholds within the category thresholds of 'completeness of scope' thresholds within the category Readily applicable. Most important
Applicability A
'environmental relevance' were not and 'environmental relevance' were 'environmental relevance' were not acidifying chemical are covered.
reached not reached reached
The method meets the science based
The method lacks of sufficient
The method is not in line with the The method is not in line with the criteria. It Includes atmospheric and Soil
environmental relevance. It fully
scope of LCA and lack of sufficient scope of LCA and lack of fate factors distinguishing between load to
considers atmospheric fate, but not
environmental relevance as the environmental relevance is obvious A sensitive area and insensitive area for
Science based the soil sensitivity to acidifying
E E increase in deposition in sensitive E as it does not consider atmospheric / biodiversity. It could be applicable
criteria deposition. It needs to be at least
area is discounted, i.e. change in fate and soil sensitivity, thus it B worldwide at continental level if
complemented by average soil fate
area above critical load is not doesn't enable any regional complemented by a global atmospheric fate
factors distinguishing for sensitive and
relevant enough in the LCA context differentiation model and expert estimate on soil sensitive
non-sensitive areas
area.
AE-type calculations are used for policy
purposes in Europe by the European
Stakeholders Not further evaluated, because the Not further evaluated, because the Not further evaluated, because the
Commission and the UNECE LRTAP
acceptance thresholds within the science based thresholds within the science based thresholds within the science based B
Convention, but models and data are
criteria criteria were not reached criteria were not reached criteria were not reached
difficult to understand without expert
knowledge
59
CML 2002 ReCiPe midpoint LIME midpoint

(Huijbregts et al.2001) (van Zelm et al., 2007a) (Hayashi et al. 2004)
The scope of the model for the evaluation of
The scope of the model for the evaluation of B The scope of the model for the evaluation of acidifying
acidifying chemicals is largely applicable,
Completeness of scope B acidifying chemicals on the European scale is / chemicals on the European scale is fully applicable, but B
although the model is parameterized for
fully applicable C factors are based on acidification of forests soil only
Japan.
High environmental relevance for biodiversity:
High environmental relevance for biodiversity: Full
Full atmospheric and soil assessment
atmospheric and soil fate considered for forests Extrapolated Limited environmental relevance. The CF
Environmental considered. Sensitive to emission scenario and
B B to other ecosystems only in a second step. Sensitive to D fully considers atmospheric fate, but not the
relevance current critical load. Relevance of the dose-
emission scenario and current critical load. It further includes soil sensitivity
response being curve being dependent of the
an effect factor
buffer capacity itself has to be verified
Model components extensively reviewed and
B B Model components extensively reviewed; uncertainty Not further evaluated, because the
Scientific robustness & uncertainty estimates available in term of spatial
/ / estimates not provided, but discussed in term of temporal thresholds within the category
Certainty variation and emission scenarios. Not full up-to-
C C emission scenarios 'environmental relevance' were not reached
date data and models
The method and the CFs are well documented
and accessible. Characterization model and
The method and the CFs are well documented and
input data not easily accessible. Can potentially
Documentation & B accessible. Characterization model and input data not easily Not further evaluated, because the
be adapted to generate CFs for different
Transparency & B / accessible. Can potentially be adapted to generate CFs for thresholds within the category
continents if complemented with a global
Reproducibility C different continents if complemented with a global 'environmental relevance' were not reached
atmospheric model. Feasibility of the dose-
atmospheric model and expert judgment on sensitive areas
response curve should be verified for other
continents
Readily applicable. Most important acidifying Readily applicable. Most important acidifying chemical are
Applicability A A thresholds within the category
chemical are covered. covered.
'environmental relevance' were not reached
The method meets the science based criteria. The method meets the science based criteria. Base
The method generally meets the science
Conclusions for Accumulated Exceedence saturation factor is likely to provide an interesting alternative
based criteria, but the selected midpoint
method apply. Relevance and feasibility of the to the critical load based methods. Extension of the concept
indicator lack of sufficient environmental
Overall evaluation of dose-response curve should be verified for other to other ecosystems than forests is required. The possibility
B B E relevance. In fact the cause-effect chain is
science based criteria continents (data requirement higher). To be seen to determine proxies of the changes in base saturation and in
only modelled up to the deposition of acid
if/how expert judgment can be incorporated. It dose-response for various continents need also to be further
eq. Potentials and do not account for a
needs to be complemented by a global explored. It needs to be complemented by a global
sensitive and non sensitive area.
atmospheric fate model. atmospheric fate model.
Based on models and data used in the

Overall evaluation of C Moderate stakeholder acceptance, Dutch government behind Not further evaluated, because the
evaluation of acidifying impacts for the EC (apart
stakeholders C / the method. Expert knowledge is required to understand the C thresholds within the science based criteria
the final category indicator), but difficult to
acceptance criteria D model were not reached
understand without expert knowledge
60
Table 17 Summary table documenting the analysis of the endpoint characterisation methods against the adapted criteria for acidification.
ReCiPe Endpoint Ecoindicator 99 LIME endpoint

(van Zelm et al., 2007a) (Goedkoop and Spriesma, 2000) (Hayashi et al. 2004)
The scope of the model for the evaluation of acidifying The scope of the model for the evaluation of The scope of the model for the evaluation of
B B
Completeness of chemicals on the European scale is applicable, but acidifying chemicals on the European scale is acidifying chemicals is largely applicable, although
/ C /
scope factors are based on acidification of European forests applicable, but factors are based on acidification the model is parameterized for Japan, factors are
C C
only of Dutch forests only based on acidification of forests only
High environmental relevance for ecosystem quality: Limited environmental relevance. Does not enable High environmental relevance for ecosystem
Full atmospheric and soil fate considered for forests discriminating between the atmospheric fates of quality: Full atmospheric and soil assessment for
Environmental
B and extrapolated to other ecosystems. Sensitive to C chemicals. Soil fate considered for forests and B forests considered and extrapolated to other
relevance
emission scenario and current critical load. It further extrapolated to other ecosystems. It further ecosystems. Sensitive to emission scenario and
includes an effect factor includes an effect factor sensitive area
The method itself has not been peer reviewed, but Model components extensively reviewed, but the
Scientific B Model components extensively reviewed and
the underlying models components are. However, dose-response model is poorly representative as
robustness & / uncertainty estimates available in term of spatial C C
poor scientific quality for the fate model. Expert the plant growth rate is based on 1 species.
Certainty C variation and emission scenario
judgment on uncertainty estimates Uncertainty is not discussed
Documentation accessible. Characterization model and input data not The method and the CFs are well documented
B accessible. Characterization model and input data
& Transparency easily accessible. Can potentially be adapted to and accessible. Characterization model and input
B C / are not easily accessible. Need to be
& generate CFs for different continents if complemented data not easily accessible. Data requirement
C complemented by a global fate model, effect model
Reproducibility with a global atmospheric model and expert judgment higher.
should be verified for other continents
on sensitive areas
Readily applicable. Most important acidifying chemical Readily applicable. Most important acidifying Readily applicable. Most important acidifying
Applicability A A A
are covered. chemical are covered. chemical are covered.
The method meets the science based criteria in some
aspects. Base saturation factor and dose-response
slopes are likely to provide an interesting basis for the The method meets the science based criteria in
next generation of acidification methods. Additional The method meets the science based criteria in some aspects. The effect factor is poorly
Overall
B studies are, however, required to verify the approach some aspects. Poor scientific quality for the fate B representative as based on 1 plant species. The
evaluation of
/ chosen by Van Zelm et al, and extension of the C model and rather old soil fate and effect models / model is parameterized for Japan. To generate
science based
C concept to other ecosystems than forests is required. are used. It is therefore difficult to generate effect C CFs for other region, it needs to be complemented
criteria
The possibility to determine proxy of the changes in data for other continents than Europe. by a global fate model and effect model should be
base saturation and in dose-response for various verified for other continents.
continents need to be further explored. It needs to be
complemented by a global atmospheric fate model.
Overall
evaluation of C Moderate stakeholder acceptance, Dutch government Low stakeholder acceptance, no authoritative B Moderate stakeholder acceptance, Japanese
stakeholders / behind the method. Expert knowledge is required to C body behind the method. Pure scientific work, / government behind the method. Expert knowledge
acceptance D understand the model understood with expert knowledge C is required to understand the model
criteria
61

As stated in the LCIA - Framework and requirements document (EC-JRC, 2010b), some
specific criteria of ‘Environmental relevance’ and ‘scientific robustness’, have been specified
for acidification and represent the basis for the evaluation. The main criteria focus on: the
presence of an atmospheric fate and transport model; a fate sensitivity factor discriminating
between sensitive (including areas with limited buffer capacity) and insensitive areas;
acidification potential considered at midpoint; updating emission data and temporal changes
evaluation for future emission. For endpoint methods, also the presence of a dose-response
model for biodiversity/bioproductivity is considered relevant.
3.7.4 Discussion of uncertainties and the importance of spatial

differentiation
The uncertainties in different methods are mainly expressed in term of spatial and
temporal variability or by expert judgment estimates (Eco-indicator 99). The intrinsic
uncertainty of the fate and effect models is not reported for any of the recommended models.
Temporal variability is usually taken into account through present and future emission
scenarios. Characterization factors for acidification increase up to a factor of 13 from 20
years to a 500 years’ time horizon and influence the difference between chemicals up to a
factor 4 as shown by Van Zelm and colleagues (2007a).
In general spatial variability in atmospheric fate can model differences in deposition of
acidifying chemicals on a few tens or hundreds kilometres scale (typically 100 x100 km
scale). Soil fate modelling can be even more detailed up to a few square kilometres. This
source of uncertainty could result in differences in CFs up to two or three orders of
magnitude among individual European countries with different integrated sensitivity (Potting
et al. 1998a; Posch et al. 2008). This difference is very important compared to the variability
of the CFs among the chemicals, which is typically ranging within one order of magnitude. In
case the emissions take place only in specific locations or countries, it makes therefore little
sense to distinguish between individual substances such as SO2 and NOx and disregarding
the spatial variability.

Following midpoint methods evaluation, the method of AE (Accumulated Exceedence)
(Seppälä et al. 2006) was chosen.
TRACI, EDIP 2003, MEEuP and LIME (at midpoint) were not recommended because they
didn’t reach some evaluation thresholds.
The scope of the TRACI model for the evaluation of acidifying chemicals is largely
applicable. At the moment, the model is parameterized only for US and addresses terrestrial
acidification only. The method was not chosen because it presents a limited environmental
relevance: it fully considers atmospheric fate, but not the soil sensitivity to acidifying
deposition. Environmental relevance can be improved by complementing the model with an
estimation of soil fate factors distinguishing between sensitive and non-sensitive areas.
62
Finally, the ASTRAP model for atmospheric fate, which is used in the TRACI model, is
considered outdated.
The scope of the EDIP 2003 model for the evaluation of acidifying chemicals is not
compatible with LCA. The method discounts effects of acidifying deposition occurring in
areas above the critical load. As for TRACI, the atmosphere deposition model (RAINS) is
dated before 2000.
MEEuP is completely lacking environmental relevance because it disregards atmospheric
fate and soil sensitivity. Furthermore it doesn't enable any regional differentiation.
LIME (at midpoint) generally meets the science based criteria, but the selected midpoint
indicator lacks of sufficient environmental relevance. In fact the cause-effect chain is only
modelled up to the deposition of acid equivalents potentials and does not account for
sensitive and non sensitive areas.
CML2002 reaches a good evaluation with the exception of being less up-to-date and
showing less stakeholder importance than others.
RECIPE (at midpoint) sets an interesting basis for the next generation of acidification
methods based on Base saturation factor (an alternative to the critical load based methods).
Nevertheless, an extension of the concept to other ecosystems than forests is required and
the feasibility to generate a set of consistent CFs for each continent still has to be further
explored (e.g. the possibility to determine proxies for the effect factor for various continents).
AE (Accumulated Exceedence) is to be preferred as default method for midpoint
evaluation of acidification. The updated factors provided by Posch and colleagues (2008)
should be used. The method meets the science based criteria, and it shows a good
stakeholder acceptance as AE-type calculations are used for policy purposes and by the
UNECE Convention on Long-range Transboundary Air Pollution (LRTAP). It includes
atmospheric and soil fate factors sensitive to emission scenario and distinguishes between
load to non-sensitive and sensitive areas. This is probably the most readily adaptable
method that can be used in further research to generate Global default Characterisation
Factors (CFs) or a set of consistent CFs for each continent if complemented by a set of
regional/continental models which are consistent with each other (that could eventually be
integrated in one global model, although not required) and expert estimate on soil sensitive
area.
Similar conclusions apply for CML and ReCiPe (midpoint) methods, but they both suffer
from a weaker stakeholder importance. In addition CML is based on less up-to-date data and
models, and for ReCiPe the feasibility to generate a set of consistent CFs for each continent
still has to be further explored (e.g. the possibility to determine proxies for the effect factor for
various continents).

At endpoint level, no method is recommended to be use because no methods are
sufficiently mature to be recommended.
As interim, among other the method developed by van Zelm and colleagues (2007a), as
used in Recipe, can be used but only for internal applications.
63
All the methods need to be complemented by a global fate model and effect models
should be verified for other continents.
LIME (at endpoint) presents model components extensively reviewed, and shows a high
environmental relevance for ecosystem quality. Full atmospheric and soil assessment for
forests is considered and extrapolated to other ecosystems. Sensitive to emission scenario
and sensitive area are present but the dose-response model is poorly representative as the
plant growth rate is based on 1 species.
Ecoindicator 99 meets the science based criteria in some aspects but it has a poor
scientific quality for the fate model and rather outdated soil fate and effect models are used.
However, among the evaluated endpoint methods, the one developed by van Zelm and
colleagues (2007a) (as described in ReCiPe methodology) sets the most interesting basis for
the next generation of acidification methods. Therefore this method is qualified as interim,
due to the fact that the dose-response model and the integration with the recommended
midpoint model need to be further evaluated and the feasibility to adapt the effect factor for
different ecosystems needs to be further explored (currently based on European forest only).
Furthermore, it doesn’t consider terrestrial acidification in other ecosystem apart from forest
and acidification on aquatic ecosystem

The recommended midpoint method for acidification provides an indicator based on the
exceedance of an acidifying deposition over the critical load. This indicator is not fully
consistent with the interim endpoint indicator as (1) it is based on a different characterization
model linking emission to the change in soil parameters and (2) it is selected on a 'side track'
of the cause-effect chain to the damages. Therefore the dose-response model and indicator
of the interim method by van Zelm et al. is currently independent of the AE indicator and its
integration with this latter still need to be further evaluated.

At midpoint, the Accumulated Exceedence model (Seppälä et al. 2006 and Posh et al.
2008) is classified as “recommended with some improvements needed” (Level II out of III).
At endpoint level, no method is recommended to be used.
If an endpoint method is required, the method proposed by van Zelm et al. (2007a) can be
use, but this method is classified as an interim method, because the method is not mature
enough to be recommended.

Additional midpoint characterisation factors cannot be calculated by the LCA practitioner
but require access to and expertise in the underlying model. The number of substances
contributing to acidification is quite limited and hence the need for additional factors is not
foreseen to be an issue.
64
3.8 Eutrophication
Characterization factors for eutrophication are traditionally calculated at midpoint level, as
it is the case for the majority of the LCIA methods considered in this analysis (Guinée et al.,
2002, Potting et al. 2005; Norris 2003; Seppälä et al. 2006). Others are damage oriented
LCIA methods and relate emissions of eutrophying substances to impacts on the endpoint
biodiversity (Steen, 1999a,b; Goedkoop and Spriensma, 2000; Payet, 2006; Goedkoop et al.
2009, Itsubo et al., 2008a).
To the extent the methods consider impacts from biological material (BOD or COD), the
characterisation factor is typically calculated from the characterisation factor for N or P based
on the amount of biological material (expressed as BOD or COD) which would on average be
produced by natural (primary) production of biological material per input of N or P in aquatic
systems.
Several of the analysed characterisation models have separate treatment of terrestrial and
aquatic systems, and most of them only address one of the two.
In LCIA- Analysis document (EC-JRC, 2010a) the pre-selection of characterisation models
for the terrestrial and aquatic eutrophication is shown separately for the two sub categories
terrestrial and aquatic eutrophication.
3.8.1.1 Terrestrial eutrophication

Midpoint
 Accumulated Exceedance (AE): Provides European country-dependent
characterisation factors for Acidification and Terrestrial Eutrophication. The
atmospheric transport and deposition model to land area is determined using the
EMEP model combined with a European critical load database. The eutrophication
potential is expressed in accumulated exceedance. The dose-response curve
implicitly equals 1 (Seppälä et al., 2006).
 CML2002: Enrichment of terrestrial ecosystems with the macronutrients N and P.
Characterisation model based on the stoichiometry given by the Redfield ratio
between N and P (derived from the average composition of algae). Also provides
characterisation factors for organic material emissions to water presented as BOD
or COD. Expressed as PO43--equivalents (Guinée et al., 2002). EDIP97 applies
similar approach expressing results in NO3--equivalents (Hauschild and Wenzel,
1998b), but the two models are so similar that only CML2002 is analysed here.
 EDIP2003: Increase in area of terrestrial ecosystem exposed above critical load
for N, site-dependent at country-level in Europe – expressed as m2 unprotected
ecosystem, calculated using the RAINS model (Hauschild and Potting, 2005).
65
Endpoint
 Eco-indicator 99: Uses a simplified fate assumption to determine the fraction of an
acidifying or eutrophying emission that is deposited on Europe (equal for all the
chemicals). The effect factor is determined applying the Dutch Nature Planner
model that focuses on the percentage of threatened species in The Netherlands
caused by acidifying and eutrophying emissions. The indicator expresses the
change in PDF m2 yr per marginal change in deposition (Goedkoop and
Spriensma, 2000)
 EPS2000: Covers both terrestrial and aquatic eutrophication. Assumes equal
distribution between deposition on natural areas, agricultural areas and water.
Damage factor for terrestrial ecosystems is based on estimate of eutrophication’s
share in number of endangered species in Sweden which is assumed to be valid
globally (Steen, 1999 a,b).
IMPACT 2002+, LIME and ReCiPe do not include terrestrial eutrophication impacts.
3.8.1.2 Aquatic eutrophication

Midpoint
 CML2002: Enrichment of aquatic ecosystems with the macronutrients N and P. No
fate model. Characterisation model based on the stoichiometry given by the
Redfield ratio between N and P (derived from the average composition of algae).
Also provides characterisation factors for organic material emissions to water
presented as BOD or COD. Indicator results expressed as PO43--equivalents
(Guinée et al., 2002). EDIP97 applies similar approach expressing results in NO3--
equivalents (Hauschild and Wenzel, 1998b), but the two models are so similar that
only CML2002 is analysed here.
 EDIP2003: Combination of EDIP97 characterisation factors (NO3—equivalents)
with exposure factors expressing the degree to which the emitted nutrient reaches
the aquatic end compartment after removal processes active in the fate model. For
waterborne emissions the exposure factors are calculated using the CARMEN
model, for airborne emissions of NOx and NH3, the exposure factor is calculated
using RAINS. (Hauschild and Potting, 2005, Potting and Hauschild, 2005).
 LIME midpoint: Increase in nutrients and COD and resulting oxygen depletion and
impacts on benthic communities modelled in four Japanese closed marine water
bodies as consequence of emissions of N and P- compounds and organic material
to water (no consideration of impacts in freshwater). (Itsubo et al., 2008a)
 ReCiPe midpoint: Approach similar to EDIP2003 but using EUTREND for
atmospheric emissions and distinguishes freshwater systems (only P-emissions
considered) and marine systems (only N considered) (Struijs et al., 2009b)
 TRACI: Characterisation factor product of nutrient factor, determined by the
substance’s content of N or P, and a transport factor, which reflects the probability
that the emission arrives in an aquatic environment to which it is the limiting
nutrient. Nutrient factors are identical to CML2002 characterisation factors
66
(including factors for COD and BOD). Transport factors are spatially differentiated
at the level of US states and developed for both atmospheric and waterborne
transport. Indicator results expressed as PO43--equivalents (Norris, 2003).
Endpoint
 EPS2000: Covers both terrestrial and aquatic eutrophication and includes factor
for BOD and COD emissions to water. Very simple fate model for waterborne
emissions assuming a fixed global distribution of N and P between N- and P-
limited systems and disregarding removal processes. For emissions of NOx a fixed
fraction is assumed to be deposited on water. Damage factor for waterborne
emissions is extrapolated from Scandinavian situation to the world. Damage factor
for airborne NOx emissions is based on the estimated contribution in the Baltic
Sea region, which is assumed to be globally representative (Steen, 1999a,b).
 IMPACT2002+ endpoint: Uses CML2002 characterisation factors at midpoint and
considers damage to freshwater systems using a damage model developed to
represent the relationship between P-exposure and species diversity in terms of
fraction of affected, or disappeared species. This method is not readily operational
and it has still not been validated with field measurements and needs to be
complemented by a fate model. Nevertheless, it could set an interesting basis for
further developments in assessing the effect of eutrophication on aquatic
ecosystems (Payet, 2006).
 LIME: Increase in nutrients and COD and resulting oxygen depletion and impacts
on benthic communities modelled in four Japanese closed marine water bodies
(no consideration of impacts in freshwater). Damage calculated for airborne N-
emissions and waterborne emissions of N and COD and expressed as loss of
benthos biomass and loss of fishery catches (Itsubo et al., 2008a).
 ReCiPe endpoint: Predicted P concentration increases in freshwater systems at
midpoint are linked to ecosystem damage (potentially disappeared fraction of
species) using database correlating P concentrations and macro fauna species
diversity in Dutch ecosystems to predict damage in terms of potentially
disappeared fraction of species (Struijs et al., 2009b)
For the evaluation the following methods are left out: LUCAS (same as TRACI but applied
in Canada), IMPACT2002+ midpoint (taken directly from CML2002, but distinguishes
between N- and P limited watersheds), MEEUP (identical to CML2002 but adds factors for
BOD, DOC, TOC and suspended solids derived from the CML2002 factor for COD by scaling
it in accordance with EU legislation, e.g. EU Directive on urban wastewater treatment) and
Swiss Ecoscarcity (in accordance with Swiss regulation, targets set for compounds or total N
and P, no characterisation modelling). Eco-indicator 99 does not cover aquatic
eutrophication.
Figure 9 shows the cause-effect chain for eutrophication of the aquatic and terrestrial
environment from air- and waterborne emissions of nutrients (N and P) and biological
material (COD or BOD) with the most important pathways highlighted (bold arrows). The
analysed LCIA methods are reported according to their position along the cause-effect chain.
67
Figure 9 Flow diagram of the cause-effect chain for eutrophication Method evaluation

The methods have been rated against the criteria developed in guidance document LCIA -
Framework and requirements document (EC-JRC, 2010b). The results are summarized in
the table below29. Discussions of specific issues in the evaluation of the models against the
criteria are presented in the background documentation30 (Eutrophication.xls), where a table
with the detailed evaluation including the whole sets of sub-criteria is given.
29
E: no compliance
30
68
Table 18 Summary of the evaluation results of the midpoint characterisation methods against the criteria for aquatic eutrophication. ( at mid and endpoint)
TRACI
Criteria CML 2002 EDIP2003 aquatic LIME midpoint ReCiPe midpoint
Norris, 2003
The scope of the model for the The scope of the model for the The scope of the model
The scope of the model for
evaluation of eutrophying evaluation of eutrophying for the evaluation of
the evaluation of eutrophying
substances is limited to aquatic substances is limited to aquatic eutrophying substances is
The scope of the model for the substances is applicable for
ecosystems and only ecosystems where it addresses applicable for aquatic
evaluation of eutrophying aquatic ecosystems on the
B A addresses issues related to all relevant issues. The model B ecosystems, but not for
Completeness substances is applicable for European scale. No
- - C oxygen depletion. No B represents European - terrestrial ecosystems.
of scope aquatic as well as terrestrial consideration of terrestrial
C B consideration of terrestrial freshwaters and marine coastal C The model is
ecosystems. Global validity, no ecosystems. Spatial
ecosystems. The model waters. Spatial differentiation parameterized for US and
temporal differentiation differentiation at the level of
represents Japanese coastal according to archetype emission spatially differentiated at
countries, no temporal
waters, freshwater systems situations, no temporal the level of US states, no
differentiation
ignored differentiation temporal differentiation
Environmental relevance is Environmental relevance

low, most important fate high, removal processes in Environmental relevance is Environmental relevance high, Environmental relevance
D A B A A
Environmental processes determining aquatic system modelled, high although removal removal processes modelled, is high although removal
- - - - -
relevance availability and exposure of but no distinction between processes for nutrients are distinction between N- and P- processes for nutrients
E B C B B
sensitive environments are freshwater and marine missing limited systems. are missing
missing systems.
Underlying fate model and Model components based on Midpoint model reviewed,
Midpoint model of limited adaptation to LCIA use Model components based on existing European models and further components
Scientific D B
environmental relevance due reviewed, uncertainty from existing Japanese models and reviewed quantification of derived from reviewed
robustness & - B C B -
to missing fate considerations, spatial variability quantified partially reviewed (?). No spatially determined uncertainty information, some
Certainty E C
no treatment of uncertainty and several emission uncertainty considerations. range and characterisation of treatment of uncertainty in
situations covered different emission situations resulting CFs.
The method principles
The method principles and the The method is documented
The method and CFs and the CFs are
CFs are documented and and accessible, the CFs only The method is documented and
Documentation documented and accessible documented and
accessible for use in a B available in Japanese, accessible with all CFs for use
& Transparency for use in a reproducible accessible for use in a
A reproducible way. B - reproducibility not clear. B in a reproducible way. C
& way. Characterization model reproducible way.
Characterization model and C Characterization model and Characterization model and
Reproducibility and input data not easily Characterization model
input data easily accessible input data not easily input data not easily accessible.
accessible. and input data not easily
and applicable. accessible.
accessible.
Characterisation factors
Characterisation factors for Characterisation factors for Characterisation factors for
for most relevant
Characterisation factors for most relevant compounds most relevant compounds most relevant compounds
compounds available and
most relevant compounds available and easy to available and easy to available and easy to
Applicability A A A A A easy to supplement -
available and easy to supplement - update supplement - update depends supplement - update depends
update depends on
supplement depends on developers of on developers of underlying on developers of underlying
developers of ASTRAP
underlying model model model
model
69
TRACI
Criteria CML 2002 EDIP2003 aquatic LIME midpoint ReCiPe midpoint
Norris, 2003
Based on models for
Based on models for European
European conditions, Parameterised to US
conditions, addresses all aspect
addresses all aspects of Parameterised to Japanese conditions. Fate model
of aquatic eutrophication for
Global validity and very limited aquatic eutrophication for conditions, addressing oxygen well reviewed, but NH3
both airborne and waterborne
uncertainty due to nearly total both airborne and depletion in coastal waters in not covered. Further
emissions. Spatial differentiation
absence of fate modelling, waterborne emissions. Japan (a bit narrow scope) components derived from
found of low importance but can
Science based B which also means limited Spatial differentiation Model components reviewed reviewed information,
be developed for European
criteria overall - environmental relevance. B supported for European B (?). No uncertainty B B some treatment of
countries. No treatment of
evaluation C Method principles and CFs countries. Spatially considerations. Method spatially determined
uncertainty in resulting CFs, but
documented and accessible determined uncertainty principles and CFs uncertainty in resulting
factors developed for different
for all main contributing discussed, found to be low. documented and accessible for CFs. Method principles
emission sources. Method and
substances. Site-generic and site- all main contributing and CFs documented and
CFs documented and
dependent CFs documented substances. accessible for all main
accessible for N-total, P-total,
and accessible for all contributing substances.
NOx and NH3.
relevant substances.
Moderate stakeholder Moderate stakeholder
Stakeholders Moderate stakeholder Moderate stakeholder
B Limited stakeholder acceptance, method accepted B B acceptance, method
acceptance: acceptance, official Danish acceptance, method accepted
- acceptance. Model easily B C by Japanese government(?). - - accepted by US EPA.
Overall LCIA methodology. Model by Dutch government. Model
C understandable Expert knowledge is required to C C Model reasonably
evaluation reasonably understandable reasonably understandable
understand the model understandable
At endpoint
EPS2000 IMPACT 2002+ endpoint LIME endpoint ReCiPe endpoint
Criteria
Steen, 1999
A The scope of the model for the B- The scope of the model for the C The scope of the model for the evaluation B The scope of the model for the
- evaluation of eutrophying C evaluation of eutrophying substances is of eutrophying substances is limited to evaluation of eutrophying substances
B substances is applicable for applicable for aquatic freshwater aquatic ecosystems and only addresses is limited to aquatic ecosystems where
aquatic as well as terrestrial ecosystems, but not for terrestrial issues related to oxygen depletion. No it addresses all relevant issues.
Completeness of
ecosystems. Lacks an ecosystems. The damage model is consideration of terrestrial ecosystems. Spatial differentiation according to
scope
atmospheric fate model. No based on European database. Spatial The model represents Japanese coastal archetype emission situations, no
spatial or temporal differentiation, differentiation at the level of countries in waters, freshwater systems ignored temporal differentiation. Effect model
global validity Europe, no temporal differentiation based on Dutch data for freshwaters
and marine coastal waters
C Environmental relevance is D- Environmental relevance is high on effect B Environmental relevance is high although A Environmental relevance high,
limited, no real fate model, but E side but low on fate side for freshwater removal processes for nutrients are - important removal processes in water
Environmental global average situation systems, marine systems not missing B modelled, distinction between
relevance estimated for both fate and effect considered. exposure of N- and P-limited systems
based on Swedish/Scandinavian for damage modelling, only damage
data model for the latter.
B Very simple fate model which has C- Endpoint model internally peer reviewed B Some model components have been B Model components based on existing
- not been reviewed, effect model D in project, no treatment of uncertainty - reviewed but for other the situation is not European models and reviewed.
Scientific
C based on rough empirically based C clear. The model addresses the main Quantification of spatially determined
robustness &
estimates. Consistent uncertainty aspects of oxygen depletion in estuaries uncertainty range and characterisation
Certainty
considerations, but not of all and coastal waters. No uncertainty of different emission situations
aspects. considerations.
70
EPS2000 IMPACT 2002+ endpoint LIME endpoint ReCiPe endpoint

Criteria
Steen, 1999
B The method principles and the B The method principles and the CFs are B The method is documented and B The method is documented and
Documentation CFs are documented and documented and accessible for use in a - accessible, the CFs only available in accessible with all CFs for use in a
& Transparency accessible for use in a reproducible way. Characterization C Japanese, reproducibility not clear. reproducible way. Characterization
& reproducible way. model and input data easily accessible Characterization model and input data not model and input data not easily
Reproducibility Characterization model and input and applicable. easily accessible. accessible.
data not easily accessible.
B Not all relevant substances have A Characterisation factors for most relevant A Characterisation factors for most relevant A All relevant substances included
characterisation factors compounds available and easy to compounds available and easy to -
Applicability
supplement supplement - update depends on B
developers of underlying model
B Very simple fate modelling, no B- Very simple fate modelling, no removal B Parameterised to Japanese conditions, B Based on models for European
- removal mechanisms considered. C mechanisms considered. Damage to addressing damage to benthic conditions, addresses all aspect of
C Damage to natural environment natural environment considers impacts communities of coastal waters in Japan (a aquatic eutrophication for both
very simple based on empirical from P compounds on freshwater bit narrow scope) Model components airborne and waterborne emissions.
data. Consistent uncertainty ecosystems only. No uncertainty reviewed (?). No uncertainty Spatial differentiation found of low
Science based
considerations, but not of all considerations. Method principles and considerations. Method principles and CFs importance but can be developed for
criteria overall
aspects. Method principles and CFs documented and accessible for all documented and accessible for all main European countries. No treatment of
evaluation
CFs documented and accessible main contributing substances (P contributing substances. uncertainty in resulting CFs, but
for a limited selection of compounds). factors developed for different
contributing substances. emission sources. Method and CFs
documented and accessible for N-
total, P-total, NOx and NH3.
Stakeholders D Low stakeholder acceptance, C Limited stakeholder acceptance. Model B Moderate stakeholder acceptance, C Moderate stakeholder acceptance,
acceptance: model easily understandable easily understandable - method accepted by Japanese method accepted by Dutch
Overall C government (?). Expert knowledge is government. Model reasonably
evaluation required to understand the model understandable
71
Table 19 Summary of the evaluation results of the endpoint characterisation methods against the criteria for terrestrial eutrophication.
Accumulated Exceedence CML 2002 EDIP2003 terrestrial EPS2000 Eco-indicator 99
Criteria
(Seppälä et al., 2006) Steen, 1999
The scope of the The scope of the model for
The scope of the model for model for the the evaluation of eutrophying The scope of the model for the
the evaluation of evaluation of substances is applicable for evaluation of eutrophying
evaluation of eutrophying on the
eutrophying substances is eutrophying terrestrial ecosystems on the substances is applicable for
A B B A B European scale is applicable, but
Completeness of applicable for terrestrial substances is European scale. No aquatic as well as terrestrial
- - - - - factors are based on combined
scope ecosystems on the applicable for aquatic consideration of aquatic ecosystems. Lacks an
B C C B C eutrophication/acidification of Dutch
European scale. No as well as terrestrial ecosystems. Spatial atmospheric fate model. No
forests only. Lacks an atmospheric
consideration of aquatic ecosystems. Global differentiation at the level of spatial or temporal
fate model
ecosystems validity, no temporal countries, temporal differentiation, global validity
differentiation differentiation included
Environmental
High environmental
relevance is low, most High environmental relevance Limited environmental relevance.
relevance for natural Environmental relevance is
important fate for natural environment. Full Does not enable discriminating
environment. Full limited, no real fate model, ,but
A D processes B atmospheric and soil between the atmospheric fates of
Environmental atmospheric and soil global average situation
- - determining - assessment considered. C B chemicals.
relevance assessment considered. estimated for both fate and
B E availability and C Sensitive to emission Soil fate considered for forests and
Sensitive to emission effect based on
exposure of sensitive scenario and current critical extrapolated to other ecosystems. It
scenario and current critical Swedish/Scandinavian data
environments are load further includes an effect factor
load
missing
Midpoint model of Very simple fate model which
Model components Underlying fate model and The method itself has not been peer
limited environmental has not been reviewed, effect
extensively reviewed and adaptation to LCIA use reviewed, but the underlying model
Scientific D relevance due to B model based on rough
uncertainty estimates reviewed, uncertainty from components have. However, poor
robustness & B - missing fate B - empirically based estimates. C
available in term of spatial spatial variability quantified scientific quality for the fate model.
Certainty E considerations, no C Consistent uncertainty
variation and emission and several temporal Expert judgment on uncertainty
treatment of considerations, but not of all
scenarios scenarios investigated estimates
uncertainty aspects.
The method and the CFs
are well documented and
The method principles
accessible.
and the CFs are
Characterization model and The method and CFs The method principles and the
documented and The method and the CFs are well
Documentation input data not easily documented and accessible CFs are documented and
B accessible for use in a B B documented and accessible for use
& Transparency accessible. Can potentially for use in a reproducible way. accessible for use in a
- A reproducible way. - B - in a reproducible way.
& be adapted to generate Characterization model and reproducible way.
C Characterization C C Characterization model and input
Reproducibility CFs for different continents input data not easily Characterization model and
model and input data data not easily accessible.
if complemented with a accessible. input data not easily accessible.
easily accessible and
global atmospheric model
applicable.
and expert judgment on
sensitive areas
Characterisation factors for Characterisation factors for
Characterisation
most relevant compounds most relevant compounds
factors for most Readily applicable. Most important
available and easy to available and easy to Not all relevant substances
Applicability A A relevant compounds A B A eutrophying substances are
supplement - update supplement - update depends have characterisation factors
available and easy to covered.
depends on developers of on developers of underlying
supplement
underlying model model
72
Accumulated Exceedence CML 2002 EDIP2003 terrestrial EPS2000 Eco-indicator 99

Criteria
(Seppälä et al., 2006) Steen, 1999
The method meets the Based on models for
Global validity and
science based criteria. It European conditions,
very limited Very simple fate modelling, no
Includes atmospheric and addresses terrestrial
uncertainty due to removal mechanisms
soil fate factors eutrophication for airborne The method meets the science
nearly total absence considered. Damage to natural
distinguishing between emissions. Spatial based criteria in some aspects.
of fate modelling, environment very simple based
load to sensitive area and differentiation supported for Poor scientific quality for the fate
Science based B which also means B on empirical data. Consistent B
insensitive area for European countries. model and rather old soil fate and
criteria overall A - limited environmental B - uncertainty considerations, but -
biodiversity. It could be Quantification of spatially effect models are used. It is
evaluation C relevance. Method C not of all aspects. Method C
applicable worldwide at determined uncertainty in therefore difficult to generate effect
principles and CFs principles and CFs documented
continental level if resulting site-generic CFs. data for other continents than
documented and and accessible for a limited
complemented by a global Site-generic and site- Europe.
accessible for all main selection of contributing
atmospheric fate model dependent CFs documented
contributing substances.
and expert estimate on soil and accessible for all relevant
substances.
sensitive area. substances.
High stakeholder Moderate stakeholder

Stakeholders Low stakeholder acceptance, no
acceptance, but models B Limited stakeholder B acceptance, official Danish C
acceptance: Low stakeholder acceptance, authoritative body behind the
B and data are difficult to - acceptance. Model - LCIA methodology. Expert D -
Overall model easily understandable method. Pure scientific work,
understand without expert C easily understandable C knowledge is required to D
evaluation understood with expert knowledge
knowledge understand the model
73

Specific criteria of ‘Environmental relevance’ and ‘scientific robustness’ have been
specified for aquatic and terrestrial eutrophication in the LCIA - Framework and requirements
document (EC-JRC, 2010b).
The main criteria focus on: the presence of a fate and transport model; advection out of a
region not considered a final loss; influential fate processes considered (for aquatic systems:
denitrification, precipitation and sedimentation of P; for terrestrial systems: oxidation,
deposition); for damages on ecosystems, a fate sensitivity factor discriminating between
sensitive and insensitive recipients is considered; magnitude of exceedance for exposure
above critical level is considered; potency or dose-response is included; distinction of
individual N- and P-compounds; latest knowledge for the cause-effect chain with the critical
links are covered (Atmospheric fate and transport model, exposure model, potency or dose-
response model); coverage of the impacts in the modelling from midpoint to endpoint is
complete.
It has to be noted that for LIME, some information is missing or partially incomplete in the
documentation available for this guideline due to difficulties in accessing the parts of the
background information that was not provided in English.
3.8.4 Discussion of uncertainties and the importance of spatial

differentiation
The uncertainties are mainly expressed in term of spatial and temporal variability or by
expert judgment estimates (Eco-indicator 99). The intrinsic uncertainty of the fate and effect
models is not reported for any of the recommended models.
Temporal variability is taken into account for some of the terrestrial eutrophication models
through present and future emission scenarios.
Spatial differentiation in atmospheric fate can model differences in deposition of
eutrophying substances on a few tens or hundreds kilometres scale (typically 100 x100 km
scale). Soil fate modelling can be even more detailed up to a few square kilometres. This
source of uncertainty could be up to two or three orders of magnitude between individual
European countries (Potting et al. 1998 b; Posch et al. 2008) and is very important compared
to the variability between chemicals, which is typically ranging within one order of magnitude.
It therefore makes little sense to assess terrestrial eutrophication, distinguishing between
individual substances like NOx and NH3 and disregarding the spatial variability.
For aquatic eutrophication, the spatially determined variation between countries in Europe
(Potting and Hauschild, 2005, Goedkoop et al., 2009) or states within USA (Norris, 2003) is
found to be less than one order of magnitude making this level of spatial differentiation less
important for aquatic eutrophication. Here, however, the distinction of aquatic receiving
environments according to their limiting nutrient makes a crucial difference, and a distinction
between freshwater systems (generally P-limited) and marine water systems (generally N-
limited) is seen as very important.
It is questioned above whether organic material emissions should be counted as
contributing to eutrophication. If they are classified as eutrophying, their factors should be
74
derived from the factors for N or P assuming a standard primary production of BOD per
added nutrient.

AE (Accumulated Exceedence) is to be preferred as recommended model for midpoint
evaluation of terrestrial eutrophication. The method meets the science based criteria, and it
shows a good stakeholder acceptance as AE-type calculations are used for policy purposes
in Europe by the European Commission and by the United Nations Economic Commission
for Europe’s Long-range Transboundary Air Pollution Convention (UNECE LRTAP). It
includes atmospheric and soil fate factors sensitive to emission scenario and distinguish
between load to non-sensitive and sensitive areas. This is probably the most readily
adaptable method that can be used to generate a set of consistent CFs for each continent (or
for a generic one) if complemented by a set of regional/continental models which are
consistent with each other (that could eventually be integrated in one global model) and
expert estimate on soil sensitive area. Recommendation of the AE approach will also ensure
consistency between the treatment of terrestrial eutrophication and terrestrial acidification for
which impact category it is also recommended at midpoint level.
CML2002 and EDIP97 apply a similar approach, so only CML2002 was considered here.
The weak points of CML2002 are that it does not include any spatial and temporal
differentiation; it considers a worst case scenario, and there is no treatment of uncertainty.
Environmental relevance is low, most important fate processes determining availability and
exposure of sensitive environments are missing;
EDIP2003 is based on models for European conditions, and it addresses terrestrial
eutrophication for airborne emissions. Spatial differentiation is supported at the level of
European countries. It provides site-generic CFs with a quantification of their spatially
determined uncertainty. It quantifies the area exposed to critical level but disregards
exposure above thresholds. Compared to AE it has a moderate stakeholder acceptance.

Most of the characterisation models for aquatic eutrophication have a rather weak
modelling of the fate and ignore some of the important removal processes for both N and P.
The best modelling of aquatic fate at midpoint level is performed using the CARMEN model
(Klepper et al., 1995), as applied in two methodologies, ReCiPe and EDIP2003.The model is
likewise restricted to a European validity.
ReCiPe uses a more recent model for atmospheric fate and adopts a more consistent
framework presenting the characterisation factors as nutrient concentration increases
distinguishing aquatic receiving compartments according to the limiting nutrient. Therefore
the approach used in ReCiPe is preferred as recommended default method at midpoint level
for aquatic eutrophication.
CML, TRACI and EDIP2003 methods have the strength of addressing both terrestrial and
aquatic eutrophication. CML lacks a fate model, while TRACI and EDIP both suffer from a
weaker performance in some of the central science based criteria concerning the quality of
75
the underlying models and for EDIP concerning the distinction of receiving water bodies
according to limiting nutrient.
LIME has a very restricted focus on oxygen depletion in marine environment which makes
it unsuitable in a global, let alone a European context.

At endpoint, no methods are recommended to be used for terrestrial eutrophication.
None of the evaluated endpoint models (Eco-indicator 99 and EPS2000) reach the
sufficient scientific quality and consensus when linking midpoint to damage indicators,
therefore no formal recommendation has been done as at that stage.
EPS2000 has a very simple fate model (not reviewed) and effect model. Environmental
relevance is limited and a global average situation is estimated from Swedish data. It does
provide consistent uncertainty considerations, but not for all aspects.
Eco-indicator 99 addresses only terrestrial eutrophication and does so together with
acidification. It has a very simple fate model (not reviewed) and effect model. It represents
only Dutch conditions. Uncertainty considerations are limited.

At endpoint, no methods are recommended to be used for aquatic eutrophication.
As interim, the damage model to freshwater ecosystems (from P exposure) based on
empirical data for a large selection of Dutch ecosystems, as implemented in ReCiPe, can be
used. This approach is seen as the most relevant and scientifically sound for damage
modelling in freshwater eutrophication.
LIME is, among the evaluated endpoint models, the only one that addresses marine
eutrophication in a scientifically sound way, but its focus on oxygen depletion effects on
benthic communities that is too restricted to support a recommendation and the model is not
straightforward to be extended beyond the present Japanese setting. Therefore no formal
recommendation has been done for marine eutrophication.
EPS2000 also addresses aquatic eutrophication at endpoint level but it has no real fate
model and its effect model is based on questionable and undocumented extrapolation from
Swedish/Scandinavian damage data to estimate global damage data. It therefore performs
too weakly in the science-based criteria to support a recommendation.

For the models for freshwater eutrophication there is a fine consistency between midpoint
and the interim method for endpoint as they have been developed applying the same model
in a consistent framework. No endpoint model is recommended for terrestrial and aquatic
(both freshwater and marine) eutrophication.
76

Terrestrial eutrophication
As midpoint charachterisation method for terrestrial eutrophication it is recommended the
use of the Accumulated Exceedence (Seppälä et al. 2006 and Posh et al. 2008), classified
as being “recommended with some improvements needed” (Level II out of III).
No endpoint method is recommended for terrestrial eutrophication.
Aquatic eutrophication
The midpoint method recommended for aquatic eutrophication (both freshwater and
marine) is the method developed by Struijs et al., 2009b that uses the EUTREND model for
atmospheric emissions and distinguishes freshwater systems (only P-emissions considered)
and marine systems (only N considered). It is classified as being “recommended with some
improvements needed” (Level II out of III)
No endpoint method is recommended for aquatic eutrophication.
If an endpoint method is required, the method for aquatic eutrophication as developed in
ReCiPe can be used as interim for damage in freshwater systems, being the most
appropriate among the existing approaches but still with considerable shortcomings and
uncertainties.
No endpoint method can be used as interim for marine eutrophication.

Additional midpoint characterisation factors cannot be calculated by the LCA practitioner
but require access to and expertise in the underlying model. The number of substances
contributing to eutrophication is quite limited and hence the need for additional factors is not
foreseen to be an issue.
77
3.9 Ecotoxicity
The pre-selection of characterisation models for the ecotoxicity impact category is
presented in LCIA- Analysis document (EC-JRC, 2010a) and is summarized in the table
below.
Table 20 Selected midpoint methods and underlying models for ecotoxicity.
Midpoint method Underlying model Reference
USEtox Model developed under auspices of Rosenbaum et al. (2008)

UNEP/SETAC Life Cycle Initiative
a
ReCiPe USES-LCA version 2.0 Huijbregts and Van Zelm (2009)
b
IMPACT 2002+ IMPACT2002 Jolliet et al. (2003)
TRACI CalTOX 4.0 Bare et al. (2003)

c
EDIP2003 EDIP1997, combined with site dependent Tørsløv et al. (2005)
factors
Swiss Ecoscarcity Based on a combination of actual emissions Frischknecht et al. (2008)

and emission limit values
MEEuP Based on emission limit values Kemna et al. (2005)
Endpoint method
EPS2000 Based on empirical information of red list Steen (1999a, b)

species supposed to be threatened by
chemicals and total emission loads
a
ReCiPe USES-LCA version 2.0 Huijbregts and Van Zelm (2009)
b
IMPACT 2002+ IMPACT2002 Jolliet et al. (2003)
a
The most recent version of the model USES-LCA is the underlying model for the calculations of characterisation
factors for ecotoxicity in ReCiPe. Previous versions of the model family USES-LCA and EUSES, employed in
CML2002 and Eco-indicator99, were not included in the evaluation.
b
The European version of the model IMPACT2002 is the underlying model for the calculations of characterisation
factors for ecotoxicity in IMPACT2002+. LUCAS and LIME contain respectively Canadian and Japanese versions
of IMPACT2002 and were not included in the evaluation.
c
The most recent version of the EDIP method is evaluated (2003 version). A previous version, EDIP1997, was
not included in the evaluation.
The methods that were analysed can be divided into three groups according to their fate
modelling: 1. Full multimedia fate modelling (USEtox, ReCiPe, IMPACT2002+, Caltox
(TRACI), 2. Partial fate modelling - Environmental key properties (EDIP) and 3. no fate
modelling (Swiss Ecoscarcity and MEEuP). Methods within Group 1 and 2 model impacts at
the same level in the impact pathway predicting Potentially Affected Fraction of species
(PAF) in some form, while methods in group 3 are not showed in the figure: they do not really
target PAF’s at all as they are not based on fate assessment.
Figure 10 illustrates the environmental mechanism of ecotoxicological impacts and
corresponds to the framework of fate and ecotoxicological effect assessment.
78
emissions
air Vegetation crop
fate
marine water environmental

freshwater groundwater Soil
concentration
exposure
individual trophic ecosystem and effects
species level, e.g. level
Algae
species
individual crustacae multiple species ocurrence
species 1,2,...n & ecosystem
fish severity,
endpoint
modeling
Damage on marine Damage on freshwater Damage on terrestrial PAF, PDF

ecosystems ecosystems ecosystems biodiversity
loss
Damage on ecosystem health
Figure 10 Flow diagram for ecotoxicity

The methods have been rated against the criteria developed in guidance document LCIA -
Framework and requirements document (EC-JRC, 2010b). The results are summarized in
the table below31. Table 21 summarizes the evaluation of the five most prominent midpoint
models for ecotoxicological impact assessment in LCIA, while Table 22 gives the summary of
the evaluation of the two remaining midpoint models and the three endpoint models
considered. Discussions of specific issues in the evaluation of the models against the criteria
are presented in the background documentation32 (Ecotoxicity.xls), where a table with the
detailed evaluation including the whole sets of sub-criteria is given.
31
E: no compliance
32
79
Table 21 Summary table documenting the analysis of five midpoint characterisation methods against the adapted criteria for ecotoxicity.
USEtox (midpoint) ReCiPe (midpoint) IMPACT2002+ (midpoint) TRACI (midpoint) EDIP2003 (midpoint)
The scope of the model for
The scope of the model for the The scope of the model for the The scope of the model for the
the generic evaluation of
The scope of the model for the A generic evaluation of chemicals is A A generic evaluation of chemicals generic evaluation of chemicals is
Completeness of chemicals is fully applicable, A/
A generic evaluation of / fully applicable, except that the / / is fully applicable, except that fully applicable, but not adaptable to
scope except that the model is B
chemicals is fully applicable B model is parameterised for B B the model is parameterised for a spatial and temporal explicit
parameterised for European
European circumstances US circumstances evaluation
circumstances
Environmental relevance is Environmental relevance is high, Environmental relevance is high,
Environmental relevance is Environmental relevance is
B high for freshwater ecotoxicity, although ecotoxicity data are A but marine compartment and fate
Environmental high. Terrestrial ecotoxicity is, high. Ecotoxicity effect factors B/
/ except for the exclusion of B based on acute EC50 and / B processes, such as advective
relevance however, based on aquatic are, however, based on NOEC C
C marine and terrestrial terrestrial ecotoxicity is based on B transport of chemicals, are
data instead of EC50 data
ecotoxicity aquatic data excluded
Chemical input data and model Model components Model components extensively Effect assessment is scientifically
Model components extensively
Scientific components extensively extensively reviewed and reviewed and uncertainty robust, but intermedia transport is
reviewed and uncertainty C/
robustness & B reviewed by a large group of B B uncertainty estimates B estimates available, but not comprehensively included nor
estimates available, but chemical D
Certainty model developers, but no available, but chemical data chemical data not always verification of model results and
data not always reviewed
uncertainty estimates available not always reviewed reviewed uncertainty estimates
The model, documentation and The model, documentation The model, documentation and
Documentation & The model, documentation and The model, documentation and
results are published and the and results are published and results are published and the
Transparency & A A results are published and the A A A results are published and the model
model can be used free of the model can be used free of model can be used free of
Reproducibility model can be used free of charge can be used free of charge
charge charge charge
Database with > 2000
ecotoxicological CF for Database with > 400 Database with > 100
Database with > 2000 Database with > 100
B freshwater ecotoxicity is A ecotoxicological ecotoxicological
ecotoxicological characterisation ecotoxicological characterisation
Applicability / available, can be easily applied A / characterisation factors is B characterisation factors is B
factors is available that can be factors is available that can be
C and updated. Data for marine B available that can be easily available that can be easily
easily applied and updated easily applied and updated
and terrestrial ecotoxicity is applied and updated applied and updated
lacking
The model addresses the The model addresses the
freshwater and terrestrial freshwater and terrestrial
The model addresses the
The model addresses the The model addresses the environments, includes all vital environments, includes the effect
freshwater, marine and
freshwater environment, freshwater, marine and terrestrial fate model elements in a part in a scientifically sound way,
Overall evaluation terrestrial environments,
includes all vital model environments, includes all vital scientifically sound way, except except for metals, and is well
of science based B B B includes all vital model B C
elements in a scientifically model elements in a scientifically for metals, and is well documented. The fate assessment
criteria elements in a scientifically
sound way, except for metals, sound way, except for metals, and documented. The is, however, very simplified and no
sound way, except for metals,
and is sufficiently documented is well documented ecotoxicological effect information is available on the
and is well documented
assessment can be further uncertainties involved in the model
improved by using EC50 data results
Principles of the model are easy to
understand and based on the
Principles of the model are Principles of the model are
Overall evaluation A EUSES-system applied in the EU B Principles of the model are easy Principles of the model are easy to
easy to understand and the easy to understand, but the
of stakeholders / B to evaluate new and existing / B to understand and endorsed by C understand, but the model is not
UNEP encourages its use by model is not endorsed by an
acceptance B chemicals, but the LCA version is C the USEPA endorsed by an authoritative body
businesses and governments. authoritative body
not officially endorsed by an
international authoritative body
80
Table 22 Summary table documenting the analysis of the two remaining midpoint models and three endpoint models for ecotoxicity.
Swiss Ecoscarcity (midpoint) MEEuP (midpoint) EPS2000 (endpoint) ReCiPe (endpoint) IMPACT2002+ (endpoint)
The scope of the model
No ecotoxicological The scope of the model for
No ecotoxicological impact for the generic evaluation
impact mechanisms The scope of the model the generic evaluation of
mechanisms included. A A of chemicals is fully
Completeness of included. Indicators for the generic chemicals is fully applicable,
E Indicators derived from E A / / applicable, except that
scope derived from policy- evaluation of chemicals except that the model is
policy-based emission limit B B the model is
based emission limit is fully applicable parameterised for European
values parameterised for
values circumstances
European circumstances
No specific focus on No specific focus on No modelling of
Environmental relevance for Environmental relevance
ecotoxicological impacts, ecotoxicological chemical-specific
Environmental endpoint assessment of for endpoint assessment
E as actual emissions and E impacts, as emission E ecotoxicological impacts C C
relevance effects on ecosystem of effects on ecosystem
emission limit values are limit values are used as along the environmental
biodiversity is rather low biodiversity is rather low
used as impact indicator impact indicator cause-effect chain
Scientific Not further evaluated due Not further evaluated Not further evaluated Hardly any validation data No validation data
robustness & to lack of environmental due to lack of due to lack of D available for the endpoint E available for the endpoint
Certainty relevance environmental relevance environmental relevance effect factors effect factors
The model,
The model, documentation
Documentation & Not further evaluated due Not further evaluated Not further evaluated documentation and
and results are published
Transparency & to lack of environmental due to lack of due to lack of A A results are published and
and the model can be used
Reproducibility relevance environmental relevance environmental relevance the model can be used
free of charge
free of charge
Database with > 400
Database with > 2000
ecotoxicological
Not further evaluated due Not further evaluated Not further evaluated ecotoxicological A
characterisation factors is
Applicability to lack of environmental due to lack of due to lack of A characterisation factors is /
available that can be
relevance environmental relevance environmental relevance available that can be easily B
easily applied and
applied and updated
updated
No compliance with
No compliance with No compliance with
science-based criteria There is little compliance for There is little compliance
science-based criteria for science-based criteria
for the evaluation of the endpoint effect part of for the endpoint effect
Overall evaluation the evaluation of for the evaluation of
ecotoxicological the method, as the overall part of the method, as
of science based E ecotoxicological impacts. E E ecotoxicological D D
impacts. Political concept of the endpoint the overall concept of the
criteria Political emission targets impacts. Cause-effect
emission targets are effect factors is hardly endpoint effect factors is
are used in the indicator chain of individual
used in the indicator validated hardly validated
development chemicals not included
development
Principles of the model are Principles of the model
Not further evaluated, Not further evaluated,
Not further evaluated, relatively easy to are relatively easy to
Overall evaluation because the thresholds because the thresholds
because the thresholds understand, but the endpoint understand, but the
of stakeholders within the science based within the science based C C
within the science based part of the model is not endpoint part of the
acceptance criteria were not criteria were not
criteria were not reached endorsed by an authoritative model is not endorsed by
reached reached
body. an authoritative body
81

As stated in the LCIA- Framework and requirements document (EC-JRC, 2010b), LCA
characterisation models and factors for ecotoxicity effects must be based on models that
account for a chemical’s fate in the environment, species exposure, and differences in
toxicological response (likelihood of effects and severity).
Therefore, two of the midpoint methods (Swiss Ecoscarcity and MEEuP) were not further
evaluated since the proposed approaches haven’t an ecotoxicity model behind the
calculations. Amongst the five remaining ecotoxicity methods, four of them (USEtox,
IMPACT, ReCiPe, TRACI) show compliance with criteria in all essential aspects for the
science-based criteria, while EDIP has a compliance only in some aspects due to a rather
simplistic fate assessment. For the evaluation of stakeholders’ acceptance criteria, the
USEtox model stands out compared to the other models, as the principles of the model are
easy to understand and the UNEP encourages its use by businesses and governments.
For all the three evaluated endpoint methods (EPS2000, ReCiPe, IMPACT2002+), there
is little or no compliance with the scientific and stakeholder acceptance criteria, as the overall
concept of the endpoint effect factors is hardly validated and the endpoint part of the
methods is not endorsed by an authoritative body.
Posthuma and De Zwart (2006) indicated for responses of fish species assemblages that
the observed loss of species due to mixture toxicity matches the predicted risks based on
EC50-data, at least in a relative sense (slope 1:1), and with a maximum observed fraction of
lost species equal to the EC50-based ecotoxicity predictor variable. Nevertheless, so far the
methods are immature to be used even at interim.
Also for terrestrial and marine ecotoxicity none of the methods is recommended.

differentiation
Uncertainty in the fate factors of organic chemicals is mainly caused by uncertainty in the
degradation rates (Rosenbaum et al., 2008), and for e.g. metals the lack of addressing true
metal bioavailability in the fate calculations (see e.g. Chapman et al., 2003 and Chapman,
2008 for a critical review on this aspect). In the effect factor calculations, uncertainty is
mainly due to the lack of toxicity data for species of various trophic levels (Van Zelm et al.,
2007b).
As discussed in Rosenbaum et al. (2008), the characterisation factors for ecotoxicity must
be used in a way that reflects the large variation of 12 orders of magnitude between chemical
impacts per unit emission as well as the 2 orders of magnitude uncertainty on the individual
characterisation factors. In practice, this means that for the LCA practitioner, these
characterisation factors for ecotoxicity can be useful to identify the 10 or 20 most important
chemicals pertinent for their application. The life-cycle ecotoxicity scores enable thus the
identification of all chemicals contributing more than e.g. one thousandth to the total score. In
most applications, this will allow the practitioner to identify 10 to 20 chemicals to look at in
priority and perhaps more importantly to disregard 400 other substances whose impact is not
82
significant for the considered application. The same types of conclusions are also found in
screening risk assessments (e.g. Harbers et al., 2006).
Research towards the importance of including spatial differentiation in the calculation of
characterisation factors for ecotoxicity has been hardly addressed within LCA, although e.g.
Tørsløv et al. (2005) indicate that excluding spatial variability is probably less influential
compared to the importance of parameter uncertainty, for instance, in degradation rates and
toxicity data, in the calculation of characterisation factors for ecotoxicity.

USEtox is preferred as the recommended default method for the midpoint evaluation of
freshwater ecotoxicity impacts. This is equally consistent with the model recommended for
toxicity impacts for humans. It results from a consensus building effort amongst related
modellers and, hence, the underlying principles reflect common and agreed
recommendations from these experts. The model accounts for all important parameters in
the impact pathway as identified by a systematic model comparison within the consensus
process. The model addresses the freshwater part of the environment problem and includes
the vital model elements in a scientifically up-to-date way. USEtox has also been set up to
model a global default continent.
In USEtox, a distinction is made between interim and recommended characterization
factors, reflecting the level of expected reliability of the calculations in a qualitative way
(Rosenbaum et al. 2008). Ecotoxicological characterisation factors for ‘metals’, ‘dissociating
substances’ and ‘amphiphilics’ (e.g. detergents) are all classified as interim in USEtox. The
providers argue that this is due to the relatively high uncertainty of addressing fate and
effects for all chemicals within these substance groups at this time. For the remaining set of
chemicals, recommended aquatic ecotoxicological characterisation factors are based on
effect data of at least three different species covering at least three different trophic levels (or
taxa)
No available method is recommended to address marine and terrestrial ecotoxicity. It
should be noted that the use of indicators for freshwater ecosystems is not a proxy for
marine and terrestrial ones and, in many cases, only accounts for part of the long-term fate
and ecosystem exposure of emissions. Actually, chemicals that doesn’t remain long in
freshwater and have a high persistence may imply terrestrial or marine effects not yet
addressed by USEtox.

The USEtox midpoint model is recommended for the midpoint calculations for freshwater
ecotoxicity. No model
No method is recommended for the endpoint assessment of ecotoxicity, as no method is
mature enough.
Both at midpoint and at endpoint, no method is recommended for marine and terrestrial
ecotoxicity.
83

At midpoint, USEtox is a satisfactory method for freshwater aquatic ecotoxicity,
recommended for non polar organics but needing minor improvements (Level II). At the
moment, it is not considered to apply very well to metals, dissociating substances and
amphiphilics (e.g. detergents). These substances are classified as Level III.
The endpoint characterisation ecotoxicity models for all chemicals are classified as
immature to be recommended due to the preliminary nature of the results available and the
assumptions made between the midpoint indicator and impacts on ecosystems. Substantial
research still needs to be carried out on this issue before general conversion rules can be
developed to address toxicity effects on biodiversity.

In case a midpoint characterisation factor is missing for an important elementary flow in
the inventory, it can be determined using the model as documented in Rosenbaum et al.
(2008). The latest version of the USEtox model may be downloaded at www.usetox.org to
calculate characterization factors for new substances.
The calculation requires the availability of the needed substance properties among which
particularly the toxicity and biodegradability data can be uncertain and difficult to find. These
are normally the input parameters contributing most to the overall uncertainty of the
characterisation factor.
3.10 Land use

Extensive research has been done on the impact category land use (e.g. Müller-Wenk,
1998a; Köllner, 2001; van der Voet, 2001; Weidema and Lindeijer, 2001). In 2007, a
framework for land use impact assessment in LCA was published by Milà i Canals et al.
(2007a), which not only gives a description of land use impacts, but also suggests possible
indicators at both midpoint and endpoint levels, and includes guidelines on how to address
the reference land system.
The midpoint characterisation factor for land use, in the earliest stage of the cause-effect
chain, is mostly taken as the amount and quality deficit of land occupied or transformed.
Some midpoint methods use indicators like soil structure, soil pH or soil organic carbon. The
endpoint characterisation factor mostly refers to the amount of species lost due to land use
or to the change in Net Primary Production (NPP) of the land used.
The pre-selection of characterisation models for the land-use impact category has been
explained in another ILCD document: “Analysis of existing Environmental Impact
Assessment methodologies for use in Life Cycle Assessment” and is summarized in the table
below.
84
Table 23 Selected midpoint methods and underlying models for land use.
ReCiPe Not based on a specific model De Schryver and Goedkoop (2009b)
Milà i Canals Based on Soil Organic Matter (SOM) Mila i Canals et al. (2007b)
Baitz Based on seven quality indicators Baitz (2002) further developed by Bos
and Wittstock (2008)
Endpoint method
EPS2000 Based on species diversity loss and production Järvinen and Miettinen (1987)
of wood
Eco-Indicator 99 Based on species diversity loss Köllner (2000) in Goedkoop and

(EI99) Spriensma (2000)
ReCiPe Based on species diversity loss De Schryver and Goedkoop (2009b)
LIME Based on species diversity loss and production Itsubo et al. (2008b)
of wood loss?
Swiss Ecoscarcity Based on species diversity loss Köllner (2001), Köllner and Scholz
(2008)
Additional recent developments exist but have not yet resulted in available
characterisation factors and, therefore, have not been further evaluated. These include:
 Michelsen (2007): The main value seems to lie in assessing biodiversity from an
ecosystem rather than species diversity point of view. The study is focused on
forestry, but a similar approach could be developed for other land–use types
 A new project group under the UNEP/SETAC Life Cycle Initiative on land use.33
The methods considered at the midpoint level are:

 ReCiPe
This method is a follow-up of the CML2002 method by Guiné et al. (2002). The
surface area occupied or transformed is taken into account, without any further
characterisation. In that sense, ReCiPe is not a characterisation model but rather a
selection of LCI parameters, like the method of Baitz (2002).
 Milà i Canals (2007b)
This method considers Soil Organic Matter (SOM) as a soil quality indicator. SOM is
qualified as a keystone soil quality indicator, especially for assessing the impacts on
fertile land use (agriculture and forestry systems). It influences properties like buffer
capacity, soil structure and fertility. However, it must be noted that in LCIA it should be
combined with biodiversity indicators. In highly acidified or waterlogged soils the SOM
may not correlate directly with soil quality. The LCA practitioner is expected to know
See http://fr1.estis.net/builder/includes/page.asp?site=lcinit&page_id=337831BE-0C0A-4DC9-
33
AEE5-9DECD1F082D8
85
the location, the timeframe, and the SOM values before and after the land occupation,
the SOM value of the reference land system, the relaxation rate, and associated SOM
values. Based on this, the LCA practitioner is expected to calculate the
characterisation factors for the foreground system. Characterisation factors for certain
land use flows in the background system are provided in Milà i Canals et al. (2007c).
 Baitz (2002)
The method proposed by Baitz (2002) and further developed by Bos and Wittstock
(2008) is based on an inventory of seven indicators that can be used to describe the
impacts related to land occupation and transformation. For each indicator, a
description and a classification is given for its dependence on a set of fundamental
quality parameters, such as the main types of soil, the slope of the landscape, the
carbon content and the maturity of the landscape. The LCA practitioner is expected to
investigate which conditions apply for a certain area (assuming this is known) and
assess in which class the landscape falls under. When no site-specific information is
available, data are taken from a background database as country-specific averages.
The following indicators are to be used:
(1) Erosion stability,
(2) Filter, buffer and transformation function for water,
(3) Groundwater availability and protection (against leaching into the groundwater -
partially dependent on water permeability)
(4) Net Primary Production (NPP),
(5) Water permeability and absorption capacity,
(6) Emission filtering absorption and protection, and
(7) Ecosystem stability and biodiversity34.
Until now, the different indicators cannot be combined or weighted at the midpoint
level. All indicators are calculated as elementary flows that in a next step should be
used as indicators to characterize impact categories which are yet to be defined. The
method is relatively unknown, partly because most information is available in German
and just recently released in English.
At the endpoint level, the following methods have been evaluated:
34
To illustrate how the method works, the indicator for the ecosystem stability and biodiversity (one of the seven
indicators) is described. The practitioner is expected to determine a value for the following parameters:
1. Maturity (MG)
2. Naturalness (NK)
3. Species richness (AR)
4. Diversity of land structures (SV)
6. Level of anthropogenic interference (AB)
To help users, there are default values, and often these default values depend on the country in which the land is
used or other relatively easily-identifiable factors. Once values have been chosen, the resulting factor is
calculated with the formula MG + NK + (AR+SV)/2 + AB. The result is an Ecosystem stability and biodiversity
factor with a value between 1.5 and 22.
86
 EPS2000
This method considers the use of land and its effects on the production of wood. For
land use, only regional effects are considered. The characterisation factors for land
use are expressed in Normalized EXtinction of species (NEX), while the factors for
wood productivity are expressed in kg of dry wood. Both units are added together
using Willingness to Pay as the basis for conversion.
 Eco-indicator 99
This method considers land transformation and occupation in Central Europe. Both
local and regional effects are taken into account. Unlike other methods discussed
here, possible double counting with other impact categories is avoided by adapting
other impact categories. The reason for doing this is that these land-use models are
based on “observed” effects and not on modelled effects, as is the case for
eutrophication and ecotoxicity. It is therefore difficult to link the disappearance of
species to either direct land-use impacts, or to the impacts from the use of herbicides /
manure. For example, in the eco-indicator 99, direct effects from manure on land are
considered to be taken into account by land use and excluded from the impact
category eutrophication.
The characterisation factors are expressed in potentially disappeared fraction of
species: PDF*occupation time for occupation, and PDF*restoration time for
transformation.
 ReCiPe
This method considers land transformation and occupation in Northwest Europe. Both
local and regional effects are taken into account. Three levels of land-use intensity are
considered. The characterisation factors are expressed in potentially disappeared
fraction of species (PDF) for occupation, and PDF*restoration time for transformation.
The underlying mathematical calculations are based on the work of Köllner (2001),
although some different assumptions are applied. The model is not reviewed and does
not include uncertainty data.
 LIME
This method considers the effects of land use in Japan, based on biodiversity changes
and effects on primary production. Primary production effects are calculated for Japan
according to the adopted land-use classification system and applying the Chikugo
Model (Uchijima and Seino 1985). The biodiversity loss is based on extinction
probability of vascular plants on the red-species list of Japan. The model considers the
life expectancy of the target species by calculating the amount of species throughout
Japan.
 Swiss Ecoscarcity
This method is based on the work of Köllner (2001) to analyse the effects of land use.
It considers land occupation, based on plant species loss in the Swiss plateau. Both
local and regional effects are taken into account. The original characterisation factors
are expressed in ecosystem damage potentials (EDP), which are based on a
nonlinear effect-damage function. The publication of Köllner in 2008 is based on his
87
PhD thesis produced in 2001 and on the underlying mathematical framework. In both
publications several types of characterisation factors are produced. However, the
factors used by Swiss Ecoscarcity are the total damages (local + regional), while the
publication in 2008 only presents the local damage factors. As the work of 2001 is
used today by the Swiss Ecoscarcity method, this is used here.
Figure 11 Flow diagram of the cause-effect chain of land use (adapted from Weidema and Lindeijer
(2001))35.

The methods have been rated against the criteria developed in the LCIA - Framework and
Requirements document (EC-JRC, 2010b). Table 24 summarizes the results of the
evaluation of the three most prominent midpoint models for land-use impact assessment in
LCIA while Table 25 summarizes the evaluation of the five analysed endpoint models.
Discussions of specific issues in the evaluation of the models against the criteria36 are
presented in the background documentation37 (Land use.xls), where a table with the detailed
evaluation, including the whole set of sub-criteria, is given.
In the figure: NPP=Net Primary Production; SOM= Soil Organic Matter; * Land occupation does not entail land
35
transformation but is responsible for maintaining an altered state; ** Amount of area transformed or occupied.
36
E: no compliance
37
88
Table 24 Summary of the evaluation results of three midpoint models that assess land use in an LCA context.
Baitz (2002), further developed by

ReCiPe Midpoint Milà i Canals et al. (2007b)
Bos and Wittstock (2008)
Limited impact indicator, based on soil organic matter
Seven different indicators describing soil
(SOM). Case-specific characterisation factors (CF)
Completeness of quality. Most data must be collected by the
E No impact mechanism included. B C should be developed by the user. Site-specific data are
scope practitioner. When no site-specific data are
needed. CFs available for elementary flows based on
available, country-average data are used.
the land-use classification system CORINE+.
Characterisation model does not distinguish Characterisation model includes seven
Environmental The charactersation model includes one indicator. Only
E different species composition between land use C different land use effects. Only local effects C
relevance local effects are considered..
types. are considered.
The main scientific value is in the
Scientific robustness Not further evaluated due to lack of comprehensive selection of indicators, and Characterisation model is reviewed and is recent
0 B A
& Certainty environmental characterisation model. the pragmatic guidance to users for (2007).
calculating characterisation factors.
A general bakground database is available.
Documentation,Transp
Documentation is available in both German The model documentation and characterisation factors
arency & 0 Not further evaluated. D A
and English, although the latter is not publicly are published and available free of charge.
Reproducibility
available yet.
Characterisation factors should be Default factors are available for background processes.
determined by the user. The method Case-specific characterisation factors should be
Applicability 0 Not further evaluated. E E
provides guidance. Already implemented and produced by the practitioner. Considerable information
tested in some databases. is needed.
Seven quality indicators describing different
Only one indicator describing soil quality. Case-specific
soil-quality aspects; there is no way to
Overall evaluation of No compliance with science-based criteria for characterisation factors should be produced by the
E D aggregate these at midpoint level. C
science-based criteria the evaluation of land use impacts. practitioner. Model is reviewed and good for agro- and
Characterisation factors are not available;
forestry-systems.
normalisation is not available.
Principles of the model are relatively easy to
Overall evaluation of Complex method that produces different
No compliance with science-based criteria for understand, but not endorsed by an authoritative body.
stakeholders’ E D indicators. Not endorsed by an authoritative C
the evaluation of land use impacts. Exclusion of biodiversity is a limitation for several
acceptance body.
relevant stakeholders.
89
Table 25 Summary of the evaluation results of five endpoint models that assess land use in an LCA context.
ReCiPe Eco-Indicator 99 EPS2000 LIME Swiss Ecoscarcitiy

Indicator based on biodiversity
Valid for mid-Europe.
Valid for Northwest Europe. (red list species) and wood
Indicator based on Valid for Japan. Indicator Valid for mid-Europe. Indicator
Completeness of Indicator based on productivity. Biodiversity only
C C biodiversity. Double-counting D D based on biodiversity and C based on biodiversity and adopts
scope biodiversity. Possible double- based on Swedish data.
with pesticides and fertilisers NPP. the CORINE classification.
counting not considered. Possible double-counting not
considered.
considered.
Characterisation model
Characterisation model
reflects loss of species based No characterisation model used. Characterisation models Characterisation model reflects
Environmental reflects loss of species based
on species-area relationship. Inclusion of biodiversity and include effects on primary biodiversity loss. Transformation
relevance: Overall C D on species-area relationship. D C D
Considers land use primary production effects production and biodiversity not available in Ecoscarcity
evaluation Exclusion of effects on
intensiveness. Exclusion of based on empirical data. loss. implementation.
primary production.
effects on primary production.
Indicators cannot be
Scientific robustness Only input data reviewed. No Only input data reviewed. Only input data reviewed.
confirmed due to lack of Characterisation model is reviewed.
& Certainty: Overall C uncertainty figures available. C Uncertainty figures available. C Uncertainty figures available. E B
documentation. No model Uncertainty figures available.
evaluation: Most recent data used. Relatively old data employed. Relatively old data employed.
uncertainties considered.
Documentation,Transp
The model documentation and The model documentation The model documentation and The model documentation and
arency & English documentation does
A results are published and A and results are published and A results are published and E A results are published and available
Reproducibility: not exist.
available free of charge. available free of charge available free of charge. free of charge.
Overall evaluation
Characterisation factors are Characterisation factors are Characterisation factors are Characterisation factors are
Applicability: Overall Characterisation factors are
B available, can be easily B available and can be easily B available and can be easily D B available and can be easily applied
evaluation not available in English .
applied and updated applied and updated. applied and updated. and updated.
Based on recent data and
Based on most recent data Based on old data, does not
No characterisation model used, The characterisation model knowledge, considers several land-
and knowledge, considers consider land-use intensity,
Overall evaluation of considers NPP and biodiversity produced only applies to use types (only for occupation). It
C land-use intensiveness, but D but takes into account effects D D C
science based criteria effects. Based on old data, Japan. Lacks English does not take into account double-
does not take into account of double-counting and
uncertainty data included. documentation. counting effects. The model is
double-counting effects uncertainty data.
reviewed.
The principles of the model The principles of the model The principles of the model
The principles of the model are The principles of the model are
Overall evaluation of are relatively easy to are relatively easy to are relatively easy to
relatively easy to understand, relatively easy to understand, but
stakeholders C understand, but the model is C understand, but the model is C C understand, but the model is C
but the model is not endorsed the model is not endorsed by an
acceptance not endorsed by an not endorsed by an not endorsed by an
by an authoritative body. authoritative body.
authoritative body. authoritative body. authoritative body.
90

In general, land-use methods do not score highly against most criteria, as developments
are still ongoing. The ReCiPe Midpoint method is only a simple addition of the land area
used (in square metres), without any characterisation (and, hence, differentiation) of the
different land-use types and associated environmental impact. Mila I Canals et al. (2007b) go
one step further and include a characterisation model based on changes in soil organic
matter (SOM) . Baitz (2002) midpoint method requires rather extensive set of LCI
parameters, despite the availability of default values, and excludes a characterisation model.
All endpoint models use “observed” damages to biodiversity, depending on the way the
land is managed or used, while LIME and EPS2000 also use “observed” productivity
indicators. The use of “observed” data is an important difference from other impact
categories where a clear cause-effect mechanism is used. In land-use models, researchers
try to reason back from the observed damages. This means the quality differentiation is
strongly correlated with the quality, the interpretation and the scope of the “observed” data.
3.10.4 Recommended default method at the midpoint level

At the midpoint level, two interesting approaches have been identified. The method of
Baitz (2002), which is further developed (Bos and Wittstock, 2008), has a low score on
applicability, as only parameters that can be used as inventory items are provided and no
characterisation factors given - the user has to determine these. The method produces five to
seven indicators describing soil quality as a whole. However, weighting is necessary in order
to aggregate these at the endpoint level, which is difficult, as there is no normalisation data
available and all units differ. The method by Milà i Canals et al. (2007b) produce only one
indicator describing soil quality as a whole. However, it does not cover biodiversity impacts.
In this method, the user is expected to determine the characterisation factors of relevance for
the foreground system. Milà i Canals et al. (2007c) give characterisation factors for some
land use flows in the background system.
The midpoint method implemented in ReCiPe (De Schryver and Goedkoop, 2009b) simply
adds up all land occupation and transformation. It is simple and robust, but misses
environmental relevance.
Based on the above information, the method by Milà i Canals (2007b) is chosen as the
most appropriate among the existing approaches, even though its scope is currently limited.
3.10.5 Recommended default method at the endpoint level

At the endpoint level, all methods evaluated are too immature to be recommended.
However, the ReCiPe method may be used as an interim solutionThe ReCiPe method
considers land occupation and transformation, but only for 12 different land-use types,
specifically chosen to be the most stable according to the model used and most relevant for
LCA. The model distinguishes three types of arable land-use intensity. It is based on the
most recent British data and inventory data by Köllner (2001) as additional information.
The Swiss Ecoscarcity model, also based on the work of Köllner (2001), is not
recommended as it does not consider land-transformation impacts. Nevertheless, the more
recent work of Köllner (2008) contains elements and data which can be used for further
91
research. All land-use methods available at the moment work with data that represent only a
limited region of Europe with specific vegetation types. As a result, these cannot be easily
transferred to other ecosystems and continents.

As the recommended method at the midpoint level and the interim method at the endpoint
level operate with different environmental impact pathways, there is poor consistency
between them. This is identified as a research need for this impact category (see Annex 2).

At midpoint, the method by Milà i Canals et al. 2007 is classified as recommended, but to
be applied with caution (Level III).
At endpoint level, no method is recommended to be used. However, if an endpoint
method is required, the ReCiPe method can be used as an interim, as it is not mature
enough for recommendation.

The recommended midpoint method has a number of default characterisation factors for
several land-use elementary flows that are based on the land classification system CORINE
+. Additional characterisation factors may be calculated provided data on SOC exists for
further land use types in other regionsThe interim endpoint method provides characterisation
factors for a range of land-use types and conversions. The user can choose different time
horizons for land transformations. A separate section in the report is devoted to the linkage
between the LCI parameters and the characterisation factors.
3.11 Resource depletion

Several authors address the effects of resource use and propose ways to integrate
resource depletion into the LCA framework, e.g. Müller-Wenk (1998b), Meadows et al.
(2004), Steen (2006), and Stewart and Weidema (2005).
The pre-selection of characterisation models for the resource-depletion impact category
has been explained in another ILCD document (“Analysis of existing Environmental Impact
Assessment methodologies for use in Life Cycle Assessment”) and is summarized in Table
26. Not all models initially proposed in that document cover exactly the same impacts arising
from the use of resources. As a result, it is difficult to analyse all different models on resource
depletion as one single group.
Following the impact pathway, resource depletion impacts are suggested to be divided
into four categories reflecting the lack of consensus on what the issue is for this impact
92
category (see also discussion on the AoP Natural Resources in another ILCD document:
“Framework and requirements”,EC-JRC, 2010b).
Category 1 methods are at the first step of the impact pathway. They use an inherent
property of the material as a basis for the characterisation. The environmental relevance is
low in terms of expressing resource depletion, but the characterisation factors are relatively
robust and certain. As described in the AoP Natural Resources, those methods that do not
include the concept of resource scarcity are not considered. Therefore, these methods were
considered incompatible with the AoP Natural Resources (irrespective of the quality of the
method).
Category 2 methods address the scarcity of the resource. They have a higher
environmental relevance, and potentially also a higher uncertainty.
Category 3 methods focus on water and are treated as a separate category due to the
regional dependence of this resource issue, which the characterisation model needs to
consider.
Category 4 describes the endpoint methods. These aim to cover the entire environmental
mechanism.
The different models are grouped in the analysis on the basis of the resources they take
into account and on the level in which they are located in the cause-effect chain of resource
depletion.
Figure 12 gives an overview of the classification of the different methods analysed in this
section, according to the impacts they cover and their position in the cause-effect chain. A
recommendation will be considered for each of the four categories analysed.
Resource use
(extracted amount) MEEUP (water)
Exergy
Category 1 Common resource
characteristic
Eco-scarcity
(energy and gravel
Category 2 Decreased availability

CML/EDIP
Eco-scarcity (water)
Regeneration (natural Recovery (urban and

growth) waste mining)
Damage to availability for

future extraction Eco-indicator-99
ReCiPe (endpoint)
Category 4 (endpoint) Future availability and
effort needed
Impact 2002
(minerals)
Future provision of needs

EPS
1
Figure 12 Overview of methods classification for resource depletion
Table 26 Selected methods and underlying models for resource depletion. (see description of each
method below)
93
Swiss Ecoscarcity Frischknecht et al. (2008)

(energy and gravel)
Exergy CEENE: Cumulative exergy extraction from Dewulf et al. (2007)

the natural environment
CML2002 Guinée and Heijungs, 1995 Guinée et al. (2002)
EDIP1997 (2004 EDIP 1997 (Nedermark) Hauschild (1998, updated in 2004)

update)
MEEUP (water) Kemna et al. (2005)
Swiss Ecoscarcity Frischknecht et al. (2006b)

(water)
Endpoint method
Eco-indicator 99 Müller-Wenk (1998b) Müller-Wenk (1998b) and Goedkoop

(EI99) (1999)
EPS2000 Steen (1999)
IMPACT 2002+ Fossil fuels: IMPACT2002+; Minerals as in Jolliet (2003)

EI99
ReCiPe Goedkoop and De Schryver (2008), De

Schryver and Goedkoop (2009c)
3.11.2 Pre-selection of midpoint methods

The following models were pre-selected.
 Exergy
This method is based on Dewulf et al. (2007). Exergy38 values have been determined
for a list of resources covering fossil fuels, minerals, nuclear energy, land resources,
renewable resources (e.g. wind power and hydropower), atmospheric resources and
water resources. The method addresses several shortcomings of earlier exergy
methods, like double counting in bio-based fuels and confusing exergy loss in ores
with exergy loss in the minerals that actually contain the metals being exploited.
 Swiss Ecoscarcity 2007 (energy)
The Ecoscarcity method covers a number of resource depletion categories. We
selected the “energy” resource impact categories, to remain within the scope of this
assessment. Fossil depletion is characterised by using the net calorific value of the
fuels as the basis of the characterisation. This is a common practice in many other
methods (e.g. IMPACT2002+ at midpoint, and ReCiPe at midpoint). The renewable
energy is characterised by the amount produced. For example, for solar input, it is not
the use of solar energy that falls on the PV panels, but the actual electric power that
can be effectively used. Fossil and renewable energy can be combined using the
38
Exergy of a resource expresses the maximum amount of useful work the resource can provide. Energy is never
destroyed (first law of thermodynamics), but the energy contained in, for example, lukewarm water can provide
much less work than the same amount of energy in overheated steam.
94
distance-to-target approach, where the difference between the actual use and the
desired target use, set by the Swiss government form the basis of the weights
adopted. For non-renewable resources, the energy content (net calorific value, in MJ)
is multiplied by a factor of 3.3, while the renewable energy resources are multiplied
with a factor 1.1. As a consequence, this impact category concerns not only one but
two midpoints. For non-renewable resources, MJ (higher heating value) per kg is used
as a characterisation factor. For renewable resources, a correction factor is sometimes
used for the ratio between primary energy input and produced energy. Wood is only
considered to be renewable if there is an appropriate forest-management regime.
 CML 2002
This method includes non-renewable resources (fossil fuels and minerals). In Guinée
et al. (2002) only the ultimate stock reserves are included, which refers to the quantity
of resources that is ultimately available, estimated by multiplying the average natural
concentration of the resources in the earth’s crust by the mass of the crust (Guinée,
1995). In Oers et al. (2002), additional characterisation factors have been listed on the
basis of USGS economic reserve and reserve base figures in addition to the ultimate
reserve. The characterisation factors are named ‘abiotic depletion potentials’ (ADP)
and expressed in kg of antimony equivalent, which is the adopted reference element.
The abiotic depletion potential is calculated for elements and, in the case of economic
reserves and reserve base, several mineral compounds.
 EDIP 1997
This method was updated in 2004 and includes non-renewable resources (fossil fuels
and minerals). The amount of the resource extracted is divided by the 2004 global
production of the resource and weighted according to the quantity of the resources in
economically-exploitable reserves. Effectively, this means that the global annual
production drops out, so that the characterisation model is based on the economic
reserves only. The characterisation factors are expressed in person-reserve, meaning
the quantity of the resource available to an average world citizen.
 MEEUP
The part of this method that concerns only water as a resource has been selected. It
includes the use of both process and cooling water. The characterisation factor
expresses the amount of water used (litres). MEEUP also addresses other resource
categories, but these are directly taken from other methods, e.g. CML 2002.
 Swiss Ecoscarcity (water)
This method concerns only water. Its characterisation factors distinguish six levels of
water scarcity in a given region. As such, it is the first method that differentiates the
regional severity of water availability.
3.11.3 Pre-selection of endpoint methods

The following endpoint methods were pre-selected:
 Eco-indicator 99
95
This method includes non-renewable resources (fossil fuels and minerals). The
characterisation factor is expressed as Surplus Energy. This expresses the additional
energy requirements due to mining resources with a decreased grade at some point in
the future. This point is arbitrarily chosen as the time mankind has mined 5 times the
historical extraction up to 1990. Current technology is assumed. The method
calculates the depletion of elements, not ores.
 EPS2000
This method includes non-renewable resources (fossil fuels and minerals) and
renewable resources (water, fish, meat and wood). The amount of resource depleted
is directly normalized and weighted using monetization. Characterisation factors are
expressed in Willingness to Pay, indicating the costs of extracting and purifying the
element. The characterisation is based on future technologies and abundance of
metal ores in the Earth’s crust. It is based on depletion of element concentrations, and
expressed in amount of element in ore concentrations.
 IMPACT 2002+
This method includes non-renewable resources (fossil fuels and minerals). The
mineral depletion is modelled as in Eco-Indicator 99. The characterisation factor of
fossil fuels is expressed as total primary energy, including feedstock energy for energy
carriers (higher heating value). The surplus energy and the actual fossil fuel energy
contents are added using a weighting factor of 1; there is no clear justification.
 ReCiPe
This method includes non-renewable resources (fossil fuels and minerals). For
minerals, the marginal increase of costs due to the extraction of an amount of ore is
the basis of the model. Furthermore, mineral depletion is based on depletion of ores,
instead of elements. This is an advantage because most minerals come from different
ores, and each ore usually produces several minerals. Some minerals are almost
exclusively co-products and with the ReCiPe method these can now be modelled in
greater detail. For fossil fuels, the marginal increase of oil production costs (due to the
need to mine non-conventional oils) is used. Characterisation factors are expressed as
Surplus Costs. These are the costs incurred due to the fact that, after the extraction of
the “best” (highest grade) resources, future mining becomes more expensive. In this
cost calculation, a depreciation rate of 3% is chosen. Current technology is assumed
to determine the costs.

Tables 27 and 28 summarize the evaluation39 of the six most prominent midpoint models
for resource depletion in LCIA, and give the summary of the evaluation of the four endpoint
models considered. The detailed scores are reported in a separate excel file 40(Resource
depletion.xls).
39
E: no compliance
40
96
Table 27 Summary of the analysis of six midpoint characterisation methods against the adapted criteria for resources.
Exergy Swiss Ecoscarcity CML2002 EDIP2003 MEEuP Swiss Ecoscarcity water
energy
Category Category 1 Category 1 Category 2 Category 2 Category 3 Category 3
The model is
The model is very The model includes adding
relatively complete for The model is relatively
complete. It covers up water amounts, but
energy resources, The model is relatively complete for mineral and The model is relatively
Completeness minerals, fossil fuels and does not differentiate
A C with an interesting but C complete for mineral and C fossil fuel-depletion. An E C complete for water depletion,
of scope flow resources (including, according to regional
Swiss specific fossil-fuel depletion. attempt for water use and in a regionally-specified way.
solar, wind, hydropower differences in water
correction factor for wood extraction is made.
and water). scarcity.
renewability.
Based on 1990 extraction
rates and economically-
Very complete The renewability Characterisation factors for exploitable reserves. Does
implementation of the factor is a new economic reserves, reserve not capture importance of a Simplistic environmental The model assesses water
Environmental
C exergy concept. However, C concept, but needs B base , and ultimate reserves C resource well, since D model for assessing the B depletion on a regional basis.
relevance
this method does not reflect elaboration to become are available. Antimony is the extraction rates are not impact of water. Recovery rates are included.
scarcity. useful. reference resource adopted. included. Water impact is not
applicable, only one CF for all
types of wood.
The paper is not reviewed
The paper is reviewed by
yet, proposed by the UNEP-
The paper is reviewed by The paper is reviewed by external experts. High
Scientific There is only a very SETAC Life Cycle Initiative
external experts. external experts. Uncertainties uncertainties arise in the There is no scientific
robustness & B E rudimentary scientific B C E C but suggested in SETAC
Uncertainties are described are described but not economically-based reserves model.
Certainty model. UNEP results. Uncertainties
but not quantified. quantified. calculations, but these are not
are discussed but not
quantified.
quantified.
The model
Documentation, Documentation is available The model documentation The model documentation
The model and results are documentation and The documentation is
Transparency & A B A online. The website has A and results are easy A B and results are so far only
very well documented. results are so far only easily available.
Reproducibility descriptions and factors. available. available in German.
available in German.
Characterisation
Characterisation factors are Characterisation factors are Characterisation factors are Characterisation factors Characterisation factors are
factors are available
Applicability A available and can be easily A A available and can be easily A available and can be easily A are available and can be B available and can be applied
and can be easily
applied. applied. applied. easily applied. when country is specified.
applied.
The model is very Robust method for mineral Robust method for non-
complete. However, there Mixture of science resources. characterisation renewable resource Too simplistic for
Science-based Promising approach for water
B are different views on C and Distance-to- B factors for available for B depletion, which is based on D consideration as a science B
criteria use.
whether exergy is a Target. economic reserve, reserve economically-exploitable based method.
relevant indicator. base, and ultimate reserves. reserves.
The principles of the method The principles of the method The principles of the method
It is not clear whether
This method is mainly are relatively easy to are relatively easy to Simple method, not are relatively easy to
Stakeholders policy-makers are
C D interesting for Swiss B understand, but the model is B understand, but the model is E endorsed by an B understand, but the model is
acceptance interested in using exergy
policymaking. not endorsed by an not endorsed by an authoritative body. not endorsed by an
as a resource indicator.
authoritative body. authoritative body. authoritative body.
97
Table 28 Summary of the analysis of four endpoint characterisation methods against the adapted criteria for resources.
EPS2000 ReCiPe Eco-Indicator 99 (EI99) IMPACT2002+
Category Category 4 Category 4 Category 4 Category 4
The model is relatively complete
The model is relatively complete for
Completeness of The model includes minerals, energy for minerals and fossil. More The model is relatively complete for minerals
A B minerals and fossil. Additional B B
scope resources, wood and fish extraction. substance flows can always be and fossil
substance flows can be added.
added.
Models potential situation in distance The model focuses on deposit depletion This model adopts surplus energy
Environmental This model adopts surplus energy for future
B future, when average rock is used as an C and, from this, mineral depletion. It has D for future extraction efforts as an D
relevance extraction efforts as an indicator.
ultimate resource. a short time-horizon. indicator.
Relatively novel approach that develops
Assuming the very long time theory on a basis of data from 500 For a medium-term perspective,
Scientific
perspective is chosen, the method is mines, and takes into account the the method is reasonably Similar to eco-indicator 99, but without
robustness & C B C D
relatively consistent, but uncertainties important co-products from deposits. consistent although quite considering different scenarios.
Certainty
are high. Uncertainties due to economic-based dependent on one reference.
weighting exist.
Documentation,
The model documentation and results The model documentation and results The model documentation and The model documentation and results are
Transparency & A A A A
are easily available. are easily available results are easily available. easily available.
Reproducibility
Characterisation factors are
Characterisation factors are available Characterisation factors are available Characterisation factors are available and can
Applicability B B B available and can be easily B
and can be easily applied. and can be easily applied. be easily applied.
applied.
Relatively simple model, based on
Relatively complete scientific model Relatively simple model, based on estimated
Science based Method based on very long time estimated slope factors.
C B described in all details, based on large C C slope factors. Combination with fossil fuels
criteria scenarios, with many assumptions Combination with fossil fuels
dataset of mining data somewhat problematic
somewhat problematic
The principles of the method are The principles of the method are
The principles of the method are
Stakeholders complex. The model is recent and thus understandable, but not well Relatively simple method, but not well accepted
C understandable but not well accepted or C C C
acceptance not accepted yet. Not endorsed by an accepted or endorsed by an or endorsed by an authoritative body.
endorsed by an authoritative body.
authoritative body. authoritative body.
98

As discussed above, there is some unclarity as to what exactly the Impact Category
Resource Depletion should reflect (see also the discussion on the Area of Protection in the
ILCD: “Framework and requirements document” (EC-JRC, 2010b). This has highlighted the
need to distinguish four categories of methods:
1. methods based on a property of a resource, irrespective of the level of depletion. It is
proposed that, regardless of the quality of such methods, they should not be
considered to express “resource depletion” because these do not express scarcity
(which is the intended aspect of resource use that has to be captured) ;
2. methods based on the “use to availability ratio” or “use to availability - current rate of
extraction ratio”. These are midpoint methods that express the scarcity of a resource;
3. methods for water. These are treated separately, as scarcity of water is very
dependent on location; there is no global market for water, and thus no global scarcity
as that e.g. for oil; and
4. endpoint methods. These attempt to quantify (some of) the consequences of the
depletion of resources for society.
The lack of a clear definition of the AoP Natural Resources results in a similar vagueness in
the scope of this impact category. Consequently, different methods present different ranges
of scope. For example, some methods take into account biotic resources, especially the so-
called “fund” resources (e.g. wood and fish), whereas others consider biotic resources
covered by the land-use impact category.
3.11.6 Recommended default method for category 1

Exergy, as a midpoint indicator, is considered to be the most mature method in category
1. The exergy method is based on an inherent property of a resource. For bio-based
products, the exergy in the solar radiation absorbed is used.
This impact pathway does not describe the depletion processes, but the consumption of
exergy. This value does not depend on the scarcity of the resource (i.e. even if the last tonne
of the resource is depleted, the exergy value remains the same). The method does not
address the question of whether or not exergy losses from solar energy are as important as
exergy losses in the stock of minerals in the earth. These different types of exergy
consumption are added without further weighting. If this assumption is accepted, and if there
is agreement on the fact that exergy losses represent the midpoint of interest, the exergy
method is well developed, easy to apply and overcomes several limitations of alternative
methods.
The most important question is whether any method in this category has sufficient
environmental relevance to be recommended, as the factor does not take into account the
future scarcity of a resource, while it somehow considers the aspect of dispersion which is
also an indicator of availability. As justified under the description of the AoP Natural
Resources (see ILCD Framework and requirements document), the method to be
recommended must have an element that reflects the scarcity of the resource. Given that
99
category 1 methods do not fulfil this criterion, no method within this category is
recommended.

The CML method uses the Abiotic Depletion Potential (ADP), given in kg of antimony
equivalents, to be multiplied with the amount of a given resource extracted. For ADP, the
annual production of the resource (the extraction rate) is divided by the reserves squared,
and the result divided by the same ratio for the reference resource, antimony. The value for
reserves is squared to take into account the fact that a simple ratio of annual production over
reserve may, in the case of higher production rates corresponding to larger reserves and vice
versa, fail to reflect the impact that e.g. 1 kg of resource extraction has on overall scarcity. By
including the annual production rate, CML also captures the current importance of a given
resource.
The CML method is recommended in the ILCD framework since it captures scarcity by
including extraction as well as reserves of a given resource. Characterization factors are
given for metals, fossil fuels and, in the case of reserve base and economic reserves,
mineral compounds (van Oers et al. 2002). In addition, the method covers most of the
substances/materials identified as critical by the European Commission’s Ad-hoc Working
Group on defining critical raw materials (European Commission 2010).
Data on reserves and production are taken from the US Geological Survey
(http://minerals.usgs.gov/minerals/pubs/mcs/).
Oers et al. (2002) give characterization factors for economic reserves, reserve base, and
ultimate reserves. The characterization factors given for the reserve base are recommended,
as this reflects a longer time horizon and the possibility of improvement in mining technology,
making feasible the exploitation of previously sub-economic deposits. The reserve base
includes deposits which meet certain minimal chemical and physical requirements to
potentially become economically exploitable within planning horizons (Oers et al. 2002).
3.11.8 Recommended default method for category 3 for water

The Swiss Ecoscarcity (water) method is preferred for a midpoint evaluation which
considers the impact of water depletion. The method has a very rudimentary environmental
model because it relates water use to local scarcity of water. Thi enables differentiation
between situations where water extraction causes different levels of impact.

At the endpoint level, all methods evaluated are too immature to be recommended.
However, the ReCiPe method may be used as an interim solution.
This model’s indicator is based on the marginal increase of extraction costs, due to the
extraction of a certain amount of a resource. This indicator allows a combination of fossil and
mineral depletion. The environmental mechanism is built upon a large dataset covering 500
mines, and it primarily determines the effect on deposits. A deposit usually contains many
100
different minerals, and many minerals are almost exclusively mined as co-products of other
metals. Consequently, the model does more justice to the real world than methods that only
address the depletion of single minerals. The extraction of single metals is calculated using a
system of price allocation. This is one of the uncertain factors as metal prices fluctuate
significantly.
For fossil resources, the gradual change from conventional (liquid) oil to unconventional
oils is taken as the factor that drives up extraction costs. The damage pathway is only
developed for oil. Other fossil fuels are added by using the energy content of the fuel as a
basis. The model has only recently become available and, thus, it is too early to evaluate its
acceptance.
The EPS2000 method incurs in high uncertainties due to its focus on a very long time
frame. However, in addition to mineral and fossil depletion, it considers water, wood and fish
depletion. These items should be considered in further research and development.

The recommended midpoint method is not compatible with the interim endpoint method.
This is because midpoint indicators reflect an early stage of the cause-effect chain and, in
this case, the information becomes too aggregated to be used in a further modelling step.
Eco-indicator99 and IMPACT2002+ do not have a midpoint for minerals, but Impact 2002
does have a (category 1) midpoint for fossil fuels. ReCiPe has the same midpoint for fossil
fuels, and a midpoint for minerals that is close (along the environmental mechanism) to the
endpoint. EPS2000 has a category 1 indicator for the midpoints.
For water, no endpoint model has been evaluated.

For category 1, the exergy method is the best in its category, but it is not seen appropriate
for recommendation, as it does not address the scarcity of the resource.
For category 2, the CML 2002 method is classified as recommended method with some
improvement needed (Level II), preferably with a sensitivity analysis on ultimate versus
economic reserves.
For category 3 (water), the Swiss Ecoscarcity method is classified as a Level III,
recommended, but to be applied with caution.
For category 4, the ReCiPe method is classified as an interim, immature for
recommendation, but the most appropriate among the existing approaches.
No method is recommended for renewable resources.

The recommended midpoint method has a very complete coverage of the commonly-used
metals, but additional indicators may need to be calculated for rare earth metals. Sufficient
guidance is given in the methodology documentation. Users have to collect data on annual
production.
101
Adding characterisation factors to the endpoint models for metals and minerals is
considerably difficult. This is important because only the most important metals are covered.
In ReCiPe, factors for Phosphate, Indium, Lithium and other minor metals are missing. The
reason for this is that these metals were not in the original mining dataset used. When a
different dataset is used, it may be difficult to get characterisation factors that are comparable
with the existing set.
3.12 Other impacts

3.12.1 State of the field
Besides the core set of impact categories that are always or often included in LCIA, there
is a number of impact categories that are only occasionally or never addressed. This is so for
various reasons:
 for some impact categories (e.g., noise, accidents), no appropriate inventory data
are available in most case studies or LCI-databases
 for some impact categories (e.g., noise, impacts of Genetically Modified
Organisms (GMO)) and for some specific material (such as nanomaterial) no
characterisation factors are available
 for some impact categories (e.g., erosion, impacts of GMO), there is no consensus
on the characterisation model or even on the main principles for characterisation.
For many of these “other impact categories”, two or three of these reasons apply. As
such, they cannot yet be recommended as a mandatory impact category for the European
Platform on LCA. But on the other hand, some of these impact categories may be extremely
important. For instance, it has been suggested that disturbance by noise is responsible for a
substantial part of the DALY-score for human health (e.g., Hofstetter & Müller-Wenk, 2005),
and that erosion, desiccation and other physical disturbances can play a decisive role in
ecosystem degradation (Cotler & Ortega-Larrocea, 2006). Below is given a short description
of some of these “other impact categories”, followed by some recommendations for future
research.
What exactly counts as one of the “other impact categories” is not immediately clear. In
the context of the present project, an a priori restriction has been made to noise, accidents,
desiccation, erosion, and salination.
3.12.2 Framework
3.12.2.1 Introduction
The common LCA-framework in terms of inventory-characterisation model-impact
assessment appears to be difficult to apply for many of these other impact categories.
Whereas emissions (in kg) of the same pollutant from different unit processes can be
meaningfully aggregated and subsequently fed into a characterisation model, noise (in dB)
from different unit processes cannot simply be added. The same applies to impact from GMO
102
and erosion; here we even don’t know the appropriate metric for reporting inventory aspects
(elementary flows). For a few impact categories, this may be easier. Accidents, at least
conceived in terms of number of casualties, can be calculated for a unit process, aggregated
over the life cycle, and converted into some generic impact number, perhaps DALY. Below,
we will briefly discuss the most important “other impact categories” as to methods proposed
and research prospects.
3.12.2.2 Noise
Noise, or noise nuisance, refers to the environmental impacts of sound. In principle, these
impacts could cover at least human health and ecosystem health, but the environmental
mechanisms are complex, non-linear and highly dependent upon local circumstances.
Moreover, noise is similar to odour in that a given level of exposure is experienced differently
by different individuals. Hence, whether or not sound waves will lead to ’nuisance’ depends
partly on the actual situation and partly on the person interviewed. On the other hand, even
when noise is not experienced as nuisance, or when it does not lead to hearing loss, it may
still impair human health, e.g. by inducing cardiovascular diseases (Babisch, 2006)
Most LCIA methodologies do not have an impact category ‘noise’. This runs counter to the
observed fact that most people deem noise to be a major environmental problem, but is
probably due to the unavailability of an appropriate and practically feasible impact
assessment method for noise.
Two lines of approach, which have recently been elaborated, are by Müller-Wenk (2004)
and Meijer et al. (2006).
The paper by Müller-Wenk describes a method for a quantitative assessment in LCA of
noise impacts on human health originating from road vehicle noise. An adaptation to rail
noise is planned. The method starts out from the following data: transport distance in km,
quantity transported, category of vehicle, time (day/night) and country of transport. The
magnitude of health impairment due to noise is determined separately for each vehicle class
(cars, trucks, etc) and is calculated per vehicle-kilometre driven during the day or at night
time on the Swiss road network. This health impairment is expressed in cases of sleep
disturbance or communication disturbance, and furthermore aggregated in DALY (Disability
Adjusted Life Years) units representing the number, duration and severity of the health
cases. The method is modelling the full cause-effect chain from the noise emissions of a
single vehicle up to the health damage. As in some other modern concepts of environmental
damage assessment, the analysis is subdivided into the four modules of fate analysis,
exposure analysis, effect analysis and damage analysis. The fate analysis yielding the noise
level increment due to an additional road transport over a given distance is conducted for
transports with known or with unknown routing, the latter case being more important in LCA
practice. The current number of persons subject to specific background levels of noise is
determined on the basis of the road traffic noise model, LUK, developed by the Swiss canton
of Zürich. The number of additional cases of health impairment due to incremental noise is
calculated with data out of the Swiss Noise Study 90 (cf. Müller-Wenk, 2002). An
assessment of the severity of sleep disturbance and communication disturbance, in
comparison to other types of health impairment, was performed by a panel consisting of
physicians experienced in the field of severity weighting of disabilities.
103
Meijer et al. 2006 build on Müller-Wenk, 2004 and elaborate this for indoor exposure to
noise by outdoor road transport. They have developed a methodology to calculate damages
to human health of occupants due to indoor exposure to noise emitted by neighbourhood car
traffic. The goal of the study was to assess the influence of the location of the dwelling on the
health of the occupants, compared to the damage to human health associated with the rest
of the life cycle of that dwelling. Fate, exposure and human health effects were addressed in
the calculation procedure. Fate factors for noise were based on noise levels generated by
traffic. Effect factors for noise were based on linear relationships between noise level
changes and health effects, while taking into account threshold values for noise levels for
negative impacts. Damage factors were calculated on the basis of disability adjusted life
years (DALYs). A default noise reduction due to the dwelling itself is included in the
calculations. The indoor exposure models used to calculate health damages are based on
the work of Müller-Wenk for noise. In the fate calculations, noise levels are calculated for a
scenario rather than on a per-vehicle base because of the non-linear relationship between
traffic density and noise level, and because there are threshold values for the noise levels
above or under which a change in noise level has no effect on the human health. For the
calculation of the effect factors for traffic noise, data from epidemiological researches – as
obtained by Müller-Wenk – was used. In these works, a linear dose-response relationship
between average noise levels and negative impacts was adopted.
Both approaches focus on road transport. As this is only one part of the life-cycle,
application for LCA in general will lead to biased results (e.g., air transport will be much
better than road transport). Of course, road transport is a major source of noise disturbance,
and the approaches can in principle be extended to include other sources of noise. Another
criticism is that the DALY-conversion has been made by a panel that concentrated on noise
issues, so that a too strong emphasis on noise is reflected in high DALY-values. On some of
these issues, improvements have been made the last few years. For instance, there is an EU
directive (2002/49/EC) in which noise maps of exposure are to be made by the member
states, progress has been made in understanding the cause-effect mechanisms (see, e.g.,
Babisch, 2006) and the DALY-weighting has improved (Mathers et al., 2003)
Although the need is stressed for developing a method to incorporate noise (from
transportation or otherwise) into LCA in a generally applicable way, and we think that the last
few years the state of knowledge has improved, no recommended approach is available for
LCA in general at this moment.
3.12.2.3 Accidents
The term “accidents” has many meanings. The accidental spill of chemicals does not
require a separate impact category, but may be taken into account along with the regular
emissions of chemicals, likewise, for nuclear facilities. Here, this impact category is taken to
refer to casualties resulting from accidents. The area of protection is human health, no
casualties that cause effects on ecosystems (such as car collisions with wildlife) are
considered. Most LCIA-methodologies do not include an impact category ‘casualties’.
Schmidt et al. (2004) describe a very useful method for including the working environment in
LCA, which encompasses casualties. This casualty analysis is based on a database
developed by EDIP in which the working environment impacts per kilo of produced goods are
listed for a number of economic activities. Hofstetter & Norris (2003) discuss injuries
104
(casualties) related to working environment accidents and these are part of the analysis of
“indoor & occupational exposure”. Also here, we might take out of their full method for the
working environment a method for assessing casualties. However, all approaches mentioned
focus on the occupational part, which is only one part of the life-cycle wide casualties. As
such, no approach can be recommended at the present stage of development.
3.12.2.4 Desiccation
Desiccation refers to local impacts due to water use, mainly in agricultural areas. Water
use as a resource has been discussed in the document on resource depletion. The impacts
on landscape, vegetation, soil productivity, etc. have, as far as we know, not been addressed
in the context of LCA. However, desiccation is an extremely important issue with impacts on
all three areas of protection. Hence, an LCA that provides answers to a question on, say,
cotton versus wool, without addressing desiccation is of a very restricted value. Desiccation
obviously has much to do with land use, and it could be assessed under that name.
However, none of the methods for land use in LCIA take into account the water use, but
focus on issues like vegetation before and after.
3.12.2.5 Erosion
For erosion three approaches have been found in literature: Cowell & Clift (2000), Muys &
Garcia Quijjano (2002) and Mattsson et al. (2000). According to Cowell & Clift (2000) the loss
of soil mass is an indicator for depletion of resources (soil as a resource). As a
characterisation model the soil static reserve life is proposed (SSRL = R/E). The soil static
reserve life is a function of global reserves of agricultural soil (R, i.e. total topsoil in the world
in tonnes) and current annual global net loss of topsoil mass by erosion (E in tonnes/year).
The necessary inventory data to calculate the impact score is the loss of soil mass (in
tonnes), either measured or estimated (e.g. using erosion models like USLE; see
http://topsoil.nserl.purdue.edu/usle/; http://www.fao.org//docrep/t1765e/t1765e0e.htm). At this
moment no operational factors are available. To derive such factors information is necessary
on the reserve of the topsoil, i.e. area and depth of topsoil suitable for agriculture.
Furthermore also worldwide erosion data should be available. Finally because soil is not
globally available (i.e. not shipped all over the world like ores and fossil fuels) a differentiation
of the factors for different regions is recommendable, using regional reserves and regional
erosion rates. Moreover, it is not quite clear what the inventory items are to which erosion
applies. So, even when the LCIA part is solved, additional work might be needed to connect
it to the LCI databases.
The method of Muys & Garcia Quijano (2002) describes the land use impact by 17
quantitative indicators divided over 4 impact subcategories: soil, water, vegetation structure
and biodiversity. The indicator soil erosion is a sub-indicator in the sub-impact category soil.
In this method it is proposed to transform the loss of soil mass into a loss of soil depth (in m)
using the bulk density of the soil. Finally, the loss of soil depth over a period of 100 years is
compared to the total rootable soil depth up to 1m. A complete loss of the soil within a period
of less than 100 years leads to the maximum impact score. (Erosion risk factor = E (kg/ha/yr)
× 100 yr/ Total Rootable Soil Depth (1m)). The necessary inventory data to calculate the
impact score is the loss of soil mass (in tonnes), either measured or estimated (e.g. using
erosion models like USLE).
105
At this moment no operational factors are available. To derive such set information, more
or less the same information is necessary as described for the method of Cowell & Clift,
(2000). The method of Mattson et al. (2000) describes the land use impact by 9 indicators for
3 impact sub categories soil fertility (7), biodiversity (1) and landscape (1). Most indicators
are described qualitative. The indicator soil erosion is a sub-indicator in the impact sub
category soil fertility. In this method it is proposed to use the loss of soil mass (kg) as an
indicator for erosion impact without using characterisation factors.
None of the methods discussed above is elaborated in an operational set of
characterisation factors. As such, no approach can be recommended at the present stage of
development.As part of future research, it is recommended to determine the problem of
erosion and the associated interventions properly first. Erosion basically is a natural
phenomenon that will occur any way. Human activities may, due to their nature and intensity,
enhance erosion. In a systems analysis as LCA, soil is considered to be environment by
most practitioners as also done in toxicity models. When soil is part of the environment
system, soil loss cannot be the intervention, as it doesn’t cross the economy-environment
boundary.41 Similar to global warming, the impact category should actually be something as
“enhanced erosion or enhanced soil loss”. Then, the proper interventions still need to be
determined. Man can enhance erosion by removing terraces, cutting hedgerows on steep
slopes, by deep ploughing and other agricultural practices. Interventions could thus be soil
disturbance by ploughing or cutting hedgerows. What one needs to know is thus extent of
natural loss of soil as a reference and the marginally increased loss of soil (kg/ha/yr) due to
all kinds of soil disturbance interventions (ploughing, cutting hedgerows etc.). All these
different interventions should be linked to characterisation factors indicating the marginally
increased soil loss due to that specific intervention, fully comparable to global warming, for
example. Perhaps, a generic characterisation factor per agricultural activity type could be
developed. To what extent this would be possible in practice, remains to be investigated but
some work has already been done that can be used as starting point of development (see
Guinée et al., 2006). Like desiccation, erosion could be taken into account as an aspect of
land use. This, however, requires a large restructuring of the way land use is now being
assessed in most LCIA methods.
3.12.2.6 Salination
Salination may refer to an effect with two different causes: the deposition of ions, and the
removal of water. Current LCIA methods do not adequately characterize the effects of
common ions associated with salinity impacts. Salination (or salinisation) of water resources
and of agricultural plots is of strategic concern in countries as South Africa and Australia, and
the need for life-cycle assessments to be able to incorporate salinity effects is apparent.
There is sufficiently clear cause-effect relationships between the sources (deposition of ions
and removal of water) and impacts of salinity, and impacts are claimed to be sufficiently
different in nature from existing categories to warrant a separate salinity impact category. For
41
That is, it contradicts the basic approach of LCA, where interventions by the economic system lead to a deficit
or excess in the environment, which on its turn leads to impacts such as resource depletion or toxicity.
Considering a physical activity, such as ploughing, clearing and levelling, as the intervention, requires a radical
departure from the present elementary flow based LCA.
106
the second pathway, removing water, a clear overlap with desiccation is present. To what
extent these two issues can and should be separated is not yet clear.The references include
only specific methods as developed for soil salinisation in Australia (Feitz & Lundie, 2002)
and water and soil salinisation in South Africa (Leske & Buckley, 2003, 2004 a, b). Salination
may, like desiccation and erosion, be included in a revised land use methodology. This is,
however, a research recommendation, not a recommended practice for now.
3.12.3 Cause-effect chains

For some of these “other impact categories”, a cause-effect chain can be drafted. For
noise, for instance, we have a clear idea of a facility (the unit process) which generates a
sound (the elementary flow), which propagates into the environment (fate), and leads to
impacts on man or ecosystems (effect). However, even here we already meet some
problems. One of the approaches discussed includes in the LCA of a house the impacts due
to road traffic noise on the street where the house is located. In the life cycle inventory of the
house, we would probably account for the sound emissions of mining, construction, and
disposal, but it is unclear if the sound emissions of a road transport which is part of a
different product life cycle should be included in the impacts of the life cycle of a house.For
some other impact categories, the situation is already more confusing at the outset. For
erosion, for instance, it is already not exactly clear what the elementary flow is. If that is
unclear, we cannot connect it to a unit process, and not connect it to an impact pathway.It
appears that more research is needed to bring more structure into the cause-effect chains of
the other impact categories prior to proposing characterisation factors.
3.12.4 Method selection

No models have been selected for further analysis.

As no models were analyzed, no evaluation can be made besides the overall remark that
it is premature to include the other impact categories in general purpose LCAs.
3.12.6 Conclusion
The other impact categories (here restricted to noise, accidents, desiccation, erosion, and
salination) are definitely important categories, and they deserve attention in future LCA
developments. In particular for specific LCAs (e.g., comparing different transport means,
comparing different agricultural practices) differences in impacts on noise, erosion,
desiccation, etc. may be critical. There is, however, no generally applicable model for these
impact categories. Development of the cause-effect framework and models to address the
associated midpoint or endpoint categories need to be developed. Moreover, it is important
to develop other missing categories as well; examples include the impact of genetic pollution,
landscape and aesthetic issues, and impacts of electromagnetic fields and light. Research
needs are described in Annex 2.
107
4 References
[1] Amann, M., Cofala, J., Heyes, C., Klimont, Z., Schöpp, W. (1999). The RAINS model:
A tool for assessing regional emission control strategies in Europe. Pollution
Atmosphérique 20(41-46).
[2] Ahbe, S., Braunschweig, A., Müller-Wenk, R. (1990). Methodology for Ecobalances
Based on Ecological Optimization, BUWAL (SAFEL) Environment Series No. 133,
Bern.
[3] Andersson-Sköld, Y., Grennfelt, P., Pleijel K. (1992). Photochemical Ozone Creation
Potentials: A study of Different Concepts. J. Air Waste Manage. Assoc. Vol. 42, No. 9,
pp. 1152-1158.
[4] Babisch, W. (2006). Transportation Noise and Cardiovascular Risk. Review and
Synthesis of Epidemiological Studies. Dose-effect Curve and Risk Estimation.
Umweltbundesamt, Dessau.
[5] Baitz, M. (2002): Die Bedeutung der funktionsbasierten Charakterisierung von
Flächen-Inanspruchnahmen in industriellen Prozesskettenanalysen. Phd-Thesis,
University of Stuttgart. See www.shaker.de (German).
[6] Bare, J.C., Pennington, D.W. and Udo de Haes, H.A. (1999). Life Cycle Impact
Assessment Sophistication – International Workshop. International Journal of Life
Cycle Assessment 4(5): 299-306.
[7] Bare, J. C., Hofstetter, P., Pennington, D.W. and Udo de Haes. H. A. (2000). Life
cycle impact assessment midpoints vs. endpoints: The sacrifices and the benefits.
International Journal of Life Cycle Assessment 5(5): 319–326.
[8] Bare J.C. (2002). Developing a Consistent Decision-Making Framework by Using the
U.S. EPA's TRACI Systems Analysis Branch, Sustainable Technology Division,
National Risk Management; Research Laboratory, US Environmental Protection
Agency, Cincinnati, OH;
http://www.epa.gov/ORD/NRMRL/std/sab/traci/aiche2002paper.pdf
[9] Bare J.C., Norris G.A., Pennington, D.W., McKone, T.E. (2003). TRACI, The Tool for
the Reduction and Assessment of Chemical and Other Environmental Impacts.
Journal of Industrial Ecology. 6 (3–4): 49-78.
[10] Beaugelin, K. (2006). EDEN, Modelling radiological dose in non-human species:
Principles, computerization, and application. Health Physics, 90 (5), pp. 485-493.
[11] Bos, U. and Wittstock, B. (2007). Land use methodology. Report to summarize the
current situation of the methodology to quantify the environmental effects of Land
Use. Report, Lehrstuhl für Bauphysics, University of Stuttgart.
[12] Brand, G., Braunschweig, A., Scheidegger, A., Schwank, O. (1998). Weighting in
Ecobalances with the Ecoscarcity Method – Ecofactors 1997. BUWAL (SAFEL)
Environment Series No. 297, Bern.
[13] Carter, W.P. (2000). Updated Maximum Incremental Reactivity scale for regulatory
applications. Sacramento, CA. California Air Resources Board.
[14] Chapman, P.M. (2008). Environmental risks of inorganic metals and metalloids: a
continuing, evolving scientific odyssey. Human Ecol. Risk Assess. 14: 5 – 40.
[15] Chapman, P.M., Wang, F., Janssen, C.R., Goulet, R.R., Kamunde, C.N. (2003).
Conducting ecological risk assessments of inorganic metals and metalloids: current
status. Human Ecol. Risk Assess. 9: 641 – 697.
108
4 Reference
[16] Cotler, H., Ortega-Larrocea, M.P. (2006). Effects of land use on soil erosion in a
tropical dry forest ecosystem, Chamela watershed, Mexico. Catena 65 (2006): 107-
117.
[17] Cowell, S.J., Clift, R. (2000). A methodology for assessing soil quantity and quality in
life cycle assessment. Journal of Cleaner Production 8: 321-331.
[18] Crettaz, P., Pennington, D, Rhomberg, L, Brand, K, Jolliet, O. (2002). Assessing
human health response in life cycle assessment using ED10s and DALYs: Part 1-
Cancer effects. Risk Anal. 22: 931–946.
[19] Den Hollander HA, Van Eijkeren JCH, Van de Meent, D. (2004). SimpleBox 3.0:
multimedia mass balance model for evaluating the fate of chemicals in the
environment. RIVM Report no 601200003, Bilthoven, The Netherlands
[20] Den Outer P.N., van Dijk A., Slaper H. (2008). Validation of ultraviolet radiation
budgets using satellite observations from the OMI instrument. RIVM Report no
610002002, Bilthoven, The Netherlands, pp. 59
[21] De Schryver and Goedkoop (2009a). Climate Change. Chapter 3 in: Goedkoop, M.,
Heijungs, R., Huijbregts, M.A.J., De Schryver, A., Struijs, J., Van Zelm, R. (2009).
ReCiPe 2008 A life cycle impact assessment method which comprises harmonised
category indicators at the midpoint and the endpoint level. Report I: Characterisation
factors, first edition.
[22] De Schryver and Goedkoop (2009b). Land Use. Chapter 10 in: Goedkoop, M.,
[23] De Schryver and Goedkoop (2009c). Mineral Resource. Chapter 12 in: Goedkoop, M.,
ReCiPe 2008. A life cycle impact assessment method which comprises harmonised
[24] De Schryver A.M., Brakkee K.W., Goedkoop M.J., Huijbregts M.A.J. (2009).
Characterization Factors for Global Warming in Life Cycle Assessment Based on
Damages to Humans and Ecosystems. Environ Sci Technol 43 (6): 1689–1695.
[25] Dewulf, J. et al. (2007). Cumulative Exergy extraction from the Natural environment
(CEENE): a comprehensive Life Cycle Impact Assessment Method for resource
depletion. Env. Science & Technology, accepted for publication Oct 2 2007.
[26] Derwent, R.G., Jenkin, M.E., Saunders, S.M., Pilling, M.J. (1998). Photochemical
ozone creation potentials for organic compounds in Northwest Europe calculated with
a master chemical mechanism. Atmospheric Environment 32(14/15), 2429-2441.
[27] Donaldson K., Stone V., Borm P.J., Jimenez L.A., Gilmour P.S., Schins R.P.,
Knaapen A.M., Rahman I., Faux S.P., Brown D.M. and MacNee W. (2003) Oxidative
stress and calcium signaling in the adverse effects of environmental particles (PM10).
Free Radical Biology and Medicine 34(11), 1369-1382.
[28] Dreicer, M., Tort, V., Manen, P. (1995). ExternE, Externalities of Energy, Vol. 5
Nuclear, Centr d'étude sur l'Evaluation de la Protection dans le domaine nucléaire
(CEPN), edited by the European Commission DGXII, Science, Research and
development JOULE, Luxembourg.
[29] EC, (2003). Communication on Integrated Product Policy. COM(2003) 302)
109
4 Reference
[30] EC, (2005). ExternE. Externalities of Energy Methodology 2005 Update (Edited by
Bickel P.and Friedrich R.)
[31] EC, (2008). Communication on the sustainable consumption and production and
sustainable industrial policy action plan. COM, 2008; 397 final.
[32] EC-JRC (2010a) ILCD Handbook. Analysis of existing Environmental Impact
Assessment methodologies for use in Life Cycle Assessment. p115. Available at
http://lct.jrc.ec.europa.eu
[33] EC- JRC (2010b), ILCD Handbook. Framework and Requirements for LCIA models
and indicators p112. Available at http://lct.jrc.ec.europa.eu
[34] EEA (1997). Biodiversity. Europe's environment: the second assessment. European
Environment Agency. Copenhagen, DK., Elsevier Science: 145-178.
[35] Ecotech (2001) Ecological Footprinting: Final study. Working Document for the STOA
Panel. PE number: 297.571/Fin.St. European Parliament, Directorate General for
Research, Luxembourg
[36] Environmental Impact Assessment Review, Volume 20, Number 2, April 2000, pp.
159-189(31).
[37] EPA (1997). Exposure Factors Handbook. U.S. Environmental Protection Agency,
National Center for Environmental Assessment, Office of Research and Development.
www.epa.gov/ncea/efh/
[38] Escher, B.I., Hermens, J.L.M. (2004). Internal exposure: linking bioavailability to
effects. Environ. Sci. Technol. 38: 455A-462A.
[39] Feitz, A., Lundie, S. (2002). Soil Salinisation: A Local Life Cycle Assessment Impact
Category. International Journal of Life Cycle Assessment, 7 (4): 244-249.
[40] Forster, P., Ramaswamy, V., Artaxo, P., Berntsen, T., Betts, R., Fahey, D.W.,
Haywood, J., Lean, J., Lowe, D.C., Myhre, G., Nganga, J., Prinn, R., Raga, G.,
Schulz, M. and Van Dorland, R., (2007). Changes in Atmospheric Constituents and in
Radiative Forcing. In: Climate Change 2007: The Physical Science Basis IPCC 2007.
Contribution of Working Group I to the Fourth Assessment Report of the
Intergovernmental Panel on Climate Change.
[41] Frischknecht, R., Braunschweig, A., Hofstetter P., Suter P. (2000), Modelling human
health effects of radioactive releases in Life Cycle Impact Assessment. Environmental
Impact Assessment Review, 20 (2) pp. 159-189.
[42] Frischknecht, R., Steiner, R., Jungbluth, N. (2008). Methode der ökologischen
Knappheit – Ökofaktoren 2006, ö.b.u. und Bundesamt für Umwelt, Bern.
[43] Frischknecht, R., Steiner, R., Braunschweig, A., Egli, N., Hildesheimer, G. (2006a).
Swiss Ecological Scarcity Method: the new version 2006. Proceedings of the 7th
International Conference on EcoBalance, Tsukuba, Japan, November 2006.
[44] Frischknecht R, Steiner R, & Jungbluth N, The Ecological Scarcity Method - Eco-
Factors (2006b): A method for impact assessment in LCA. 2009, Federal Office for
the Environment FOEN: Zürich und Bern. Retrieved from
www.bafu.admin.ch/publikationen/publikation/01031/index.html?lang=en
[45] Garnier-Laplace J. C., Beaugelin-Seiller K, Gilbin R, Della-Vedova C, Jolliet O, Payet
J, 2008. A Screening Level Ecological Risk Assessment and ranking method for liquid
radioactive and chemical mixtures released by nuclear facilities under normal
operating conditions. Proceedings of the International conference on radioecology
and environmental protection, 15-20 june 2008, Bergen.
110
4 Reference
[46] Garnier-Laplace J. C., Beaugelin-Seiller K, Gilbin R, Della-Vedova C, Jolliet O, Payet

J, (2009).A Screening Level Ecological Risk Assessment and ranking method for
liquid radioactive and chemical mixtures released by nuclear facilities under normal
operating conditions Radioprotection 44 (5) 903-908 DOI: 10.1051/radiopro/20095161
[47] Garnier-Laplace, J.C., Della-Vedova, R. Gilbin, D. Copplestone, J. Hingston and P.
Ciffroy, (2006). First derivation of Predicted-No-Effect Values for freshwater and
terrestrial ecosystems exposed to radioactive substances. Environ. Sci.&Techno.,40,
6498-6505.
[48] Goedkoop and De Schryver (2009). Fossil Resource. Chapter 13 in: Goedkoop, M.,
[49] Goedkoop, M. and Spriensma, R. (2000). The Eco-indicator 99: A damage oriented
method for Life Cycle Impact Assessment Methodology. Ministry of VROM, The
Hague, The Netherlands. (updated version: www.pre.nl/eco-indicator99 ).
[50] Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., De Schryver, A., Struijs, J., Van Zelm,
R. (2009): ReCiPe 2008 A life cycle impact assessment method which comprises
harmonised category indicators at the midpoint and the endpoint level. Report I:
Characterisation factors, first edition.
[51] Greco, S.L., Wilson, A.M., Spengler J.D., and Levy J.I. (2007). Spatial patterns of
mobile source particulate matter emissions-to-exposure relationships across the
United States. Atmospheric Environment (41), 1011-1025.
[52] Guinėe JB, Heijungs R. (1993). A proposal for the classification of toxic substances
within the framework of life cycle assessment of products. Chemosphere 26(10),
1925-1944.
[53] Guinée, J.B. (1995). Development of a methodology for the environment life-cycle
assessment of products; with a case study on margarines. Thesis, Leiden University.
[54] Guinée, J.B. (Ed.), Gorrée, M., Heijungs, R., Huppes, G., Kleijn, R., de Koning, A.,
Van Oers, L., Wegener Sleeswijk, A., Suh, S.,. Udo de Haes, H.A, De Bruijn, J.A.,
Van Duin R., Huijbregts, M.A.J. (2002). Handbook on Life Cycle Assessment:
Operational Guide to the ISO Standards. Series: Eco-efficiency in industry and
science. Kluwer Academic Publishers. Dordrecht (Hardbound, ISBN 1-4020-0228-9;
Paperback, ISBN 1-4020-0557-1).
[55] Guinée, J.B., Van Oers, L., A. de Koning; W. Tamis (2006). Life cycle approaches for
Conservation Agriculture - Part I: A definition study for data analysis; Part II: Report of
the Special Symposium on Life Cycle Approaches for Conservation Agriculture on 8
May 2006 at the SETAC-Europe 16th Annual Meeting at The Hague. Research
commissioned by Syngenta Crop Protection AG, Basel. CML report 171, Leiden
(http://www.leidenuniv.nl/cml/ssp/index.html).
[56] Harbers, J.V., Huijbregts, M.A.J., Posthuma, L., Van de Meent, D. (2006). Estimating
the impact of high-production-volume chemicals on remote ecosystems by toxic
pressure calculation. Environ. Sci. Technol. 40: 1573-1580.
[57] Hauschild, M.Z. (2005). Assessing environmental impacts in a life cycle perspective.
ES&T, 39(4), 81A-88A.
[58] Hauschild, M.Z. and Wenzel, H. (1998a). Environmental assessment of product. Vol.
2 -Scientific background, Chapman & Hall, United Kingdom, Kluwer Academic
Publishers, ISBN 0412 80810 2, Hingham, MA., USA. (2004 update figures
http://www.lca-center.dk/cms/site.aspx?p=1378 ).
111
4 Reference
[59] Hauschild, M.Z. and Wenzel, H. (1998b). Nutrient enrichment as criterion in the
environmental assessment of products. Chapter 5 of Hauschild, M.Z. and Wenzel, H.
Environmental assessment of products. Vol. 2 - Scientific background, 565 pp.
Chapman & Hall, United Kingdom, Kluwer Academic Publishers, Hingham, MA. USA.
ISBN 0412 80810 2.
[60] Hauschild, M. and Potting, J. (2005). Spatial differentiation in life cycle impact
assessment – the EDIP2003 methodology. Environmental News no. 80. The Danish
Ministry of the Environment, Environmental Protection Agency, Copenhagen.
[61] Hauschild, M.Z., Potting, J., Hertel, O., Schöpp, W., and Bastrup-Birk, A., (2006).
Spatial differentiation in the characterisation of photochemical ozone formation – The
EDIP2003 methodology. Int.J.LCA, 11(Special Issue 1), 72-80.
[62] Hauschild, M., Huijbregts, M., Jolliet, O., MacLeod, M., Margni, M., Payet, J.,
Schuhmacher, M., van de Meent, D., McKone, T. (2008). Building a Consensus Model
for Life Cycle Impact Assessment of Chemicals: the Search for Harmony and
Parsimony. Environ. Sci. Technol., 2008, 42 (19), 7032-7037.
[63] Hayashi, K., Itsubo, N. and Inaba, A. (2000). Development of Damage Function for
Stratospheric Ozone Layer Depletion - A Tool Towards the Improvement of the
Quality of Life Cycle Impact Assessment IJLCA 5(5), 265-172.
[64] Hayashi, K., Okazaki, M., Itsubo, N., Inaba, A. (2004). Development of Damage
Function of Acidification for Terrestrial Ecosystems Based on the Effect of Aluminum
Toxicity on Net Primary Production. International Journal of Life Cycle Assessment
9(1): 13-22.
[65] Hayashi, K., Nakagawa, A., Itsubo, N., Inaba, A. (2006). Expanded Damage Function
of Stratospheric Ozone Depletion to Cover Major Endpoints Regarding Life Cycle
Impact Assessment. International Journal of Life Cycle Assessment 11, 150-161.
[66] Hertwich, E.G., Pease, and Koshland, C.P. (1997). Evaluating the environmental
impact of products and production processes: A comparison of six methods. The
Science of the Total Environment 196: 13-29.
[67] Hertwich, E.G., Pease, W.S. and McKone, T.E. (1998). Evaluating toxic impact
assessment methods: What works best? Environmental Science and Technology
32:A138 – A144.
[68] Hertwich, E., McKone, T. and Pease, W. (1999). Parameter uncertainty and variability
in evaluative fate and exposure models. Risk Analysis 19: 1193 – 1204.
[69] Hertwich, E., Matales, S.F., Pease, W.S., McKones, T.E. (2001): Human Toxicity
Potentials for Life-Cycle Assessment and Toxics Release Inventory Risk Screening.
Environmental Toxicology and Chemistry 20, 928-939
[70] Hettelingh, J.P., Posch, M., Slootweg, J., Reinds, G. J., Spranger, T., Tarrason, L.
(2007). Critical loads and dynamic modelling to assess European areas at risk of
acidification and eutrophication. Water Air Soil Pollution Focus 7: 379–384.
[71] Heyes C., Schöpp, W., Amann, M. And Unger, S. (1996). A reduced-form model to
predict long-term ozone concentrations in Europe. Interim report WP-96-
12/December, IIASA, Vienna.
[72] Hofstetter, P. (1998). Perspectives in Life Cycle Impact Assessment. A Structure
Approach to Combine Models of the Technosphere, Ecosphere and Valuesphere.
Kluwer Academic Publishers, Dordrecht, The Netherlands, 484 pp.
[73] Hofstetter, P., Müller-Wenk, R.(2005). Monetization of health damages from road
noise with implications for monetizing health impacts in life cycle assessment. Journal
of Cleaner Production 13:1235-1245
112
4 Reference
[74] Hofstetter, P., Norris, G.A. (2003). Why and how should we assess occupational
health impacts in integrated product policy? Environmental Science & Technology,
37(10): 2025-2035.
[75] Houghton, J.T., Jenkins, G.J., Ephraums, J.J. (eds.) (1990). Climate Change: The
IPCC Scientific Assessment IS92a. Cambridge University Press, Cambridge, 365 pp.
[76] Huijbregts, M.A.J., Rombouts, L.J.A., Ragas A.M.J., Van de Meent, D. (2005a).
Human-toxicological effect and damage factors of carcinogenic and noncarcinogenic
chemicals for life cycle impact assessment. Integrated Environ. Assess. Manag. 1:
181-244.
[77] Huijbregts, M.A.J., Struijs, J., Goedkoop, M., Heijungs, R., Hendriks, A.J., Van de
Meent, D. (2005b). Human population intake fractions and environmental fate factors
of toxic pollutants in life cycle impact assessment. Chemosphere 61: 1495-1504.
[78] Huijbregts, M.A.J., Thissen, U., Guinée, J.B., Jager, T., Van de Meent, D., Ragas,
A.M.J., Wegener Sleeswijk, A., Reijnders, L. (2000). Priority assessment of toxic
substances in life cycle assessment, I: Calculation of toxicity potentials for 181
substances with the nested multi-media fate, exposure and effects model USES-LCA.
Chemosphere 41:541-573.
[79] Huijbregts, M.A.J., Van Zelm, R. (2009). Ecotoxicity and human toxicity. Chapter 7 in:
Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., Struijs, J., De Schryver, A., Van Zelm,
R. (2009): ReCiPe 2008 A life cycle impact assessment method which comprises
harmonised category indicators at the midpoint and the endpoint level. Report I:
Characterisation factors, first edition.
[80] Huijbregts, M.A.J., Verkuijlen, S.W.E., Heijungs, R., Reijnders, L. (2001). Spatially
explicit characterization of acidifying and neutrophying air pollution in life-cycle
assessment. Journal of Industrial Ecology 4(3): 75-92.
[81] Humbert, S. (2009). Geographically Differentiated Life-cycle Impact Assessment of
Human Health. Doctoral dissertation, University of California, Berkeley, Berkeley,
California, USA.
[82] ICRP, (1999). Risk Estimation for Multifactorial Diseases, Ann. ICRP 29(3-4),
http://www.icrp.org/annals_list.asp
[83] ICRP, (2007). P103: The 2007 Recommendations of the International Commission on
Radiological Protection. Annals of the ICRP 37(2-4) 1-332.
http://www.sciencedirect.com/science/journal/01466453; table A.4.4
[84] IER, (2008). EcoSense. Institut für Energiewirtschaft und Rationelle
Energieanwendung (IER), Universität Stuttgart, 70550 Stuttgart, Germany.
http://ecosenseweb.ier.uni-stuttgart.de/ .
[85] IPCC (2000), Land Use, Land-Use Change, and Forestry. A Special Report of the
IPCC, (eds. Watson,R.T., Noble, I.R., Bolin, B., Ravindranath, N.H. Verardo, D.J.,
Dokken, D.J.), Cambridge University Press, Cambridge, United Kingdom and New
York, NY, USA.
[86] IPCC (2007). IPCC Climate Change Fourth Assessment Report: Climate Change
2007. http://www.ipcc.ch/ipccreports/assessments-reports.htm
[87] ISO (1997). Environmental management–life cycle assessment– principles and
framework, (ISO14040), Paris.
[88] ISO (2000). Environmental management - life cycle assessment - life cycle impact
assessment (ISO 14042). ISO, Geneva.
[89] ISO (2006). Environmental management - life cycle assessment - life cycle impact
assessment (ISO 14044). ISO, Geneva.
113
4 Reference
[90] Itsubo, N. and Inaba, A. (2003): A New LCA Method: LIME has been completed. Int J
LCA 8 (5) 305
[91] Itsubo, N., Sakagami, M., Washida, T., Kokubu, K. and Inaba, A. (2004). Weighting
Across Safeguard Subjects for LCIA through the Application of Conjoint Analysis.
IJLCA 9(3), 196-2005.
[92] Itsubo, N. et al. (2008a). LIME documentation. Eutrophication. Chapter 2.8 of draft
English translation of LIME documentation (figures and tables in Japanese).
[93] Itsubo, N. et al. (2008b). LIME documentation. Land use. Chapter 2.9 of draft English
translation of LIME documentation (figures and tables in Japanese).
[94] Itsubo, N. et al. (2008c). LIME documentation. Ozone layer depletion. Chapter 2.1 of
draft English translation of LIME documentation (figures and tables in Japanese).
[95] Itsubo, N. et al. (2008d). LIME documentation. Photochemical oxidants. Chapter 2.5
of draft English translation of LIME documentation (figures and tables in Japanese).
[96] Itsubo, N., Li, R., Abe, K., Nakagawa, A., Hayashi, K., Inaba, A. (2003). Biodiversity
Damage Assessment – Applying the Theory of Biology in LCIA. Slides from platform
presentation (WE2/8) at the SETAC Europe 13th Annual Meeting at Hamburg,
Germany, April 27 – May 1, 2003
[97] Jarvinen, O. and Miettinen, K. (1987): "Sista paret ut" Naturskyddsforeningen.
Miljoforlaget, Sweden.
[98] Jolliet, O., Margni, M., Charles, R., Humbert, S., Payet, J., Rebitzer, G., Rosenbaum,
R. (2003). IMPACT 2002+: A New Life Cycle Impact Assessment Methodology.
International Journal of Life Cycle Assessment 8 (6): 324-330.
[99] Jolliet, O., R. Müller-Wenk, J. C. Bare, A. Brent, M. Goedkoop, R. Heijungs, N. Itsubo,
C. Peña, D. Pennington, J. Potting, G. Rebitzer, M. Stewart, H. Udo de Haes and B.
Weidema (2004). The LCIA Midpoint-damage Framework of the UNEP/SETAC Life
Cycle Initiative. International Journal of Life Cycle Assessment 9(6): 394-404
[100] Jolliet, O., Shaked S., and Humbert S. (2006). Comparative Identification of Most
Significant Toxics Affecting Human Health. SETAC Europe, 16th Annual Meeting, The
Hague, The Netherlands, May 2006.
[101] Kemna, R., Van Elburg, M., Li W., Van Holsteijn, R. (2005). MEEUP – Methodology
Report. EC, Brussels. (Final version, 28-11-2005).
[102] Klasmeier, J.; Matthies, M.; MacLeod, M.; Fenner, K.; Scheringer, M.; Stroebe, M.; Le
Gall, A. C.; McKone, T.; Van de Meent, D.; Wania, F., (2006).Application of
multimedia models for screening assessment of long-range transport potential and
overall persistence. Environ. Sci. Technol. 40, 53-60.
[103] Klepper O, Beusen AHW, Meinardi CR (1995): Modelling the flow of nitrogen and
phosphorus in Europe: from loads to coastal seas. RIVM report 451501004, RIVM,
Bilthoven, the Nederlands.
[104] Klepper, O. and Van de Meent, D. (1997). Mapping the Potentially Affected Fraction
(PAF) of species as an indicator of generic toxic stress. Report 607504001. National
Institute of Public Health and the Environment (RIVM), Bilthoven, The Netherlands.
[105] Michelsen, O. (2007). Assessment of land use impact on biodiversity. International
Journal of LCA, 13 (1), pp. 22-31.
[106] Köllner, T. (2000). Species-pool Effect Potentials (SPEP) as a yardstick to evaluate
land-use impacts on biodiversity. Journal of Cleaner Production 8 (4), pp 293-311
114
4 Reference
[107] Köllner, T. (2001). Land Use in Product Life Cycles and its Consequences for
Ecosystem Quality. PhD thesis No. 2519, University St. Gallen. And: Countryside
Survey 2000: Survey of Broad Habitats and Landscape features. ISBN: 1 85112 460
8.
[108] Koellner, T., & Scholz, R. W. (2008). Land Use in LCA (Subject: Editor Llorenç Milà i
Canals) Assessment of Land Use Impacts on the Natural Environment Part 2: Generic
Characterization Factors for Local Species Diversity in Central Europe. International
Journal of Life Cycle Assessment 13(1): 32 - 48.
[109] Krewitt, W., Pennington, D., Olsen, S.I., Crettaz, P., Jolliet, O. (2002). Indicators for
human toxicity in life cycle impact assessment. Chapter 5 in: Udo de Haes, H.A. (ed.).
Life-cycle impact assessment: striving toward best practice. Pensacola (FL), USA.
Society of Environmental Toxicology and Chemistry (SETAC).
[110] Krewitt, W., Trukenmüller, A., Bachmann, T.M., Heck, T. (2001). Country-specific
damage factors for air pollutants. A step towards site dependent life cycle impact
assessment. Int. Journal of Life Cycle Assessment, 6(4), 199-210.
[111] Kros, J. (2002). Evaluation of Biogeochemical Models at Local and Regional Scale.
Wageningen (The Netherlands), Wageningen University.
[112] Krol, M., (1993). Netherlands "Changes in crop yield" due to climate change.
Presentation at conference on comparative risk analysis and priority setting, Keyston,
Colorado, June 7-11
[113] Leske, T., Buckley, C. (2003). Towards the development of a salinity impact category
for South African environmental life-cycle assessments: Part 1 - A new impact
category. Water SA, 29, 3, 289-296.
[114] Leske, T., Buckley, C. (2004a). Towards the development of a salinity impact category
for South African life cycle assessments: Part 2 - A conceptual multimedia
environmental fate and effect model. Water SA, 30, 2, 241-251.
[115] Leske, T., Buckley, C. (2004b). Towards the development of a salinity impact category
for South African life cycle assessments: Part 3 – Salinity potentials. Water SA, 30, 2,
253-265.
[116] Ligthart, .T. N., Jongbloed, R.H.,and J. E. Tamis. (2010).A method for improving
Centre for Environmental Studies (CML) characterisation factors for metal
(eco)toxicity — the case of zinc gutters and downpipes.Int. Journal of Life Cycle
Assessment, 15 (8): 745-756
[117] Lindeijer, E., Muller-Wenk, R., Steen, B. (2002). Impact assessment on resources and
land use. In: Udo de Haes et al. Life cycle impact assessment: Striving towards best
practice. SETAC, Pensacola, Florida.
[118] MacLeod, M., Woodfine, D.G., Mackay, D., McKone, T.E., Bennett, D.H. Maddalena,
R. (2001). BETR North America: A regionally segmented multimedia contaminant fate
model for North America. Environmental Science Pollution 8: 156-163.
[119] Meinshausen, M., 2005. Emission & Concentration Implications of long-term Climate
Targets, Dissertation 15946 for the Swiss federal Institute of Technology, Zurich
http://www.up.ethz.ch/publications/dissertations/MalteMeinshausen_2005_dissertation
.pdf
[120] Margni, M., D. W. Pennington, C. Amman and O. Jolliet (2004). "Evaluating
multimedia/multipathway model intake fraction estimates using POP emission and
monitoring data." Environmental Pollution 128(1-2): 263-277.
115
4 Reference
[121] Margni, M., Gloria, T., Bare, J., Seppälä, J., Steen, B., Struijs, J., Toffoletto, L., Jolliet,
O. (2008). Guidance on how to move from current practice to recommended practice
in Life Cycle Impact Assessment. Paris, UNEP/SETAC Life Cycle Initiative.
[122] Marshall, J.D., Teoh, S.K., Nazaroff, W.W. (2005). Intake fraction of nonreactive
vehicle emissions in US urban areas. Atmospheric Environment 39, 1363-1371.
[123] Mathers, C.D., Bernard, C., Moesgaard Iburg, K., Inoue, M., Ma Fat, D., Shibuya, K.,
Stein, C., Tomijima, N., Xu, H. (2003) Global Burden of Disease in 2002: data
sources, methods and results. Global Programme on Evidence for Health Policy
Discussion Paper No. 54, World Health Organization (revised February 2004).
[124] Mattsson, B., Cederberg, C., Blix, L. (2000). Agricultural land use in life cycle
assessment (LCA): case studies of three vegetable oil crops. Journal of Cleaner
Production 8: 283-292.
[125] McKone, T., Bennett, D., Maddalena, R. (2001). CalTOX 4.0 Technical Support
Document, Vol. 1. LBNL - 47254, Lawrence Berkeley National Laboratory, Berkeley,
CA.
[126] McKone, T. E.; MacLeod, M. (2004).Tracking multiple pathways of humanexposure to
persistent multimedia pollutants: Regional, continental, and global scale models.
Annu. Rev. Environ. Resour. 28, 463–492.
[127] McMichael, A.J., Campbell-Lendrum, D.H., Corvalan, C.F., Ebi, K.L., Githeko, A.,
Scheraga, J.D., Woodward, A., (2003). Climate change and human health. Risk and
responses. Word Health Organization, Geneva. 322p.
[128] Meadows, D. (Dennis), Meadows, D. (Donella), Randers, J. (2004). Limits to Growth:
The 30-Year Update. White River Junction, Vermont: Chelsea Green Publishing
Company.
[129] Meijer, A., Huijbregts, M.A.J., Hertwich, E.G., Reijnders, L. (2006). Including Human
Health Damages due to Road Traffic in Life Cycle Assessment of Dwellings.
International Journal of Life Cycle Assessment, 11 (Special Issue 1): 64-71.
[130] Meinshausen, M., (2005). Emission & Concentration Implications of long-term Climate
Targets, Dissertation 15946 for the Swiss federal Institute of Technology, Zurich.
http://www.up.ethz.ch/publications/dissertations/MalteMeinshausen_2005_dissertation
.pdf .
[131] Milà i Canals, L., Bauer, C., Depestele, J., Dubreuil, A., Freiermuth Knuchel, R.,
Gaillard, G., Michelsen, O., Müller-Wenk, R., Rydgren B. (2007a). Key elements in a
framework for land use impact assessment within LCA. Int J LCA 12:5-15
[132] Milà i Canals L, Romanyà J, Cowell SJ (2007b). Method for assessing impacts on life
support functions (LSF) related to the use of ‘fertile land’ in Life Cycle Assessment
(LCA). J Clean Prod 15 1426-1440
[133] Milà i Canals L, Muñoz I, McLaren SJ. (2007c). LCA Methodology and Modelling
Considerations for Vegetable Production and Consumption. CES Working Papers
02/07 Available from http://www.ces-surrey.org.uk/
[134] Milà i Canals L, Chenoweth J, Chapagain AK, Orr S, Antón A, Clift R (2009)
Assessing freshwater use impacts in LCA Part 1: inventory modelling and
characterisation factors for the main impact pathways. Int J Life Cycle Ass 14(1):28–
42
[135] Müller-Wenk, R. (1994). The Ecoscarcity Method as a Valuation Instrument within the
SETAC-Framework, in: Udo de Haes/Jensen/Klöpffer/Lindfors (Ed.): Integrating
Impact Assessment into LCA, SETAC-Europe, Brussels 1994, p. 115-120.
116
4 Reference
[136] Müller-Wenk, R. (1998a) Land use – The main threat to species. How to include land
use in LCA. Institute for Economy and the Environment (IWO), University St. Gallen.
[137] Müller-Wenk, R. (1998b) Depletion of Abiotic Resources Weighted on the Base of
"Virtual" Impacts of Lower Grade Deposits in Future. IWÖ Diskussionsbeitrag Nr. 57,
Universität St. Gallen, Switzerland, ISBN 3-906502-57-0.
[138] Müller-Wenk, R. (2002). Attribution to road traffic of the impact of noise on health.
Environmental Series No. 339. Swiss Agency for the Environment, Forests and
Landscape, Bern.
[139] Müller-Wenk, R. (2004). A Method to Include in LCA Road Traffic Noise and its Health
Effects. International Journal of Life Cycle Assessment, 9 (2): 76-85.
[140] Murray, C.J.L., Lopez, A.D. (1996). The global burden of disease: a comprehensive
assessment of mortality and disability from diseases, injuries, and risk factors in 1990
and projected to 2020. Global Burden of Disease and Injury Series Volume I. Harvard
School of Public Health, World Bank, World Health Organisation, USA. 990 p.
[141] Muys, B., Garcia Quijano, J. (2002). A new method for Land Use Impact Assessment
in LCA based on ecosystem exergy concept. Internal report. Laboratory for Forest,
Nature and Landscape Research, KU. Leuven.
[142] Narita, N., Nakahara, Y., Morimoto, M., Aoki, R., Suda, S. (2004). Current LCA
Database Development in Japan – Results of the LCA Project. Int J LCA 9 (6) 355-
359.
[143] Nishioka, Y., Levy, J.I., Norris, G.A., Wilson, A., Hofstetter, P., and Spengler, J.D.
(2002). Estimating the public health benefits of increased residential insulation for new
housing. Risk Analysis 22(5): 1003–1017.
[144] Nordhaus, W.D. (1994). The 'DICE' Model: Background and Structure of a Dynamic
Integrated Climate-Economy Model of the Economics of Global Warming. Cowles
Foundation Discussion Paper. New Haven, Conn.: Cowles Foundation for Research
in Economics.
[145] Norris, G. (2002). Impact characterisation in the tool for the reduction and assessment
of chemical and other environmental impacts: Methods for acidification,
eutrophication, and ozone formation. Journal of Industrial Ecology 6(3/4): 83 – 105.
[146] Norris, G.A. (2003). Impact Characterization in the Tool for the Reduction and
Assessment of Chemical and other Environmental Impacts: Methods for Acidification,
Eutrophication, and Ozone Formation. Journal of Industrial Ecology 6 (3&4): 79-101.
[147] Oers, L. van, A. de Koning, J.B. Guinée & G. Huppes, 2002. Abiotic resource
depletion in LCA - Improving characterisation factors for abiotic resource depletion as
recommended in the new Dutch LCA Handbook. DWW report, Delft; see
http://www.cml.leiden.edu/research/industrialecology/researchprojects/finished/abiotic
-depletion-lcia.html
[148] Payet, J. (2004). Assessing toxic impacts on aquatic ecosystems in LCA. Doctoral
thesis Thesis 3112, Ecole Polytechnique Fédérale de Lausanne (EPFL, CH-1015
Lausanne), pp.214
[149] Payet, J. (2006). Deliverable D.4.1.4 Report describing a method for the quantification
of impacts on aquatic freshwater ecosystems resulting from different stressors (e.g.,
toxic substances, eutrophication, etc). Report from EU FP6 project no. 003956,
NOMIRACLE.
[150] Pennington D. W, Crettaz P, Tauxe A, Rhomberg L, Brand B, Jolliet O (2002):
Assessing Human Health Response in Life Cycle Assessment Using ED 10s and
DALYs: Part 2-Noncancer Effects. Risk Analysis 22, 947–963
117
4 Reference
[151] Pennington, D.W., Margni, M., Ammann, C., Jolliet, O. (2005): Multimedia fate and
human intake modeling: Spatial versus nonspatial insights for chemical emissions in
Western Europe. Environmental Science and Technology 39 (4), 1119-1128.
[152] Pennington, D.W., Margni, M., Payet, J., and Jolliet, O. (2006). Risk and Regulatory
Hazard-Based Toxicological Effect Indicators in Life-Cycle Assessment (LCA). Human
and Ecological Risk Assessment, Vol. 12, No. 3. (June 2006: best HERA paper of
year 2006 in Integrated Risk Assessment), pp. 450-475.
[153] Pfister, S.; Koehler, A.; Hellweg, S. Assessing the environmental impacts of
freshwater consumption in LCA. Environ. Sci. Technol. 2009, 43, 4098–4104
[154] Pope, C.A., Burnett R.T., Thun, M.J., Calle, E.E., Krewski,D., Ito, K., Thurston, G.D.
(2002). Lung cancer, cardiopulmonary mortality, and long-term exposure to fine
particulate air pollution. Journal of the American Medical Association 287, 1132-1141.
[155] Posch, M., Seppälä, J., Hettelingh, J.P., Johansson, M., Margni M., Jolliet, O. (2008).
The role of atmospheric dispersion models and ecosystem sensitivity in the
determination of characterisation factors for acidifying and eutrophying emissions in
LCIA. International Journal of Life Cycle Assessment (13) pp.477–486
[156] Posthuma L., De Zwart D. (2006). Predicted effects of toxicant mixtures are confirmed
by changes in fish species assemblages in Ohio, USA, rivers. Environmental
Toxicology and Chemistry, 25 (4), pp. 1094-1105.
[157] Potting, J. and Hauschild, M. (2005). Background for spatial differentiation in life cycle
impact assessment – the EDIP2003 methodology. Environmental project no. 996,
Danish Environmental Protection Agency, Copenhagen, 2005.
[158] Potting, J., Schöpp, W., Blok, K., Hauschild, M.Z. (1998a). Comparison of the
acidifying impact from emissions with different regional origin in life-cycle assessment.
Journal of Hazardous Materials 61(1-3): 155-162.
[159] Potting, J., Schöpp, W., Blok, K., Hauschild, M.Z. (1998b). Site-Dependent Life-Cycle
Impact Assessment. Journal of Industrial Ecology 2(2): 63-87.
[160] Potting, J., Trukenmüller, A., Christensen, F.M., Van Jaarsveld, H., Olsen, S.I.,
Hauschild, M.Z. (2005). Human toxicity. In: Hauschild, M.Z. and Potting, J. (2005).
Spatial differentiation in life cycle impact assessment – the EDIP2003 methodology.
Environmental News no. 80. The Danish Ministry of the Environment, Environmental
Protection Agency, Copenhagen. Chapter 8.
[161] Preiss P., Klotz V. (2008). Revised Description of updated and extended toolsfor the
detailed site-dependent assessment of External costs EcoSenseWeb. NEED project
(EU Project no: 502687), Deliverable n° 7.1 - RS 1b, public available on www.needs-
project.org
[162] Rabl, A. and Spadaro, J.V. (2004). The RiskPoll software, version is 1.051 (dated
August 2004). www.arirabl.com.
[163] Ravishankara AR, Daniel JS, Portmann RW. Nitrous Oxide (N2O) (2009): The
dominant ozone-depleting substance emitted in the 21st Century Science 326:123-
125
[164] Rees, W. E. (1992) "Ecological footprints and appropriated carrying capacity: what
urban economics leaves out," Environment and Urbanisation. 4(2): 121-130,
[165] Rochat, D., Margni, M., Jolliet, O. (2006). Continent-Specific Characterization Factors
and Intake Fractions for Toxic Emissions: Does it make a Difference? International
Journal of Life Cycle Assessment 11(Special issue 1) :55-63.
118
4 Reference
[166] Ron E, Muirhead C., 1998. The carcinogenic effects of ionizing radiation:
epidemiological evidence. In: LowDoses of Ionizing Radiation: Biological Effects and
Regulatory Control. Proceedings of a Conference, Seville, Spain, 17–21 November
1997. Vienna: IAEA and WHO, pp. 165–80.
[167] Rosenbaum, R. (2006). Multimedia and food chain modelling of toxics for comparative
risk and life cycle impact assessment. Thesis 3539, Chapter 5 on uncertainties, Ecole
Polytechnique Fédérale de Lausanne (EPFL, CH-1015 Lausanne).
[168] Rosenbaum, R., Margni, M. and Jolliet, O. (2007). A flexible matrix algebra framework
for the multimedia multipathway modeling of emission to impacts. Environment
international. 33 (5): 624-634.
[169] Rosenbaum, R.K., Bachmann, T.M., Gold, L.S., Huijbregts, M.A.J., Jolliet, O.,
Juraske, R., Köhler, A., Larsen, H.F., MacLeod, M., Margni, M., McKone, T.E., Payet,
J., Schuhmacher, M., van de Meent, D., Hauschild, M.Z. (2008): USEtox - The UNEP-
SETAC toxicity model: recommended characterisation factors for human toxicity and
freshwater ecotoxicity in Life Cycle Impact Assessment. International Journal of Life
Cycle Assessment, 13(7): 532-546, 2008
[170] Schere, K.L. and Demerjian, K.L., (1984). User's guide for the photochemical box
model (PBM), EPA-600/8-84-022a.
[171] Schmidt, A. Bruun Rasmussen, P., Andreasen, J., Fløe, T and Poulsen, K.E. (2004),
LCA and the working environment. Environmental Project no. 907, Danish
Environmental Protection Agency, Copenhagen, Denmark.
[172] Seppälä, J., Posch, M., Johansson, M., Hettelingh, J.P. (2006). Country-dependent
Characterisation Factors for Acidification and Terrestrial Eutrophication Based on
Accumulated Exceedance as an Impact Category Indicator. International Journal of
Life Cycle Assessment 11(6): 403-416.
[173] Steen, B (1999a). A Systematic approach to environmental priority strategies in
product development (EPS). Version 2000-general system characteristics; CPM
report 1999:4, Chalmers University of Technology, Gothenburg, Sweden
[174] Steen, B (1999b). A Systematic approach to environmental priority strategies in
product development (EPS). Version 2000-Models and data of the default method;
CPM report 1999:5, Chalmers University of Technology, Gothenburg, Sweden.
[175] Steen, B. (2006). Abiotic resource depletion, different perception of the problem with
mineral deposits. International journal of Life Cycle Assessment, special issue vol. 11
(1): 49-54.
[176] Stewart, M. and Weidema, B. (2005). A consistent framework for assessing the
impacts from resource use, a focus on resource functionality. International journal of
Life Cycle Assessment 10(4): 240-247.
[177] Stoeser, D.B. and Heran, W.D. (eds.) (2000). USGS Mineral deposit models. USGS
Digital Data Series DDS-064, U.S. Geological Survey, Denver, CO, United States.
[178] Struijs, J., van Wijnen, H.J., van Dijk, A. and Huijbregts, M.A.J. (2009a). Ozone layer
depletion. Chapter 4 in: Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., De Schryver,
A., Struijs, J., Van Zelm, R. (2009). ReCiPe 2008 A life cycle impact assessment
method which comprises harmonised category indicators at the midpoint and the
endpoint level. Report I: Characterisation factors, first edition.
[179] Struijs, J., Beusen, A., van Jaarsveld, H. and Huijbregts, M.A.J. (2009b). Aquatic
Eutrophication. Chapter 6 in: Goedkoop, M., Heijungs, R., Huijbregts, M.A.J., De
Schryver, A., Struijs, J., Van Zelm, R. (2009). ReCiPe 2008 A life cycle impact
assessment method which comprises harmonised category indicators at the midpoint
and the endpoint level. Report I: Characterisation factors, first edition.
119
4 Reference
[180] Struijs J., van Dijk A., SlaperH., van Wijnen H.J., VeldersG. J. M., Chaplin G.,
HuijbregtsM. A. J. (2010). Spatial- and Time-Explicit Human Damage Modeling of
Ozone Depleting Substances in Life Cycle Impact Assessment. Environmental
Science & Technology 44 (1): 204-209
[181] Tarrason, L., Fagerli, H., Klein, H., Simpson, D., Benedictow, A., Vestreng, V.,
Riegler, E., Emberson, L., Posch, M., Spranger T. (2006). Transboundary
acidification, eutrophication and ground level ozone in Europe from 1990 to 2004 in
support for the review of the Gothenburg protocol. Oslo, Norway, Norwegian
Meteorological Institute.
[182] Toffoletto, C. Bulle, J. Godin, C. Reid and L. Deschênes (2007). "LUCAS – A New
LCIA Method Used for a Canadian-Specific Context." International Journal of Life
Cycle Assessment 12(2): 93-102
[183] Thomas, C.D., Cameron, A., Green, R.E., Bakkenes, M., Beaumont, L.J., Collingham,
Y.C., Erasmus, B.F.N., Ferreira de Siqueira, M., Grainger, A. (2004). Extinction risk
from climate change. Nature; vol. 427 (6970), pp 145-147.
[184] Thompson, M., Ellis R., Wildavsky A. (1990). Cultural Theory, Westview Print Boulder.
[185] Tol, R.S.J. (1999). The Marginal Costs of Greenhouse Gas Emissions. Energy
Journal 20 (1), 61-81.
[186] Tørsløv, J., Hauschild, M.Z., Rasmussen, D. (2005). Ecotoxicity. In: Potting, J.,
Hauschild, M.Z. (eds.) Spatial Differentiation in Life Cycle Impact Assessment – The
EDIP2003 Methodology. Environmental News no. 80. The Danish Ministry of the
Environment, Environmental Protection Agency, Copenhagen.
[187] Udo de Haes, H.A., Finnveden, G., Goedkoop, M., Hauschild, M.Z., Hertwich, E.G.,
Hofstetter, P., Jolliet, O., Klöpffer, W., Krewitt, W., Lindeijer, E., Mueller-Wenk, R.,
Olsen, I., Pennington, D.W., Potting, J., Steen, B. (2002). Life-Cycle Impact
Assessment: Striving towards Best Practice. Society of Environmental Toxicology and
Chemistry (SETAC). ISBN 1-880661-64-6.
[188] Udo de Haes, H.A., Jolliet, O., Finnveden, G., Hauschild, M.Z., Krewitt, W., Müller-
Wenk, R.(1999). Best Available Practice Regarding Impact Categories and Category
Indicators in Life Cycle Impact Assessment. International Journal of Life Cycle
Assessment 4(2), 66-74.
[189] Uchijima, Z., Seino, H. (1985). Agroclimatic evaluation of net primary productivity of
natural vegetations, (1) Chikugo model for evaluating net primary productivity. Journal
of Agricultural Metereology, 40 (4), 343-352.
[190] UNSCEAR, 2000. Sources and effects of ionizing radiation: UNSCEAR 2000 report to
the General Assembly, with scientific annexes. Vol. 1: Sources. United Nations, New
York. p. 654. ISBN: 92-1-142238-8. Online publication at:
http://www.unscear.org/unscear/en/publications/2000_1.html.
[191] Uno, I., Wakamatsu, S., (1992). Analysis of wintertime high concentration of NO 2
using a photochemical box model. J. Jpn. Soc. Air Pollut. 27, 246–257 (Japanese).
[192] UNSCEAR, 1993. United Nations Scientific Committee on the Effects of Atomic
Radiation (UNSCEAR), editor. Sources and Effects of Ionizing Radiation. Report to
the General Assembly, with Scientific Annexes. New York: United Nations, 1993.
[193] Van de Meent, D., Huijbregts, M.A.J. (2005). Calculating life-cycle assessment effect
factors from potentially affected fraction-based ecotoxicological response functions.
Environ. Toxicol. Chem. 24: 1573-1578.
[194] Van der Voet, E. (2001); Land use in LCA. CML-SSP working paper 02.002.
120
4 Reference
[195] Van Dijk, A., Den Outer, P., Van Wijnen, H., Slaper, H. (2007). Manual of AMOUR
2.0: Assessment MOdel for Ultraviolet Radiation and Risks. RIVM Report no.: 410
200.7; Bilthoven, The Netherlands.
[196] Van Dijk, A. Den Outer, P. N.; Slaper, H.(2008) Climate and Ozone change Effects on
Ultraviolet radiation and Risks (COEUR) using and validating earth observations;
RIVM Report 61000 2001/ 2008; Bilthoven, The Netherlands, 2008.
[197] Van Jaarsveld, J.A., (1995). Modelling the long-term atmospheric behaviour of
pollutants on various spatial scales. Ph.D. Thesis, University of Utrecht, Utrecht.
[198] Van Jaarsveld, J.A., Van Pul, W.A.J., De Leeuw, F.A.A.M. (1997). Modelling transport
and deposition of persistent organic pollutants in the European region. Atmospheric
Environment 7: 1011-1024.
[199] Van Loon M., Vautard R., Schaap M., Bergstrom R., Bessagnet B., Brandt J., Builtjes
P., Christensen J.H., Cuvelier K., Graf A., Jonson J., Krol M., Langner J., Roberts P.,
Rouil L., Stern R., Tarrason L., Thunis P., Vignati E., White L.,Wind P., (2007).
Evaluation of long-term ozone simulations from seven regional air quality models and
their ensemble average, Atmos. Environ. 41, pp. 2083–2097
[200] Van Zelm R., Huijbregts M.A.J., Van Jaarsveld H.A., Reinds G.J., De Zwart D., Struijs
J., Van de Meent D. (2007a). Time horizon dependent characterisation factors for
acidification in life-cycle impact assessment based on the disappeared fraction of
plant species in European forests. Environmental Science and Technology 41(3): 922-
927.
[201] Van Zelm R., Huijbregts M.A.J., Harbers J.V., Wintersen A., Struijs J., Posthuma L.,
Van de Meent D. (2007b). Uncertainty in msPAF-based ecotoxicological freshwater
effect factors for chemicals with a non-specific mode of action in life cycle impact
assessment. Integrated Environmental Assessment and Management 3 (2): 203-210.
[202] Van Zelm, R., Huijbregts, M.A.J., Den Hollander, H.A., Van Jaarsveld, H.A., Sauter,
F.J., Struijs, J., Van Wijnen, H.J., Van de Meent, D. (2008). European characterization
factors for human health damage of PM10 and ozone in life cycle impact assessment.
Atmospheric Environment 42, 441-453.
[203] Vautard, R., Builtjes, P.J.H., Thunis, P., Cuvelier, C., Bedogni, M., Bessagnet, B.,
Honore´ , C., Moussiopoulos, N., Pirovano, G., Schaap, M., Stern, R., Tarraso´ n, L.,
Wind, P., (2007). Evaluation and intercomparison of ozone and PM10 simulations by
several chemistry transport models over four European cities within the CityDelta
project. Atmospheric Environment 41, 173–188.
[204] Wackernagel, M., Monfreda, C., Moran, D., Wermer, P., Goldfinger, S., Deumling, D.,
Murray, M., (2005). “National Footprint and Biocapacity Accounts 2005: The
underlying calculation method”. Global Footprint Network, Oakland, California, USA.
[205] Wegener Sleeswijk A., Van Oers L., Guinée J., Struijs J., Huijbregts M.A.J. (2008).
Normalisation in product life cycle assessment: An LCA of the global and European
economic systems in the year 2000. Science of the Total Environment 390 (1): 227-
240.
[206] Weihe, W.H. (1986). In "Urban climatology and its applications with special regard to
tropical areas". WMO-report No. 652.Proceedings of the technical conference, Mexico
d.F, 26-30 November 1984, , Geneva 1986,
[207] Weidema, B., Lindeijer (2001). Physical impacts of land use in product life cycle
assessment. Final report of the EURENVIRON-LCAGAPS sub-project on land use.
Department of Manufacturing Engineering and Management, Technical University of
Denmark, Lyngby.
121
4 Reference
[208] Wenzel, H., Hauschild M.Z. and Alting, L. (1997). Environmental assessment of
products. Vol. 1 - Methodology, tools and case studies in product development, 544
pp. Chapman & Hall, United Kingdom, Kluwer Academic Publishers, Hingham, MA.
USA. ISBN 0 412 80800 5.
[209] Wolff, S. (2000). Evaluation of fine particle exposures, health risks, and control
options. Doctoral thesis. Boston, MA: Department of Environmental Health, Harvard
School of Public Health.
[210] WHO (1987). Ozone. In: Air Quality Guidelines for Europe, pp 315-326. WHO
Regional Publications, European Series No 23, Copenhagen.
[211] WHO (1989). Indoor air quality: organic pollutants. EURO Reports and Studies 111,
WHO Regional Office for Europe, Copenhagen.
[212] WHO (2002). Global Burden of Disease statistics 2002. Available online at
http://www.who.int/healthinfo/bodestimates/en/index.html
[213] WHO (2004). Health aspects of air pollution—answers to follow-up questions from
CAFE. Report on a WHO working group meeting, Bonn, Germany, 15–16 January
2004.
[214] WMO (1999). Scientific Assessment of Ozone Depletion: 1998. Global Ozone
Research and Monitoring Project - Report No. 44, ISBN 92-807-1722-7, Geneva.
[215] WMO (2003). Scientific Assessment of Ozone Depletion: Global Ozone Research and
Monitoring Project – Report No. 47.
122
4 Reference
5 Annex 1 - Consistency across midpoint and

endpoint indicators
During the analysis of the different methods for addressing a variety of impact categories
by a number of authors, it is inevitable that inconsistencies are introduced, such as:
 The same criterion can be adhered to more strictly for one impact category than for
another one. For instance, the criterion “Ability for third parties to freely generate
additional, consistent factors and to further develop models” will perhaps be more
important for toxicity than for climate change.
 Authors know their own work better, and they also know it when it has not yet been
fully published. For instance, some of the methods within ReCiPe or IMPACT2002+ are
known to some authors and not to others.
 The cause-effect chains of the different impact categories sometimes show
inconsistencies. For instance, models for human toxicity do not take into account that
people will change their diet to prevent exposure to contaminants, while endpoint models
for ozone depletion are explicitly based on reduction scenarios.
Sometimes, such inconsistencies are justifiable. The need to generate extra
characterisation factors for toxicity models is large, while it is small for climate change. But
sometimes, it is not. Differences in background knowledge of the different authors have been
levelled through the internal and external review process. A central task has therefore been
to identify inconsistencies in the description and assessment of methods and impact
categories, and where needed to remove such inconsistencies. Some questions that came
up during this consistency analysis were the following:
 Do midpoint and endpoint methods have intrinsic differences in their assessment?
For instance, do all midpoint methods show less environmental relevance, and more
scientific robustness, than endpoint methods? In the analysis we have decided that the
really interesting comparison is across impact categories at midpoint and across impact
categories at endpoint. Giving all midpoint methods an E (lowest score) on relevance and
all endpoint methods an E (lowest score) on robustness would then not be discriminating
within the two subsets. Therefore, it was decided that midpoint methods can still give a
good score on environmental relevance, and endpoint methods on scientific robustness.
 The newer methods, like EDIP 2003 and ReCiPe, have an advantage over the older
ones, like Eco-indicator 99 and EPS2000, in that they build on the experiences and
models of the older ones. They stand on the shoulders of giants, so to say. But that has a
price: often not all of their elements have been published in the open literature, and the
experience with and acceptance of these methods is limited because they are still very
new. Since there is a rapid development towards better and more complete methods in
LCIA today, we have decided in the interest of the durability of the recommendations to
give priority to the newer methods in this issue, provided that we assess the underlying
science as being good based on our own reviews.
Fundamental discussions about the types of mechanisms that should ideally be included
in a model for a certain impact category have not played a role in the assessment, but have
been placed in the chapters on the Areas of Protection.
123
Annex 1: Consistency across midpoint and endpoint indicators
Apart from the analysis, also the recommendations on characterisation methods have
been checked to ensure consistency across impact categories at midpoint and endpoint
level. It is analysed whether the impact pathways, which are modelled by the recommended
characterisation models, are complementary at midpoint level and at endpoint level or
whether they present overlap or insufficient coverage of the relevant environmental
mechanisms. Detected inconsistencies are corrected if possible, or the recommendations are
modified with the aim of ensuring complementarity between the impact categories to the
extent possible.
5.1 Handling of cultural perspectives and time

perspective
Some of the recommended characterisation methods have been developed in three
versions in accordance with different mindsets or perspectives inspired from cultural theory
and reflecting the uncertainty of the data/method (Thomson, 1990, and introduced in LCA by
Hofstetter, 1998). The theory of cultural perspectives is used to manage the relatively wide
range of choices that have to be made in endpoint models. Each choice has an impact on
the characterisation factors. The Cultural theory is a social science based and widely used
method to group this wide range of choices into three consistent, but hypothetical
stakeholder perspectives, the Egalitarian, the Individualist and the Hierarchist perspective.
Each perspective can be linked to one of three fundamentally different perceptions of the
world. For example there are different views on the time perspective to be used. The
question whether future generations are equally important as present generations does not
have a wrong or right answer, but it depends on the worldview of the stakeholder who has to
decide whether an assessment is acceptable or not. Another choice is the required level of
scientific proof in order to accept that there is a problem. Another example is to what extent
we can assume that predicted damages can be partially avoided by proper management or
by technology development. The Egalitarian perspective assumes that future generations are
very important, it is pessimistic about the role of management and technology, and assumes
the precautionary principle regarding the inclusions of the effects. On the other end of the
scale, the Individualist perspectives combines the short time perspective with an optimistic
view on what market forces and innovation can prevent, while it has a low confidence in
management solutions: it assumes that only proven effects are relevant. The Hierarchist
perspective is consensus driven. It assumes that a long time perspective is relevant, and
bases itself on scientific consensus in many choices. The recommended LCIA framework
does not operate with different perspectives, and where the recommended method does, the
hierarchical perspective is recommended, while the other perspectives are used in a
sensitivity assessment. In terms of time perspective, the recommendation is to choose a time
perspective which includes the full impact, i.e. in principle an infinite time perspective
(corresponding to the hierarchical or egalitarian time perspective for those methods which
are developed with different perspectives). In the concrete case of climate change the
recommended time horizon is 100 years due to a lower uncertainty. Factors with a 500 years’
time perspective are provided but factors for a much shorter time horizon are also provided
for sensitivity analysis.
124
Annex 1: Consistency across midpoint and endpoint indicators
6 Annex 2 - Research needs

The research recommendations accompanying the model recommendations are
presented for each impact category and differentiated according to their priority, and the
estimated amount of work required is estimated.
In addition to the research needs outlined below, there is a general need to identify where
spatial and temporal variations from global defaults are statistically justifiable to warrant
distinction in practice. Furthermore, where trully global models are not available there is a
need to further investigate the importance of this discrepancy and, if necessary, default more
global models.
For all impact categories, there is equally a general need for a practical framework and
methods to identify and quantify the main sources of parameter (stochastic) and model
(systematic) uncertainty. These can then be used to further define where developments are
essential. These methods must be compatible with approaches used in the inventory phase
of LCA.
6.1 Climate change

For the midpoint method, based on IPCC there are no further research recommendations
for e.g. the LCIA community, only suggesting to follow the updates provided (on average
every 4 years). For endpoint methods the following research issues are identified:
High priority
 Improve the link between the emission of CO2 equivalent emissions and the expected
temperature increase. A relatively low effort task could be to review climate models,
similar to the study of Meinshausen (2005) used in ReCiPe (De Schryver and
Goedkoop, 2009a), that allow for establishing this link and for further interpreting the
causes if these models have a significant difference in results.
 Improve and update the link between temperature increase and ecosystem damage,
which are still being debated in the scientific literature. A relatively low effort task
would be a further literature study on models that predict ecosystem damages. A
clear point of attention is in a number of assumptions on if species can migrate, and if
so how fast, and on which scientific basis we can indeed make a link between
temperature increase and habitat alteration (Thomas et al., 2004)
 Extend the environmental mechanisms in the LCIA method, to cover other pathways
than via temperature increases, such as humidity increases and rainfall.
 Improve and further develop the link between climate change and change in biomass
and food productivity, which is still being debated. Within this field the scientific work
performed is limited.
 Improve and extend the range of linkages between temperature increase and
expected human health damages (like emerging and re-emerging diseases) such as
done in WHO assessments (McMichael, 2003). An important issue is to clearly
distinguish different adaptation scenarios, possible policy measures and potential
125
Annex 2: Research needs
technological breakthroughs, as these all have significant effect. Also here a further
review of existing studies would be a first priority, and will need a relatively low effort.
 Develop a procedure to deal with the different sets of assumptions on adaptation and
other responses, similarly to, but better than the perspectives used in ReCiPe, based
on cultural theory (Thompson et al., 1990)
 Consider heating related stress not only for human health but also for impact to
ecosystem
The main recommendation for the long term research is to develop links to ongoing
climate change impact research centres, and request to provide up to date data that reflects
the effect of releasing a kilo of CO2 equivalents, instead of assessing entire energy
scenarios.
6.2 Ozone depletion

High priority
 Develop a better understanding of whether it is desirable to base the fate model on
the sharply declining emission scenarios.
 Develop a better understanding of the way the damage to ecosystems (vegetation,
marine plankton in Arctic and Antarctic regions) can be incorporated. A first step
would be to further study (and translate) the LIME method, and to further investigate
other publications that link stratospheric ozone layer thickness to damages on
ecosystems and crops. (e.g. Ravishankara et al.2009)
6.3 Human toxicological effects

The following research recommendations are relevant for the further development of
human toxicological effect methods (models and factors):
High priority
 Analysis of uncertainty and the provision of practical guidance for straightforward use
in LCIA/LCA based e.g. on clusters of similar chemicals
 Development of straightforward methods to fill data gaps based e.g. on similarity of
physical/chemical parameters
 Analysis of spatial and temporal distinctions necessary to reduce uncertainty based
e.g. on clusters of similar chemicals and emission scenarios
 Development of compatible factors to accommodate indoor air emissions including in
the work environment (large amount of work)
 Development of compatible factors to better model e.g. metals and their speciation in
fate, exposure and effects (large amount of work)
 Development of compatible factors for pesticides and other agricultural improving
chemicals focusing on exposure associated with crop residues (medium amount of
work)
126
Medium priority
 Identification of when spatial and possibly temporal differentiation is important.
 Improvement of human exposure models (high amount of work, long-term research)
Low priority
 Inclusion of dermal route of exposure. With the possible exception of e.g. cosmetics
and some nanoparticles, this is however likely to be of limited importance for the
intake fraction according to Hertwich et al.(2001) (low amount of work).
 Improvement of endpoint modelling for non-cancer effects (medium amount of work,
long-term research)
6.4 Particulate matter/Respiratory Inorganic

High priority
 Spatial differentiation of fate and exposure needs to be improved to capture
emissions in other types of environment (oceans, etc.) (medium workload).
 Differentiate effect factors between PM depending on the source and size distribution
(diesel, gasoline, coal, etc.) (high workload)
 Include NH4+ as secondary aerosol coming from NH3 emissions. For agricultural
practice, NH3 emissions and health effects coming from NH4+ secondary aerosols
can be very relevant. On the short term, the possibility to have an Interim factor for
NH3 extrapolated from van Zelm et al. (2008) will be explored.
 Modelling of effect factors needs to consider surface and number instead of solely
mass as a proxy for adverse health effects (medium workload).
Evaluate whether ‘chronic bronchitis (adults)’ is an important endpoint and what is the
severity factor associated to ‘chronic bronchitis (adults)’.
Medium priority
 Modelling of fate and exposure needs to consider the evolution of Particle Size
Distribution (medium workload).
 Modelling of CF for sizes smaller than 2.5 μm (medium workload).
 Investigate the possibility to use the EcoSense model to cover respiratory inorganic
impact in LCA in a consistent way with other criteria pollutants (medium workload)
6.5 Ionizing radiation

High priority
 Ensure a better compatibility of the fate and exposure model developed for the
ionising radiation impact category with USEtox for both human and ecotoxicity, for
similar population densities and ecosystem characteristics (low workload)
 Cover Radon for indoor emission in a compatible way to outdoor exposure
(medium workload)
127
Extend the number of radionuclides covered for both human health and

ecosystems (high workload)
Medium priority
 Update of the 1990 DALY to latest WHO 2002 statistics, ensuring a full
consistency with the treatment of cancers used for the human toxicity category
(with average of 11.5 DALY per case of cancer against 17 DALY/case as reported
by Frischknecht et al., 2000). (low workload)
 Include the marine and terrestrial environment for the ecosystem damage
assessment (high workload)
 Consider the radioactive waste disposal in underground facilities (disregarded in
Frishknecht et al, 2000 due to lack of data availability) (high workload)
Low priority
 Further develop the endpoint modelling for ecosystems, in conjunction with
ecosystem impacts, including long-term genomic research to study genomic
instabilities and indirect effect of radiation (high workload)
Methods/ research needs highlighted during the public consultation
 A different approach to handling the calculation of characterisation factors was
suggested and needs to be further assessed. Using latest UNSCEAR reports for
the assessment of human exposure towards ionising radiation (as specified by
Preiss and Klotz 2008).
 Using latest (ICRP, 2007) recommendations regarding risk factors for impact
assessment.
 Note that no exposure factor (i.e., collective dose) is provided for C-14
releases into freshwater, while those into soil and surface ocean are stated to be
about the same as those into air (UNSCEAR, 2000 paragraphs 233-236). This
could be due to a lack of the global C model that does not distinguish a freshwater
compartment and can therefore not assess corresponding releases. But there is
no reason to believe that releases into freshwater are irrelevant if releases into soil
and surface oceans should lead to the same exposure.
 The severity factors as given in Frischknecht et al. (2000) could be used in
combination with the exposure factor and the effect factor from Preiss & Klotz
(2008), even though the DALYs are dated.
6.6 Photochemical ozone formation

High priority
 As a research recommendation, ReCiPe might be expanded at endpoint to include
vegetation impacts by combining the LOTOS-EUROS model with information of
vegetation distribution and sensitivities in Europe (high amount of work).
128
 Investigate the current updates of the EcoSense model and their ability to provide
both geographically differentiated and site-generic characterisation factors addressing
both human health impacts and crop impacts from photochemical ozone formation
 A damage model for vegetation might also be based on EDIP2003 midpoint results,
which would require linkage of the present time and area integrated exposure above
a threshold for vegetation effects to damage on the AOP Natural environment (low
amount of work)
 The decision to only include acute mortality effects of ozone in modelling should be
further consolidated. In essence it means that the human health issue of
photochemical ozone is a local smog issue in heavily populated/industrialised areas
since it is unlikely that regional ozone concentrations should reach a level where
acute effects may be observed. This distinguishes the human health impacts from the
vegetation impacts which are indeed of a regional character (low amount of work)
 The assumption of linear dose-response curve for human health effects without any
threshold as applied in the LOTOS-EUROS should be further consolidated.
Moreover, the application of SOMO35 , i.e. including threshold as implemented in
EcoSense, should also be further consolidated (low amount of work)
 Midpoint characterisation factors for CO and CH4 should be calculated with the
LOTOS-EUROS and EcoSense model (low amount of work)
The importance of human health damages from photochemical ozone formation
compared to damage from particulate matter/respiratory inorganics should be
investigated. Preliminary results based on calculations of normalisation references for
the two impact categories suggest several orders of magnitude lower damage from
photochemical ozone, which would argue that photochemical ozone formation is
primarily of concern due to its damage to vegetation (low amount of work)
Medium priority
 The recommended models should be adapted to other continents by modifying their
fate, exposure and effect models or replacing them with similar models already
developed for these continents. (high amount of work)
6.7 Acidification
High priority:
 Provide/derive CF factors explicitly for SO3 and NO and NO2 (low to medium amount
of work)
 Determine a set of consistent global default or continental/regional CFs at midpoint
level using the Seppälä et al. (2006) method by (medium amount of work):
- Calculating a set of consistent regional/continental emission-deposition fate
factors, which could ideally be calculated from a global atmospheric fate and
transport model
- Mapping (changes) in sensitive area at the global scale, expert judgment
could also be applied
129
- Calculating a set of characterization factors for each region/continent, by

connecting global emission-deposition matrices with the global sensitive area
information
 Quantify uncertainties: model uncertainties and variability (low amount of work):
- between individual emission countries
- between different regions/continents in respect to a global generic factor
- as a function of emission time and changes in current emission level.
Medium priority
 Determine a clear link between the midpoint Accumulated Exceedance calculations
and the endpoint assessment, by building a scientific consensus on the dose-
response model proposed by van Zelm and colleagues (2007) (high amount of work)
and particularly:
- Seeking for independent confirmation whether base saturation is a suitable
starting point instead currently used rapidly changing variables (pH, Al
concentrations, etc.)
- Modelling the changes of key soil parameters (base saturation or pH) on a
global scale for different soil types
- Expanding the effect model approach, based on the dose-response curve of
European forest, to various ecosystem types
- Calculating a set of regional/continental characterization factors by connecting
atmospheric soil and fate calculations with a regionalized effect model
 Investigate the need to address the waterborne emissions of acidifying substances
within the proposed framework and calculate CFs (high amount of work)
 Ocean acidification via CO2 immission from atmospheric point of view is getting
prominent in the climate change community + what comes in via the rivers (high
amount of work)
 Investigate the aspect of neglecting deposition in areas below the critical load in an
LCA perspective (high amount of work)
 Investigate effect of acidification damage on climate change for endpoint
characterisation (high amount of work)
 Develop characterisation factors for relevant strong and weak acids, particularly HCl
and acetic and formic acid (low to medium amount of work)
6.8 Eutrophication
High priority
 Development of a model for LCIA purposes that is capable of evaluating terrestrial,
freshwater and marine fate and effects in an integrated and consistent way and
applicable to a regional as well as a global scale, building on elements in the
analysed approaches (high amount of work)
130
 Investigate damage approach proposed by Payet, 2006 for freshwater systems and
compare to approach applied in ReCiPe to develop a damage model for freshwater
systems based on a broader European data base (high amount of work)
 Compare fate models from TRACI, LIME and ReCiPe midpoint models to determine
differences and derive simplifications and generalisations which may be applied in a
global midpoint characterisation model for aquatic eutrophication (medium amount of
work)
Develop damage model for marine eutrophication linking increase in nutrient
concentrations at midpoint level to damage to biodiversity in marine ecosystems at
regional and global levels (high amount of work)
Medium priority
 Quantify uncertainties: model uncertainties and variability (low amount of work)
‐ between individual emission countries
‐ between different regions/continents in respect to a global generic factor
‐ as a function of emission time and changes in current emission level
6.9 Ecotoxicological effects

High priority
 Provide framework and methods for assessing uncertainty (see also human
toxicological effects).
 Further develop approach for addressing data gaps (see also human toxicological
effects).
 Guidance for when temporal and spatial distinctions are necessary (see also human
toxicological effects)
 Inclusion of bioavailability, where not implicitly included in toxicity data, considering

also changes in future availability due to long-term geochemical and geological
processes in the soil (see e.g. Ligthart et al 2010)
 Consider biomagnification and its relative importance
 Take into account indirect ecological effects of chemicals via food web changes
 Inclusion of the internal critical body burden concept in LCIA for ecotoxicity to support
better inter-substance interpolation
 Further evaluation of model calculations with field data on changes in species
diversity and development of endpoint methods
Medium priority
 Further develop terrestrial fate and ecotoxicity effect models and marine ecotoxicity
effect models that can have an application in LCIA (high effort and long term
research)
131
 Inclusion of fate and effects of metabolites, assuming not considered in the

toxicological effects data
 Factors for metal, in particular Zinc (Ligthart et al, 2010)
6.10 Land use

For land use midpoint characterisation the following research issues are identified:
High priority
 Immature methods (Milà i Canals, Baitz) should be extensively tested (high amount of
work)
 Characterisation factors should be developed, based on existing models (high
amount of work)
An important input to this research can come from a newly established working group on
land use42
For land use endpoint characterisation the following research issues are identified:
High priority
 Include regionalized characterisation factors (worldwide), based on more input data
and other reference land use types. Special attention should be paid to developing
countries (high amount of work)
 Continue the work on soil quality impacts, transformed to the correct unit (high
amount of work)
 Include effects on climate change, transformed to the correct unit. (high amount of
work)
 Immature methods (Recipe) should be extensively tested (low amount of work)
Medium priority
 Implement uncertainty data and distribution (low amount of work)
 Include impacts on primary production, transformed to the correct unit (high amount
of work)
 Include more relevant land use types (high amount of work)
Implement elements and available data from the work of Köllner 2008 should be
considered and used in further developments of land use models (low amount of
work)
Low priority
 Implement clear description on uncertainty in the data (low amount of work)
 Let the work be reviewed (low amount of work)
42
http://fr1.estis.net/builder/includes/page.asp?site=lcinit&page_id=337831BE-0C0A-4DC9-AEE5-
9DECD1F082D8
132
6.11 Resource depletion

High priority
 More clarity on the AoP resources, on for example which impacts it covers needs to be
developed (high amount of work)
 Discussions with stakeholders and experts are needed on what is actually important
and what does society want to protect (including the short-term vs. long-term resource
depletion) (high amount of work)
 Development of a model to assess biotic resource and species depletion at the
midpoint and endpoint level (high amount of work)
 Investigate and model the effects of resource depletion (water depletion, biotic
resources) in developing countries. The model on water depletion, considered in
Swiss Ecoscarcity 2007, is a good starting point (high amount of work)
Medium priority
 Have a critical review of the ReCiPe approach and develop a programme to further
improve the main weaknesses. This would involve contributions from geologists, and
economists (high amount of work)
 Assessment of water consumption/ freshwatere usage and the related impacts (high
amount of work)
 Investigate if the use to stock ratio can be developed with other references than
ultimate reserves. This reference may not be the most environmental relevant. In this
sense we call for a combination of strength of EDIP and CML at midpoint level (high
amount of work)
 For the exergy method, investigation is needed on the acceptability of major
conceptual choices used. For example, the assumption that all exergy losses are
equally relevant (low amount of work)

 For water depletion, Mila I Canals et al.2009 and Pfister et al.2009 were
suggested to be considered.
 It should also be investigated how the environmental damage cost for water
abstraction has been derived. This could be an alternate way for endpoint for
freshwater usage.
6.12 Other impacts

The other impact categories (noise, accidents, desiccation, erosion, and salination) are
definitely important categories, and they deserve attention in future LCA developments. In
particular for specific LCA studies (e.g., comparing different transport means, comparing
different agricultural practices) differences in impacts on noise, erosion, desiccation, etc. may
be critical. There is, however, no generally applicable model for these impact categories.
133
Development of the cause-effect framework and models to address the associated midpoint
or endpoint categories need to be developed.
The table below gives a first opinion on the research priority and an estimate of the
research effort for the five different impact categories that have been analysed.
In addition, the research efforts for desiccation and salination, and probably also erosion,
should be combined effort, probably merged under the impact category of land use impacts.
Finally, an open mind towards additional missing impact categories is needed. Especially for
activities going on in developing countries, the OECD-based impacts might be insufficient.
Table 29 Research needs for impacts covered under “other impacts”
Impact Priority amount of work
Noise high medium
Accidents low low
Desiccation high high
Erosion high high
Salination high high

 Methods that accounts for dissecation, like Pfister et al.2009
 Heat stress has only been addressed toward human health (in climate change), while
it is clearly affecting ecosystems (e.g. in cooling water releases), (Verones et al.
2010).
6.13 Ecological footprint

High priority
 How should a score based on the footprint method be interpreted in an LCA
context
 How to deal with double counting on demand for land for carbon sequestration,
timber production and biodiversity
 How to combine with other impact categories
Low priority
 Include more land use types
 Extrapolate the effects of carbon dioxide to other emissions
134
7 Annex 3 – Development of this document

Based on and considering the following documents
The background document has been drafted taking into account amongst others the following
existing sources:
 Harmonised ISO standards
- ISO 14040: 2006 Environmental management - Life cycle assessment – Principles and
framework
- ISO 14044: 2006 Environmental management - Life cycle assessment - Requirements and
guidelines
 Guidance documents in the field of Life Cycle Impact Assessment (LCIA)
This ILCD Handbook builds on two previous impact assessment documents:
ILCD - Handbook Analysis of existing Environmental Impact Assessment methodologies for
use in Life Cycle Assessment (EC-JRC 2010a)
ILCD – Handbook Framework and Requirements for LCIA models and indicators (EC-JRC,
2010b)
The methods, analyses and recommendations in this guidance document build on existing
methods and achievements made in the scientific communities, including the voluntary achievements
of task forces of the Society of Environmental Toxicology and Chemistry (SETAC) and more recently
in the joint Life Cycle Initiative facilitated by the United Nations Environment Program (UNEP) with
SETAC. We equally acknowledge the US Environmental Protection Agency (EPA) for providing
workshop documentation and other documents related to the scope and framework of LCIA.
A wealth of information and publications on the LCIA framework, methodologies and methods has
been taken into account as referenced in the document.
Drafting
This document was initially drafted by contractors (see list below) with support under the European
Commission Joint Research Centre (JRC) contract no. contract no.383163 F1SC concerning
“Definition of recommended Life Cycle Impact Assessment (LCIA) framework, methods and factors”.
This work has been funded by the European Commission, partially supported through Commission-
internal Administrative Arrangements (Nos 070402/2005/414023/G4, 070402/2006/443456/G4,
070307/2007/474521/G4, and 070307/2008/513489/G4) between DG Environment and the Joint
Research Centre.
Invited stakeholder consultations

An earlier draft version of this document has been distributed to more than 60 organisations and
groups, covering EU Member States, European Commission (EC) Services, National Life Cycle
Database Initiatives outside the European Union, business associations as members of the Business
Advisory Group, Life Cycle Assessment software and database developers and Life Cycle Impact
Assessment method developers as members of the respective Advisory Groups, as well as other
relevant institutions.
Disclaimer: Involvement in the development or consultation process does not imply an agreement
with or endorsement of this document.
135
Annex 3: Development of this document

Public consultation
A public consultation was carried out on the advance draft guidance document from October 18,
th
2010 to November 26, 2010. This included a public consultation workshop, which took place on 27
October 2010, in Brussels.
Overview of involved or consulted organisations and individuals

The following organisations and individuals have been consulted or provided comments, inputs and
feedback during the invited or public consultations in the development of this document:
Invited consultation
Internal EU steering committee

 European Commission services (EC),
 European Environment Agency (EEA),
 European Committee for Standardization (CEN),
 IPP representatives of the 27 EU Member States
National LCA database projects and international organisations:

 United Nations Environment Programme, DTIE Department (UNEP-DTIE)
 World Business Council for Sustainable Development (WBCSD)
 Brazilian Institute for Informatics in Science and Technology (IBICT)
 University of Brasilia (UnB)
 China National Institute for Standardization (CNIS)
 Sichuan University, Chengdu, China
 Japan Environmental Management Association for Industry (JEMAI)
 Research Center for Life Cycle Assessment (AIST), Japan
 SIRIM-Berhad, Malaysia
 National Metal and Material Technology Center (MTEC), Focus Center on Life Cycle
Assessment and EcoProduct Development, Thailand
Advisory group members
Business advisory group

 Alliance for Beverage Cartons and the Environment (ACE), Europe
 Association of Plastics Manufacturers (PlasticsEurope)
 Confederation of European Waste-to-Energy plants (CEWEP)
 European Aluminium Association
 European Automobile Manufacturers' Association (ACEA)
 European Cement Association (CEMBUREAU)
 European Confederation of Iron and Steel Industries (EUROFER)
 European Copper Institute
 European Confederation of woodworking industries (CEI-Bois)
136

 European Federation of Corrugated Board Manufacturers (FEFCO)

 Industrial Minerals Association Europe (IMA Europe)
 Lead Development Association International (LDAI), global
 Sustainable Landfill Foundation (SLF), Europe
 The Voice of the European Gypsum Industry (EUROGYPSUM)
 Tiles and Bricks of Europe (TBE)
 Technical Association of the European Natural Gas Industry (Marcogaz)
LCA database and tool developers’ advisory group

 BRE Building Research Establishment Ltd - Watford (United Kingdom)
 CML Institute of Environmental Science, University of Leiden (The Netherlands)
 CODDE Conception, Developement Durable, Environnement – Paris (France)
 ecoinvent centre – (Switzerland)
 ENEA – Bologna (Italy)
 Forschungszentrum Karlsruhe GmbH - Eggenstein-Leopoldshafen (Germany)
 Green Delta TC GmbH – Berlin (Germany)
 Ifu Institut für Umweltinformatik GmbH – Hamburg (Germany)
 IVL Swedish Environmental Research Institute – Stockholm (Sweden)
 KCL Oy Keskuslaboratorio-Centrallaboratorium Ab – Espoo (Finland)
 LBP, University Stuttgart (Germany)
 LCA Center Denmark c/o FORCE Technology – Lyngby (Denmark)
 LEGEP Software GmbH - Dachau (Germany)
 PE International GmbH – Leinfelden-Echterdingen (Germany)
 PRé Consultants – Amersfoort (The Netherlands)
 Wuppertal Institut für Klima, Umwelt, Energie GmbH – Wuppertal (Germany)
Life Cycle Impact Assessment method developers advisory group

 CIRAIG – Montreal (Canada)
 CML Institute of Environmental Science, University of Leiden (The Netherlands)
 Ecointesys Life Cycle Systems - Lausanne (Switzerland)
 IVL Swedish Environmental Research Institute – Stockholm (Sweden)
 PRé Consultants – Amersfoort (The Netherlands)
 LCA Center Denmark – Lyngby (Denmark)
 Musashi Institute of Technology
 Research Center for Life Cycle Assessment (AIST) (Japan)
137

Public consultation
Contributors providing written feedback in the public consultation
SURNAME Name Name of organisation/ institution
ANDRAE Anders Huawei Technologies Sweden AB

ASKHAM Cecilia Ostfold rresearch
BACHMANN Till EIFER
BENGT Davidsonn CEPI aisbl- Confederation of Europena
Paper Industries
DE BEAUFORT- Angeline FEFCO
LANGEVELD
DOKA Gabor Doka Life Cycle Assessment, Zurich
DRIELSMA Johannes Euromines
GALEANO Sergio Georgia-Pacific LLC
GLORIA Thomas Industrial ecology consultant
HAGLIND Ingrid Swedish Forest Industry Association
JOUNGBLUTH Niels ESU services Ltd
KUNZE Peter ACEA - European Automobile
Manufacturers Association
O’BANION Brian Firebox Association
PFISTER Stephan ETH Zurich
PREISS Philipp Institut für Energiewirtschaft und Rationelle
Energieanwendung, IER, Uni-Stuttgart
RICKLY Jean Pierre JPR concept
RIVET Fabrice FEVE - The European Container Glass
Association
SINDEN Graham The Carbon Trust
BENGT Steen Centre of environmental Assessment of
Prodcuts and Material Systems(CPM)
Chalmenrs University of Technology
UIHLEIN Andrea SCA Hygiene products Gmbh
VAN OVERBEKE Elisabeth RDC Environment
VIGHI Marco University of Study Milano Bicocca
Participating in public consultation workshops (registered participants):
SURNAME Name Name of organisation/ institution
AGOSTINI Alessandro EC-JRC-IE

ALEXANDRE Céline RDC Environment
ALLACKER Karen K.U.Leuven - Dept. ASRO
AMANI Pegah Department of food and resource
economics, University of Bonn
ANDRAE Anders Huawei Technologies Sweden AB
BACHMANN Till EIFER
BAUER Christian EAFA
138

BENETTO Enrico Centre de Recherche Public Henri Tudor
BOLLEN Jan ArcelorMittal
BOOGMAN Philipp Österreichisches Institut für Baubiologie und

-ökologie
BOONEN Katrien VITO
BOUREIMA Faycal VUB
BROADBENT Clare worldsteel
CASTELAN Guy PlasticsEurope
CHASTANET Pierre European Commission INFSO
CHRISTOFFERSEN Anita Norwegian Ministry of Fisheries and Coastal
Affairs
DANILA Ana Maria European Aluminium Association
DAVIDSSON Bengt CEPI
DE BEAUFORT- Angeline FEFCO
LANGEVELD
DE LATHAUWER Dieter Federal public service health, foodchain
safety and environment
DELGADO Clara Gaiker
DUTILH Chris Unilever Benelux
ERVIN Jon Leon Norvegian Coastal Adm

FRANKLIN Emma BRE
GALATOLA Michele DG ENV

GALEANO Sergio Georgia-Pacific LLC
GAZULLA SANTOS Cristina GiGa (ESCI-UPF)

GEORGALI Stella CEMBUREAU
GIROT Caroline SGS - CTS
GNANSOUNOU Edgard Swiss Federal Institute of Technology
Lausanne (EPFL)
GRILLI Franco Ferrero
GUEDEMANN Thomas AkzoNobel
HANSEN Klara
HAUSCHILD Michael Technical University of Denmark
HOHENTHAL Catharina VTT
KARAYIANNI- Joao Euromines
VASCONCELOS
KLIMEK Wojciech EC - DG RTD - I3
KOSKIMÄKI Asko VTT
KRISKVOLL Gury Sintef
KUNZE Peter ACEA - European Automobile
Manufacturers Association
LANGSHOLT Camilla Mission of Norway
LEROY Christian European Aluminium Association
LEVY Monique DG ENV
139

LINDFORS Lars-Gunnar Swedish Environmental Research Institute

LIU Ran Oeko-Institut e.V.
MARECHAL Freddy BCF Consult
MARTIN Stéphanie PSA Peugeot Citroën
MCKONE Thomas University of California
MENGER-KRUG Eve Fraunhofer ISI
MESSAGIE Maarten Vrije Universiteit Brussel
MUELLER Katrin Siemens AG
NANUKUTTAN Sreejith Queen's University Belfast
OLENJNIK Marta CEWEP
ORGELET Julie Bureau veritas CODDE
PAVANELLO Romeo Ambiente Italia srl
PIEROBON Marianna BASF
PREISS Philipp Institut für Energiewirtschaft und Rationelle
Energieanwendung, IER, Uni-Stuttgart
REINHARDT Guido Ifeu Heidelberg
RETHORE Olivier ADEME
RIEDL Karsten EON
RIVET Fabrice FEVE - The European Container Glass
Association
ROJOT Camille PricewaterhouseCoopers
SARRIEU Gael AIRBUS
SCHARFF Heijo Sustainable Landfill Foundation
SLAVIC Alexis
SOENEN Bram Federal public service health, foodchain
safety and environment
SUNDSETH Kyrre NILU- Norwegian Institute for Air Research
TOJAL Veronica Glass For Europe
VAN DE LINDE Janine Hunter Douglas Europe BV
VAN DER MEIJDEN Martin Hunter Douglas Europe BV
VAN OVERBEKE Elisabeth RDC Environment
VAXELAIRE Stéphane BRGM
VEIGA BENESCH Johan DG RTD
VERMEIREN Keen Solvay s.a.
VIEIRA Marisa PRé Consultants
VIOT Jean-François RHODIA
VISSER René Tata Steel b.v.
WALK Wolfgang Volkswagen AG
WEBERSBERGER Sheryl German Glass Federation (BV Glas)
ZOGHAÏB Jad Solinnen
140

Contractors as members of the initial drafting team

 Michael Hauschild, DTU and LCA Center Denmark
 Mark Goedkoop, PRé consultants, Netherlands
 Jeroen Guinée, CML, Netherlands
 Reinout Heijungs, CML, Netherlands
 Mark Huijbregts, Radboud University, Netherlands
 Olivier Jolliet, Ecointesys-Life Cycle Systems, Switzerland
 Manuele Margni, Ecointesys-Life Cycle Systems, Switzerland
 An De Schryver, PRé consultants, Netherlands
Coordinators and contributors from the Joint Research Centre (JRC, IES)
 Rana Pant (project coordinator)
 Serenella Sala (project coordinator)
 Raffaella Bersani
 Marc-Andree Wolf
 Miguel Brandão
 David W. Pennington
141

142
European Commission
EUR 24571 EN – Joint Research Centre – Institute for Environment and Sustainability
Title: International Reference Life Cycle Data System (ILCD) Handbook- General guide for Life Cycle
Assessment- Provisions and action steps
Author(s): -
Luxembourg: Publications Office of the European Union
2011 – 142 pp. – 21.0 x 29.7 cm
EUR – Scientific and Technical Research series – ISSN 1018-5593
ISBN 978-92-79-17451-3
doi:10.2788/33030
Cite as: European Commission-Joint Research Centre - Institute for Environment and Sustainability:
International Reference Life Cycle Data System (ILCD) Handbook- Recommendations for Life Cycle Impact
Assessment in the European context. First edition November 2011. EUR 24571 EN. Luxemburg. Publications
Office of the European Union; 2011
Abstract
Life Cycle Thinking (LCT) and Life Cycle Assessment (LCA) are the scientific approaches behind moder
environmental policies and business decision support related to Sustainable Production and Consumption
(SCP). The International Reference Life Cycle Data System (ILCD) Handbook provides governments and
businesses with a basis for assuring quality and consistency of life cycle data, methods and assessments. This
guidance document provides recommendations on models and characterisation factors that should be used in
Life Cycle Impact Assessement (LCIA) to analyse the emissions into air, water and soil, as well as the natural
resources consumed in terms of their contributions to different impacts on human health, natural environment,
and availability of resources. It supports the calculation of indicators for different impacts such as climate
change, ozone depletion, photochemical ozone formation, respiratory inorganics, ionising radiation, acidification,
eutrophication, human toxicity, ecotoxicity, land use and resource depletion for use in a common integrated
framework, such as LCA. The principle target audience for this document is the Life Cycle Impact Assessment
(LCIA) expert but also the experienced LCA practitioner and decision makers that are interested in the Impact
Assessment models and indicators used in LCA.
143
How to obtain EU publications
Our priced publications are available from EU Bookshop (http://bookshop.europa.eu), where you
can place an order with the sales agent of your choice.
The Publications Office has a worldwide network of sales agents. You can obtain their contact
details by sending a fax to (352) 29 29-42758.
The mission of the JRC is to provide customer-driven scientific and technical support
LB-NA-24571-EN-C
for the conception, development, implementation and monitoring of EU policies. As a
service of the European Commission, the JRC functions as a reference centre of
science and technology for the Union. Close to the policy-making process, it serves
the common interest of the Member States, while being independent of special
interests, whether private or national.

International Reference Life Cycle Data System Ilcd-Gp - Eudor - WEB - LBNA24571ENC - 002

Uploaded by

Copyright:

Available Formats

International Reference Life Cycle Data System Ilcd-Gp - Eudor - WEB - LBNA24571ENC - 002

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

International Reference Life Cycle Data System Ilcd-Gp - Eudor - WEB - LBNA24571ENC - 002

Uploaded by

Copyright:

Available Formats

Input products

International Reference Life Cycle Data System

Waste Waste Material

Recommendations for Life Cycle Impact Recycling

Assessment in the European context Extraction 2

- based on existing environmental impact assessment models and factors

Address: Via E. Fermi, 2749– 21027 Ispra (VA) - Italy

Europe Direct is a service to help you find answers

© European Union, 2011

About Life Cycle Impact Assessment (LCIA)

Role of this Guidance Document within the ILCD Handbook

Approach and key issues addressed in this document

EXECUTIVE SUMMARY .................................................................................................................. IV

3.4.8 Calculation principles ........................................................................................ 41

3.10.6 Consistency between midpoint and endpoint methods ...................................... 92

Table 1 Recommended methods and their classification at midpoint .......................................... 5

AG Advisory Group of the European Platform on LCA

Characterisation Throughout this document an “LCIA methodology” refers to a

Classification A step of Impact assessment, in which environmental interventions are

 Consensus on the need to integrate midpoint and endpoint3 models in a consistent

See the glossary.

see http://lct.jrc.ec.europa.eu/assessment/partners for details

4. Characterisation factors of the recommended methods, accessible as spreadsheet via

1.1 Summary of Recommended Methods

Table 1 Recommended methods and their classification at midpoint

Impact category Recommended default LCIA method Indicator Classific

Ozone depletion Steady-state ODPs 1999 as in WMO Ozone Depletion Potential I

Ionising radiation, No methods recommended Interim

Photochemical ozone LOTOS-EUROS (Van Zelm et al, 2008) Tropospheric ozone II

Acidification Accumulated Exceedance (Seppälä et al. Accumulated Exceedance II

Eutrophication, Accumulated Exceedance (Seppälä et al. Accumulated Exceedance II

Ecotoxicity (terrestrial No methods recommended

Resource depletion, CML 2002 (Guinée et al., 2002) Scarcity II

Table 2 Recommended models and their classification at endpoint

Recommendation from midpoint to endpoint

Impact category Recommended default LCIA method Indicator Classific

Climate change No methods recommended interim

Ozone depletion No methods recommended interim

Human toxicity, non- No methods recommended interim

Ionising radiation, No methods recommended interim

Ionising radiation, No methods recommended

Acidification No methods recommended interim

Eutrophication, No methods recommended

Eutrophication, No methods recommended interim

Ecotoxicity No methods recommended

Land use No methods recommended interim

Resource depletion, No methods recommended

Resource depletion, No methods recommended interim

2 Recommended methods for impact categories

2.1 Procedure for analysis and classification of methods

2.1.1 Application of criteria and sub criteria

Stakeholder acceptance criterion

2.2 Recommendation levels

Level I: Recommended and satisfactory

2.3 Research needs

2.4 Geographic scope

3 Background information on the evaluation of

3.1 Climate change

3.1.2 Pre-selection of methods for further evaluation