Geosci. Model Dev.

, 9, 1937–1958, 2016
www.geosci-model-dev.net/9/1937/2016/
doi:10.5194/gmd-9-1937-2016
© Author(s) 2016. CC Attribution 3.0 License.

Overview of the Coupled Model Intercomparison Project Phase 6

(CMIP6) experimental design and organization
Veronika Eyring1 , Sandrine Bony2 , Gerald A. Meehl3 , Catherine A. Senior4 , Bjorn Stevens5 , Ronald J. Stouffer6 , and
Karl E. Taylor7
1 Deutsches Zentrum für Luft- und Raumfahrt (DLR), Institut für Physik der Atmosphäre, Oberpfaffenhofen, Germany
2 Laboratoirede Météorologie Dynamique, Institut Pierre Simon Laplace (LMD/IPSL), CNRS,
Université Pierre et Marie Curie, Paris, France
3 National Center for Atmospheric Research (NCAR), Boulder, CO, USA
4 Met Office Hadley Centre, Exeter, UK
5 Max-Planck-Institute for Meteorology, Hamburg, Germany
6 Geophysical Fluid Dynamics Laboratory/NOAA, Princeton, NJ, USA
7 Program for Climate Model Diagnosis and Intercomparison (PCMDI),

Lawrence Livermore National Laboratory, Livermore, CA, USA

Correspondence to: Veronika Eyring ([email protected])

Received: 3 December 2015 – Published in Geosci. Model Dev. Discuss.: 14 December 2015
Revised: 15 April 2016 – Accepted: 27 April 2016 – Published: 26 May 2016

Abstract. By coordinating the design and distribution of data standards, will be the entry cards for models participat-
global climate model simulations of the past, current, and ing in CMIP. Participation in CMIP6-Endorsed MIPs by in-
future climate, the Coupled Model Intercomparison Project dividual modelling groups will be at their own discretion and
(CMIP) has become one of the foundational elements of will depend on their scientific interests and priorities. With
climate science. However, the need to address an ever- the Grand Science Challenges of the World Climate Research
expanding range of scientific questions arising from more Programme (WCRP) as its scientific backdrop, CMIP6 will
and more research communities has made it necessary to re- address three broad questions:
vise the organization of CMIP. After a long and wide com-
munity consultation, a new and more federated structure has – How does the Earth system respond to forcing?
been put in place. It consists of three major elements: (1) a – What are the origins and consequences of systematic
handful of common experiments, the DECK (Diagnostic, model biases?
Evaluation and Characterization of Klima) and CMIP his-
torical simulations (1850–near present) that will maintain – How can we assess future climate changes given inter-
continuity and help document basic characteristics of mod- nal climate variability, predictability, and uncertainties
els across different phases of CMIP; (2) common standards, in scenarios?
coordination, infrastructure, and documentation that will fa-
cilitate the distribution of model outputs and the characteriza- This CMIP6 overview paper presents the background and ra-
tion of the model ensemble; and (3) an ensemble of CMIP- tionale for the new structure of CMIP, provides a detailed
Endorsed Model Intercomparison Projects (MIPs) that will description of the DECK and CMIP6 historical simulations,
be specific to a particular phase of CMIP (now CMIP6) and and includes a brief introduction to the 21 CMIP6-Endorsed
that will build on the DECK and CMIP historical simulations MIPs.
to address a large range of specific questions and fill the sci-
entific gaps of the previous CMIP phases. The DECK and
CMIP historical simulations, together with the use of CMIP

Published by Copernicus Publications on behalf of the European Geosciences Union.

1938 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

1 Introduction contains a description of a small set of simulations for CMIP

which are intended to be common to all participating mod-
The Coupled Model Intercomparison Project (CMIP) orga- els (Sect. 3), details of which are provided in the Appendix.
nized under the auspices of the World Climate Research Pro- We then present a brief overview of CMIP6 that serves as
gramme’s (WCRP) Working Group on Coupled Modelling an introduction to the other contributions to this special issue
(WGCM) started 20 years ago as a comparison of a handful (Sect. 4), and we close with a summary.
of early global coupled climate models performing experi-
ments using atmosphere models coupled to a dynamic ocean,
a simple land surface, and thermodynamic sea ice (Meehl et 2 CMIP design – a more continuous and distributed
al., 1997). It has since evolved over five phases into a ma- organization
jor international multi-model research activity (Meehl et al.,
2000, 2007; Taylor et al., 2012) that has not only introduced In preparing for CMIP6, the CMIP Panel (the authors of this
a new era to climate science research but has also become paper), which traditionally has the responsibility for direct
a central element of national and international assessments coordination and oversight of CMIP, initiated a 2-year pro-
of climate change (e.g. IPCC, 2013). An important part of cess of community consultation. This consultation involved
CMIP is to make the multi-model output publicly available in the modelling centres whose contributions form the sub-
a standardized format for analysis by the wider climate com- stance of CMIP as well as communities that rely on CMIP
munity and users. The standardization of the model output in model output for their work. Special meetings were orga-
a specified format, and the collection, archival, and access of nized to reflect on the successes of CMIP5 as well as the sci-
the model output through the Earth System Grid Federation entific gaps that remain or have since emerged. The consulta-
(ESGF) data replication centres have facilitated multi-model tion also sought input through a community survey, the scien-
analyses. tific results of which are described by Stouffer et al. (2015).
The objective of CMIP is to better understand past, Four main issues related to the overall structure of CMIP
present, and future climate change arising from natural, un- were identified.
forced variability or in response to changes in radiative forc- First, we identified a growing appreciation of the scientific
ings in a multi-model context. Its increasing importance and potential to use results across different CMIP phases. Such
scope is a tremendous success story, but this very success approaches, however, require an appropriate experimental
poses challenges for all involved. Coordination of the project design to facilitate the identification of an ensemble of mod-
has become more complex as CMIP includes more models els with particular properties drawn from different phases of
with more processes all applied to a wider range of ques- CMIP (e.g. Rauser et al., 2014). At the same time, it was
tions. To meet this new interest and to address a wide vari- recognized that an increasing number of Model Intercompar-
ety of science questions from more and more scientific re- ison Projects (MIPs) were being organized independent of
search communities, reflecting the expanding scope of com- CMIP, the data structure and output requirements were often
prehensive modelling in climate science, has put pressure on inconsistent, and the relationship between the models used in
CMIP to become larger and more extensive. Consequently, the various MIPs was often difficult to determine, in which
there has been an explosion in the diversity and volume of context measures to help establish continuity across MIPs or
requested CMIP output from an increasing number of ex- phases of CMIP would also be welcome.
periments causing challenges for CMIP’s technical infras- Second, the scope of CMIP was taxing the resources of
tructure (Williams et al., 2015). Cultural and organizational modelling centres making it impossible for many to consider
challenges also arise from the tension between expectations contributing to all the proposed experiments. By providing a
that modelling centres deliver multiple model experiments to better basis to help modelling centres decide exactly which
CMIP yet at the same time advance basic research in climate subset of experiments to perform, it was thought that it might
science. be possible to minimize fragmented participation in CMIP6.
In response to these challenges, we have adopted a more A more federated experimental protocol could also encour-
federated structure for the sixth phase of CMIP (i.e. CMIP6) age modelling centres to develop intercomparison studies
and subsequent phases. Whereas past phases of CMIP were based on their own strategic goals.
usually described through a single overview paper, reflect- Third, some centres expressed the view that the punctu-
ing a centralized and relatively compact CMIP structure, this ated structure of CMIP had begun to distort the model devel-
GMD special issue describes the new design and organiza- opment process. Defining a protocol that allowed modelling
tion of CMIP, the suite of experiments, and its forcings, in a centres to decouple their model development from the CMIP
series of invited contributions. In this paper, we provide the schedule would offer additional flexibility, and perhaps en-
overview and backdrop of the new CMIP structure as well as courage modelling centres to finalize their models and sub-
the main scientific foci that CMIP6 will address. We begin mit some of their results sooner on their own schedule.
by describing the new organizational form for CMIP and the Fourth and finally, many groups expressed a desire for par-
pressures that it was designed to alleviate (Sect. 2). It also ticular phases of CMIP to be more than just a collection of

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1939

Figure 1. CMIP evolution. CMIP will evolve but the DECK will provide continuity across phases.

MIPs, but rather to reflect the strategic goals of the climate the development of procedures for soliciting and evaluating
science community as, for instance, articulated by WCRP. MIPs in light of the scientific focus chosen for CMIP6. These
By focusing a particular phase of CMIP around specific sci- procedures were developed and implemented by the CMIP
entific issues, it was felt that the modelling resources could Panel. The responses to the CMIP5 survey helped inform a
be more effectively applied to those scientific questions that series of workshops and resulted in a draft experiment de-
had matured to a point where coordinated activities were ex- sign for CMIP6. This initial design for CMIP6 was published
pected to have substantial impact. in early 2014 (Meehl et al., 2014) and was open for com-
A variety of mechanisms were proposed and intensely de- ments from the wider community until mid-September 2014.
bated to address these issues. The outcome of these discus- In parallel to the open review of the design, the CMIP Panel
sions is embodied in the new CMIP structure, which has three distributed an open call for proposals for MIPs in April 2014.
major components. First, the identification of a handful of These proposals were broadly reviewed within WCRP with
common experiments, the Diagnostic, Evaluation and Char- the goal to encourage and enhance synergies among the dif-
acterization of Klima (DECK) experiments (klima is Greek ferent MIPs, to avoid overlapping experiments, to fill gaps,
for “climate”), and CMIP historical simulations, which can and to help ensure that the WCRP Grand Science Challenges
be used to establish model characteristics and serves as its en- would be addressed. Revised MIP proposals were requested
try card for participating in one of CMIP’s phases or in other and evaluated by the CMIP Panel in summer 2015. The se-
MIPs organized between CMIP phases, as depicted in Fig. 1. lection of MIPs was based on the CMIP Panel’s evaluation
Second, common standards, coordination, infrastructure, and of ten endorsement criteria (Table 1). To ensure community
documentation that facilitate the distribution of model out- engagement, an important criterion was that enough mod-
puts and the characterization of the model ensemble, and elling groups (at least eight) were willing to perform all of
third, the adoption of a more federated structure, building on the MIP’s highest priority (Tier 1) experiments and provid-
more autonomous CMIP-Endorsed MIPs. ing all the requested diagnostics needed to answer at least
Realizing the idea of a particular phase of CMIP being one of its leading science questions. For each of the selected
centred on a collection of more autonomous MIPs required CMIP6-Endorsed MIPs it turned out that at least ten mod-

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1940 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Table 1. Main criteria for MIP endorsement as agreed with representatives from the modelling groups and MIPs at the WGCM 18th Session
in Grainau, Germany in October 2014.

No. MIP endorsement criterion

1 The MIP and its experiments address at least one of the key science questions of CMIP6.
2 The MIP demonstrates connectivity to the DECK experiments and the CMIP6 historical simulations.
3 The MIP adopts the CMIP modelling infrastructure standards and conventions.
4 All experiments are tiered, well defined, and useful in a multi-model context and do not overlap with other
CMIP6 experiments.
5 Unless a Tier 1 experiment differs only slightly from another well-established experiment, it must already have
been performed by more than one modelling group.
6 A sufficient number of modelling centres ( ∼ 8) are committed to performing all of the MIP’s Tier 1 experiments
and providing all the requested diagnostics needed to answer at least one of its science questions.
7 The MIP presents an analysis plan describing how it will use all proposed experiments, any relevant observa-
tions, and specially requested model output to evaluate the models and address its science questions.
8 The MIP has completed the MIP template questionnaire.
9 The MIP contributes a paper on its experimental design to the GMD CMIP6 special issue.
10 The MIP considers reporting on the results by co-authoring a paper with the modelling groups.

elling groups indicated their intent to participate in Tier 1 ex- use changes) continue to evolve significantly, influencing the
periments at least, thus attesting to the wide appeal and level simulated climate. In order to distinguish among the histor-
of science interest from the climate modelling community. ical simulations performed under different phases of CMIP,
the historical simulations are labelled with the phase (e.g.
“CMIP5 historical” or “CMIP6 historical”). A similar ar-
3 The DECK and CMIP historical simulations gument could be made to exclude the AMIP experiments
from the DECK. However, the AMIP experiments are sim-
The DECK comprises four baseline experiments: (a) a his-
pler, more routine, and the dominating role of sea surface
torical Atmospheric Model Intercomparison Project (amip)
temperatures and the focus on recent decades means that for
simulation, (b) a pre-industrial control simulation (piCon-
most purposes AMIP experiments from different phases of
trol or esm-piControl), (c) a simulation forced by an abrupt
CMIP are more likely to provide the desired continuity.
quadrupling of CO2 (abrupt-4×CO2) and (d) a simulation
The persistence and consistency of the DECK will make
forced by a 1 % yr−1 CO2 increase (1pctCO2). CMIP also
it possible to track changes in performance and response
includes a historical simulation (historical or esm-hist) that
characteristics over future generations of models and CMIP
spans the period of extensive instrumental temperature mea-
phases. Although the set of DECK experiments is not ex-
surements from 1850 to the present. In naming the experi-
pected to evolve much, additional experiments may become
ments, we distinguish between simulations with CO2 con-
enough well established as benchmarks (routinely run by
centrations calculated and anthropogenic sources of CO2
modelling groups as they develop new model versions) so
prescribed (esm-piControl and esm-hist) and simulations
that in the future they might be migrated into the DECK.
with prescribed CO2 concentrations (all others). Hereafter,
The common practice of including the DECK in model de-
models that can calculate atmospheric CO2 concentration
velopment efforts means that models can contribute to CMIP
and account for the fluxes of CO2 between the atmosphere,
without carrying out additional computationally burdensome
the ocean, and biosphere are referred to as Earth System
experiments. All of the DECK and the historical simulations
Models (ESMs).
were included in the core set of experiments performed under
The DECK experiments are chosen (1) to provide conti-
CMIP5 (Taylor et al., 2012), and all but the abrupt-4×CO2
nuity across past and future phases of CMIP, (2) to evolve
simulation were included in even earlier CMIP phases.
as little as possible over time, (3) to be well established, and
Under CMIP, credentials of the participating atmosphere–
incorporate simulations that modelling centres perform any-
ocean general circulation models (AOGCMs) and ESMs are
way as part of their own development cycle, and (4) to be rel-
established by performing the DECK and CMIP historical
atively independent of the forcings and scientific objectives
simulations, so these experiments are required from all mod-
of a specific phase of CMIP. The four DECK experiments
els. Together these experiments document the mean climate
and the CMIP historical simulations are well suited for quan-
and response characteristics of models. They should be run
tifying and understanding important climate change response
for each model configuration used in a CMIP-Endorsed MIP.
characteristics. Modelling groups also commonly perform
A change in model configuration includes any change that
simulations of the historical period, but reconstructions of
might affect its simulations other than noise expected from
the external conditions imposed on historical runs (e.g. land-

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1941

different realizations. This would include, for example, a simulations that are proposed as part of DAMIP. In any case
change in model resolution, physical processes, or atmo- it is important that all forcing data sets are documented and
spheric chemistry treatment. If an ESM is used in both CO2 - are made available alongside the model output on the ESGF.
emission-driven mode and CO2 -concentration-driven mode Likewise to the extent modelling centres simplify forcings,
in subsequent CMIP6-Endorsed MIPs, then both emission- for instance by regridding or smoothing in time or some other
driven and concentration-driven control, and historical simu- dimension, this should also be documented.
lations should be done and they will be identical in all forc- For the future scenarios selected by ScenarioMIP, forcings
ings except the treatment of CO2 . are provided by the integrated assessment model (IAM) com-
The forcing data sets that will drive the DECK and CMIP6 munity for the period 2015–2100 (or until 2300 for the ex-
historical simulations are described separately in a series of tended simulations). For atmospheric emissions and concen-
invited contributions to this special issue. These articles also trations as well as for land use, the forcings are harmonized
include some discussion of uncertainty in the data sets. The across IAMs and scenarios using a similar procedure as in
data will be provided by the respective author teams and CMIP5 (van Vuuren et al., 2011). This procedure ensures
made publicly available through the ESGF using common consistency with historical forcing data sets and between the
metadata and formats. different forcing categories. The selection of scenarios and
The historical forcings are based as far as possible on ob- the main characteristics are described elsewhere in this spe-
servations and cover the period 1850–2014. These include: cial issue, while the underlying IAM scenarios are described
in a special issue in Global Environmental Change.
– emissions of short-lived species and long-lived green- An important gap identified in CMIP5, and in previous
house gases (GHGs), CMIP phases, was a lack of careful quantification of the ra-
– GHG concentrations, diative forcings from the different specified external forcing
factors (e.g. GHGs, sulphate aerosols) in each model (Stouf-
– global gridded land-use forcing data sets, fer et al., 2015). This has impaired attempts to identify rea-
sons for differences in model responses. The effective ra-
– solar forcing, diative forcing or ERF component of the Radiative Forcing
– stratospheric aerosol data set (volcanoes), MIP (RFMIP) includes fixed SST simulations to diagnose
the forcing (RFMIP-lite), which are further detailed in the
– AMIP sea surface temperatures (SSTs) and sea ice con- corresponding contribution to this special issue. Although
centrations (SICs), not included as part of the DECK, in recognition of this de-
ficiency in past phases of CMIP we strongly encourage all
– for simulations with prescribed aerosols, a new ap-
CMIP6 modelling groups to participate in RFMIP-lite. The
proach to prescribe aerosols in terms of optical prop-
modest additional effort would enable the radiative forcing to
erties and fractional change in cloud droplet effective
be characterized for both historic and future scenarios across
radius to provide a more consistent representation of
the model ensemble. Knowing this forcing would lead to a
aerosol forcing, and
step change in efforts to understand the spread of model re-
– for models without ozone chemistry, time-varying grid- sponses for CMIP6 and contribute greatly to answering one
ded ozone concentrations and nitrogen deposition. of CMIP6’s science questions.
An overview of the main characteristics of the DECK and
Some models might require additional forcing data sets (e.g. CMIP6 historical simulations appears in Table 2. Here we
black carbon on snow or anthropogenic dust). Allowing briefly describe these experiments. Detailed specifications
model groups to use different forcing1 data sets might better for the DECK and CMIP6 historical simulations are provided
sample uncertainty, but makes it more difficult to assess the in Appendix A and are summarized in Table A1.
uncertainty in the response of models to the best estimate of
the forcing, available to a particular CMIP phase. To avoid 3.1 The DECK
conflating uncertainty in the response of models to a given
forcing, it is strongly preferred for models to be integrated The AMIP and pre-industrial control simulations of the
with the same forcing in the entry card historical simulations, DECK provide opportunities for evaluating the atmospheric
and for forcing uncertainty to be sampled in supplementary model and the coupled system, and in addition they establish
1 Here, we distinguish between an applied input perturbation a baseline for performing many of the CMIP6 experiments.
(e.g. the imposed change in some model constituent, property, or
Many experiments branch from, and are compared with, the
boundary condition), which we refer to somewhat generically as pre-industrial control. Similarly, a number of diagnostic at-
a “forcing”, and radiative forcing, which can be precisely defined. mospheric experiments use AMIP as a control. The idealized
Even if the forcings are identical, the resulting radiative forcing de- CO2 -forced experiments in the DECK (abrupt-4×CO2 and
pends on a model’s radiation scheme (among other factors) and will 1pctCO2), despite their simplicity, can reveal fundamental
differ among models. forcing and feedback response characteristics of models.

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1942 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Table 2. Overview of DECK and CMIP6 historical simulations providing the experiment short names, the CMIP6 labels, brief experiment
descriptions, the forcing methods, as well as the start and end year and minimum number of years per experiment and its major purpose.
The DECK and CMIP6 historical simulation are used to characterize the CMIP model ensemble. Given resource limitations, these entry
card simulations for CMIP include only one ensemble member per experiment. However, we strongly encourage model groups to submit at
least three ensemble members for the CMIP historical simulation as requested in DAMIP. Large ensembles of AMIP simulations are also
encouraged. In the “forcing methods” column, “All” means “volcanic, solar, and anthropogenic forcings”. All experiments are started on
1 January and end on 31 December of the specified years.

Experiment CMIP6 label Experiment description Forcing methods Start End Minimum Major purpose
short name year year no. years
per
simulation
DECK experiments
AMIP amip Observed SSTs All; CO2 concen- 1979 2014 36 Evaluation, variability
and SICs prescribed tration prescribed
Pre-industrial piControl or Coupled atmosphere– CO2 concentration n/a n/a 500 Evaluation, unforced
control esm-piControl ocean pre-industrial prescribed or variability
control calculated
Abrupt abrupt-4×CO2 CO2 abruptly quadru- CO2 concentration n/a n/a 150 Climate sensitivity,
quadrupling of pled and then held prescribed feedback, fast responses
CO2 concen- constant
tration
1 % yr−1 CO2 1pctCO2 CO2 prescribed to CO2 concentration n/a n/a 150 Climate sensitivity,
concentration increase at 1 % yr−1 prescribed feedback, idealized
increase benchmark
CMIP6 historical simulation
Past ∼ 1.5 historical or Simulation of the All; CO2 concen- 1850 2014 165 Evaluation
centuries esm-hist recent past tration prescribed
or calculated

For nearly 3 decades, AMIP simulations (Gates et al., representative of the period prior to the onset of large-scale
1999) have been routinely relied on by modelling centres industrialization, with 1850 being the reference year. Histor-
to help in the evaluation of the atmospheric component of ically, the industrial revolution began in the 18th century, and
their models. In AMIP simulations, the SSTs and SICs are in nature the climate in 1850 was not stable as it was al-
prescribed based on observations. The idea is to analyse and ready changing due to prior historical changes in radiative
evaluate the atmospheric and land components of the climate forcings. In CMIP6, however, as in earlier CMIP phases, the
system when they are constrained by the observed ocean con- control simulation is an attempt to produce a stable quasi-
ditions. These simulations can help identify which model er- equilibrium climate state under 1850 conditions. When dis-
rors originate in the atmosphere, land, or their interactions, cussing and analysing historical and future radiative forcings,
and they have proven useful in addressing a great variety of it needs to be recognized that the radiative forcing in 1850
questions pertaining to recent climate changes. The AMIP due to anthropogenic greenhouse gas increases alone was al-
simulations performed as part of the DECK cover at least the ready around 0.25 W m−2 (Cubasch, 2013) although aerosols
period from January 1979 to December 2014. The end date might have offset that to some extent. In addition, there were
will continue to evolve as the SSTs and SICs are updated other pre-1850 secular changes, for example, in land use
with new observations. Besides prescription of ocean con- (Hurtt et al., 2011), and as a result, global net annual emis-
ditions in these simulations, realistic forcings are imposed sions of carbon from land use and land-use change already
that should be identical to those applied in the CMIP histor- were responsible in 1850 for about 0.6 Pg C yr−1 (Houghton,
ical simulations. Large ensembles of AMIP simulations are 2010). Under the assumptions of the control simulation, how-
encouraged as they can help to improve the signal-to-noise ever, there are no secular changes in forcing, so the con-
ratio (Li et al., 2015). centrations and/or sources of atmospheric constituents (e.g.
The remaining three experiments in the DECK are GHGs and emissions of short-lived species) as well as land
premised on the coupling of the atmospheric and oceanic cir- use are held fixed, as are Earth’s orbital characteristics. Be-
culation. The pre-industrial control simulation (piControl or cause of the absence of both naturally occurring changes in
esm-piControl) is performed under conditions chosen to be forcing (e.g. volcanoes, orbital or solar changes) and human-

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1943

induced changes, the control simulation can be used to study ing variability and century timescale trends (e.g. Flato et al.,
the unforced internal variability of the climate system. 2013). These simulations can also be analysed to determine
An initial climate spin-up portion of a control simulation, whether climate model forcing and sensitivity are consis-
during which the climate begins to come into balance with tent with the observational record, which provides opportu-
the forcing, is usually performed. At the end of the spin-up nities to better bound the magnitude of aerosol forcing (e.g.
period, the piControl starts. The piControl serves as a base- Stevens, 2015). In addition they, along with the control run,
line for experiments that branch from it. To account for the provide the baseline simulations for performing formal de-
effects of any residual drift, it is required that the piCon- tection and attribution studies (e.g. Stott et al., 2006) which
trol simulation extends as far beyond the branching point help uncover the causes of forced climate change.
as any experiment to which it will be compared. Only then As with performing control simulations, models that in-
can residual climate drift in an experiment be removed so clude representation of the carbon cycle should normally
that it is not misinterpreted as part of the model’s forced re- perform two different CMIP historical simulations: one with
sponse. The recommended minimum length for the piControl prescribed CO2 concentration and the other with prescribed
is 500 years. CO2 emissions (accounting explicitly for fossil fuel combus-
The two DECK climate change experiments branch from tion). In the second, CO2 concentrations are predicted by
some point in the 1850 control simulation and are designed the model. The treatment of other GHGs should be identi-
to document basic aspects of the climate system response to cal in both simulations. Both types of simulation are useful
greenhouse gas forcing. In the first, the CO2 concentration in evaluating how realistically the model represents the re-
is immediately and abruptly quadrupled from the global an- sponse of the carbon cycle anthropogenic CO2 emissions, but
nual mean 1850 value that is used in piControl. This abrupt- the prescribed concentration simulation enables these more
4×CO2 simulation has proven to be useful for characterizing complex models to be evaluated fairly against those models
the radiative forcing that arises from an increase in atmo- without representation of carbon cycle processes.
spheric CO2 as well as changes that arise indirectly due to
the warming. It can also be used to estimate a model’s equi- 3.3 Common standards, infrastructure, and
librium climate sensitivity (ECS, Gregory et al., 2004). In documentation
the second, the CO2 concentration is increased gradually at
a rate of 1 % per year. This experiment has been performed A key to the success of CMIP and one of the motivations
in all phases of CMIP since CMIP2, and serves as a consis- for incorporating a wide variety of coordinated modelling
tent and useful benchmark for analysing model transient cli- activities under a single framework in a specific phase of
mate response (TCR). The TCR takes into account the rate CMIP (now CMIP6) is the desire to reduce duplication of
of ocean heat uptake which governs the pace of all time- effort, minimize operational and computational burdens, and
evolving climate change (e.g. Murphy and Mitchell, 1995). establish common practices in producing and analysing large
In addition to the TCR, the 1 % CO2 integration with ESMs amounts of model output. To enable automated processing
that include explicit representation of the carbon cycle allows of output from dozens of different models, CMIP has led the
the calculation of the transient climate response to cumula- way in encouraging adoption of data standards (governing
tive carbon emissions (TCRE), defined as the transient global structure and metadata) that facilitate development of soft-
average surface temperature change per unit of accumulated ware infrastructure in support of coordinated modelling ac-
CO2 emissions (IPCC, 2013). Despite their simplicity, these tivities. The ESGF has capitalized on this standardization to
experiments provide a surprising amount of insight into the provide access to CMIP model output hosted by institutions
behaviour of models subject to more complex forcing (e.g. around the world. As the complexity of CMIP has increased
Bony et al., 2013; Geoffroy et al., 2013). and as the potential use of model output expands beyond
the research community, the evolution of the climate mod-
3.2 CMIP historical simulations elling infrastructure requires enhanced coordination. To help
in this regard, the WGCM Infrastructure Panel (WIP) was
In addition to the DECK, CMIP requests models to simu- set up, and is now providing guidance on requirements and
late the historical period, defined to begin in 1850 and ex- establishing specifications for model output, model and sim-
tend to the near present. The CMIP historical simulation and ulation documentation, and archival and delivery systems for
its CO2 -emission-driven counterpart, esm-hist, branch from CMIP6 data. In parallel to the development of the CMIP6
the piControl and esm-piControl, respectively (see details in experiment design, the ESGF capabilities are being further
Sect. A1.2). These simulations are forced, based on observa- extended and improved. In CMIP5, with over 1,000 differ-
tions, by evolving, externally imposed forcings such as so- ent model/experiment combinations, a first attempt was also
lar variability, volcanic aerosols, and changes in atmospheric made to capture structured metadata describing the models
composition (GHGs and aerosols) caused by human activ- and the simulations themselves. Based upon the Common In-
ities. The CMIP historical simulations provide rich oppor- formation Model (CIM, Lawrence et al., 2012), tools were
tunities to assess model ability to simulate climate, includ- provided to capture documentation of models and simula-

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1944 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

tions. This effort is now continuing under the banner of the

international ES-DOC activity, which establishes agreements
on common Controlled Vocabularies (CVs) to describe mod-
els and simulations. Modelling groups will be required to
provide documentation following a common template and
adhering to the CVs. With the documentation recorded uni-
formly across models, researchers will, for example, be able
to use web-based tools to determine differences in model ver-
sions and differences in forcing and other conditions that af-
fect each simulation. Further details on the CMIP6 infras-
tructure can be found in the WIP contribution to this special
issue.
A more routine benchmarking and evaluation of the mod-
els is envisaged to be a central part of CMIP6. As noted
above, one purpose of the DECK and CMIP historical sim-
ulations is to provide a basis for documenting model sim-
ulation characteristics. Towards that end an infrastructure
is being developed to allow analysis packages to be rou-
tinely executed whenever new model experiments are con-
tributed to the CMIP archive at the ESGF. These efforts uti-
lize observations served by the ESGF contributed from the
obs4MIPs (Ferraro et al., 2015; Teixeira et al., 2014) and Figure 2. Schematic of the CMIP/CMIP6 experiment design. The
ana4MIPs projects. Examples of available tools that target inner ring and surrounding white text involve standardized func-
routine evaluation in CMIP include the PCMDI metrics soft- tions of all CMIP DECK experiments and the CMIP6 historical
ware (Gleckler et al., 2016) and the Earth System Model simulation. The middle ring shows science topics related specifi-
Evaluation Tool (ESMValTool, Eyring et al., 2016), which cally to CMIP6 that are addressed by the CMIP6-Endorsed MIPs,
brings together established diagnostics such as those used with MIP topics shown in the outer ring. This framework is super-
in the evaluation chapter of IPCC AR5 (Flato et al., 2013). imposed on the scientific backdrop for CMIP6 which are the seven
WCRP Grand Science Challenges.
The ESMValTool also integrates other packages, such as the
NCAR Climate Variability Diagnostics Package (Phillips et
al., 2014), or diagnostics such as the cloud regime metric to change from one CMIP phase to the next. To maximize
(Williams and Webb, 2009) developed by the Cloud Feed- the relevance and impact of CMIP6, it was decided to use
back MIP (CFMIP) community. These tools can be used to the WCRP Grand Science Challenges (GCs) as the scientific
broadly and comprehensively characterize the performance backdrop of the CMIP6 experimental design. By promoting
of the wide variety of models and model versions that will research on critical science questions for which specific gaps
contribute to CMIP6. This evaluation activity can, compared in knowledge have hindered progress so far, but for which
with CMIP5, more quickly inform users of model output, as new opportunities and more focused efforts raise the possi-
well as the modelling centres, of the strengths and weak- bility of significant progress on the timescale of 5–10 years,
nesses of the simulations, including the extent to which these GCs constitute a main component of the WCRP strat-
long-standing model errors remain evident in newer models. egy to accelerate progress in climate science (Brasseur and
Building such a community-based capability is not meant Carlson, 2015). They relate to (1) advancing understanding
to replace how CMIP research is currently performed but of the role of clouds in the general atmospheric circulation
rather to complement it. These tools can also be used to com- and climate sensitivity (Bony et al., 2015), (2) assessing the
pute derived variables or indices alongside the ESGF, and response of the cryosphere to a warming climate and its
their output could be provided back to the distributed ESGF global consequences, (3) understanding the factors that con-
archive. trol water availability over land (Trenberth and Asrar, 2014),
(4) assessing climate extremes, what controls them, how they
have changed in the past and how they might change in the
4 CMIP6 future, (5) understanding and predicting regional sea level
change and its coastal impacts, (6) improving near-term cli-
4.1 Scientific focus of CMIP6 mate predictions, and (7) determining how biogeochemical
cycles and feedback control greenhouse gas concentrations
In addition to the DECK and CMIP historical simulations, and climate change.
a number of additional experiments will colour a specific
phase of CMIP, now CMIP6. These experiments are likely

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1945

These GCs will be using the full spectrum of observa-

tional, modelling and analytical expertise across the WCRP,
and in terms of modelling most GCs will address their spe-
cific science questions through a hierarchy of numerical
models of different complexities. Global coupled models ob-
viously constitute an essential element of this hierarchy, and
CMIP6 experiments will play a prominent role across all
GCs by helping to answer the following three CMIP6 science
questions: How does the Earth system respond to forcing?
What are the origins and consequences of systematic model
biases? How can we assess future climate change given inter-
nal climate variability, climate predictability, and uncertain-
ties in scenarios?
These three questions will be at the centre of CMIP6. Sci-
ence topics related specifically to CMIP6 will be addressed
through a range of CMIP6-Endorsed MIPs that are organized
by the respective communities and overseen by the CMIP
Panel (Fig. 2). Through these different MIPs and their con-
nection to the GCs, the goal is to fill some of the main scien-
tific gaps of previous CMIP phases. This includes, in particu-
lar, facilitating the identification and interpretation of model
systematic errors, improving the estimate of radiative forc-
ings in past and future climate change simulations, facilitat-
ing the identification of robust climate responses to aerosol
forcing during the historical period, better accounting of the
impact of short-term forcing agents and land use on climate,
better understanding the mechanisms of decadal climate vari-
ability, along with many other issues not addressed satisfac-
torily in CMIP5 (Stouffer et al., 2015). In endorsing a num- Figure 3. Contributions of CMIP6-Endorsed MIPs to the three
ber of these MIPs, the CMIP Panel acted to minimize over- CMIP6 science questions and the WCRP Grand Science Chal-
laps among the MIPs and to reduce the burden on modelling lenges. A filled circle indicates highest priority and an open circle,
second highest priority. Some of the MIPs additionally contribute
groups, while maximizing the scientific complementarity and
with lower priority to other CMIP6 science questions or WCRP
synergy among the different MIPs.
Grand Science Challenges.

4.2 The CMIP6-Endorsed MIPs

tributions will include a description of the experimental de-
Close to 30 suggestions for CMIP6 MIPs have been re- sign and scientific justification of each of the experiments for
ceived so far, of which 21 MIPs were eventually endorsed Tier 1 (and possibly beyond), and will link the experiments
and invited to participate (Table 3). Of those not selected and analysis to the DECK and CMIP6 historical simulations.
some were asked to work with other proposed MIPs with They will additionally include an analysis plan to fully jus-
overlapping science goals and objectives. Of the 21 CMIP6- tify the resources used to produce the various requested vari-
Endorsed MIPs, 4 are diagnostic in nature, which means that ables, and if the analysis plan is to compare model results to
they define and analyse additional output, but do not require observations, the contribution will highlight possible model
additional experiments. In the remaining 17 MIPs, a total diagnostics and performance metrics specifying whether the
of around 190 experiments have been proposed resulting in comparison entails any particular requirement for the simula-
40 000 model simulation years with around half of these in tions or outputs (e.g. the use of observational simulators). In
Tier 1. The CMIP6-Endorsed MIPs show broad coverage addition, possible observations and reanalysis products for
and distribution across the three CMIP6 science questions, model evaluation are discussed and the MIPs are encour-
and all are linked to the WCRP Grand Science Challenges aged to help facilitate their use by contributing them to the
(Fig. 3). obs4MIPs/ana4MIPs archives at the ESGF (see Sect. 3.3).
Each of the 21 CMIP6-Endorsed MIPs is described in a In some MIPs, additional forcings beyond those used in the
separate invited contribution to this special issue. These con- DECK and CMIP6 historical simulations are required, and
tributions will detail the goal of the MIP and the major scien- these are described in the respective contribution as well.
tific gaps the MIP is addressing, and will specify what is new
compared to CMIP5 and previous CMIP phases. The con-

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1946 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Table 3. List of CMIP6-Endorsed MIPs along with the long name of the MIP, the primary goal(s) and the main CMIP6 science theme as
displayed in Fig. 2. Each of these MIPs is described in more detail in a separate contribution to this special issue. MIPs marked with ∗ are
diagnostic MIPs.

Short name of Long name of MIP Primary goal(s) in CMIP6 Main CMIP6
MIP science theme
AerChemMIP Aerosols and (a) Diagnosing forcings and feedback of tropospheric aerosols, tropo- Chemistry/
Chemistry Model spheric ozone precursors and the chemically reactive WMGHGs; (b) doc- Aerosols
Intercomparison umenting and understanding past and future changes in the chemical
Project composition of the atmosphere; (c) estimating the global-to-regional cli-
mate response from these changes.
C4 MIP Coupled Climate Understanding and quantifying future century-scale changes in the global Carbon cycle
Carbon Cycle carbon cycle and its feedback on the climate system, making the link
Model Intercom- between CO2 emissions and climate change.
parison Project
CFMIP Cloud Feedback Improving assessments of cloud feedback via (a) improved understanding Clouds/
Model Intercom- of cloud-climate feedback mechanisms and (b) better evaluation of clouds Circulation
parison Project and cloud feedback in climate models. Also improving understanding of
circulation, regional-scale precipitation, and non-linear changes.
DAMIP Detection and (a) Estimating the contribution of external forcings to observed global Characterizing
Attribution Model and regional climate changes; (b) observationally constraining future cli- forcings
Intercomparison mate change projections by scaling future GHG and other anthropogenic
Project responses using regression coefficients derived for the historical period.
DCPP Decadal Climate Predicting and understanding forced climate change and internal vari- Decadal
Prediction Project ability up to 10 years into the future through a coordinated set of hindcast prediction
experiments, targeted experiments to understand the physical processes,
and the ongoing production of skilful decadal predictions.
FAFMIP Flux-Anomaly- Explaining the model spread in climate projections of ocean climate Ocean/Land/
Forced Model change forced by CO2 increase, especially regarding the geographical Ice
Intercomparison patterns and magnitude of sea level change, ocean heat uptake, and ther-
Project mal expansion.
GeoMIP Geoengineering Assessing the climate system response (including on extreme events) to Geoengineering
Model Intercom- proposed radiation modification geoengineering schemes by evaluating
parison Project their efficacies, benefits, and side effects.
GMMIP Global Monsoons (a) Improving understanding of physical processes in global monsoons Regional
Model Intercom- system; (b) better simulating the mean state, interannual variability, and phenomena
parison Project long-term changes of global monsoons.
HighResMIP High-Resolution Assessing the robustness of improvements in the representation of impor- Regional
Model Intercom- tant climate processes with weather-resolving global model resolutions phenomena
parison Project (∼ 25 km or finer), within a simplified framework using the physical cli-
mate system only with constrained aerosol forcing.
ISMIP6 Ice Sheet Model Improving confidence in projections of the sea level rise associated with Ocean/Land/
Intercomparison mass loss from the ice sheets of Greenland and Antarctica. Ice
Project for CMIP6
LS3MIP Land Surface, Providing a comprehensive assessment of land surface, snow, and soil Ocean/Land/
Snow and Soil moisture-climate feedback, and diagnosing systematic biases in the land Ice
Moisture modules of current ESMs using constrained land-module-only experi-
ments.
LUMIP Land-Use Model Quantifying the effects of land use on climate and biogeochemical cy- Land use
Intercomparison cling (past–future), and assessing the potential for alternative land man-
Project agement strategies to mitigate climate change.

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1947

Table 3. Continued.

OMIP Ocean Model In- Providing a framework for evaluating, understanding, and improving Ocean/Land/
tercomparison ocean, sea ice, and biogeochemical, including inert tracers, components Ice
Project of climate and Earth system models contributing to CMIP6. Protocols are
provided to perform coordinated ocean/sea ice/tracer/biogeochemistry
simulations forced with common atmospheric data sets.
PMIP Paleoclimate Mod- (a) Analysing the response to forcings and major feedback for past cli- Paleo
elling Intercompar- mates outside the range of recent variability; (b) assessing the credibility
ison Project of climate models used for future climate projections.
RFMIP Radiative Forcing (a) Characterizing the global and regional effective radiative forcing for Characterizing
Model Intercom- each model for historical and 4×CO2 simulations; (b) assessing the abso- forcings
parison Project lute accuracy of clear-sky radiative transfer parameterizations; (c) identi-
fying the robust impacts of aerosol radiative forcing during the historical
period.
ScenarioMIP Scenario Model (a) Facilitating integrated research on the impact of plausible future sce- Scenarios
Intercomparison narios over physical and human systems, and on mitigation and adap-
Project tation options; (b) addressing targeted studies on the effects of particular
forcings in collaboration with other MIPs; (c) help quantifying projection
uncertainties based on multi-model ensembles and emergent constraints.
VolMIP Volcanic Forcings (a) Assessing to what extent responses of the coupled ocean–atmosphere Characterizing
Model Intercom- system to strong volcanic forcing are robustly simulated across state-of- forcings
parison Project the-art coupled climate models; (b) identifying the causes that limit robust
simulated behaviour, especially differences in their treatment of physical
processes
CORDEX∗ Coordinated Re- Advancing and coordinating the science and application of regional cli- Impacts
gional Climate mate downscaling (RCD) through statistical and dynamical downscaling
Downscaling of CMIP DECK, CMIP6 historical, and ScenarioMIP output.
Experiment
DynVarMIP∗ Dynamics and Va- Defining and analysing diagnostics that enable a mechanistic approach Clouds/
riability Model In- to confront model biases and understand the underlying causes behind Circulation
tercomparison circulation changes with a particular emphasis on the two-way coupling
Project between the troposphere and the stratosphere.
SIMIP∗ Sea Ice Model Understanding the role of sea ice and its response to climate change by Ocean/Land/
Intercomparison defining and analysing a comprehensive set of variables and process- Ice
Project oriented diagnostics that describe the sea ice state and its atmospheric
and ocean forcing.
VIACS AB∗ Vulnerability, Im- Facilitating a two-way dialogue between the CMIP6 modelling commu- Impacts
pacts, Adaptation nity and VIACS experts, who apply CMIP6 results for their numerous re-
and Climate search and climate services, towards an informed construction of model
Services Advisory scenarios and simulations and the design of online diagnostics, metrics,
Board and visualization of relevance to society.

A number of MIPs are developments and/or continuation paleoclimate (PMIP), which assesses the credibility of the
of long-standing science themes. These include MIPs specif- model response to forcing outside the range of recent vari-
ically addressing science questions related to cloud feedback ability. These MIPs reflect the importance of key forcing and
and the understanding of spatial patterns of circulation and feedback processes in understanding past, present, and future
precipitation (CFMIP), carbon cycle feedback, and the un- climate change and have developed new experiments and sci-
derstanding of changes in carbon fluxes and stores (C4 MIP), ence plans focused on emerging new directions that will be
detection and attribution (DAMIP) that newly includes 21st- at the centre of the WCRP Grand Science Challenges. A few
century GHG-only simulations allowing the projected re- new MIPs have arisen directly from gaps in understanding
sponses to GHGs and other forcings to be separated and in CMIP5 (Stouffer et al., 2015), for example, poor quantifi-
scaled to derive observationally constrained projections, and cation of radiative forcing (RFMIP), better understanding of

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1948 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

ocean heat uptake and sea level rise (FAFMIP), and under- diagnostic MIP requests certain key variables of interest to
standing of model response to volcanic forcing (VolMIP). the VIACS community be delivered in a timely manner to be
Since CMIP5, other MIPs have emerged as the modelling used by climate services and in impact studies.
community has developed more complex ESMs with inter- All MIPs define output streams in the centrally coordi-
active components beyond the carbon cycle. These include nated CMIP6 data request for each of their own experiments
the consistent quantification of forcings and feedback from as well as the DECK and CMIP6 historical simulations (see
aerosols and atmospheric chemistry (AerChemMIP), and, for the CMIP6 data request contribution to this special issue for
the first time in CMIP, modelling of sea level rise from land details). This will ensure that the required variables are stored
ice sheets (ISMIP6). at the frequency and resolution required to address the spe-
Some MIPs specifically target systematic biases focusing cific science questions and evaluation needs of each MIP and
on improved understanding of the sea ice state and its at- to enable a broad characterization of the performance of the
mospheric and oceanic forcing (SIMIP), the physical and CMIP6 models.
biogeochemical aspects of the ocean (OMIP), land, snow We note that only the Tier 1 MIP experiments are overseen
and soil moisture processes (LS3MIP), and improved un- by the CMIP Panel, but additional experiments are proposed
derstanding of circulation and variability with a focus on by the MIPs in Tiers 2 and 3. We encourage the modelling
stratosphere–troposphere coupling (DynVarMIP). With the groups to participate in the full suite of experiments beyond
increased emphasis in the climate science community on the Tier 1 to address in more depth the scientific questions posed.
need to represent and understand changes in regional circula- The call for MIP applications for CMIP6 is still open and
tion, systematic biases are also addressed on a more regional new proposals will be reviewed at the annual WGCM meet-
scale by the Global Monsoon MIP (GMMIP) and a first ings. However, we point out that the additional MIPs sug-
coordinated activity on high-resolution modelling (High- gested after the CMIP6 data request has been finalized will
ResMIP). have to work with the already defined model output from the
For the first time, future scenario experiments, previously DECK and CMIP6 historical simulations, or work with the
coordinated centrally as part of the CMIP5 core experiments, modelling group to recover additional variables from their
will be run as an MIP ensuring clear definition and well- internal archives. We also point out that some experiments
coordinated science questions. ScenarioMIP will run a new proposed by CMIP6-Endorsed MIPs may not be finished un-
set of future long-term (century timescale) integrations en- til after CMIP6 ends.
gaging input from both the climate science and integrated
assessment modelling communities. The new scenarios are
based on a matrix that uses the shared socioeconomic path- 5 Summary
ways (SSPs, O’Neill et al., 2015) and forcing levels of the
Representative Concentration Pathways (RCP) as axes. As a CMIP6 continues the pattern of evolution and adaptation
set, they span the same range as the CMIP5 RCPs (Moss et characteristic of previous phases of CMIP. To centre CMIP at
al., 2010), but fill critical gaps for intermediate forcing levels the heart of activities within climate science and encourage
and questions, for example, on short-lived species and land links among activities within the World Climate Research
use. The near-term experiments (10–30 years) are coordi- Programme (WCRP), CMIP6 has been formulated scientif-
nated by the decadal climate prediction project (DCPP) with ically around three specific questions, amidst the backdrop
improvements expected, for example, from the initialization of the WCRP’s seven Grand Science Challenges. To meet
of additional components beyond the ocean and from a more the increasingly broad scientific demands of the climate-
detailed process understanding and evaluation of the predic- science community, yet be responsive to the individual prior-
tions to better identify sources and limits of predictability. ities and resource limitations of the modelling centres, CMIP
Other MIPs include specific future mitigation options, e.g. has adopted a new, more federated organizational structure.
the land use MIP (LUMIP) that is for the first time in CMIP CMIP has now evolved from a centralized activity involv-
isolating regional land management strategies to study how ing a large number of experiments to a federated activity, en-
different surface types respond to climate change and di- compassing many individually designed MIPs. CMIP6 com-
rect anthropogenic modifications, or the geoengineering MIP prises 21 individual CMIP6-Endorsed MIPs and the DECK
(GeoMIP), which examines climate impacts of newly pro- and CMIP6 historical simulations. Four of the 21 CMIP6-
posed radiation modification geoengineering strategies. Endorsed MIPs are diagnostic in nature, meaning that they
The diagnostic MIP CORDEX will oversee the downscal- require additional output from models, but not additional
ing of CMIP6 models for regional climate projections. An- simulations. The total amount of output from CMIP6 is es-
other historic development in our field that provides, for the timated to be between 20 and 40 petabytes, depending on
first time in CMIP, an avenue for a more formal communi- model resolution and the number of modelling centres ulti-
cation between the climate modelling and user community mately participating in CMIP6. Questions addressed in the
is the endorsement of the vulnerability, impacts, and adapta- MIPs are wide ranging, from the climate of distant past to
tion and climate services advisory board (VIACS AB). This the response of turbulent cloud processes to radiative forc-

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1949

Figure 4. CMIP6 timeline for the preparation of forcings, the realization of experiments and their analysis.

ing, from how the terrestrial biosphere influences the uptake Data availability
of CO2 to how much predictability is stored in the ocean,
from how to best project near-term to long-term future cli- The model output from the DECK and CMIP6 historical sim-
mate changes while considering interdependence and differ- ulations described in this paper will be distributed through
ences in model performance in the CMIP6 ensemble, and the Earth System Grid Federation (ESGF) with digital object
from what regulates the distribution of tropospheric ozone, identifiers (DOIs) assigned. As in CMIP5, the model out-
to the influence of land-use changes on water availability. put will be freely accessible through data portals after reg-
The last 3 years have been dedicated to conceiving and istration. In order to document CMIP6’s scientific impact
then planning what we now call CMIP6. Starting in 2016, the and enable ongoing support of CMIP, users are obligated
first modelling centres are expected to begin performing the to acknowledge CMIP6, the participating modelling groups,
DECK and uploading output on the ESGF. Forcings for the and the ESGF centres (see details on the CMIP Panel web-
DECK and CMIP6 historical simulations will be ready be- site at http://www.wcrp-climate.org/index.php/wgcm-cmip/
fore mid-2016 so that these experiements can be started, and about-cmip). Further information about the infrastructure
by the end of 2016 the diverse forcings for different scenarios supporting CMIP6, the metadata describing the model out-
of future human activity will become available. Past experi- put, and the terms governing its use are provided by the
ence suggests that most centres will complete their CMIP WGCM Infrastructure Panel (WIP) in their invited contribu-
simulations within a few years while the analysis of CMIP6 tion to this special issue. Along with the data, the provenance
results will likely go on for a decade or more (Fig. 4). of the data will be recorded, and DOIs will be assigned to col-
Through an intensified effort to align CMIP with spe- lections of output so that they can be appropriately cited. This
cific scientific questions and the WCRP Grand Science Chal- information will be made readily available so that published
lenges, we expect CMIP6 to continue CMIP’s tradition of research results can be verified and credit can be given to the
major scientific advances. CMIP6 simulations and scientific modelling groups providing the data. The WIP is coordinat-
achievements are expected to support the IPCC Sixth Assess- ing and encouraging the development of the infrastructure
ment Report (AR6) as well as other national and international needed to archive and deliver this information. In order to
climate assessments or special reports. Ultimately scientific run the experiments, data sets for natural and anthropogenic
progress on the most pressing problems of climate variability forcings are required. These forcing data sets are described in
and change will be the best measure of the success of CMIP6. separate invited contributions to this special issue. The forc-
ing data sets will be made available through the ESGF with
version control and DOIs assigned.

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1950 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Appendix A: Experiment specifications equilibrium conditions consistent with the model by, for ex-
ample, running with ocean conditions starting earlier in the
A1 Specifications for the DECK 1970s or cycling repeatedly through year 1979 before simu-
lating the official period. Results from the spin-up period (i.e.
Here we provide information needed to perform the DECK, prior to 1979) should be discarded, but the spin-up technique
including specification of forcing and boundary conditions, should be documented.
initialization procedures, and minimum length of runs. This For CMIP6, AMIP simulations should cover at least the
information is largely consistent with but not identical to the period from January 1979 through December 2014, but mod-
specifications for these experiments in CMIP5 (Taylor et al., elling groups are encouraged to extend their runs to the end
2009). of the observed period. Output may also be contributed from
The DECK and CMIP6 historical simulations are re- years preceding 1979 with the understanding that surface
quested from all models participating in CMIP. The expec- ocean conditions were less complete and in some cases less
tation is that this requirement will be met for each model reliable then.
configuration used in the subsequent CMIP6-Endorsed MIPs The climate found in AMIP simulations is largely de-
(an entry card). For CMIP6, in the special case where the termined by the externally imposed forcing, especially the
burden of the entry card simulations is prohibitive but the ocean conditions. Nevertheless, unforced variability (noise)
scientific case for including a particular model simulation is within the atmosphere introduces some non-deterministic
compelling (despite only partial completion of the entry card variations that hamper unambiguous interpretation of ap-
simulations), an exception to this policy can be granted on a parent relationships between, for example, the year-to-year
model-by-model basis by the CMIP Panel, which will seek anomalies in SSTs and their consequences over land. To as-
advice from the chairs of the affected CMIP6-Endorsed MIP. sess the role of unforced atmospheric variability in any par-
CMIP6 is a cooperative effort across the international cli- ticular result, modelling groups are encouraged to generate
mate modelling and climate science communities. The mod- an ensemble of AMIP simulations. For most studies, a three-
elling groups have all been involved in the design and imple- member ensemble, where only the initial conditions are var-
mentation of CMIP6, and thus have agreed to a set of best ied, would be the minimum required, with larger size ensem-
practices proposed for CMIP6. Those best practices include bles clearly of value in making more precise determination
having the modelling groups submit the DECK experiments of statistical significance.
and the CMIP6 historical simulations to the ESGF, as well as
any CMIP6-Endorsed MIP experiments they choose to run. A1.2 Multi-century pre-industrial control simulations
Additionally, the modelling groups decide what constitutes a
new model version. The CMIP Panel will work with the MIP Like laboratory experiments, numerical experiments are de-
co-chairs and the modelling groups to ensure that these best signed to reveal cause and effect relationships. A standard
practices are followed. way of doing this is to perform both a control experiment
and a second experiment where some externally imposed ex-
A1.1 AMIP simulation periment condition has been altered. For many CMIP experi-
ments, including the rest of the experiments discussed in this
As in the first simulations performed under the Atmospheric Appendix, the control is a simulation with atmospheric com-
Model Intercomparison Project (AMIP, Gates et al., 1999), position and other conditions prescribed and held constant,
SSTs and SICs in AMIP experiments are prescribed con- consistent with best estimates of the forcing from the histor-
sistent with observations (see details on this forcing data ical period.
set in the corresponding contribution to this special issue). Ideally the pre-industrial control (piControl) experiment
Land models should be configured as close as possible to the for CMIP would represent a near-equilibrium state of the cli-
one used in the CMIP6 historical simulation including tran- mate system under the imposed conditions. In reality, sim-
sient land use and land cover. Other external forcings includ- ulations of hundreds to many thousands of years would be
ing volcanic aerosols, solar variability, GHG concentrations, required for the ocean’s depths to equilibrate and for biogeo-
and anthropogenic aerosols should also be prescribed consis- chemical reservoirs to fully adjust. Available computational
tent with those used in the CMIP6 historical simulation (see resources generally preclude integrations long enough to ap-
Sect. A2 below). Even though in AMIP simulations models proach equilibrium, so in practice shorter runs must suffice.
with an active carbon cycle will not be fully interactive, sur- Usually, a piControl simulation is initialized from the control
face carbon fluxes should be archived over land. run of a different model or from observations, and then run
AMIP integrations can be initialized from prior model in- until at least the surface climate conditions stabilize using
tegrations or from observations or in other reasonable ways. 1850 forcings (see Stouffer et al., 2004, for further discus-
Depending on the treatment of snow cover, soil water con- sion). This spin-up period can be as long as several hundred
tent, the carbon cycle, and vegetation, these runs may require years and variables that can document the spin-up behaviour
a spin-up period of several years. One might establish quasi- should be archived (under the experiment labels piControl-

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1951

spinup or esm-piControl-spinup). At the very least the length we recommend a specification of this forcing that attempts to
of the spin-up period should be documented. balance conflicting objectives to
Although equilibrium is generally not achieved, the
changes occurring after the spin-up period are usually found – minimize artificial climate responses to discontinuities
to evolve at a fairly constant rate that presumably decreases in radiative forcing at the time a historical simulation is
slowly as equilibrium is approached. After a few centuries, initiated, and
these drifts of the system mainly affect the carbon cycle and – minimize artefacts in sea level change due to thermal
ocean below the main thermocline, but they are also manifest expansion caused by unrealistic mismatches in condi-
at the surface in a slow change in sea level. The climate drift tions in the centennial-scale averaged forcings for the
must be removed in order to interpret experiments that use pre- and post-1850 periods. Note that any preindus-
the pre-industrial simulation as a control. The usual proce- trial multi-centennial observed trend in global-mean
dure is to assume that the drift is insensitive to CMIP exper- sea level is most likely to be due to slow changes in
iment conditions and to simply subtract the control run from ice-sheets, which are likely not to be simulated in the
the perturbed run to determine the climate change that would CMIP6 model generation.
occur in the absence of drift.
Besides serving as controls for numerical experimentation, The first consideration above implies that radiative forcing
the piControl and esm-piControl are used to study the natu- in the control run should be close to that imposed at the be-
rally occurring, unforced variability of the climate system. ginning of the CMIP historical simulation (i.e. 1850). The
The only source of climate variability in a control arises second implies that a background volcanic aerosol and time-
from processes internal to the model, whereas in the more averaged solar forcing should be prescribed in the control
complicated real world, variations are also caused by exter- run, since to neglect it would cause an apparent drift in sea
nal forcing factors such as solar variability and changes in level associated with the suppression of heat uptake due to
atmospheric composition caused, for example, by human ac- the net effect of, for instance, volcanism after 1850, and this
tivities or volcanic eruptions. Consequently, the physical pro- has implications for sea level changes (Gregory, 2010; Gre-
cesses responsible for unforced variability can more easily be gory et al., 2013). We recognize that it will be impossible
isolated and studied using the control run of models, rather to entirely avoid artefacts and artificial transient effects, and
than by analysing observations. practical considerations may rule out conformance with ev-
A DECK control simulation is required to be long enough ery detail of the control simulation protocol stipulated here.
to extend to the end of any perturbation runs initiated from With that understanding, here is a summary of the recom-
it so that climate drift can be assessed and possibly removed mendations for the imposed conditions on the spin-up and
from those runs. If, for example, a historical simulation (be- control runs, followed by further clarification in subsequent
ginning in 1850) were initiated from the beginning of the paragraphs:
control simulation and then were followed by a future sce-
– Conditions must be time invariant except for those asso-
nario run extending to year 2300, a control run of at least
ciated with the mean climate (notably the seasonal and
450 years would be required. As discussed above, control
diurnal cycles of insolation).
runs are also used to assess model-simulated unforced cli-
mate variability. The longer the control, the more precisely – Unless indicated otherwise (e.g. the background vol-
can variability be quantified for any given timescale. A con- canic forcing), experiment conditions (e.g. greenhouse
trol simulation of many hundreds of years would be needed gas concentrations, ozone concentration, surface land
to assess variability on centennial timescales. For CMIP6 conditions) should be representative of Earth around the
it is recommended that the control run should be at least year 1850.
500 years long (following the spin-up period), but of course
the simulation must be long enough to reach to the end – Orbital parameters (eccentricity, obliquity, and longi-
of the experiments it spawns. It should be noted that those tude of the perihelion) should be held fixed at their 1850
analysing CMIP6 simulations might also require simulations values.
longer than 500 years to accurately assess unforced variabil- – Land use should not change in the control run and
ity on long timescales, so modelling groups are encouraged should be fixed according to reconstructed agricultural
to extend their control runs well beyond the minimum rec- maps from 1850. Due to the diversity of model ap-
ommended number of years. proaches in ESMs for land carbon, some groups might
Because the climate was very likely not in equilibrium deviate from this specification, and again this must be
with the forcing of 1850 and because different components clearly documented.
of the climate system differentially respond to the effects of
the forcing prior to that time, there is some ambiguity in de- – The solar constant should be fixed at its mean value (no
ciding on what forcing to apply for the control. For CMIP6 11-year solar cycle) over the first two solar cycles of the
historical simulation (i.e. the 1850–1873 mean).

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1952 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

– A background volcanic aerosol should be specified that is likely that it will be impossible to eliminate all artefacts in
results in radiative forcing matching, as closely as pos- quantities such as historical sea level change. For this reason,
sible, that experienced, on average, during the historical and because some models may deviate from these specifica-
simulation (i.e. 1850–2014 mean). tions, it is recommended that groups perform an additional
simulation of the historical period but with only natural forc-
– Models without interactive ozone chemistry should ing included. With this additional run, which is already called
specify the pre-industrial ozone fields from a data set for under DAMIP, the purely anthropogenic effects on sea
produced from a pre-industrial control simulation that level change can be isolated.
uses 1850 emissions and a mean solar forcing averaged The forcing specified in the piControl also has implica-
over solar cycles 8–10, representative of the mean mid- tions for simulations of the future, when solar variability and
19th century solar forcing. volcanic activity will continue to exist, but at unknown lev-
– For models with interactive chemistry and/or aerosols, els. These issues need to be borne in mind when designing
the CMIP6 pre-industrial emissions dataset of reactive and evaluating future scenarios, as a failure to include vol-
gases and aerosol precursors should be used. For models canic forcing in the future will cause future warming and
without internally calculated aerosol concentrations, a sea level rise to be over-estimated relative to a piControl ex-
monthly climatological dataset of aerosol physical and periment in which a non-zero volcanic forcing is specified.
optical properties should be used. This is accounted for by introducing a time-invariant non-
zero volcanic forcing (e.g. the mean volcanic forcing for the
In the CO2 -concentration-driven piControl, the value of the piControl) into the scenarios. This is further specified in the
global annual mean 1850 atmospheric CO2 concentration is ScenarioMIP contribution to this special issue.
prescribed and held fixed during the entire experiment. There These issues, and the potential of different modelling cen-
are some special considerations that apply to control simula- tres adopting different approaches to account for their partic-
tions performed by emission-driven ESMs (i.e. runs with at- ular constraints, highlight the paramount importance of ade-
mospheric concentrations of CO2 calculated prognostically quately documenting the conditions under which this and the
rather than being prescribed). In the esm-piControl simula- other DECK experiments are performed.
tion, emissions of CO2 from both fossil fuel combustion and
land-use change are prescribed to be zero. In this run any A1.3 Abruptly quadrupling CO2 simulation
residual drift in atmospheric CO2 concentration that arises
from an imbalance in the exchanges of CO2 between the at- Until CMIP5, there were no experiments designed to quan-
mosphere and the ocean and land (i.e. by the natural carbon tify the extent to which forcing differences might explain
cycle in the absence of anthropogenic CO2 emissions) will differences in climate response. It was also difficult to diag-
need to be subtracted from perturbation runs to correct for nose and quantify the feedback responses, which are medi-
a control state not in equilibrium. It should be emphasized ated by global surface temperature change (Sherwood et al.,
that the esm-piControl is an idealized experiment and is not 2015). In order to examine these fundamental characteristics
meant to mimic the true 1850 conditions, which would have of models – CO2 forcing and climate feedback – an abrupt
to include a source of carbon of around 0.6 Pg C yr−1 from 4×CO2 simulation was included for the first time as part
the already perturbed state that existed in 1850. of CMIP5. Following Gregory et al. (2004), the simulation
Due to a wide variety of ESMs and the techniques they use branches in January of the CO2 -concentration-driven piCon-
to compute land carbon fluxes, it is hard to make statements trol and abruptly the value of the global annual mean 1850
that apply to all models equally well. A general recommen- atmospheric CO2 concentration that is prescribed in piCon-
dation, however, is that the land carbon fluxes in the emission trol is quadrupled and held fixed. As the system subsequently
and concentration-driven control simulations should be sta- evolves toward a new equilibrium, the imbalance in the net
ble in time and in approximate balance so that the net carbon flux at the top of the atmosphere can be plotted against global
flux into the atmosphere is small (less than 0.1 Pg C yr−1 ). temperature change. As Gregory et al. (2004) showed, it is
Further details on ESM experiments with a carbon cycle are then possible to diagnose both the effective radiative forc-
provided in the C4 MIP contribution to this special issue. ing due to a quadrupling of CO2 and also effective equilib-
The historical time-average volcanic forcing stipulated rium climate sensitivity (ECS). Moreover, by examining how
above for the control run is likely to approximate the much individual flux components evolve with surface temperature
longer term mean. The volcanic aerosol radiative forcing es- change, one can learn about the relative strengths of differ-
timates of Crowley (2000) for the historical period and the ent feedback, notably quantifying the importance of various
last millennium are −0.18 and −0.22 W m−2 , respectively. feedback associated with clouds.
Because the mean volcanic forcing between 1850 and 2014 In the abrupt-4×CO2 experiment, the only externally im-
is small, the discontinuity associated with transitioning from posed difference from the piControl should be the change
a mean forcing to a time-varying volcanic forcing is also ex- in CO2 concentration. All other conditions should remain as
pected to be small. Even though this is the design objective, it they were in the piControl, including any background vol-

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1953

canic aerosols. By changing only a single factor, we can un- surface fluxes and stores of carbon will become a key diag-
ambiguously attribute all climatic consequences to the in- nostic from which one can infer emission rates that are con-
crease in CO2 concentration. sistent with a 1 % yr−1 increase in model CO2 concentration.
The minimum length of the simulation should be This DECK baseline carbon cycle experiment is built upon
150 years, but longer simulations would enable investiga- in C4 MIP to diagnose the strength of model carbon climate
tions of longer-timescale responses. Also there is value, as in feedback and to quantify contributions to disruption of the
CMIP5, in performing an ensemble of short (∼ 5-year) simu- carbon cycle by climate and by direct effects of increased
lations, all prescribing global annual mean 1850 atmospheric CO2 concentration.
CO2 concentration but initiated at different times throughout
the year (in addition to the abrupt-4×CO2 simulation initi- A2 The CMIP6 historical simulations
ated from the piControl in January). Such an ensemble would
reduce the statistical uncertainty with which the effective CMIP6 historical simulations of climate change over the pe-
CO2 radiative forcing could be quantified and would allow riod 1850–2014 are forced by common data sets that are
more detailed and accurate diagnosis of the fast responses largely based on observations. They serve as an important
of the system under an abrupt change in forcing (Bony et benchmark for assessing model performance through evalu-
al., 2013; Gregory and Webb, 2008; Kamae and Watanabe, ation against observations. The historical integration should
2013; Sherwood et al., 2015). Different groups will be able be initialized from some point in the control integration (with
to afford ensembles of different sizes, but in any case each historical branching from the piControl and the esm-hist
realization should be initialized in a different month and the branching from esm-piControl) and be forced by varying
months should be spaced evenly throughout the year. time, externally imposed conditions that are based on obser-
vations. Both naturally forced changes (e.g. due to solar vari-
A1.4 1 % CO2 increase simulation ability and volcanic aerosols) and changes due to human ac-
tivities (e.g. CO2 concentration, aerosols, and land use) will
The second idealized climate change experiment was intro- lead to climate variations and evolution. In addition, there is
duced in the early days of CMIP (Meehl et al., 2000). It is unforced variability which can obscure the forced changes
designed for studying model responses under simplified but and lead to expected differences between the simulated and
somewhat more realistic forcing than an abrupt increase in observed climate variations (Deser et al., 2012).
CO2 . In this 1pctCO2 experiment, the simulation is branched The externally imposed forcing data sets that should be
from the piControl, and the global annual mean CO2 concen- used in CMIP6 cover the period 1850 through the end of
tration is gradually increased at a rate of 1 % yr−1 (i.e. expo- 2014 and are described in detail in various other contribu-
nentially), starting from its 1850 value that is prescribed in tions to this special issue. In the CO2 -concentration-driven
the piControl. A minimum length of 150 years is requested historical simulations, time-varying global annual mean con-
so that the simulation goes beyond the quadrupling of CO2 centrations for CO2 and other long-lived greenhouse gases
after 140 years. Note that in contrast to previous definitions, are prescribed. If a modelling center decides to represent ad-
the experiment has been simplified so that the 1 % CO2 in- ditional spatial and seasonal variations in prescribed green-
crease per year is applied throughout the entire simulation house gas forcings, this needs to be adequately documented.
rather than keeping it constant after 140 years as in CMIP5. Recall from Sect. A1.2 that the conditions in the control
Since the radiative forcing is approximately proportional to should generally be consistent with the forcing imposed near
the logarithm of the CO2 increase, the radiative forcing lin- the beginning of the CMIP historical simulation. This should
early increases over time. Drawing on the estimates of ef- minimize artificial transient effects in the first portion of the
fective radiative forcing (for definitions see Myhre et al., CMIP historical simulation. An exception is that for the CO2 -
2013) obtained in the abrupt-4×CO2 simulations, analysts emission-driven experiments, the zero CO2 emissions from
can scale results from each model in the 1 % CO2 increase fossil fuel and the land-use specifications for 1850 in the esm-
simulations to focus on the response differences in models, piControl could cause a discontinuity in land carbon at the
largely independent of their forcing differences. In contrast, branch point.
in CMIP6 historical simulations (see Sect. A2), the forcing As described in Sect. A1.2, the 1850 esm-piControl should
and response contributions to model differences in simulated be developed for an idealized case that is stable in time and
climate change cannot be easily isolated. balance so that the net carbon flux into the atmosphere is
As in the abrupt-4×CO2 experiment, the only externally small. Meanwhile, the start of the esm-hist in 1850 should
imposed difference from the piControl should be the change be as realistic as possible and attempt to account for the fact
in CO2 concentration. The omission of changes in aerosol the land surface was not in equilibrium in 1850 due to prior
concentrations is the key to making these simulations easier land-use effects (Houghton, 2010; Hurtt et al., 2011). Some
to interpret. modelling groups have developed methods to achieve these
Models with a carbon cycle component will be driven twin goals in a computationally efficient manner, for exam-
by prescribed CO2 concentrations, but terrestrial and marine ple, by performing pre-1850 off-line land model simulations

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1954 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Table A1. Specifications in the DECK and CMIP6 historical simulations.

Experiment Volcanic stratospheric aerosol Solar variability Anthropogenic forcings

amip Time-dependent observations Time-dependent observations Time-dependent observations
piControl Background volcanic aerosol that Fixed at its mean value (no Given that the historical simulations
results in radiative forcing matching, as 11-year solar cycle) over the start in 1850, the piControl should have
closely as possible, that was experi- first two solar cycles of the fixed 1850 atmospheric composition,
enced, on average, during the historical historical simulation (i.e. the not true pre-industrial
simulation (i.e. 1850–2014 mean) 1850–1873 mean)
esm-piControl As in piControl As in piControl As in piControl but with CO2 concen-
tration calculated, rather than prescribed.
CO2 from both fossil fuel combustion
and land-use change are prescribed to be
zero.
abrupt-4×CO2 As in piControl As in piControl As in piControl except CO2 that is 4
times that of piControl
1pctCO2 As in piControl As in piControl As in piControl except CO2 that is
increasing at 1 % yr−1
historical Time-dependent observations Time-dependent observations Time-dependent observations
esm-hist As in historical As in historical As in historical but with CO2 emissions
prescribed and CO2 concentration calcu-
lated (rather than prescribed)

to account for the land carbon cycle disequilibrium before used, the experiment for 1850–2014 will be labelled histori-
1850 and to adequately simulate carbon stores at the start of cal (esm-hist in the case of the emission-driven run) and the
the historical simulation (Sentman et al., 2011). Due to the period from 2015 through near-present will likely be labelled
wide diversity of modelling approaches for land carbon in the historical-ext (esm-hist-ext).
ESMs, the actual method applied by each group to account Even if the CMIP6 historical simulations are extended be-
for these effects will differ and needs to be well documented. yond 2014, all future scenario simulations (called for by Sce-
As discussed earlier, there will be a mismatch in the spec- narioMIP and other MIPs) should be initiated from the end
ification of volcanic aerosols between control and historical of year 2014 of the CMIP6 historical simulation since the
simulations that especially affect estimates of ocean heat up- “future” in CMIP6 begins in 2015.
take and sea level rise in the historical period. This can be Due to interactions within and between the components
minimized by prescribing a background volcanic aerosol in of the Earth system, there is a wide range of variability
the pre-industrial control that has the same cooling effect as on various time and space scales (Hegerl et al., 2007). The
the volcanoes included in the CMIP6 historical simulation. timescales vary from shorter than a day to longer than sev-
Any residual mismatch will need to be corrected, which re- eral centuries. The magnitude of the variability can be quite
quires a special supplementary simulation (see Sect. A1.2) large relative to any given signal of interest depending on the
that should be submitted along with the CMIP6 historical time and space scales involved and on the variable of inter-
simulation. est. To more clearly identify forced signals emerging from
For model evaluation and for detection and attribution natural variability, multiple model integrations (comprising
studies (the focus of DAMIP) there would be considerable an ensemble) can be made where only the initial conditions
value in extending the CMIP6 historical simulations beyond are perturbed in some way which should be documented. A
the nominal 2014 ending date. To include the more recent common way to do this is to simply branch each simulation
observations in model evaluation, modelling groups are en- from a different point in the control run. Longer intervals be-
couraged to document and apply forcing data sets represent- tween branch points will ensure independence of ensemble
ing the post-2014 period. For short extensions (up to a few members on longer timescales. By averaging many different
years) it may be acceptable to simply apply forcing from one ensemble members together, the signal of interest becomes
of the future scenarios defined by ScenarioMIP. To distin- clear because the natural variations tend to average out if the
guish between the portion of the historical period when all ensemble size and averaging period are long enough. If the
models will use the same forcing data sets (i.e. 1850–2014) variability in the models is realistic, then the spread of the
from the extended period where different data sets might be ensemble members around the ensemble average is caused

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1955

by unforced (i.e. internal) variability. To minimize the num-

ber of years included in the entry card simulations, only one
ensemble member is requested here. However, we strongly
encourage model groups to submit at least three ensemble
members of their CMIP historical simulation as requested in
DAMIP.

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1956 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Acknowledgements. We thank the scientific community for their Nauels, A., Xia, Y., Bex, V., and Midgley, P. M., Cambridge Uni-
engagement in the definition of CMIP6 and for the broad partici- versity Press, Cambridge, UK and New York, NY, USA, 2013.
pation in the CMIP5 survey in 2013. We thank the co-chairs and Deser, C., Knutti, R., Solomon, S., and Phillips, A. S.: Communi-
steering committee members of the CMIP6-Endorsed MIPs for cation of the role of natural variability in future North American
their continuous engagement in defining CMIP6, and the modelling climate, Nat. Clim. Change, 2, 775–779, 2012.
groups and wider community for reviewing the CMIP6 design and Eyring, V., Righi, M., Lauer, A., Evaldsson, M., Wenzel, S., Jones,
organization. We thank the WGCM Infrastructure Panel (WIP) C., Anav, A., Andrews, O., Cionni, I., Davin, E. L., Deser, C.,
for overseeing the CMIP6 infrastructure, Martin Juckes for taking Ehbrecht, C., Friedlingstein, P., Gleckler, P., Gottschaldt, K.-D.,
the lead in preparing the CMIP6 data request, and the group of Hagemann, S., Juckes, M., Kindermann, S., Krasting, J., Kunert,
scientists who are producing forcing datasets for CMIP6. Thanks D., Levine, R., Loew, A., Mäkelä, J., Martin, G., Mason, E.,
to Jonathan Gregory for raising awareness about the treatment of Phillips, A. S., Read, S., Rio, C., Roehrig, R., Senftleben, D.,
volcanic forcing in the pre-industrial control experiment and its Sterl, A., van Ulft, L. H., Walton, J., Wang, S., and Williams,
consequence for sea level changes, and to Pierre Friedlingstein, K. D.: ESMValTool (v1.0) – a community diagnostic and perfor-
George Hurtt, Chris Jones, and David Lawrence for help in mance metrics tool for routine evaluation of Earth system models
defining carbon cycle and land-use specifications in the DECK in CMIP, Geosci. Model Dev., 9, 1747–1802, doi:10.5194/gmd-
experiments and CMIP6 historical simulations. Norbert Noreiks 9-1747-2016, 2016.
is thanked for help in drafting the figures. Thanks to our topical Ferraro, R., Waliser, D. E., Gleckler, P., Taylor, K. E., and
editor Julia Hargreaves, to Gavin Schmidt and the other two Eyring, V.: Evolving obs4MIPs to Support the Sixth Coupled
anonymous reviewers, and to everyone who contributed to the Model Intercomparison Project (CMIP6), B. Am. Meteorol.
open discussions for constructive comments. GM and KET were Soc., doi:10.1175/BAMS-D-14-00216.1, online first, 2015.
supported by the Regional and Global Climate Modeling Program Flato, G., Marotzke, J., Abiodun, B., Braconnot, P., Chou, S. C.,
(RGCM) of the U.S. Department of Energy’s Office of Biological & Collins, W., Cox, P., Driouech, F., Emori, S., Eyring, V., Forest,
Environmental Research (BER) (through Cooperative Agreement C., Gleckler, P., Guilyardi, E., Jakob, C., Kattsov, V., Reason, C.,
no. DE-FC02-97ER62402 for GM), and GM received additional and Rummukainen, M.: Evaluation of Climate Models, in: Cli-
support from the U.S. National Science Foundation. The National mate Change 2013: The Physical Science Basis. Contribution of
Center for Atmospheric Research is sponsored by the National Working Group I to the Fifth Assessment Report of the Intergov-
Science Foundation. ernmental Panel on Climate Change, edited by: Stocker, T. F.,
Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Boschung, J.,
The article processing charges for this open-access Nauels, A., Xia, Y., Bex, V., and Midgley, P. M., Cambridge Uni-
publication were covered by a Research versity Press, Cambridge, UK and New York, NY, USA, 2013.
Centre of the Helmholtz Association. Gates, W. L., Boyle, J. S., Covey, C., Dease, C. G., Doutriaux, C.
M., Drach, R. S., Fiorino, M., Gleckler, P. J., Hnilo, J. J., Marlais,
Edited by: J. Hargreaves S. M., Phillips, T. J., Potter, G. L., Santer, B. D., Sperber, K. R.,
Taylor, K. E., and Williams, D. N.: An Overview of the Results
of the Atmospheric Model Intercomparison Project (AMIP I), B.
Am. Meteorol. Soc., 80, 29–55, 1999.
References Geoffroy, O., Saint-Martin, D., Olivie, D. J. L., Voldoire, A., Bellon,
G., and Tyteca, S.: Transient Climate Response in a Two-Layer
Bony, S., Bellon, G., Klocke, D., Sherwood, S., Fermepin, S., and Energy-Balance Model. Part I: Analytical Solution and Parame-
Denvil, S.: Robust direct effect of carbon dioxide on tropical ter Calibration Using CMIP5 AOGCM Experiments, J. Climate,
circulation and regional precipitation, Nat. Geosci., 6, 447–451, 26, 1841–1857, 2013.
2013. Gleckler, P. J., Doutriaux, C., Durack P. J., Taylor K. E., Zhang,
Bony, S., Stevens, B., Frierson, D. M. W., Jakob, C., Kageyama, M., Y., Williams, D. N., Mason, E., and Servonnat, J.: A More
Pincus, R., Shepherd, T. G., Sherwood, S. C., Siebesma, A. P., Powerful Reality Test for Climate Models, Eos Trans. AGU,
Sobel, A. H., Watanabe, M., and Webb, M. J.: Clouds, circulation doi:10.1029/2016EO051663, 2016.
and climate sensitivity, Nat. Geosci., 8, 261–268, 2015. Gregory, J. and Webb, M.: Tropospheric adjustment induces a cloud
Brasseur, G. and Carlson, D.: Future directions for the World Cli- component in CO2 forcing, J. Climate, 21, 58–71, 2008.
mate Research Programme, Eos, Transactions American Geo- Gregory, J. M.: Long-term effect of volcanic forcing on
physical Union, 96, doi:10.1029/2015EO033577, 2015. ocean heat content, Geophys. Res. Lett., 37, L22701,
CMIP Panel: WCRP Coupled Model Intercomparison Project doi:10.1029/2010gl045507, 2010.
(CMIP), available at: http://www.wcrp-climate.org/index.php/ Gregory, J. M., Ingram, W. J., Palmer, M. A., Jones, G. S.,
wgcm-cmip/about-cmip, 2016. Stott, P. A., Thorpe, R. B., Lowe, J. A., Johns, T. C., and
Crowley, T. J.: Causes of climate change over the past 1000 years, Williams, K. D.: A new method for diagnosing radiative forc-
Science, 289, 270–277, 2000. ing and climate sensitivity, Geophys. Res. Lett., 31, L03205,
Cubasch, U., Wuebbles, D., Chen, D., Facchini, M. C., Frame, doi:10.1029/2003gl018747, 2004.
D., Mahowald, N., and Winther, J.-G.: Introduction, in: Cli- Gregory, J. M., Bi, D., Collier, M. A., Dix, M. R., Hirst, A. C.,
mate Change 2013: The Physical Science Basis. Contribution of Hu, A., Huber, M., Knutti, R., Marsland, S. J., Meinshausen, M.,
Working Group I to the Fifth Assessment Report of the Intergov- Rashid, H. A., Rotstayn, L. D., Schurer, A., and Church, J. A.:
ernmental Panel on Climate Change, edited by: Stocker, T. F., Climate models without preindustrial volcanic forcing underes-
Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Boschung, J.,

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/

V. Eyring et al.: Overview of the CMIP6 experimental design and organization 1957

timate historical ocean thermal expansion, Geophys. Res. Lett., Murphy, J. M. and Mitchell, J. F. B.: Transient-Response of the
40, 1600–1604, 2013. Hadley-Center Coupled Ocean-Atmosphere Model to Increasing
Hegerl, G. C., Zwiers, F. W., Braconnot, P., Gillett, N. P., Luo, Y., Carbon-Dioxide. 2. Spatial and Temporal Structure of Response,
Marengo Orsini, J. A., Nicholls, N., Penner, J. E., and Stott, J. Climate, 8, 57–80, 1995.
P. A.: Understanding and Attributing Climate Change, in: Cli- Myhre, G., Shindell, D., Breìon, F.-M., Collins, W., Fuglestvedt,
mate Change 2007: The Physical Science Basis. Contribution of J., Huang, J., Koch, D., Lamarque, J.-F., Lee, D., Mendoza,
Working Group I to the Fourth Assessment Report of the Inter- B., Nakajima, T., Robock, A., Stephens, G., Takemura, T., and
governmental Panel on Climate Change, edited by: Solomon, S., Zhang, H.: Anthropogenic and Natural Radiative Forcing, in:
Qin, D., Manning, M., Chen, Z., Marquis, M., Averyt, K. B., Tig- Climate Change 2013: The Physical Science Basis. Contribution
nor, M., and Miller, H. L., Cambridge University Press, Cam- of Working Group I to the Fifth Assessment Report of the Inter-
bridge, UK and New York, NY, USA, 2007. governmental Panel on Climate Change, edited by: Stocker, T. F.,
Houghton, R. A.: How well do we know the flux of CO2 from land- Qin, D., Plattner, G.-K., Tignor, M., Allen, S. K., Boschung, J.,
use change?, Tellus B, 62, 337–351, 2010. Nauels, A., Xia, Y., Bex, V., and Midgley, P. M., Cambridge Uni-
Hurtt, G. C., Chini, L. P., Frolking, S., Betts, R. A., Feddema, J., versity Press, Cambridge, UK and New York, NY, USA, 2013.
Fischer, G., Fisk, J. P., Hibbard, K., Houghton, R. A., Janetos, O’Neill, B. C., Kriegler, E., Ebi, K. L., Kemp-Benedict, E., Ri-
A., Jones, C. D., Kindermann, G., Kinoshita, T., Goldewijk, K. ahi, K., Rothman, D. S., van Ruijven, B. J., van Vuuren, D. P.,
K., Riahi, K., Shevliakova, E., Smith, S., Stehfest, E., Thomson, Birkmann, J., Kok, K., Levy, M., and Solecki, W.: The roads
A., Thornton, P., van Vuuren, D. P., and Wang, Y. P.: Harmoniza- ahead: Narratives for shared socioeconomic pathways describ-
tion of land-use scenarios for the period 1500–2100: 600 years of ing world futures in the 21st century, Global Environ. Chang.,
global gridded annual land-use transitions, wood harvest, and re- doi:10.1016/j.gloenvcha.2015.01.004, online first, 2015.
sulting secondary lands, Climatic Change, 109, 117–161, 2011. Phillips, A. S., Deser, C., and Fasullo, J.: Evaluating Modes of Vari-
IPCC: Climate Change 2013: The Physical Science Basis. Contri- ability in Climate Models, Eos Trans. AGU, 95, 453–455, 2014.
bution of Working Group I to the Fifth Assessment Report of the Rauser, F., Gleckler, P., and Marotzke, J.: Rethinking the Default
Intergovernmental Panel on Climate Change, Cambridge Univer- Construction of Multimodel Climate Ensembles, B. Am. Meteo-
sity Press, Cambridge, UK and New York, NY, USA, 2013. rol. Soc., 96, 911–919, 2014.
Kamae, Y. and Watanabe, M.: Tropospheric adjustment to increas- Sentman, L. T., Shevliakova, E., Stouffer, R. J., and Malyshev, S.:
ing CO2 : its timescale and the role of land-sea contrast, Clim. Time Scales of Terrestrial Carbon Response Related to Land-
Dynam., 41, 3007–3024, 2013. Use Application: Implications for Initializing an Earth System
Lawrence, B. N., Balaji, V., Bentley, P., Callaghan, S., DeLuca, C., Model, Earth Interact., 15, 1–16, 2011.
Denvil, S., Devine, G., Elkington, M., Ford, R. W., Guilyardi, Sherwood, S. C., Bony, S., Boucher, O., Bretherton, C., Forster, P.
E., Lautenschlager, M., Morgan, M., Moine, M.-P., Murphy, S., M., Gregory, J. M., and Stevens, B.: Adjustments in the Forcing-
Pascoe, C., Ramthun, H., Slavin, P., Steenman-Clark, L., Tous- Feedback Framework for Understanding Climate Change, B.
saint, F., Treshansky, A., and Valcke, S.: Describing Earth system Am. Meteorol. Soc., 96, 217–228, 2015.
simulations with the Metafor CIM, Geosci. Model Dev., 5, 1493– Stevens, B.: Rethinking the Lower Bound on Aerosol Radiative
1500, doi:10.5194/gmd-5-1493-2012, 2012. Forcing, J. Climate, 28, 4794–4819, 2015.
Li, C., Stevens, B., and Marotzke, J.: Eurasian winter cooling in the Stott, P. A., Mitchell, J. F. B., Allen, M. R., Delworth, T. L., Gre-
warming hiatus of 1998–2012, Geophys. Res. Lett., 42, 8131– gory, J. M., Meehl, G. A., and Santer, B. D.: Observational con-
8139, 2015. straints on past attributable warming and predictions of future
Meehl, G. A., Boer, G. J., Covey, C., Latif, M., and Stouffer, R. J.: global warming, J. Climate, 19, 3055–3069, 2006.
Intercomparison makes for a better climate model, Eos, Transac- Stouffer, R. J., Weaver, A. J., and Eby, M.: A method for obtaining
tions American Geophysical Union, 78, 445–451, 1997. pre-twentieth century initial conditions for use in climate change
Meehl, G. A., Boer, G. J., Covey, C., Latif, M., and Stouffer, R. studies, Clim. Dynam., 23, 327–339, 2004.
J.: The Coupled Model Intercomparison Project (CMIP), B. Am. Stouffer, R. J., Eyring, V., Meehl, G. A., Bony, S., Senior, C.,
Meteorol. Soc., 81, 313–318, 2000. Stevens, B., and Taylor, K. E.: CMIP5 Scientific Gaps and Rec-
Meehl, G. A., Covey, C., Taylor, K. E., Delworth, T., Stouffer, R. ommendations for CMIP6, B. Am. Meteorol. Soc., submitted,
J., Latif, M., McAvaney, B., and Mitchell, J. F. B.: THE WCRP 2015.
CMIP3 Multimodel Dataset: A New Era in Climate Change Re- Taylor, K. E., Stouffer, R. J., and Meehl, G. A.: A Sum-
search, B. Am. Meteorol. Soc., 88, 1383–1394, 2007. mary of the CMIP5 Experiment Design, available at: http:
Meehl, G. A., Moss, R., Taylor, K. E., Eyring, V., Stouffer, R. //cmip.llnl.gov/cmip5/docs/Taylor_CMIP5_design.pdf (last ac-
J., Bony, S., and Stevens, B.: Climate Model Intercompar- cess: 13 May 2016), 2009.
isons: Preparing for the Next Phase, Eos Trans. AGU, 59, 77, Taylor, K. E., Stouffer, R. J., and Meehl, G. A.: An Overview of
doi:10.1002/2014EO090001, 2014. Cmip5 and the Experiment Design, B. Am. Meteorol. Soc., 93,
Moss, R. H., Edmonds, J. A., Hibbard, K. A., Manning, M. R., Rose, 485–498, 2012.
S. K., van Vuuren, D. P., Carter, T. R., Emori, S., Kainuma, M., Teixeira, J., Waliser, D., Ferraro, R., Gleckler, P., Lee, T., and
Kram, T., Meehl, G. A., Mitchell, J. F. B., Nakicenovic, N., Riahi, Potter, G.: Satellite Observations for CMIP5: The Genesis of
K., Smith, S. J., Stouffer, R. J., Thomson, A. M., Weyant, J. P., Obs4MIPs, B. Am. Meteorol. Soc., 95, 1329–1334, 2014.
and Wilbanks, T. J.: The next generation of scenarios for climate
change research and assessment, Nature, 463, 747–756, 2010.

www.geosci-model-dev.net/9/1937/2016/ Geosci. Model Dev., 9, 1937–1958, 2016

1958 V. Eyring et al.: Overview of the CMIP6 experimental design and organization

Trenberth, K. and Asrar, G.: Challenges and Opportunities in Water Williams, D. N., Balaji, V., Cinquini, L., Denvil, S., Duffy, D.,
Cycle Research: WCRP Contributions, Surv. Geophys., 35, 515– Evans, B., Ferraro, R., Hansen, R., Lautenschlager, M., and Tren-
532, 2014. ham, C.: A Global Repository for Planet-Sized Experiments and
van Vuuren, D. P., Edmonds, J., Kainuma, M., Riahi, K., Thomson, Observations, B. Am. Meteorol. Soc., doi:10.1175/bams-d-15-
A., Hibbard, K., Hurtt, G. C., Kram, T., Krey, V., Lamarque, J. 00132.1, online first, 2015.
F., Masui, T., Meinshausen, M., Nakicenovic, N., Smith, S. J., Williams, K. and Webb, M.: A quantitative performance assessment
and Rose, S. K.: The representative concentration pathways: an of cloud regimes in climate models, Clim. Dynam., 33, 141–157,
overview, Climatic Change, 109, 5–31, 2011. 2009.

Geosci. Model Dev., 9, 1937–1958, 2016 www.geosci-model-dev.net/9/1937/2016/