ERDC/CRREL M-00-1

US Army Corps
of Engineers®
Engineer Research and
Development Center

Remote Sensing of In-Flight Icing Conditions

Operational, Meteorological, and Technological
Considerations
Charles C. Ryerson January 2000
and Engineering Laboratory
Cold Regions Research

Approved for public release; distribution unlimited.

To Contents
Abstract: Remote-sensing systems that map aircraft opment is needed; differential attenuation and neural
icing conditions in the flight path from airports or aircraft network assessment of multiple-band radar returns are
would allow icing to be avoided and exited. Icing remote- most promising to date. Airport-based radar or radio-
sensing system development requires consideration of meters are the most viable near-term technologies. A
the operational environment, the meteorological envi- radiometer that profiles cloud liquid water, and experi-
ronment, and the technology available. mental techniques to use radiometers horizontally, are
Operationally, pilots need unambiguous cockpit icing promising.
displays for risk management decision-making. Human The most critical operational research needs are to
factors, aircraft integration, integration of remotely assess cockpit and aircraft system integration, develop
sensed icing information into the weather system infra- avoid-and-exit protocols, assess human factors, and
structures, and avoid-and-exit issues need resolution. integrate remote-sensing information into weather and
Cost, maintenance, power, weight, and space concern air traffic control infrastructures. Improved spatial charac-
manufacturers, operators, and regulators. terization of cloud and precipitation liquid-water content,
An icing remote-sensing system detects cloud and drop-size spectra, and temperature are needed, as well
precipitation liquid water, drop size, and temperature. as an algorithm to convert sensed conditions into a
An algorithm is needed to convert these conditions into measure of icing potential. Technology development also
icing potential estimates for cockpit display. Specifica- requires refinement of inversion techniques. These goals
tion development requires that magnitudes of cloud can be accomplished with collaboration among federal
microphysical conditions and their spatial and temporal agencies including NASA, the FAA, the National Center
variability be understood at multiple scales. for Atmospheric Research, NOAA, and the Department
The core of an icing remote-sensing system is the of Defense. This report reviews operational, meteoro-
technology that senses icing microphysical conditions. logical, and technological considerations in developing
Radar and microwave radiometers penetrate clouds and the capability to remotely map in-flight icing conditions
can estimate liquid water and drop size. Retrieval devel- from the ground and from the air.

Cover: Artist’s rendering of possible cockpit display as aircraft enters regions of hazardous icing
conditions. The display would be created from information gathered from either on-board
remote sensors or ground-based sensors uplinking information to the aircraft.

How to get copies of CRREL technical publications:

Department of Defense personnel and contractors may order reports through the Defense Technical Informa-
tion Center:
DTIC-BR SUITE 0944
8725 JOHN J KINGMAN RD
FT BELVOIR VA 22060-6218
Telephone (800) 225-3842
E-mail [email protected]
[email protected]
WWW http://www.dtic.mil/
All others may order reports through the National Technical Information Service:
NTIS
5285 PORT ROYAL RD
SPRINGFIELD VA 22161
Telephone (703) 487-4650
(703) 487-4639 (TDD for the hearing-impaired)
E-mail [email protected]
WWW http://www.ntis.gov/index.html
A complete list of all CRREL technical publications is available from
USACRREL (CEERD-IM-HL)
72 LYME RD
HANOVER NH 03755-1290
Telephone (603) 646-4338
E-mail [email protected]
For information on all aspects of the Cold Regions Research and Engineering Laboratory, visit our
World Wide Web site:
http://www.crrel.usace.army.mil

To Contents
Monograph
ERDC-CRREL M-00-1
US Army Corps
of Engineers®
Cold Regions Research &
Engineering Laboratory

Remote Sensing of In-Flight Icing Conditions

Operational, Meteorological, and Technological
Considerations
Charles C. Ryerson January 2000

Prepared for
NASA GLENN RESEARCH CENTER

Approved for public release; distribution is unlimited.

To Contents
PREFACE
This report was prepared by Charles C. Ryerson, Research Physical Scientist, Snow and Ice Divi-
sion, U.S. Army Cold Regions Research and Engineering Laboratory (CRREL), Engineer Research
and Development Center.
Funding for this report was provided by the NASA Glenn Research Center, Icing Technology
Branch, under NASA customer order no. C-73343-E, In-Flight Remote Sensing Icing Avoidance.
The author thanks A. Reehorst of the NASA Glenn Research Center and G. Koenig of CRREL
for extensive reviews of this report and constructive comments, A. Clifford and the CRREL Library
for literature acquisition, T. Vaughan for the cover art, and M. Bergstad of CRREL for editing.
The contents of this report are not to be used for advertising or promotional purposes. Citation of
brand names does not constitute an official endorsement or approval of the use of such commercial
products.

ii
To Contents
 CONTENTS
Preface ....................................................................................................................................... ii
Glossary ..................................................................................................................................... v
1.0 Executive Summary ............................................................................................................. 1
2.0 Introduction .......................................................................................................................... 2
2.1 Purpose ........................................................................................................................... 2
2.2 Scope .............................................................................................................................. 2
2.3 Goal ................................................................................................................................ 2
2.4 Relevance ....................................................................................................................... 3
2.5 Philosophy ..................................................................................................................... 4
2.6 Organization of this report ............................................................................................. 4
3.0 Operational requirements ..................................................................................................... 5
3.1 Summary ........................................................................................................................ 5
3.2 Introduction .................................................................................................................... 6
3.3 Pilot needs and human factors ....................................................................................... 6
3.4 Manufacturers and operators.......................................................................................... 8
3.5 Regulatory issues, weather forecasting, and traffic management .................................. 9
3.6 Test beds and platforms .................................................................................................. 10
4.0 Meteorological sensing requirements .................................................................................. 11
4.1 Summary ........................................................................................................................ 11
4.2 Introduction .................................................................................................................... 13
4.3 Characterization needs ................................................................................................... 14
4.4 In-situ instrumentation ................................................................................................... 24
4.5 Terminology ................................................................................................................... 28
5.0 Technology development ..................................................................................................... 29
5.1 Summary ........................................................................................................................ 29
5.2 Introduction .................................................................................................................... 31
5.3 Radar .............................................................................................................................. 31
5.4 Passive microwave radiometers ..................................................................................... 39
5.5 Lidar ............................................................................................................................... 43
5.6 Temperature measurement ............................................................................................. 44
6.0 Recommendations ............................................................................................................... 46
6.1 Operational needs research ............................................................................................ 47
6.2 Meteorological needs research ....................................................................................... 47
6.3 Technology needs research ............................................................................................ 47
7.0 Government roles, missions, and collaborative activities ................................................... 47
7.1 NASA ............................................................................................................................. 47
7.2 FAA ................................................................................................................................ 47
7.3 NCAR ............................................................................................................................ 47
7.4 NOAA ETL .................................................................................................................... 47
7.5 Department of Defense .................................................................................................. 48
8.0 Implementation .................................................................................................................... 48
9.0 Literature cited ..................................................................................................................... 48

To Contents iii
Appendix A: Synopsis of operational information needs, state of knowledge, strengths,
and weaknesses .................................................................................................................... 59
Appendix B: Synopsis of sensing needs, state of knowledge, strengths, and weaknesses ........ 63
Appendix C: Synopsis of sensor technology needs, state of knowledge, strengths,
and weaknesses .................................................................................................................... 65
Abstract ...................................................................................................................................... 69

ILLUSTRATIONS
Figure
1. Aircraft icing paradigm ..................................................................................................... 4
2. Sensing range affects avoid-and-exit capability ................................................................ 14

TABLE
Table
1. Relationships between radar band, frequency, wavelength, and weather parameter
sensed .......................................................................................................................... 31

iv
To Contents
 GLOSSARY

AGATE: Advanced General Aviation Technology Experiment.

ceilometer: Instrument that remotely measures the height of cloud base above ground
level.
disdrometer: Instrument for measuring the sizes of raindrops.
exceedance conditions: Conditions outside of the envelope of conditions defined by FAR 25,
Appendix C.
FAR 25, Appendix C: FAA Federal Aviation Regulation 25, Appendix C, defining the range
of liquid water, drop size, temperature, and exposure conditions for
which aircraft are certified for flight in icing.
free flight: FAA concept of autonomous aircraft flight, with aircraft–ground sepa-
ration provided by onboard aircraft sensors rather than by air traffic
control (ATC).
pireps: In-flight pilot reports of flying conditions, including icing.
prediction detection: Detection of icing condition remotely; remote-sensing system predic-
tions that icing conditions are in the flight path.
range gate: Distance between consecutive radar measurements along a radial, typ-
ically about 100 to 500 m, and controlled by pulse duration.
reactive detection: Detection of icing conditions in situ; reaction of the in-situ instruments
indicates that icing is occurring.
supercooled large drops (SLDs): Drops larger than typical cloud drops defined in FAR 25, Appendix C;
typically drizzle drops ranging in diameter from 50 to 500 µm.

v
To Contents
To Contents
 Remote Sensing of In-Flight Icing Conditions
Operational, Meteorological, and Technological Considerations
CHARLES C. RYERSON

1.0 EXECUTIVE SUMMARY

In-flight icing is a significant aviation risk despite sities. The absolute range of cloud liquid-water content
improving icing forecasts and onboard ice protection in clouds of various genera is generally well known,
systems. A remote-sensing system designed to detect but its vertical and horizontal distribution within clouds
icing conditions in the flight path could allow aircraft and cloud masses is not well understood. The cloud gla-
to avoid and exit hazardous conditions. Ground-based ciation process is also imperfectly understood, as is the
near airports or airborne, such systems would be most effect of mixed-phase conditions on aircraft icing and
useful to low, slow-flying aircraft that frequently on cloud remote sensing. The shape of the drop-size
encounter icing, such as turboprops and helicopters. spectrum is not well characterized in icing conditions,
Development of an icing remote-sensing system especially in supercooled large drops, and the median
requires consideration of the operational environment volume diameter does not often provide an adequate
within which it is used, the meteorological environment description of the drop-size spectrum, especially if the
it senses, and the technology available for sensing icing distribution is bimodal. Though it is generally known
conditions. that static temperature changes more rapidly vertically
Operationally, pilots need information in the cock- than horizontally, there has been very little characteri-
pit for making risk-management decisions. Displays must zation of temperature distribution in icing conditions.
evoke proper pilot decisions and provide clear, unam- The ability to detect temperature change ahead of an
biguous warnings of severe conditions for avoidance. aircraft is critical, because the temperature within liquid
Human factors and cockpit and aircraft integration water determines whether icing will occur. Character-
issues must be developed in addition to avoid-and-exit ization of icing conditions is expensive, typically requir-
protocol and training. Dispatchers and meteorologists ing research aircraft. More reliable and less costly
need integration of icing remote-sensing systems into instrumentation is needed to replace first-generation opti-
the weather system infrastructure. Cost, maintenance, cal probes, and to reduce cost, coordination is needed
power, weight, and space are a concern of manu- with other federal programs that make cloud microphys-
facturers, operators, and regulators, as is the evalua- ical measurements. Such cooperation has started
tion of aircraft flight envelopes in icing conditions. between the Canadian Atmospheric Environment Ser-
An icing remote-sensing system detects conditions vice, Transport Canada, NASA, and the FAA.
conducive to icing, including cloud and precipitation The core of an icing avoidance system is the tech-
liquid-water content, the drop-size spectrum, and tem- nology used to sense icing microphysical conditions.
perature. An icing metric algorithm would convert these Radar and microwave radiometers are the most viable
measurements into an estimate of icing potential for technologies. Ranging capability makes radar an attrac-
cockpit display. To develop specifications, the absolute tive technology for detecting liquid water, drop size,
magnitudes of cloud microphysical conditions and the and possibly temperature. Multiple-band radars, such
spatial and temporal variability of icing weather condi- as X and Ka bands to retrieve liquid-water content using
tions, must be understood at multiple scales. Icing cloud differential attenuation techniques and X, Ka, and W
microphysics have been measured since the 1940s by bands to retrieve liquid water and drop size using neu-
NACA, NASA, the FAA, NCAR, and several univer- ral nets, currently appear most viable. Information-

To Contents
retrieval algorithms, noise, and Mie scattering present requirements of the system, the environment to be
major radar technological challenges. Although radar can sensed, and the technology available to accomplish the
be used in a variety of orientations, scanning vertically task. This report identifies the state of knowledge,
from the ground or horizontally from the air, its size, strengths, weaknesses, major issues, barriers, and oppor-
weight, and power demands make ground-based radar a tunities, and it identifies the research and investment
more viable near-term technology. needed to create a prototype icing remote-sensing sys-
Microwave radiometer development is less mature tem.
than radar technology. However, the recent introduction
of a radiometer that scans and profiles temperature, water 2.2 Scope
vapor, and cloud liquid water, and experimentation with This report provides a framework for the develop-
techniques to use radiometers in a horizontal sensing ment of a plan for creating in-flight icing-avoidance
mode, in addition to the more traditional vertical or near- remote-sensing systems for use in the national airspace.
vertical modes, are promising. Microwave radiometers There are three strategic elements:
are passive, an advantage to the military, but they lack • Identifying the operational needs of pilots, opera-
absolute ranging capability—a disadvantage. Identifica- tors, manufacturers, and regulators as to functional
tion of cloud glaciation and drop size with polarization requirements, system utilization, aircraft integra-
is an additional useful radar and microwave radiometer tion, and human factors
capability. • Identifying sensing requirements
Lidar is not considered a viable technology for remote • Identifying technologies, and their state of devel-
sensing of icing conditions because it cannot sense deeply opment, for an integrated sensing system.
into optically thick clouds. Remote detection of temper-
ature is possible with microwave radiometers and RASS 2.3 Goal
from the ground, but perhaps only by using microwave This report is intended to provide background infor-
radiometers from aircraft. Considerable development is mation to facilitate the creation of a development plan
needed in this area. to improve aircraft operational capabilities and safety
The most critical needs in operational research are to in icing environments. This will be accomplished by
assess cockpit and aircraft system integration, develop developing remote-sensing systems that provide pilots
avoid-and-exit protocol, and assess the human factors with information about the location and intensity of in-
in using remotely sensed icing information. In addition, flight icing hazards, giving them the ability to avoid
remotely sensed icing information must be integrated and exit icing expediently. This will
into weather and air traffic control infrastructures, and air-
craft flight envelopes and the hazard of icing to aircraft • Increase safety
performance in these envelopes need better definition. • Reduce delays
Improved absolute and spatial characterization of cloud • Increase aircraft utilization
and precipitation liquid-water content, drop-size spectra, • Increase military readiness.
and temperature are needed to develop remote-sensing One method of either avoiding or escaping icing is
system specifications. An icing metric must also be devel- to sense remotely, either from the ground or from air-
oped that will allow the sensed microphysical conditions craft, atmospheric icing potential (Ryerson 1996, 1997,
to be converted into a measure of icing potential for air- 1998). This requires scanning the airspace ahead of an
craft. Technology development requires refinement of aircraft for supercooled water and presenting that infor-
inversion techniques for unambiguously retrieving mation to the pilot in a manner consistent with effi-
liquid-water content, drop size, and temperature from cient cockpit risk assessment.
clouds and precipitation. These goals can be accom- Currently, no dedicated system exists for remotely
plished with strong leadership and collaboration among sensing the icing potential in a projected flight path for
federal agencies, including NASA, the FAA, the National an individual aircraft. A remote-sensing system that
Center for Atmospheric Research, NOAA, and DoD. advises pilots of the icing risk ahead of an aircraft will
Partnership between government and industry will bring be an information management system that senses the
viable technologies to prototype and to market. environment, processes the sensed information, and
presents it in a useful form. This requires that the proper
environmental parameters be sensed with an accuracy
2.0 INTRODUCTION
suitable for providing useful information, that the infor-
2.1 Purpose mation be processed with sufficient speed to assist pi-
Development of a remote-sensing icing-avoidance lots, and that information be presented in a manner that
avionics system requires assessment of the operational aids pilots in making icing risk-management decisions.

2
To Contents
 2.4 Relevance pilots in instrument meteorological conditions (IMC)—
Current methods of avoiding icing, including meteor- a pilot training problem—many of the accidents may
ological and pilot reports, are extremely ineffective be preventable with onboard icing-avoidance systems
(Erickson 1997). Icing information is not provided with (Bertorelli 1992). Even a VFR pilot may be able to avoid
the detail, accuracy, and timeliness needed for commer- the most serious of icing conditions—freezing drizzle
cial and private aircraft to avoid icing conditions effi- or freezing rain—with onboard remote-sensing icing
ciently. As a result, either aircraft cannot fly or large avoidance capability. An instrument flight rule (IFR)
areas of potentially flyable airspace must sometimes be pilot could avoid icing within IMC, or at least avoid
avoided because of inadequate spatial and temporal reso- conditions that tax aircraft ice-removal systems.
lution of forecasts. Military aviators and civil aviators Helicopters flying low-altitude missions to service
in the Far North where bush flying is common also need offshore oil rigs and in search-and-rescue operations,
to be able to avoid icing autonomously because fore- for example, are particularly vulnerable to icing. In addi-
casts are often unavailable in operational areas (Owen tion, limited altitude capability, low speeds, rotating
1997). In addition, increased use of laminar flow air- components, and generally low power reserves make
foils and more efficient engine designs less tolerant of them more susceptible to ice than fixed-wing aircraft
contaminants make some aircraft more susceptible to icing. (Manningham 1991). However, helicopters’ low speed
As air traffic increases in volume, new aircraft designs are and maneuverability may be an asset if they are
implemented, and new routes are established, more air- equipped with an icing-avoidance system because they
craft that are less tolerant to contamination may be have more flexible course- and altitude-changing capa-
exposed to icing. To increase aviation safety and bility than fixed-wing aircraft. As recently as 1995, only
efficiency, and to increase military readiness and air su- one commercial helicopter was icing-certified, the Aero-
periority, improved methods of avoiding and exiting spatiale Super Puma (AHS 1995).
icing are needed. In-flight icing is not generally considered a problem
Aircraft flying at 400 knots or greater, which includes in Army aviation, despite problems in Bosnia, because
most jets, generally do not have icing problems because most missions are flown in warm climates, missions
they typically have heated leading edges and fly above are not flown if ice is predicted, and icing is so infre-
most icing (Taylor 1991). However, jets on approach quent that readiness is little affected. However, about 9%
and departure, 300-kt turboprops, piston aircraft, and of Army medevac flights in Alaska are canceled due to
helicopters are all susceptible. Turboprops fly exclu- icing, and medevac commanders give icing avoidance
sively at lower altitudes and are thus exposed to ice for a high priority. Mayer et al. (1984) found about 525
extended periods. Few light piston-engine aircraft have icing-related mishaps in the Navy between 1964 and
deicing capability. Helicopters are probably the most 1984, with about 70% due to in-flight problems and
threatened of all aircraft because of their unique aero- nearly all due to foreign-object damage from ice. Acci-
dynamics and mission requirements and because they dent reports in recent years suggest that the Navy has
typically lack deicing capability. had fewer icing problems, with more reports of hail-
Pilots, operators, and manufacturers typically do not impact damage than airframe ice accretion problems,
know when most aircraft reach their performance limits but Lef et al. (1994) state that the Navy is concerned
in icing (Erickson 1997). There are generally few clues about the icing threat to carrier-launched aircraft and that
provided to the pilot that indicate how close an aircraft helicopter icing accidents are not infrequent. Air Force
is to those limits. This is of particular concern for air- transport aircraft have also experienced icing problems,
craft operation outside of FAA FAR 25, Appendix C, for example, in tropical cumulus clouds at high alti-
design guidelines. The result is that many pilots may tudes. The Coast Guard reports problems with icing in
unknowingly operate their aircraft at or near safety limits search-and-rescue and enforcement missions (Yatto
when in many icing situations, despite the availability 1997).
of onboard protection systems. Military and civilian unmanned aerial vehicles
Fortunately, transport-category aircraft are rarely lost (UAVs) are special cases in need of icing avoidance sys-
to icing, although there are reported incidents (Engel- tems. UAVs, especially high-altitude, long-endurance
berg and Bryant 1995), but private general aviation does UAVs, may be required to seek routes through icing
not fare as well. Aviation magazines carry many reports conditions autonomously (Siquig 1993). Onboard
of private general-aviation icing incidents and accidents. weather-sensing systems could be coupled with autono-
Between 40 and 60 private general-aviation accidents mous controls to allow UAVs to avoid or minimize
annually are attributed to in-flight structural icing, about icing, which impacts them more severely than it does
50% of which are fatal (AVEMCO 1983, Taylor 1991). conventional aircraft because of their low power and
Although about 50% of these are visual flight rule (VFR) high-efficiency airfoils.

3
To Contents
 In the future, several changes in flight activity could early in the research and development process on scen-
affect vulnerability to icing: ario and training aids, which may help establish the
direction of technology development.
• Implementation of Free Flight may reduce air traf-
Although it may not be absolutely necessary, devel-
fic control surveillance of aircraft routes, making
opment of an icing avoidance avionics system also
pilots more responsible for weather avoidance.
requires an understanding of pilot decision-making pro-
• Increased commuter aircraft activity will create
cesses. The form in which information is delivered to
more flights by smaller aircraft at lower altitudes
pilots may subconsciously affect their decision-making
and slower speeds, increasing vulnerability to
process (Hansman 1997). Understanding how pilots
icing.
make decisions for avoiding or coping with in-flight haz-
• Increased military emphasis on helicopters and
ards will affect the development of a decision-support
UAVs, both uniquely vulnerable to icing, is
system. Appropriate paradigms for icing avoidance may
expected.
be current fielded thunderstorm and wind-shear avoid-
• Increased air traffic is also expected globally, so
ance systems. Pilots generally view any information
to reduce the number of icing accidents the acci-
beyond that currently available in the cockpit as useful
dent rate must decrease (Brown and Dorr 1997).
(Erickson 1997), but the kind of information desired
2.5 Philosophy must be identified: too much or inappropriate informa-
Icing remote-sensor development is driven essen- tion could confuse pilots.
tially by one question: What icing information does the The pilot’s needs determine the information provided
pilot need to make better risk-management decisions? by a remote-sensing system, but the atmospheric envi-
A remote-sensing system that reports icing potential ronment and the physics of icing and aircraft flight deter-
ahead of an aircraft to a pilot is a decision-support sys- mine what information must be sensed to create the
tem that provides information needed to make decisions information the pilot needs. Pilots are concerned about
regarding flight safety. Pilots’ needs drive the develop- flight safety and are thus concerned about the perfor-
ment process because they are the ultimate users of mance of the aircraft should it ice. Aircraft performance
information generated by the system. The development changes in response to ice accretion on the airframe.
process, however, must work within the restrictions and Weather is the phenomenon that causes changes in air-
opportunities provided by the aircraft, regulators, sens- craft performance by providing conditions conducive to
ing technology, and meteorology. ice formation on an airframe. Thus, the ice accretion pro-
Because pilot information needs are the primary dri- cess may be viewed as an input–process–response sys-
vers of the process, an early development requirement tem (Fig. 1). Weather is processed by the aircraft to pro-
should be to determine what pilots need to know. Do duce ice on the airframe, which, in turn, influences air-
pilot information needs change with platform, mission, craft performance.

Weather
Weather Aircraft
Aircraft Performance
Performance

Input
Input Process
Process Response
Response
Figure 1. Aircraft icing paradigm.

airspace class, mode of flight (i.e., approach, departure, Sensing requirements are independent of specific air-
or cruise), or some other factor? Is a ground-based sys- craft, because different aircraft process identical weather
tem sufficient, or would an aircraft-mounted sensing conditions in different ways: The same weather condi-
system add significant value? What spatial and temporal tions may produce different icing conditions on a light
resolutions are needed, and how do pilots prefer to view piston-engine aircraft than on a jet transport or a helicop-
the data—as plan, profile, or perspective views, and as ter. And identical weather conditions may produce dif-
individual temperature and liquid-water content maps, ferent icing conditions on an aircraft in different flight
or as composite maps of icing potential? How should configurations depending upon power application, angle
icing potential be expressed, and how should hazard of attack, skin temperature, and other factors. Although
areas be identified? Though all of this information is the meteorological conditions may be identical, the pro-
not needed to begin research in all areas, it does set the cessed information provided by the remote-sensing sys-
stage and reduces the possibility of misdirecting tem should be aircraft-specific to allow the pilot to antici-
research and development. It also allows work to begin pate potential aircraft performance changes due to icing.

4
To Contents
 2.6 Organization of this report icing is most frequently encountered. Because wind-
Broadly, research and development for remotely shear alert systems have evolved since the development
detecting icing conditions can be placed in three cate- of cockpit resource-management concepts, they may
gories: operations, meteorology, and technology. Op- serve as useful analogs for designing an effective pilot
erations includes the human/machine interface, interface for icing avoid-and-exit advisory systems.
regulatory issues, avoid/escape strategies, aircraft inte- Research is needed in this area. In addition, unlike wind
gration, training, and terminology. Meteorology involves shear, which is a directly sensed threat to aircraft, icing
atmospheric environment and characteristics that must does not occur until aircraft enter icing conditions, pro-
be sensed. Technology refers to the remote-sensing cess cloud microphysical conditions, and create ice on
systems that may be able to sense icing potential. Sub- the airframe (Fig. 1). Thus, icing potential is a virtual
areas of research and development needed are identi- phenomenon, and the most appropriate methods for
fied within each primary category. This report gives an quantifying, analyzing, and displaying the virtual threat
overview of the state of the art, describes barriers and must be determined. The types of display, terminology,
opportunities to development, and recommend devel- methods of indicating potential icing intensity, sensor
opment directions. range, resolution, accuracy, and refresh rate and warning
time needed are a function of airspace class, aircraft
type and configuration, and mode of flight. This mix of
3.0 OPERATIONAL REQUIREMENTS
conditions needs to be considered in developing opti-
3.1 Summary mal pilot information systems, training protocol, and
Pilots are risk managers. When it is a question of sensing systems.
flight safety, they want clear, unambiguous informa- Cost, weight, space, power, and maintenance are
tion about the location and intensity of weather threats some of the concerns of aircraft manufacturers and
before they enter them. A top-level requirement with operators. The aircraft most needing protection—com-
regard to icing is to provide pilots with a decision-support muters, helicopters, and light aircraft—offer the fewest
system specific to the remote sensing of icing potential of these resources, so for airborne sensing systems, the
ahead of aircraft. Standoff guidance about icing poten- need is to provide the greatest benefit for the least
tial could be provided from satellite and ground-based impact. The spatial and temporal threat of icing is gener-
sensors uplinking information to the cockpit or from ally small when viewed annually, so it is probable that
aircraft-mounted remote-sensing systems. However, icing remote-sensing systems will not be installed in
satellites still do not have the capability of providing lieu of competing avionics or weapons systems. They
high spatial- and temporal-resolution icing information. will probably be used only if mandated or required
Ground-based sensing systems at airports would be because of market or extreme safety pressures.
most cost-effective per aircraft served, protect the most Remote sensors may provide pilots with the location
critical phases of flight, and are systems for which and intensity of icing potential ahead of their aircraft.
sensing technologies are most mature. Aircraft-based However, pilots must also be able to determine how
systems would be most costly per aircraft and the tech- this icing potential will affect safety, because a deci-
nologies are least mature, but aircraft would be pro- sion to enter or avoid the sensed conditions is a func-
tected in all phases of flight, especially when arriving tion of the aircraft’s ability to operate in icing. Aircraft
or departing from small, remote airports that do not have are certified for flight in icing conditions according to
remote-sensing systems. atmospheric criteria specified in FAA FAR 25, Appen-
Ground-based systems should be developed first, dix C, but pilots need to know how aircraft respond to
because of their near-term technological maturity and conditions outside of Appendix C. They also need to
cost effectiveness as an operational test bed, and to pro- know how much additional icing an iced aircraft can
tect congested airport approach and departure areas. tolerate. That is, they need to know how much ice is
Satellite-based sensors need continued development to necessary to produce unsafe operating conditions. Air-
supplement local sensing systems and to provide pro- craft may have to be tested outside of Appendix C con-
tection during cruise flight where spatial and temporal ditions to determine their operational limits. In addi-
detail may not be as critical. Aircraft-based systems need tion, the development of smart aircraft-monitoring sys-
the greatest development, but they offer the greatest tems may also be needed to guide the pilot’s decision
potential for providing the information pilots need. to enter or avoid icing.
Pilots need information, not data. Thus, remote- An important side benefit of onboard icing remote-
sensing systems must provide clear, simple displays that sensing systems is the potential for downlinking weather
reduce and do not add to flight-management demands information to other aircraft, air traffic controllers, and
in critical approach and departure flight regimes where meteorologists. Downlinked weather information, from

5
To Contents
both remote-sensing and in-situ sensors aboard aircraft, 3.3 Pilot needs and human factors
will improve temporal and spatial accuracy beyond what Pilots have three concerns about the icing environ-
is now possible with pilot reports. Accurate downlinked ment (Vigeant-Langlois and Hansman 1999):
cloud liquid-water content, drop sizes, and temperature • How to recognize that they are approaching or are
will improve icing forecasts and provide more accurate in icing conditions hazardous to their aircraft
icing warnings for other aircraft. Up- and downlinking • How to avoid icing
of weather information to and from the cockpit are areas • How to escape icing.
of research being addressed by the Advanced General
Aviation Technology Experiment (AGATE) program In-flight icing is a frequent topic in pilot safety briefs
and by NASA Langley Research Center in the Aviation and popular literature (Buck 1988, Collins 1989,
Weather Information (AvWIN) program. Schuyler 1989, Taylor 1991, Bertorelli 1992, Horne
Efficient, cost-effective methods for testing remote- 1994, and others). Over the years, rules of thumb, advice,
sensing technology, testing display and information- and regulatory requirements have created ad hoc proto-
management techniques, and developing avoid-and-exit cols for icing avoidance and escape. As a result, most
protocol and training are necessary for development and pilots manage to avoid forecasted icing by not flying,
certification. Sensors may be tested from ground-based or by making route deviations to avoid the conditions,
test beds and airborne platforms. Ground-based test beds or they encounter ice and escape safely through good
include wind tunnels, spray rigs, and mountain-top test fortune. Tales of icing mishaps and escape are com-
sites. The advantages of ground-based test beds include mon fare in winter aviation literature (Creley 1990,
cost, accessibility, and control of test conditions. Moun- McClean 1992). Many pilots do not survive icing
tain-top test sites allow less control over conditions, encounters because they did not recognize that they
but they do provide the variability of natural icing. Wind were in ice until too late, and they did not have the
tunnels and spray rigs cannot provide the spatial con- ability or capability to escape once immersed.
ditions necessary for testing remote sensors, but they One of the larger causes of this avoidance and escape
can be used to test in-situ sensors and the environmen- problem is the quality of icing forecasts. Most pilots
tal effects of icing conditions on sensors and airfoils. who want to avoid icing cannot, because icing fore-
Airborne platforms provide the best environment for casts do not have sufficient accuracy as to spatial, tem-
testing aircraft-based remote sensors. Overall, a combin- poral, and intensity criteria (Erickson et al. 1996, Green
ation of ground and airborne test beds will be necessary et al. 1996, Stack 1996, Clark 1997). Forecasts are often
to test both ground-based and airborne remote-sensing made conservatively, on the side of safety, to minimize
systems. accidents and compensate for inadequacies in forecast
procedure. However, this causes aircraft to not fly or to
3.2 Introduction divert when it may not be necessary because large areas
Operations establish the functional requirements of forecasted for icing may not have ice or even clouds.
a remote-sensing system designed to detect icing. The As a result, the aviation system is less efficient, though
goal of system development is to improve the safety safer.
and efficiency of aircraft operations. However, opera-
tions is a complex, multifaceted problem. A compre- 3.3.1 Pilot needs
hensive review of general operational needs in icing Pilots need information for making risk-management
environments is presented by Brayton and Hakala decisions about in-flight icing. They want to be able to
(1996). determine whether they are in or about to enter icing,
One element, and ultimately the most important ele- because icing clues are often not visible from the cock-
ment, of operations has to do with pilots and their needs pit windows (Erickson 1997). Pilots also need to know
(Vigeant-Langlois and Hansman 1999). Although pilots the location of icing and how intense the icing might
are on the leading edge of the icing problem because be without dipping their wings into it, as is now neces-
they are actually within the icing environment, they deal sary (Green et al. 1996). Onboard, in-situ ice detectors
with more than icing. Systems from which pilots seek (reactive detection systems) are a solution to the prob-
icing guidance must be designed in a manner that best lem of determining exposure to icing for many fixed-
suits the operational requirements of the cockpit envi- wing aircraft and a few helicopters (Bracken et al. 1996),
ronment and that effectively and efficiently helps pilots but even these systems require that the aircraft enter
make management decisions. This includes the design icing conditions before determining that they are in
of the display and information delivery system and train- hazardous conditions.
ing in its use. Pilots need clear, unambiguous guidelines as to when
severe conditions are entered or warning that severe

6
To Contents
 conditions lie ahead with sufficient lead time to avoid 3.3.2 Human factors
them (Erickson 1997). According to Coleman (1996) Development of icing-avoidance avionics is a
of the Regional Airline Association, the answer to main- human-centered development process because the intent
taining safety in icing conditions is to locate severe icing is to display information to pilots. Thus, information
accurately and then avoid it. Because icing forecasts needs must be assessed, and perceptual issues vs.
cannot provide the needed accuracy at the present time, display design must be considered (Hansman 1997,
remote detection or standoff (prediction detection) sys- Vigeant-Langlois and Hansman 1999). A useful
tems may (Green et al. 1996). approach to designing a human/machine interface for
Standoff guidance of icing conditions ahead of air- icing avoidance is to review issues addressed in other
craft could be provided in at least three ways, all using weather-avoidance areas. Recent developments of
a form of remote-sensing system. One method under onboard wind-shear alert systems and other weather-
development utilizes satellite remote-sensing to map avoidance systems, especially since cockpit resource
icing potential and uplink information to aircraft (Lee management has been recognized as a consideration in
and Clark 1995, Vivekanandan et al. 1996, Lee 1997, single- and multiple-pilot cockpits, may serve as reason-
Thompson et al. 1997, Curry and Liu 1992). Satellite- able analogs. One important finding of cockpit resource-
derived information can be used to delimit areas with management research has been that automation can
liquid water, subfreezing temperatures, and cloud cover. cause human error as well as reduce it (Helmreich
Analyses could be accomplished in near real-time and 1997). Because icing-avoidance systems are likely to
would provide a useful predictive detection capability. be in greatest demand during the approach and depar-
Another method, also under development, utilizes ture phases of flight, balancing the distraction against
ground-based sensors at airports to map icing condi- the aid provided by an advisory system is critical to
tions in approach and departure areas (Gary 1983, Deck- its usefulness. A poorly designed interface may actu-
er et al. 1986, Stankov et al. 1992). Remote sensors to ally produce a hazard to flight, so proper human/
detect temperature profiles, cloud boundaries, liquid- machine interface design is nearly as crucial to final
water content, drop-size spectra, and cloud phase from success as is the ability to sense cloud microphysics
the ground are nearly available. Walter and Moynihan accurately.
(1997) even propose a mobile system for military use. Icing is probably a more complex problem than wind
An airport-based system would serve all aircraft, util- shear and convective turbulence because it is less com-
ize largely existing technologies or technologies that pact geographically, it is a hazard in all modes of flight,
are nearing maturity, and serve the phases of flight most and there is significant variation in the ability of differ-
likely to experience icing. It would be a cost-effective ent aircraft to cope with the hazard. However, there may
approach, considering cost per aircraft served (Owen be similarities. Wanke and Hansman (1991) evaluated
1997). graphical displays of microburst alerts from both
A third system would be an in-flight, aircraft-mounted ground-based and airborne detection systems. The
remote detection system (Sand and Kropfli 1991; Fourn- issues ranged from display clarity to pilot response to
ier 1993; Siquig 1993; EWA 1996; Ryerson 1996, 1997, alerts. Questions addressed involved the visual clutter
1998). An airborne system would require elements simi- of adding alerts to existing navigational displays, val-
lar to a ground-based system, with the ability to detect ue of single- vs. multiple-level intensity display, value
temperature, liquid-water content, and drop spectra. of indicating the alert source (ground or airborne), and
However, the technologies may be quite different effect of alert source on pilot procedural response.
because of their use on a small moving platform, scan- Active airline pilots were tested in a realistic transport-
ning primarily horizontally, and operating within aircraft flight simulator. The results showed that multi-
restricted power and weight limits. level intensity displays are desirable, the source of infor-
Pilots need better information than is now available, mation was not important because confidence was
and they require information that is easily understood placed in the alert whatever the source, and correlation
and provides options (Vigeant-Langlois and Hansman with other information was not important. The study
1999). An ability to see through an icing weather sys- also suggested needed training areas, because pilots
tem and map the extent of a threat area would be opti- often responded to alerts with evasive action that was
mal, but any more information than is now available inappropriate or not necessary.
would be welcome (Clark 1997). A warning time of 1 The amount of information displayed in the Wanke
to 5 minutes, preferably integrated into an existing dis- and Hansman (1991) study did not appear to be an issue.
play system, of areas that are of risk to aircraft would That is not always the situation, however, and it may
be most useful, and even no warning time may be accept- be related to single- vs. multiple-pilot cockpit environ-
able (Erickson 1997). ments. For example, Svensson et al. (1997) evaluated

7
To Contents
the effects of information complexity on fighter pilot alert system is the warning time provided for pilot reac-
performance in the Swedish air force. They found that tion. The warning time needed may be a function of
even moderately complex information, measured by the airspace class, aircraft type, mission, and mode of flight.
amount of information provided, interfered with flight Though any warning time, no matter how short, may
tasks. The critical flight measure was an ability to main- be helpful (Erickson 1997), there may be minimum
tain altitude above undulating terrain. When informa- warning times that are more acceptable than others.
tion load increased to more than 8 to 10 items, pilots Anderson and Carbaugh (1993) address this problem
could no longer integrate information and still fly the for wind-shear-alert systems and consider it a critical
airplane. The authors indicate, through numerous exam- factor in how pilots judge the value of an alert system.
ples, that humans have severe limitations in what they Vigeant-Langlois and Hansman (1999) address warning
can receive, process, and remember, and the 8 to 10 distances for icing, with commuter pilots reporting as
information load items found in this study is consistent little as 20 nautical miles as sufficient. All other classes
with the generally 7 information load items found in of pilots wanted longer warning distances, and thus
many other studies. They conclude that modern technol- greater warning times.
ogy can provide large quantities of information that According to Hansman (1997), aircraft certified for
often make the pilot feel more confident, but perfor- flight in icing have only to avoid severe icing, whereas
mance is governed by human, not technological, limi- aircraft not certified for flight in icing require more deci-
tations. Information overload can be a serious issue sion support for strategic and tactical planning, go/no-go
because pilots will either fixate on one, perhaps trivial, decisions, and escape guidance. A simple pilot deci-
problem or lose the ability to ably accomplish any tasks. sion structure has two options if icing is forecast, either
Though flying in icing may not be as stressful as ter- to not go—the risk-adverse path, or to go—the risk-
rain-following flight in a high-performance aircraft, tolerant path. There are three outcomes for the risk-
icing often is most threatening in the busy, critical depar- tolerant path:
ture and approach phases of flight.
• Encountering no ice
Another issue for icing displays is the type of graphic
• Encountering ice but having options for avoiding
image provided to the pilot. Displays may be aircraft-
its effects
referenced or ground-referenced, and each may have
• A catastrophic outcome.
plan, profile, or perspective views. Though perspective
views look realistic, plan and profile views are better In the decision-making process, the risk of making
for decision making (Hansman 1997). Early MIT stud- a flight is weighed against the flight’s value. For high-
ies of terrain-avoidance displays indicated that the type risk flights, the incentive to make the flight must be
of display affects behavior and thus the avoidance strat- high for it to occur, or options must be available to
egy used by pilots, for example by avoiding terrain by reduce the risk. The decision to fly into potential icing
climbing vs. turning. Aircraft performance characteris- involves having options for avoidance and escape, such
tics, vertical and horizontal range, resolution, accuracy, as seeking dry or warm air or turning back or landing
scan rate, and sensor limitations also affect the displayed at alternates, to ensure a successful outcome. Other
information. possible outcomes are to reach the destination without
Pilots need information, not data, so the display must encountering ice or to have an ice-induced accident. A
be a rendition of the icing environment that allows the remote-sensing ice-avoidance system may provide
pilot to obtain the needed information unambiguously information for exercising options and reducing risk.
and in a form that promotes appropriate response. One Escape options are either vertical—finding warm or dry
of the issues is 2-D vs. 3-D displays and the way pilots air, or lateral—finding dry air (Vigeant-Langlois and
relate to each for different tasks. Cloutier (1997) pre- Hansman 1999). Research must be conducted on each
sented two potential 2-D displays showing plan and of the risk paths described above into how operators
profile views of icing potential for helicopter pilots. and pilots may use in-flight ice detection to make flight
Boyer (1994) indicated that little research had been done decisions. Vigeant-Langlois and Hansman (1999) also
evaluating the effectiveness of 3-D weather displays, indicate that pilots want escape guidance to be displayed
and he addressed the benefits and costs of 3-D vs. 2-D in the cockpit.
displays and conducted an evaluation using student
pilots for navigating around weather systems. He con- 3.4 Manufacturers and operators
cluded that 2-D displays offer advantages for navigating Aircraft operators and manufacturers are concerned
around weather, with few benefits attributable to the with the cost, weight, space, power, maintenance, and
3-D display. training requirements of placing additional avionics
A measure of effectiveness of a look-ahead weather packages on aircraft. They are also concerned about

8
To Contents
 the implications of a remote detection system being advise the pilot whether conditions are threatening to
onboard, for it implies that the aircraft cannot cope with safe flight. Systems could provide air traffic controllers
icing (Bond et al. 1997). (ATCs) and meteorologists with icing intensity
Cost is a large issue because aircraft that most need information and measured cloud microphysical
remote ice-detection systems can afford it least—the parameters. They could upload weather and satellite
regional airlines (Owen 1997). In addition, unless a information from the surface and integrate it with
system is extremely inexpensive, remote-detection sys- remote-sensor guidance (Bond et al. 1997). They could
tems will find little use on light, private aircraft also use information from onboard in-situ sensors to
(Vigeant-Langlois and Hansman 1999). This is a signifi- corroborate remotely sensed information and integrate
cant problem in the Far North where light aircraft oper- all sources of information into a comprehensive icing
ate with no ice protection and with few or inadequate advisory system. Overall, aircraft icing remote-sensing
weather advisories (Owen 1997). Even airlines operat- systems could aid pilots, meteorologists, air traffic
ing large transport-category aircraft are reluctant to use controllers, dispatchers, and ultimately the public
avionics they perceive to be of limited value because, through improved aviation safety.
beyond the initial cost, there is the cost of flying it (lost
payload) as well as maintenance and training costs. 3.5.2 Regulatory issues
Weight and space are serious problems, especially Regulatory agencies are responsible for providing
for light aircraft and on many smaller civilian and mili- leadership and procedures for maintaining, improving,
tary helicopters. Although these aircraft may not be ice- and enforcing aviation safety. As a result, other than
protected, they may be IFR-rated and thus require the requirements of individual operators and the military,
additional security provided by remote ice detection. systems for remotely detecting icing may not be used
Single-engine light aircraft have little space and weight on most aircraft without being required by regulators.
reserves and, in addition, there are few locations for Regulatory needs for operating in icing environments
sensor arrays since the engine and propeller dominate have been identified by Brayton and Hakala (1996).
the front of the fuselage. It may be possible, however, Nearly all changes in the regulatory environment that
to operate a sensor through the propeller by synchron- they recommend would be affected by implementation
izing it with the rotating propeller (Kirkpatrick 1970). of remote ice-detection systems, and all would need
Power is also a problem on many aircraft. Larger evaluation should onboard remote-sensing systems
civilian aircraft and military aircraft carry power- become available. The areas most affected, and requir-
demanding avionics and weapons systems. As a result, ing greatest study by regulators, would be weather
if space, weight, or power requirements are large for a reporting procedures between aircraft and the ground;
remote-sensing system, tradeoffs between other avion- automated substitution for standard icing pilot reports;
ics or a weapons system and the remote-sensing system handling procedures for aircraft wishing diversion;
must be considered. The icing remote-sensing system flight crew, dispatch, and air traffic control (ATC) train-
may be avoided because of the small percentage of time ing; and icing severity terminology. Aircraft certifica-
that it may actually be used. Military users may also be tion to fly in icing conditions probably would not be an
concerned about the signature provided by systems util- issue, because specific aircraft capabilities within icing
izing active rather than passive remote sensors. would not be affected by warning systems; only their
A remote-sensing system designed to detect icing ability to avoid and escape would be changed.
conditions ahead of an aircraft must be inexpensive,
small in size, low in weight, and require little power. 3.5.3 Incentives
Though different types of aircraft may use systems of Regulators determine what kinds of equipment
different capabilities to reduce the impact of some of should be mandatory on aircraft in different categories
these factors, all development should focus on mini- of operation. Because of cost and complexity, unless
mizing these liabilities. there are special needs of individual operators, regu-
lators may have to mandate installation of remote-
sensing equipment for detecting icing conditions on
3.5 Regulatory issues, weather forecasting, specific classes of aircraft. Such a mandate would be
and traffic management preceded by a thorough evaluation of remote-sensing
3.5.1 Functional requirements system capabilities, with a focus on their ability to
Above all, a system designed to detect icing conditions enhance safety. Mandates for use on aircraft, or simply
remotely is a pilot decision-support system (Clark certification for those operators voluntarily using sys-
1997). It is a system that senses conditions ahead of an tems, would require that regulators consider issues of
aircraft and translates it into an icing intensity index to system integration and protocol compatibility.

9
To Contents
3.5.4 National airspace impact tors could use the information to establish protocol and
Regulators must determine the impact of remote- requirements. Since too much downlinked data could
sensing systems on operation of the national airspace result in confusion rather than clarity and actually
system, air traffic control, and the Free Flight concept. decrease the quality of information subsequently pro-
vided back to pilots, human-factors specialists should
3.5.5 Aircraft operational limits in icing work with pilot and meteorological interests to resolve
Pilots need to know the potential intensity of icing these problems.
ahead of their aircraft. They also need to know how
their aircraft responds to icing conditions, and they need 3.5.7 Training
to know the limits of their aircraft with regard to icing. Training is critical to successful use of automated
FAR Part 25, Appendix C, defines icing cloud micro- systems. It is needed to provide familiarization with
physics for aircraft design. However, if an aircraft is systems and procedures. This will be most important
not designed specifically to fly in conditions beyond as remote-sensing systems are initially placed in the
those described in Appendix C, it is not known whether field to assure that pilots, air traffic controllers, and
it can safely operate in those exceedance conditions. If operators are aware of their operational characteristics.
aircraft are not tested in conditions beyond Appendix There is often a tendency to over-rely on technology
C, perhaps they should not be sent into those condi- because of apparent belief in its accuracy and reliability
tions (Hill 1997). (Transport Canada 1996). As a result, cockpit technol-
Icing risk varies with aircraft size, aircraft design, ogy tends to reduce vigilance and situational aware-
and airfoil type. Aircraft manufacturers must identify ness, which, in an icing environment, could be fatal.
the icing conditions that are beyond the capabilities of Since remote-sensing technology may actually
their aircraft. Consideration must be given whether to detach pilots from the icing threat because automation
expand Appendix C conditions or create a new FAR to tends to increase confidence and reduce situational
address these conditions. Presently, pilots do not know awareness, there is increasing need to promote train-
if they are flying in conditions within which the air- ing. Training is needed on how the operational charac-
craft was tested, and they do not know if the icing being teristics of remote-sensing systems operate and where
experienced will take the aircraft to its limits (Bettcher their abilities and failings lie. This training should be
et al. 1996, Parelon 1996, Erickson 1997, FAA 1997). fed by studies about the characteristics of the system,
Although it is not absolutely necessary to the function- perhaps through work that would have been done to
ing of a remote icing-detection system, providing pilots verify system capability (Baum and Seymour 1980).
with information about their aircraft’s operational limits In addition, there must be training on how to respond
and being able to relate information provided by sens- when an icing warning is displayed. Human-factors
ing systems to those limits would give pilots more con- research has addressed this issue for wind shear and
fidence about decisions to avoid or fly through icing. terrain-avoidance systems and recently for icing (Hans-
man 1997, Vigeant-Langlois and Hansman 1999). How-
3.5.6 Weather downlinking ever, research is needed to determine the most appro-
Onboard icing-sensing systems, through immediate priate avoidance and escape procedures for various
and continuous downlinking, could provide forecast- classes of aircraft in different types of airspace and
ers and numerical models with objective and timely meteorological conditions. Establishing training stan-
temperature, liquid-water content, and drop-size infor- dards and best management practices is a regulatory
mation that is accurate in position. Goals of forecasters and operator responsibility.
at the NWS Aviation Weather Center are to better iden-
tify where icing is occurring, identify areas of greatest 3.6 Test beds and platforms
risk, and determine when icing conditions disappear Efficient, cost-effective methods of testing elements
(Carle 1997). The military has similar concerns (Tucker of remote-sensing systems, and full systems, are needed
1983, Peer 1986, Goe 1997). Downlinking of informa- in the development stage under conditions representa-
tion gathered onboard would indicate the magnitude tive of the operating environment. Ground-based test
and location of icing potential, indicate where there is systems are generally less expensive than airborne plat-
no icing potential (Vigeant-Langlois and Hansman forms, so their use should be encouraged at all stages
1999), and provide improved forecast verification. A of development until full testing on aircraft is required.
program should be organized to formulate standards Airborne platforms, spray tankers, and perhaps
for integrating ground, satellite, in-situ, and aircraft- mountain-top observatories should be used to test proto-
based icing information and to establish protocol for types of individual sensors and of entire remote-sensing
auto-reporting to ground and to other aircraft. Regula- systems. Remote-sensing systems intended for place-

10
To Contents
 ment on aircraft must be capable of sensing horizontally instruments, from multicylinders that provide integrated
ahead of the aircraft and above, below, and to the sides measurements to current electro-optical systems that
of the flight path, so they should be tested in the same provide measurements with high temporal resolution.
position as they will be used aboard aircraft to provide Thousands of research flight hours have resulted in a
confident results (Ryerson et al., 2000). Mountain-top general understanding of the magnitude of liquid-water
facilities that have potential testing capabilities include contents, and their spatial patterns, that can be encoun-
the Desert Research Institute’s Storm Peak Laboratory; tered by aircraft. For example, it is understood that
Elk Mountain, operated by the University of Wyoming; higher liquid-water contents are generally found in
Whiteface Mountain Observatory, operated by the State cumuliform rather than in stratiform clouds, that sum-
University of New York, Albany; and Mt. Washington mer supercooled liquid-water contents are highest, and
Observatory, N.H. (Ryerson et al., in prep.). An advan- that liquid water is generally more “cellular” than homo-
tage of mountain-top facilities is the availability of natu- geneous over thousands of square kilometers. Studies
ral icing conditions. Likewise, a disadvantage of natu- have also demonstrated that icing conditions, for liquid
ral icing is the lack of control over conditions. Research water and drop size, are extremely variable and diffi-
aircraft available for testing may include the NASA cult to generalize. It is also recognized, though not nec-
Twin Otter, the NRC Twin Otter and Convair 580, the essarily widely, that FAR 25, Appendix C, conditions
NCAR King Air, and aircraft from the University of are only representative values for engineering design
Wyoming, the University of North Dakota, and a vari- purposes and are not intended to represent the actual
ety of private companies (Marcotte et al. 1996). The character of the icing atmosphere as encountered by an
Air Force tanker spray rig has been removed from ser- aircraft.
vice, but the Army still operates its Helicopter Icing The characterization of the icing atmosphere has
Spray System (HISS) from Fort Rucker, Alabama. been accomplished somewhat randomly because of the
Flight simulators may also provide information use- cost of airborne research projects. Each program has a
ful for establishing sensor system characteristics. For specific focus, so flight hours are typically consumed
example, pilots’ abilities to react within given warning trying to answer the primary research questions of the
times provides information for establishing sensing project. A large-scale monitoring program dedicated to
range and update frequency. the characterization of icing conditions would allow
large geographic areas, with weather conditions experi-
enced by most of the nation, to be sampled consistently
4.0 METEOROLOGICAL SENSING
and frequently to produce information that is statisti-
REQUIREMENTS
cally valid. Work by Cooper et al. (1982) and Sand et
4.1 Summary al. (1984), by the Canadian Freezing Drizzle Experi-
The development of systems to measure icing poten- ment (CFDE), and by the NASA Glenn Research Center
tial remotely and in situ requires an understanding of during the winters of 1996–1997, 1997–1998, and
the medium being sensed: the atmosphere and its ther- 1998–1999 (Miller et al. 1998) have come closest to
mal and liquid characteristics must be understood with the ideal of covering large geographic areas with mod-
regard to the absolute magnitude of conditions and their ern, carefully calibrated instrumentation. One way to
spatial distribution. This information is needed to evalu- do this would be to instrument commercial or military
ate the feasibility of sensing and avoiding icing poten- aircraft that fly large numbers of hours, as Perkins
tial, to design instruments to sense conditions, and to (1952) did, enabling a representative sample of icing
develop methods of avoiding and exiting icing condi- conditions to be made nationwide and reported through
tions. The atmosphere must be carefully characterized a system such as ACARS (Aircraft Communications
with regard to icing potential to develop sensors and and Reporting System). Ground-based remote-sensing
training protocol, to create terminology for advising systems installed at airports to protect terminal areas
pilots, and to provide better icing forecasts. Character- may also be able to provide characterization informa-
ization is needed at all scales from the submesoscale to tion similar to that of in-flight programs. This is another
the global scale, although the synoptic scale is proba- argument for accelerating airport-based remote-sensing
bly best understood with regard to icing.* icing-avoidance systems. Finally, icing radiosondes are
Attempts to characterize the icing atmosphere, available for measuring supercooled liquid water with
liquid-water content, drop size, and temperature have height within clouds (Hill 1994). Such radiosondes,
been conducted since the 1940s with a large range of fielded nationally by the NWS, could improve icing
forecasts and the characterization of supercooled cloud
* Personal communication, M. Politovich, National Center for Atmo- water.
spheric Research, Boulder, Colorado, 1997. Cloud liquid water is generally better understood

11
To Contents
than is drop size. Nevertheless, the magnitude, distri- ASOS (Automated Surface Observing System) observa-
bution, and organization of supercooled liquid water in tions at over 600 locations nationally (Ramsay 1997),
3-D space is still only generally understood, especially which observe freezing rain, could improve understand-
with regard to the conditions that aircraft typically ing of the location, spatial patterns, frequency, and mag-
encounter. A large component of the problem within nitude of SLDs.
supercooled clouds is glaciation. Though attempts have Proponents of in-flight remote-sensing systems have
been made to model and measure glaciation to develop argued that outside air-temperature measurements made
a better understanding of the process, it is still not possi- at the fuselage are adequate for temperature character-
ble to predict accurately whether a given cloud is glaci- ization ahead of the aircraft. Though this may be gener-
ated, when it will glaciate, and how much of the total water ally true in cruise at constant-altitude flight, it is not
content is ice. Some remote sensors are sensitive primari- true where icing is most likely to occur: within storms
ly to liquid water, such as microwave radiometers, and and in the climb-out and descent phases of flight. Air
for users of these systems, mixed-phase clouds are of temperature changes most rapidly in the vertical and
little concern. However, mixed-phase conditions may within storm systems. Storms and lower altitudes are
enhance the ability of radar to detect liquid water, so also where supercooled water is more frequent, so the
the glaciation process needs to be better understood. reliability of outside air-temperature measurements for
Although less important than liquid-water content for predicting temperature ahead of the aircraft is least
determining the amount of ice to form on an aircraft, where the need is greatest. Thermal lag also occurs as
drop size, and especially supercooled large drops snow falls into warm air and melts, and as rain falls
(SLDs), determine the location and shape of ice forma- into colder air and supercools, making drop tempera-
tions. Thus, drop size may have a larger impact on iced ture unknown even if air temperature is known. In addi-
aircraft aerodynamics than liquid-water content does. tion, evidence from a few studies suggests that temper-
Drop size is also a more difficult parameter to measure ature does fluctuate considerably within cloud masses
than liquid water, and characterization is therefore less and from cloud to cloud and from clear to cloud. Air-
complete than for liquid water. Cloud droplet size varies temperature fluctuation, especially near 0°C, must be
with cloud type, cloud dynamics, location within clouds, better characterized within clouds and near frontal sur-
from cloud to cloud, with the season, air-mass origin, faces. Radiosonde observations and in-flight measure-
and other factors. Drop sizes are often characterized by ments from existing flight programs can provide most
the median volume diameter (MVD), which assumes a of this information.
unimodal drop-size distribution. This may not always The spatial structure and the size of icing areas have
be the situation, especially when SLDs are present. not been characterized. Spatial patterns of icing must be
The shape of the drop-size distribution must be care- characterized at all scales, from global to submesoscale,
fully sensed and characterized. This is especially impor- but spatial patterns are perhaps best understood at the
tant for SLDs. Instruments that count drops, such as synoptic scale. The horizontal extents of icing speci-
optical array probes, have the best probability of suc- fied in FAR 25, Appendix C, do not imply the overall
cessfully characterizing drop size. Characterizations of dimensions of icing cloud systems. Overall, little work
drop size conducted concurrently with liquid-water has been conducted in this area, with the best character-
measurements will provide relationships between the izations being by Cooper at al. (1982) and by the Cana-
two and to atmospheric dynamics. Drops are usually dian CFDE program (Cober et al. 1996b).
smaller within stratiform than within cumuliform clouds, Cloud microphysics are a focus of several large fed-
but more emphasis should be placed on explicit drop- erally funded research programs, and icing remote-
size measurements and, at least, characterization of the sensing researchers should partner with these teams to
drop-size spectra. accomplish objectives more efficiently. For example,
The need to characterize SLDs is even more critical the DoE Atmospheric Radiation Measurement (ARM)
than characterizing smaller drops because of the danger Program monitors cloud microphysics to determine the
SLDs present to aircraft and because far less is currently effects of cloud cover, type, height, and phase on glo-
known about them than about smaller drops. Flight pro- bal radiation budgets. The ARM program maintains
grams to measure SLD characteristics, such as the Cana- field sites in Oklahoma and on the North Slope of
dian CFDE project and the NASA Glenn Research Cen- Alaska. Remote sensing of clouds is one of their tools,
ter SLD program, should be continued and expanded. and NOAA ETL has been a participant in this capacity.
Ground-based programs may also be useful for charac- The Global Energy and Water Cycle Experiment
terizing conditions aloft. For example, sleet, freezing (GEWEX), part of the World Climate Research Pro-
drizzle, and freezing rain at the surface are often accom- gram, also has a cloud microphysics component that
panied by freezing precipitation aloft. Utilization of may be of value to aircraft-icing remote-sensing research.

12
To Contents
 Finally, experiments are being conducted by federal, siderably within clouds and from cloud to cloud. In addi-
state, and private groups to augment precipitation (cloud tion, temperature may change rapidly within frontal
seeding). This often involves monitoring of cloud micro- systems and in the vertical, as experienced by aircraft
physics. Efforts should be made to interact with these when changing altitude rapidly upon departing or
groups and perhaps to conduct coordinated research. approaching terminal areas.
Characterization requires reliable and accurate instru- Liquid-water drop-size spectra determine the loca-
mentation for making in-situ measurements. The ideal tion of drop impingement on airframe structures, the
instrumentation for measuring cloud microphysics type of ice that forms, the shape of ice that forms, and
would be similar in character to instruments currently the location of ice on the airframe as a result of run-
used on aircraft for determining airspeed and outside back. Runback is caused by supercooled large-drop
air temperature: generally small, inexpensive, accurate, impingement and flow along the airfoil chord causing
robust, maintenance-free, and unobtrusive. A focused freezing on areas of the airfoil unprotected by deicing
effort is needed to simplify and miniaturize current equipment. Runback has become a critical problem,
instrumentation, but efforts should also be made to com- especially with respect to SLDs within the drizzle size
pletely rethink cloud microphysics instrumentation and range (typically diameters of 50 to 500 µm). Runback
to design and develop completely new concepts. and formation of an ice ridge immediately aft of the
Finally, icing terminology needs improvement. This boot-protected leading edge is believed to be the cause
is being addressed in the FAA Inflight Aircraft Icing Plan of the ATR-72 crash in Roselawn, Indiana, in October
(FAA 1997). Currently, icing reports and forecasts are 1994 (NTSB 1996). The overall characterization of
not purely meteorological but include the aircraft. SLDs vs. smaller drop sizes is relatively poorly under-
Though practical, because pilots observe how icing is stood, so special emphasis should be placed on charac-
affecting their aircraft, the current terminology is not terizing the SLD environment.*
effective because it does not utilize purely meteorologi- Liquid water, drop temperature, and the drop-size
cal information to evaluate icing intensity. The aircraft spectra are the most important indicators of in-flight
must be separated from weather to evaluate icing con- aircraft icing potential, and thus the most important
ditions objectively and unambiguously. conditions to characterize for the development of the
sensing needs of remote-sensing systems. However, it
4.2 Introduction may also be useful to sense conditions that are not criti-
Information required to assess in-flight aviation icing cal to icing potential but that may serve as surrogates
hazard is derived from measurements of the atmospheric for the other conditions. For example, range-resolved
conditions that create ice on aircraft. In order of impor- remote sensing of air temperature or droplet tempera-
tance, those conditions are cloud or precipitation liquid- ture may prove to be the most difficult remote-sensing
water content, drop temperature, and drop size.* Of these challenge. Whether clouds are glaciated or partially
three conditions, liquid water is most important because glaciated, or mixed-phase, may provide an indication
it is the material that creates ice on the aircraft. Liquid- of whether liquid-water temperatures are warmer or
water magnitude varies widely, both spatially and tem- colder than freezing, so the ability to detect ice within
porally, so it must be measured continuously. clouds may serve as a surrogate binary temperature indi-
Droplet temperature is the second most important cation of above- or below-freezing conditions. How-
atmospheric condition affecting icing; it determines in ever, little is known about the glaciation process and
part whether liquid water will freeze on an aircraft struc- the probability of glaciation at given temperatures below
ture. Air temperature (static and total temperature are freezing. Nevertheless, characterization of cloud glaci-
not distinguished here) may serve as a surrogate for drop- ation with temperature, to determine if it would be a
let temperature, especially if droplets are so small that meaningful surrogate for temperature, may be useful.
their fall speed allows them to maintain a temperature The magnitudes of liquid-water content, tempera-
nearly that of the surrounding atmosphere. However, ture, and drop-size spectra must be characterized to
snow or graupel falling into a warm layer may melt or provide specifications for remote-sensing technology.
partially melt, resulting in particle temperatures colder The spatial variability of these conditions must also be
than the air in the warm layer. If the particles or droplets characterized. Characterization is needed at the mesos-
then fall into colder air below, they will be warmer than cale (~104–106-m scale), synoptic scale (~106-m scale),
the air. Air temperature may be relatively constant over and global scale, although for different reasons, depend-
large horizontal distances, but it may also fluctuate con- ing on the scale. Characterization of spatial variability

* Personal communication, M. Politovich, National Center for Atmospheric Research, Boulder, Colorado, 1997.

13
To Contents
 tion of cloud type), allowing for larger liquid-water
contents in shorter distances, Appendix C represents
only integrated liquid-water contents over distances and
does not address maxima or minima that can occur
within those distances. That is, “pockets” of liquid-water
Proposed
content can be much larger or smaller than the inte-
Flight grated values within a distance represented by Appen-
Path
dix C. According to Masters (1983), the original intent
of Appendix C was to represent averaged liquid-water
content values during exposure over varying distances,
so Appendix C is only a starting point for remote-
sensing system specification. Appendix C is effective,
within its limits, for creating ice-protection system
Sensor-Equipped
Aircraft specifications as long as instantaneous values of liquid
water are of no concern, but it is not as effective for
Figure 2. Sensing range affects avoid-and-exit capability.
determining the maximum and minimum liquid-water
values that may be experienced along a given route of
may provide indications of the range and spatial reso- flight. For this reason, it is necessary to review field
lution needed by remote-sensing systems. The ability measurements within icing conditions to determine the
of a system to sense completely through typical icing absolute range of conditions that can be experienced.
storm areas, for example, would provide aircraft with In addition, the range of conditions with greatest impact
avoidance capability without the risk of entrapment on aircraft operations must be considered for studies of
(Kirkpatrick 1970) (Fig. 2). In addition, if icing poten- the effect of icing on in-flight aircraft performance (Jeck
tial is nearly uniform spatially at the submesoscale, then 1998).
remote-sensing capabilities may not be practical for The National Advisory Committee for Aeronautics
avoiding or escaping icing. Spatial characteristics at the (NACA) Lewis Flight Propulsion Laboratory and Ames
synoptic scale will indicate what portions of storms Aeronautical Laboratory conducted many research
provide the greatest icing threat and indicate where flights within stratus and cumulus clouds from 1945
remote-sensing systems are most needed. Global-scale through 1950 (Lewis et al. 1947, Lewis and Hoecker
patterns of icing indicate where icing threats and the 1948, Kline 1949, Kline and Walker 1951). Cloud
needs for remote-sensing capabilities are greatest. liquid-water content was measured with rotating multi-
cylinders and a rotating-disk icing-rate meter. Rotating-
4.3 Characterization needs multicylinder and icing-disk measurements are inte-
Characterization of the dynamic range of liquid grated over periods of minutes, so absolute magnitudes
water, drop size, and temperature are needed to estab- over shorter periods, and thus distances, are not well
lish sensing specifications for hardware development. identified. Nevertheless, one flight program provided
The general ranges of conditions that could be observed maximum liquid-water contents of 0.28 g m–3 and 0.76
are generally understood (Pruppacher and Klett 1997, g m–3 for stratus and cumulus clouds, respectively, and
Rogers and Yau 1989, Fletcher 1962), but the absolute 90% of the measurements measured less than 0.5 g m–3
ranges of conditions within icing clouds, and the ranges in stratus clouds and less than 1.2 g m–3 in cumulus
of conditions necessary to produce dangerous ice accre- clouds (Lewis et al. 1947). Flights over the Great Lakes
tions on aircraft under a full range of flight and design measured liquid-water contents ranging from 0.05 g m–3
conditions, are not well known. to 0.57 g m–3 with a median of 0.22 g m–3 and a mean of
0.19 g m–3 (Kline 1949). Ninety percent of liquid-water
4.3.1 Liquid-water content contents were less than 0.40 g m–3, and 50% of all cases
FAR 25, Appendix C (FAA 1991), defines liquid- were less than 0.18 g m–3. Thirty-seven research flights
water content and mean effective drop diameter (similar over most of the northern United States measured mean
to the median volume diameter [MVD]) from the sur- values of maximum liquid-water contents averaged over
face to 6707 m for stratiform clouds and from 1220 m distances of 0.5, 3.0, 15, and 60 miles of 1.05, 0.63,
to 6707 m for cumuliform clouds. This is the minimum 0.33, and 0.14 g m–3 for cumulus clouds, and 0.44, 0.27,
standard to which all aircraft ice protection is designed 0.16, and 0.08 g m–3 for stratus clouds. Finally, flights
and may be the minimum standard to which remote- in stratiform clouds yielded a maximum liquid-water
sensing systems should be designed. However, although content of 1.30 g m–3, with 90% of measurements less
horizontal extent is factored into Appendix C (as a func- than 0.54 g m–3 and 50% less than 0.30 g m–3 (Kline

14
To Contents
 and Walker 1951). Large values were typically sought different conditions experienced by light aircraft and
during these flights for establishing engineering stan- helicopters below 3049 m. Supercooled liquid-water con-
dards. tents of up to 1.7 g m–3 were found, but 99% of values
Perkins (1952) instrumented with icing rate meters were less than 1.1 g m–3, and 95% were less than 0.6 g
four United Airlines DC-4 aircraft that flew from New m–3 for all cloud types. Larger liquid-water contents are
York City to San Francisco from January through May theoretically possible below 3049 m, but values greater
1951. Of a total of 1120 hours of flight time on typical than 2.2 g m–3 are not likely. Liquid-water values were
commercial routes and altitudes, icing was encountered largest in cumuliform clouds and behind cold fronts in
1.5% of the time. Maximum liquid-water contents did maritime air.
not exceed 1.0 g m–3, and 80% of measurements were Jeck (1982) also sorted the database described above
less than 0.4 g m–3. (Jeck 1983) according to synoptic situation and air-mass
In March 1979, Jeck (1980) made in-cloud liquid- category. The resulting liquid-water contents represent
water measurements over Lake Michigan and in the average values over uniform cloud intervals of more than
vicinity of Lake Erie. He compared instrumentation 1 km, so peak liquid-water values are not represented.
used in research flights from the 1940s and 1950s to In general, modern data liquid-water contents were 1.1
then-current instrumentation and discussed sources of to 1.2 g m–3 in cumulus clouds within lake-effect areas,
measurement error when using multicylinders to meas- modified continental air masses, high-pressure areas
ure liquid water. The most significant problem was run- without fronts, and in maritime air masses. Liquid-water
off due to incomplete freezing of water impinging upon contents were lowest, 0.2 to 0.4 g m–3, in warm frontal
the cylinders. This occurred when the proper combina- stratus clouds, occluded-front stratus clouds, cold-front
tions of air temperature and liquid-water content caused cumulus clouds, and upslope stratus clouds.
the ice temperature to remain at 0°C, the so-called Jeck later expanded the database for icing conditions
“Ludlum” limit. Jeck mounted a Johnson–Williams hot- below 3049 m to include all altitudes and presented
wire liquid-water probe that measured liquid water accu- seasonal analyses of liquid-water content for design
rately only when drop sizes were smaller than 30 µm purposes (Jeck 1989). Seasonal liquid-water magnitudes
(due to design limitations) on a Lockheed Super Con- were isolated not by calendar date, but by grouping
stellation aircraft. Measured liquid-water contents were measurements by the height of the freezing level, with
somewhat smaller than historical measurements, in part lower freezing levels occurring during winter condi-
because his measurements were in the lower portions tions. Freezing levels below 1524 m agl were used for
of stratus clouds, whereas earlier measurements sought winter conditions and above 3049 m for summer con-
the largest values typically encountered, near cloud tops. ditions. Nearly 85% of stratus cloud occurrences were
On the Jeck flights, icing generally did not occur on the below 3049 m, with the largest liquid-water contents,
aircraft when liquid-water contents were less than 0.08 maximizing near 0.9 g m–3, occurring near 1524 m for
to 0.10 g m–3. In stratus clouds less than 1524 m agl, warm and cold seasons. Liquid-water content was less
95% of all liquid-water measurements were less than than 0.3 g m–3 90% of the time in the stratus. More than
0.6 g m–3. 50% of convective clouds occurred above 3049 m, and
In a comprehensive review, Cooper et al. (1982) and cold-season convective clouds typically had liquid-
Sand et al. (1984) summarized five years of flights made water contents of less than 2.0 g m–3, with this maxi-
with modern instrumentation by the University of Wyo- mum occurring near 3659 m. Summer convective
ming King Air. Over 98% of summer and winter and clouds, however, had maximum supercooled liquid-
continental and coastal cloud liquid-water measure- water contents approaching 5.0 g m–3, but only above
ments—423,787 seconds of measurements with a 6098 m.
Johnson–Williams hot-wire liquid-water probe and a for- In a review of the state of knowledge of aircraft icing
ward-scattering spectrometer probe (FSSP)—were less conditions from around the globe, Hoffman (1984)
than 1.0 g m–3, and only 0.2% of samples exceeded 2.0 g stated that icing occurs within liquid-water contents of
m–3. Liquid-water contents nearly as high as 3.0 g m–3 0.01g m–3 to 6.0 g m–3, though values larger than 2.5 g
were encountered, but in less than 0.01% of all measure- m–3 are found only in tropical cumulonimbus clouds.
ments. At any given altitude, liquid water within stratus clouds
Jeck (1983) and Masters (1983) compiled a new can vary between 0.01 to 1.0 g m–3, and in cumulus
database of supercooled cloud properties up to 3049 m clouds it can vary between 0.01 and 1.7 g m–3.
from about 12,955 km of icing observations using a Twenty-five flights measuring liquid water in strato-
mix of old and new measurement technology, from cumulus clouds in Germany with a Johnson–Williams
multicylinders to newer optical and hot-wire instrumen- probe indicated that liquid-water contents varied from
tation. This database was constructed to address the 0.05 to 0.45 g m–3 (Hoffman et al. 1986). Integration

15
To Contents
distances were 18.5 km, so the values do not represent true range of supercooled liquid-water contents that may
maxima encountered during flight. be encountered. Though general values are known, most
Telford (1988) analyzed the causes of instability and older measurements were made over rather long averag-
the loss of a Desert Research Institute research aircraft ing distances. Nearly instantaneous measurements, on
measuring layered cloud properties in the Sierra Nevada. the order of one measurement each second (150 m for
Extreme icing was associated with the crash, and liquid- a 300-kt research aircraft), provide high resolution.
water contents were measured from 0.2 g m–3 to a maxi- Though the granularity of cloud liquid water is undoubt-
mum of 1.4 g m–3 prior to the crash. Measurements edly finer, higher resolution would not be necessary
were instantaneous and were made with an FSSP and a for developing remote-sensor specifications.
Johnson–Williams probe. Liquid-water content has been measured during field
In a detailed study of a winter storm and shallow programs with modern instrumentation at as fine as 1-
cold-front passage in the Denver area, Politovich and second intervals. Examples include ASTEX, the Atlan-
Bernstein (1995) measured unusually high liquid-water tic Stratocumulus Transition Experiment; FIRE, the
contents for that area in stratiform clouds of 0.6 g m–3. First International Satellite Cloud Climatology Project
Severe icing was also reported by the research aircraft. Regional Experiment; the U.S. DoE ARM campaign,
During the second Canadian Atlantic Storms Pro- Enhanced Shortwave Experiment; and many other
gram (CASP), Cober et al. (1995) reported 3745 super- smaller programs. These data could be reanalyzed for
cooled liquid-water content measurements within an remote-sensing purposes if they could be acquired. In
800-km radius of Halifax, Nova Scotia, within strati- addition, new flights should be made with better instru-
form and “system” clouds such as through cold fronts, mentation.
warm fronts, and low-pressure areas. Clouds of oceanic
and continental origin were included. Though few com- 4.3.2 Cloud drop-size spectra
parisons of liquid water in maritime vs. continental Liquid water is delivered to aircraft surfaces as dis-
clouds have been conducted specifically with regard to crete drops varying in diameter from only a few microns
aircraft icing, in general, liquid-water contents are sim- at the smallest diameter to over 4000 µm in rain drops
ilar for continental and marine clouds of a given gen- (Fletcher 1962, Pruppacher and Klett 1997, Rogers and
era (Rogers and Yau 1989). Cober et al. (1996a) also Yau 1989, Willis and Tattelman 1989). Drop size has
found that liquid water varied little between cloud types, several important roles in aircraft icing.
except for larger liquid-water-content standard devia- One effect of drop size on airframe icing is its influ-
tions in system clouds. Median supercooled liquid-water ence on the amount of water collected. The amount of
content was 0.11 g m–3, with supercooled liquid-water liquid water delivered to an airframe surface is a func-
content exceeding 0.94 g m–3 only 0.01% of the time, tion of the collection efficiency of that surface as
similar to conditions measured by Sand et al. (1984) affected by the relative speed between the surface and
over the Great Lakes and California in winter. Stewart the drop, the radius of the surface, and the drop diameter.
et al. (1996) measured supercooled liquid-water con- As relative wind speed increases, drop size increases,
tents of warm and cold fronts off the Nova Scotia coast. and as surface radius decreases, droplet collection effi-
Cloud types are not provided, but the measurements ciency increases. As a result, smaller drops are carried
were made with modern optical instruments. Maximum over an airfoil surface to impact aft of the leading edge,
supercooled liquid-water contents were not greater than whereas larger drops impact closer to the leading edge.
0.9 g m–3, and typically they were less than 0.3 g m–3. Objects with a large radius are preferentially impacted
Pruppacher and Klett (1997) summarize character- by large drops rather than by small drops. As a result,
istic liquid-water contents found in clouds by genera, the amount of liquid water delivered to a specific por-
warning that liquid-water content typically varies tion of an airframe surface is a function of the liquid
strongly from cloud to cloud. Early-stage cumulus typi- water residing within that portion of the total cloud drop-
cally have 0.2 to 0.5 g m–3, later-stage cumulus 0.5 to size spectrum striking the surface. This ignores run-
1.0 g m–3, and stratus and stratocumulus 0.1 to 0.5 g back and other effects that occur after drops impinge
m–3. Cumulus with strong updrafts have liquid-water upon the surface.
contents up to 5.0 g m–3. Though these measurements A second effect of drop size is upon the type and
were made in warm and cold conditions, the tops of shape of ice that forms on the airframe surface (Hans-
cumulus clouds typically have the highest liquid-water man 1985, FAA 1991, Shah et al. 1998). Depending
contents, which, in the tropics, are often supercooled upon a variety of factors, including the amount of liquid
and produce significant icing. water impinging on a portion of airframe over a unit of
It is evident that modern measurements taken at short time and the collection efficiency of the icing surface
time intervals are necessary to evaluate properly the as a function of the drop-size distribution, the type and

16
To Contents
 shape of the ice accretion will vary considerably. Low the MVD alone does not describe the shape of the drop-
liquid-water contents at low temperatures and small size distribution, nor does it adequately describe where
drop size tend to create rime ice, and larger liquid-water collection efficiency effects will cause ice accretion on
contents in warmer temperatures and larger drops tend the airframe (Newton 1979, Cooper et al. 1982).
to produce clear ice (FAA 1991). Rime tends to pro- Early measurements of drop-size spectra were made
duce an ice surface that conforms generally to the shape either with oiled or soot-covered slides or with the use
of the airfoil. Clear ice may create a smooth surface, or of rotating multicylinders. Slides were difficult to use in
it creates “horns” near the leading edge that have a large high wind speeds, though they were occasionally used as
impact on drag and airfoil lift. As drop size increases late as the 1960s (Warner 1969). Usually, rotating multi-
within clear-ice conditions, the size of the accretion cylinders, developed in the early 1940s at Mt. Wash-
increases, the impingement limits increase in area, and ington Observatory, were used instead of slides, and
the horns tend to form farther back on the airfoil (FAA they are still in use (FAA 1991, Howe 1991). Multi-
1991). Overall, according to Sand et al. (1984) and Polito- cylinders provide an indication of the shape of the drop-
vich (1989), drops larger than 30 µm in diameter have size spectra by utilizing curves developed from theory
a greater effect on flight than smaller droplets. by Langmuir and Blodgett (1946) using the collection
A third effect of drop size is runback. Although run- efficiency of various-diameter cylinders for given wind
back may occur over a large range of cloud drop sizes, speeds and drop sizes. From these curves, and the
depending upon temperature and liquid-water content, amount of ice collecting on each cylinder, the MVD
runback becomes more serious when the drizzle-size can be estimated. However, serious errors in MVD esti-
regime is entered, at about 50 µm. Here, all water does mation could occur with multicylinder use in large-
not freeze near its impingement location—some runs droplet situations, where MVDs approach 30 µm or
back and freezes beyond ice-protected areas of the lead- larger (Jeck 1980).
ing edge. This often creates an ice ridge or roughens Drop-size spectra were measured coincidentally with
wing surfaces, significantly altering airfoil aerodynam- liquid-water content in most experiments. The database
ics and aircraft performance. created by Jeck (1980) at the Naval Research Labs, in
Cloud droplet size varies by cloud genera, from cloud cooperation with the FAA, is probably the most compre-
to cloud, by season, and with location within clouds. hensive available. Jeck (1983) and Masters (1983) sum-
For example, the largest drops in growing, nonprecipi- marized older and modern measurements both below
tating cumulus clouds typically occur near the center 3049 m and at all levels of the atmosphere. Jeck (1983)
and top of the cloud within updrafts. Smaller drops are indicates that below 3049 m, average MVDs measured
found near the cloud base and near the cloud perimeter with multicylinder and newer optical instruments, for
where dry air entrainment causes evaporation of drops supercooled layer clouds, are about 13 µm and for con-
(FAA 1991). Overall, drop size is controlled by evapo- vective clouds they are 18 µm. MVD also shows temper-
ration, collision–coalescence, curvature and solute ature dependence, with MVD increasing from 10 µm to
effects, the Bergeron process, and the number and type about 30 µm in stratiform clouds as temperature
of cloud condensation nuclei present (Miller and Anthes increases from –25°C to 0°C. Jeck (1982) also observed
1980). As an example, maritime clouds of a given gen- that MVD generally increases with altitude in single-
era typically exhibit broader drop-size spectra than do layer clouds below 3049 m.
continental clouds due to differences in the type, num- Jeck (1983) questions the use of a minimum MVD
ber, and size of cloud condensation nuclei (Rogers and of 15 µm in FAR 25, Appendix C, considering analy-
Yau 1989). ses of the database of cloud properties below 3049 m
Cloud drop-size spectra are typically characterized (Jeck 1980). Masters (1983) and Jeck (1983) both pro-
by the median volume diameter (MVD), the drop size vide diagrams from this database showing MVDs in
where one-half of the spectrum’s water volume resides icing clouds well below 15 µm.
within smaller-diameter droplets and the other half A summary of five years of cloud measurements by
resides within larger droplets. Internal cloud dynamics the University of Wyoming (Sand et al. 1984) showed
may create bimodal drop-size distributions, observed in MVDs ranged from 5 to 40 µm, with a characteristic
most cloud types in most climatic regimes (Pruppacher MVD of about 15 µm. The smallest MVDs were meas-
and Klett 1997, Politovich and Vali 1983). Bimodal dis- ured during the winter over the Great Lakes and the
tributions are not properly represented by a single MVD, Great Plains, with the largest MVDs in the summer over
however, which relies on a unimodal distribution. The the Great Lakes and Illinois and in the winter over Flor-
average collection efficiency of a drop-size spectrum ida. Droplets were smaller in the Great Lakes and Illi-
around a median volume diameter is generally quite nois areas because of low liquid-water contents, accord-
close to the collection efficiency of the MVD. However, ing to Sand et al. (1984). No relationship was found

17
To Contents
between icing effects on aircraft performance, a Beech sition Experiment (ASTEX), the First ISCCP (Interna-
King Air, and the MVDs that produced the ice. Only tional Satellite Cloud Climatology Project) regional
MVDs larger than 40 µm, reaching into the supercooled experiment (FIRE), and the U.S. Department of Energy
large-drop regime, affected aircraft performance. Atmospheric Radiation Measurement (ARM) program.
Roebber (1988), in a review of icing potential on The general characteristics of MVD by cloud genera
helicopters and fixed-wing aircraft off the east coast of are understood. However, less is generally understood
Canada, presents statistics of drop sizes encountered about cloud drop size than about cloud liquid-water con-
during icing and reported by the Royal Canadian Air tent. Controls of drop-size spectra are not well para-
Force. The MVDs of convective clouds were between meterized, although the general controls are believed to
18 and 21 µm and for layered clouds near 12 µm, but be understood. Changes in drop size over time within
MVDs as large as 50 µm were generally observed in con- storms, and diurnally, have been tracked and simulated,
vective clouds, and as large as 40 µm in layered clouds. but general theory explaining drop-size evolution over
Jeck (1989) summarizes MVDs from his FAA/NRL time is not mature.
icing database for clouds at all altitudes. Mean MVDs Understanding of drop size has been hindered by the
are 13 µm for layer clouds and 17 µm for convective need for improved instrumentation, the three-dimensional
clouds. The range of MVDs within the database, by gen- complexity of liquid water within clouds, the need for
eral cloud type, are 7–21 µm for layer clouds and 10–26 observation flights focusing on drop-size measurements,
µm for convective clouds. and too much emphasis on reporting only MVDs instead
Cober et al. (1995) reported on 31 flights into Cana- of the full drop-size spectrum.
dian east coast winter storms over the North Atlantic
Ocean and created a high-quality database of those 4.3.3 Supercooled large drops
flights. Flights were made into fronts, low-pressure areas, The existence of large droplets (>50-µm diameter)
and stratus clouds. The average MVD for all clouds was was well known to early NACA investigators of the
18 µm: 16 µm for low-level stratus clouds and 20 µm microphysics of icing clouds, but they were not included
for “system” clouds. These measurements compare well in the FAR 25, Appendix C tables, which include drop
with earlier measurements in the area, according to sizes from only 15 to 40 µm (FAA 1991). Sand et al.
Cober et al. (1995), and with measurements by Sand et (1984) and Politovich (1989) state that droplets larger
al. (1984). than 30 µm in diameter have a greater effect on flight
Politovich and Bernstein (1995) investigated the pro- than smaller droplets. Hansman (1985) indicates that,
duction and depletion of supercooled liquid water in a from model and wind-tunnel tests, large drops present a
February 1990 winter storm in the Denver area. Strati- much larger threat to aircraft than small drops and that
form clouds associated with a cold-front passage creat- even a small liquid-water content in large drops may be
ed mean droplet diameters of 10–13 µm, with droplets a significant icing threat. Bragg (1996) attributes large-
larger than 50 µm in diameter observed. droplet ice accretions, and the formation of ice ridges
Small diurnal changes in drop-size spectra occur as aft of ice-protected areas, as a likely cause of flow sepa-
a result of changes in cloud dynamics between night ration, aileron snatch, and loss of roll control. Shah et
and day. Modeling of marine stratocumulus clouds by al. (1998) indicate that secondary ice shapes producing
Considine (1997) demonstrated increases in MVD of a ridges spanwise along a wing can be created by super-
few microns in the afternoon and decreases at night, with cooled large drops (SLDs), even with a heated leading
minima in the morning. Much of the effect is due to edge. The larger drops also strike unprotected areas of
daytime decreases in dry air entrainment and increases the aircraft, such as the underside of the wing, increas-
in entrainment at night. ing drag (Politovich 1989). Loss of a research aircraft
An active area to watch for advances in information by the Desert Research Institute in icing conditions may
regarding drop-size spectra, outside of aircraft icing, is have been caused by SLDs, typically drops in the 50- to
climate change research. Measurements and models 500-µm-diameter size range (Telford 1988). Coffey
characterizing cloud microphysical properties have (1995) describes the hazard of SLDs as observed from
become critical for parameterizing the effects of clouds the cockpit of a research aircraft, with advice on how to
on climate change. Radiative models used to simulate avoid and exit SLD conditions.
potential climate change and isolate the effects of green- Droplets larger than about 50 µm in diameter do not
house gases are very sensitive to cloud drop-size distri- remain suspended in clouds by turbulence effects as do
bution (Choularton and Bower 1993, Telford 1996). smaller droplets. Gravitational forces cause them to fall
Experiments analyzing the roles of cloud microphysi- at greater speeds as drop size increases, producing pre-
cal properties in climate change that are either in progress cipitation. Though long recognized as a hazard, these
or completed include the Atlantic Stratocumulus Tran- large drops have been receiving more attention in recent

18
To Contents
 years. The crash of an ATR-72 at Roselawn, Indiana, in overall, and a larger drop spectral width. They attribute
October 1994, focused the attention of aviation icing the formation of drizzle drops in stratus clouds to areas
researchers on SLDs and their unique hazard to aircraft where updraft velocities are greater, which causes differ-
(NTSB 1996, Broderick 1996). ent percentages of cloud condensation nuclei at the
Early research reports describing the results of flights cloud base to be activated. They suggest that updraft
measuring the microphysical properties of icing clouds velocity can be used to predict drop concentration and
have mentioned supercooled drizzle drops, for example, the width of the drop-size spectrum.
Kline (1949), but relatively few reports focused on In one of the most ambitious drizzle measurement
SLDs. Rodert (1951) and Lewis (1951) both indicated programs to date, Cober et al. (1995; 1996a,b,c) describe
the importance of freezing drizzle and freezing rain as freezing drizzle measurements made in the Canadian
aircraft icing hazards. However, most work until recent Freezing Drizzle Experiment (CFDE) off the New-
years has been in response to needs to understand rain- foundland coast. Freezing drizzle was encountered in
drop-formation mechanisms for cloud physicists, rather four research flights within thick (~1000-m) stratiform
than for aviation needs (Fletcher 1962, Hobbs and Deepak clouds. In these four encounters, liquid-water content
1981, Cotton and Anthes 1989, Rogers and Yau 1989, varied between 0.05 and 0.2 g m–3 when MVDs were
Houze 1993, Young 1993, Pruppacher and Klett 1997). larger than 40 µm. MVDs as large as 950 µm were
Isaac and Schemenauer (1979) found supercooled measured in freezing rain below the cloud base. Within
large drops near the tops of cumulus clouds near Yellow- clouds, MVDs often exceeded 500 µm. When combined
knife, NWT. Many cloud tops between 0°C and –8°C liquid-water contents and MVDs were compared to FAR
had concentrations of supercooled drops larger than 70 25, Appendix C, 34 of 147 data points fell outside the
µm. About twice as many clouds had concentrations of envelopes. They conclude that freezing drizzle may be
large water droplets as had concentrations of ice crys- a frequent phenomenon in East Coast winter storms and
tals. Large drops were associated with low liquid-water a significant aviation hazard.
contents, and droplets larger than 150 µm never had a Jeck (1996) published the most comprehensive
concentration of more than 1 L–1. The authors could review to date of the state of knowledge about freezing
not explain why the drops existed and did not relate rain (ZR) and drizzle (ZL) with regard to aviation. He
them to aircraft icing since the purpose of the research indicates that few instrumented aircraft have flown in
was related to precipitation enhancement. ZL and ZR, and that little is known about the meteoro-
Politovich (1989) describes icing from large droplets logical conditions and geographic locations of SLD
on a research aircraft flying in California and Arizona. occurrence. Elevated ZL and ZR, encountered by air-
Eleven flights are characterized within a narrow tem- craft in flight, are a hazard that may not be experienced
perature range, between –5.5 and –9.4°C and drop con- at the surface if they freeze as sleet before reaching the
centrations of generally less than 100 cm–3. Conditions ground. Though techniques have been proposed for
had the greatest effect on aircraft performance when detecting ZL and ZR from radiosondes, no reliable
fewer than 0.1 – 1 cm–3 droplets occurred in a size range methods of prediction are available, especially for ZL,
from 30 to 400 µm. Politovich indicates that the fre- which can occur without the traditional warm layer often
quency of these occurrences is low but not rare. Ample found in ZR. ZR and ZL are typically lower-altitude
moisture and time, accompanied by lift, must be avail- phenomena, with most occurring below 3811 m agl,
able to create these large drops. She suggests that envi- making them a distinct hazard to nonpressurized air-
ronments most likely to experience SLDs are orographic craft, helicopters, and all aircraft on approach and depar-
and upslope in warm fronts and within the warm sector ture. Little is known about the frequency, depth, and
of cyclones where adequate moisture and lift are avail- horizontal extent of ZR and ZL layers, the causes of
able. ZL, and the full range of meteorological conditions asso-
Feingold et al. (1996) argue, from numerical simu- ciated with each. Jeck indicated that the use of MVD to
lations, that the production of drizzle within clouds is characterize drop spectra associated with SLDs is not
related to droplet residence time and within-cloud turbu- appropriate because the MVD provides no indication
lence. Vigorously growing clouds produce more driz- that SLDs exist.
zle because they allow longer in-cloud drop dwell times, Hobbs and Rangno (1996) observed supercooled
prolonging the collision–coalescence process. Their drizzle drops with very high liquid-water contents off
arguments are similar to that of Politovich (1989). the Washington coast. The stratocumulus clouds were
Hudson and Svensson (1995) measured drizzle-drop trapped above an inversion, preventing cloud conden-
concentrations off the Southern California coast as part sation nuclei from the marine boundary layer below
of the FIRE experiments and associated drizzle drops the inversion from reaching the clouds. As a result, drop
closely with lower drop concentrations, larger droplets concentrations were low (~ 500 L–1), liquid-water con-

19
To Contents
tents were high (up to 0.8 g m–3), and drops as large as It is evident that SLDs create uniquely hazardous
200 µm in diameter were present. The authors indicate in-flight icing conditions, yet little is known about the
that supercooled layer clouds that form in clean mari- phenomenon: its characteristics, its climatology, or what
time air (lacking cloud condensation nuclei) that is comprises an SLD condition (Shah et al. 1998). Flights
decoupled from the surface could pose a significant threat during the winters of 1996–1997, 1997–1998, and
to aircraft from supercooled drizzle or rain. 1998–1999 by NASA Glenn Research Center’s Twin
Cober et al. (1996b,c) report conditions off the east Otter aircraft into SLD should help answer some of the
coast of Canada similar to those reported by Hobbs and remaining questions (Miller et al. 1998). Jeck’s (1996)
Rangno (1996). Freezing drizzle was observed in 1100- report addresses most of the weaknesses in knowledge
m-thick stratiform clouds in temperatures between –11°C about SLDs and is probably the most complete and suc-
and –8°C. The maritime air was very clean, with con- cinct paper on the subject from an aviation perspective.
densation nuclei allowing only a few drops to grow large
and coalesce. Though the MVD was 29 µm, cloud drop- 4.3.4 Temperature
lets larger than 40 µm exceeded 300 L–1, and 500-µm- The thermal environment of an icing event deter-
diameter drops were measured near the cloud tops. This mines the type, amount, and location of ice formation
suggests one mechanism for ZL, that of isolating humid on an airframe (Cooper and Sand 1997). The thermal
air with few condensation nuclei, allowing coalescence environment is controlled by radiative, convective, con-
and drop growth to occur. ductive, latent heat and advective processes of the atmo-
In reports exploring the causes of ZL off the Canadian sphere and the airframe and by the dynamics of the air-
east coast, Isaac et al. (1996) and Cober et al. (1996b,c) craft moving through the atmosphere. When isolated
review the processes that could cause ZL and compare from the airframe and the thermodynamics of the icing
them with CFDE measurements. In Newfoundland, ZL processes, thermal processes within the atmosphere
is associated with easterly and southeasterly winds and alone determine the temperature of air and of drops.
rarely with westerly winds. Only about 15% of ZR cases The “source” of cold also affects the amount, type,
are nonclassical, but 60% of ZL cases are nonclassical. and shape of ice that forms. For example, droplets
Classical ZR and ZL result from overrunning, such as warmer than 0°C may freeze upon a cold-soaked air-
occurs within warm fronts. Nonclassical drizzle forma- frame, but supercooled droplets may not freeze effi-
tion does not involve overrunning. Mechanisms may ciently on an airframe warmed aerodynamically above
include giant aerosol initiation of large drops, wind shear freezing. Supercooled drops impinging upon an air-
leading to entrainment, mixing and coalescence, long frame that is colder than 0°C will typically produce ice.
drop lifetimes in stratiform clouds that encourage drop Of the thermal processes operating, the temperature of
growth, and high supersaturations. Eleven days of flights the droplets, or the temperature of the atmosphere sur-
in both classical and nonclassical freezing precipitation rounding the droplets, is typically most important in
situations showed no consistency of mechanism, except determining whether ice will form on an airframe.
for wind direction and the existence of inversions and According to Rodert (1951), it is tacitly assumed
wind shear near the cloud top. that cloud droplets are at the same temperature as the
Climatologies of SLD accretions at the surface have surrounding atmosphere. This may not always be true
been developed as a method of assessing where freezing for cloud or for precipitation drops, which typically cool
rain may be occurring aloft as a hazard to aircraft. Strapp to the wet-bulb temperature of the surrounding atmo-
et al. (1996), Robbins and Cortinas (1996), and Bern- sphere through evaporation (Cooper and Sand 1997).
stein and Brown (1997) completed independent climatol- Since the relative humidity within icing clouds is typi-
ogies of the frequency of SLD events in North America cally near 100%, the dew point and air temperature will
to assess where aircraft icing due to ZR and ZL may be also be similar, especially within stratiform clouds of
occurring with greater frequency. All maps indicate stable air masses. Within cumulus clouds with active
freezing precipitation at the surface as being most com- updrafts and entrainment of dry air, evaporation and
mon east of the Rocky Mountains, with frequency subsequent cooling may be greatest near the outside of
increasing from the mid-Mississippi Valley to the North- the cloud where entrainment is most active (FAA 1991).
east and Labrador, with an axis through the Great Lakes Therefore, one will find warm cores in clouds with inter-
Basin. Ahmed and Brown (1995) produced a climatolo- nal updrafts because of reduced evaporation and the
gy of in-flight ZR globally, with seasonal detail in Great release of latent heat as drops grow. In general, cloud-
Britain and Europe from the U.K. Meteorological Office’s size droplets reach thermal equilibrium with surround-
numerical model output. Their model-derived climatolo- ing air very rapidly, typically within 1 second (Borovi-
gy suggests high frequencies of ZR over the Atlantic kov et al. 1963). Precipitation drops cool to the wet-
and Pacific Oceans. bulb temperature after they have fallen into dry air

20
To Contents
 below cloud base and begin to evaporate, but lag times level flight, so during approach and departure static
can be on the order of 10 s (Fletcher 1962). outside air temperature (OAT) at the aircraft will not
Droplets may also be cooler than the surrounding air be a reliable indicator of air temperature within the flight
within warm tongues of air advected over colder sur- path ahead of the aircraft. Vertical temperatures can
face air and below colder air aloft. Snow falling into vary, from nearly isothermal over large vertical dis-
these warm layers from above may partially or com- tances to changing by tens of degrees over a few hun-
pletely melt. However, until fully melted their tempera- dreds of meters, especially when transiting inversions.
ture remains at 0°C, so they remain colder than the warm As an example, Schroeder (1990) illustrates a winter
layer until all ice melts and the drops begin to heat temperature inversion over the Denver area of about
through convection and radiation. These areas are often 22°C within a vertical distance of less than 500 m. Such
identified on radar displays as “bright bands”—zones rapid changes are not unusual during winter.
where falling ice crystals melt and coalesce into rain- This evidence suggests that OAT measured at the
drops. aircraft, though a general indicator, is insufficient for
Temperature within clouds may fluctuate several determining if liquid water in the flight path is super-
degrees over distances of only a few meters. In addi- cooled. Confidence in the representativeness of OAT
tion, within-cloud temperatures can be considerably to predict temperature ahead of the aircraft varies with
different from outside-cloud temperatures. Rapid and the meteorological conditions around the aircraft and
significant temperature fluctuation from cloud to cloud, with the mode of flight: ascent, level, or descent.
and within clouds, makes determination of supercool- Nonthermal parameters may be useful surrogates for
ing difficult. For example, data from NCAR Winter indicating temperature. Detection of glaciation within
Icing Storms Project (WISP) flights indicate that tem- a cloud suggests that any liquid water within the cloud
perature fluctuations from clear air to cloud can be as is supercooled. However, if ice crystals are not present,
much as 6°C (NCAR 1990). Flights in Poland with a the method is not effective because there is no physical
rapid-response airborne thermometer show temperature indication of supercooling.
fluctuations within clouds of 2°C in distances of less In addition to detecting temperature within the flight
than 150 m (Haman and Malinowski 1996). Time series path, range-resolved temperature must also be sensed
of temperature through the core of a warm cumulus above and below the aircraft to provide a potential route
cloud (Lawson and Cooper 1990) showed a 3 to 4°C of escape from icing into warm air. Since air tempera-
increase of temperature upon entering the cloud, with ture varies more rapidly vertically than horizontally,
similar subsequent cooling upon exit. Temperature especially within inversions, sensing temperature above
within the cloud was nearly constant. Penetrations of and below aircraft may be useful.
supercooled stratiform clouds showed, depending upon The accuracy of temperature measurement may also
the thermometer observed since several were being be critical because of its effect, with liquid-water con-
tested, a 0 to 1°C decrease in temperature when inside tent and drop size, on ice type, density, and shape on
the cloud as compared with dry air around the cloud. In leading edges (Wright 1995). Since very small changes
another case, but without identification of cloud genera, in temperature may create large changes in ice accretion
cooling of 3°C was observed within the cloud when amount and shape, it may be useful to measure temper-
compared with surrounding dry air. Lawson and Rodi ature ahead, above, and below the aircraft with high
(1992) penetrated warm cumulus humilis clouds with accuracy.
fast-response thermometers and showed immediate 6°C
cooling when entering the clouds and immediate 6°C 4.3.5 Spatial structure
warming when exiting. Spatial scales of icing conditions affect the utility of
Temperature changes can also be large and rapid in remote-sensing systems. Icing conditions that are spa-
the horizontal when an aircraft transits fronts, though tially homogeneous over thousands of square kilometers
not as rapid as upon entering or exiting clouds. For offer less potential for avoidance without climbing or
example, transiting a cold front in horizontal flight can descending. The size of icing cells and storm areas also
produce temperature changes of 0.2°C per kilometer or affects the needed sensing range of remote-sensing sys-
more (Berry et al. 1945). Smaller changes are observed tems. Storms with small icing cells may be sensed by a
when transiting warm fronts in level flight. This is ignor- remote-sensing system sufficiently to allow an aircraft
ing turbulent mixing in the shear zone along frontal to progress iteratively through the system. Storms with
surfaces, which can cause more rapid localized tem- large icing cells may be too large to be sensed through,
perature changes. potentially trapping aviators (Fig. 2) (Kirkpatrick 1970).
The most rapid temperature changes, however, are The inability to sense completely through icing reduces
experienced during ascent or descent rather than within avoidance options and may limit an aircraft to turning

21
To Contents
back along its original route to avoid icing. Most icing cal properties related to icing. The length of supercooled
areas may be scanned completely through with a detec- liquid-water content patches was measured. A patch was
tion range of 80 km or more, as suggested by Curry defined as having supercooled liquid-water content of
and Liu (1992). at least 0.025 g m–3 for at least 0.5 km of flight. Patch-
Relatively little is known about the horizontal extent es terminated when supercooled liquid-water content
of icing conditions, though more is known today because was less than 0.025 g m–3 for 0.5 km. Average patch
of modern research flights than was known in 1979 length was 4.3 km, with a mean liquid-water content of
when Milton Beheim of NASA Lewis Research Center 0.13 g m–3, and the median patch length was 1.7 km.
indicated that the horizontal extent of the icing cloud About 90% of patches were less than 7 km in length,
had not been adequately defined (Beheim 1979). He with less than 2% longer than 50 km. Flights were made
also indicated that the fine-grain structure of the icing in low-level stratus clouds and within low-pressure areas
cloud had not been well defined. and through fronts.
FAR 25, Appendix C, tries to address icing spatial Politovich (1982) describes flights through super-
scale by providing tables for continuous conditions cooled stratiform clouds over the Great Lakes and the
within stratiform clouds and intermittent conditions Great Plains in 1981. A cloud extent started when the
within cumulus clouds (FAA 1991). Jeck (1983) indi- aircraft was within supercooled liquid cloud for at least
cates, however, that the horizontal extent of icing speci- 1 km, and cloud elements “separated by less than the
fied in Appendix C has no specification for the discon- element length were combined unless the gap was
tinuity in icing and the size and frequency of any cloud greater than 6 km.” The average icing encounter in Great
gaps. He concludes that “horizontal extent,” as indicated Lakes stratiform clouds was 9 km long, and within Great
in Appendix C, does not imply the overall dimensions Plains stratiform clouds the average encounter was 24
of icing cloud systems. km long. Embedded cells of supercooled liquid water
Jeck (1983) provides two methods of expressing the within bands of frozen clouds averaged 6 km in length.
horizontal extent of icing. For engineering-design pur- The larger extents were a result of large-scale lifting of
poses, he indicates that the horizontal extent of icing air masses. Though isolated pockets of higher liquid-
encounters, consistent with Appendix C, is the “dis- water content occurred, the clouds were generally fairly
tance flown during a given icing encounter until a cloud uniform at a given flight level.
gap of some specified duration signals the end” of the Cooper et al. (1982) characterized distances of
encounter. Of more use in determining the utility of liquid-water content encountered greater than specified
remote sensing is the horizontal extent of individual thresholds, and the frequency and size of gaps between
icing events, where an icing event is the actual distance icing encounters (where evaporation or sublimation of
of icing, which ceases at a cloud gap of any length. ice accreted on an aircraft could occur). Data from 1083
Jeck reanalyzed NACA data by the horizontal extent flight hours in California, Montana, Utah, Florida, Kan-
of the icing event and presented modern data in the same sas, Illinois, Michigan, and the Great Lakes in summer
way. The analyses indicate, as is consistent with Appen- and winter conditions were used to compile the infor-
dix C, that there is an inverse relationship between mation. In all seasons, flights were in icing conditions,
liquid-water content and event horizontal extent. Hori- but at higher altitudes in summer than in winter. The
zontal extent is about 33 km at a liquid-water content flights deliberately sought the most severe icing condi-
of about 0.01 g m–3 and is about 5.5 km for the largest tions. Cooper and his colleagues present information
observed liquid-water contents, about 1.5 g m–3. This indicating exposure distance in two ways:
should not imply, however, that liquid-water content is
• Probability of exceeding a given liquid-water con-
constant for these distances. These are average values,
tent in a given distance
and individual patches of larger or smaller liquid-water
• Probability of exceeding a liquid-water content of
contents can occur within these extents. An aircraft with
0.1 g m–3 for each region.
ice protection may be able to tolerate liquid-water con-
tent to a given magnitude but may have to avoid larger As examples, when averaged over 1 km, liquid-water
liquid-water contents. Thus, it may also be helpful to content exceeding 0.1 g m–3 occurred about 5% of the
know the size of icing “patches” with larger than speci- time and exceeded 0.5 g m–3 about 1% of the time.
fied liquid-water contents. When averaged over a distance of 10 km, liquid-water
The size of liquid-water content patches may be content exceeded 0.5 g m–3 about 0.5% of the time.
ascertained from measurements during research flights. Viewed regionally, there are large differences in the
In 1992, Cober et al. (1995) flew 31 missions, as part extent of liquid-water content greater than 0.1 g m–3.
of the Canadian Atlantic Storms Project (CASP), into The Great Lakes area, which has the lowest overall
East Coast winter storms to measure cloud microphysi- liquid-water content, has the longest continuous icing

22
To Contents
 encounters, over 80 km in the winter. Kansas, Mon- Strapp et al. (1996) and Bernstein and Brown (1997)
tana, Illinois, and Florida had short encounters, most have created modern climatologies of the occurrence
being less than 36 km in length. In Kansas and Florida of freezing precipitation. Neither study, however, specu-
there was about a 1% chance that an icing encounter lates about the spatial extent of individual freezing pre-
would extend more than 10 km. Cooper and his col- cipitation storms. Bernstein (1996) indicates implic-
leagues point out that only about 10% of encounters itly that freezing precipitation does occur for hundreds
extended more than 5 km, and the majority of icing of kilometers, but broken in continuity. As an example,
measured was in cumulus clouds. he presents a map for 1800 hr on 6 March 1996, illustrat-
Gap encounters are useful for assessing the utility ing freezing precipitation as extending in a broken band
of remote sensing to avoid icing, because aircraft could from New York City through Missouri, with a maxi-
avoid ice by navigating gaps. Cooper et al. (1982) report mum width of about 250 km.
that gaps are typically short, like icing encounters, with In general, spatial patterns of icing on sub-kilome-
50% being less than 5 km in extent. A gap occurred ter to tens-of-kilometers scales, and icing’s relationship
when liquid-water content was less than 0.01 g m–3. to synoptic and mesoscale weather, are only generally
Gerber (1996) reports liquid-water content gaps, or understood. The ability to avoid icing is a function of
minima (called turbules), of a much smaller scale em- its spatial distribution. Large cloud masses that are
bedded within marine stratocumulus clouds. Turbules homogeneous with respect to icing are difficult to avoid.
are typically a few hundred meters or less across.
Kline and Walker (1951) related icing to synoptic 4.3.6 Mixed-phase clouds
patterns in stratiform clouds during 22 flights from 1948 Supercooled liquid water freezes on aircraft struc-
through 1950. In extratropical cyclones, most icing was tures, whereas ice crystals within clouds, snow, and ice
associated with post-cold-frontal situations, with most pellets typically do not adhere (Riley 1998). Never-
icing in the southwestern and northwestern quadrants theless, clouds composed of mixtures of ice crystals
of the storm. Very little icing was found in the overrun- and supercooled water are of interest for remote sens-
ning portions of warm fronts east of the storm center. ing of icing potential for several reasons. First, remote-
Most icing was typically 300–400 km behind cold sensing systems scanning clouds that are completely
fronts, and north of the center of lows, similar to pat- glaciated or mixed phase must distinguish successfully
terns reported by Ryerson (1990) at Mt. Washington, between ice and supercooled liquid water and not be
N.H., and Mt. Mansfield, Vt. These patterns are con- compromised by the presence of ice crystals. A remote-
trary to analyses of pilot reports of icing reported by sensing system must be capable of quantifying the
Politovich* that indicate that most icing is ahead of amount of supercooled liquid water mixed with ice crys-
warm fronts and near the center of lows, with least icing tals or of sensing beyond a frozen cloud in the fore-
behind warm fronts and cold fronts. ground, for example, to a more distant supercooled
Little is known about the horizontal extent of ZL liquid cloud.
and ZR, according to Jeck (1996) in a summary of The second concern for mixed-phase clouds is in
knowledge about the phenomenon. Bennett’s (1959) reference to the need to range-resolve temperature to
report indicated that most freezing rain occurs in over- determine if sensed liquid water is supercooled. A cloud
running situations, so it is associated with warm fronts made up of a mixture of liquid water and ice crystals is
in many instances. Freezing rain can extend continu- likely to contain supercooled liquid water. As a result,
ously or intermittently several hundred kilometers par- even if a method is not found for range-resolving tem-
allel to a front and short distances perpendicular to perature, it may be possible to determine whether liquid
fronts. It can also be associated with cold fronts, but water is supercooled by sensing the presence of ice crys-
then it typically is of shorter extent than in warm fronts. tals mixed with the liquid water. Thus, mixed-phase
Design values use 160 km as a representative extent. clouds may serve as a surrogate for explicit tempera-
There is no information for the extent of ZL, according ture measurement ahead of the aircraft. Mixed-phase
to Jeck, who argues that extent should be related to the clouds, however, may be less of an icing hazard than
time an aircraft must spend below 7000 ft, especially supercooled clouds without ice crystals (Guttman and
on approach and departure. This agrees with Perkins’ Jeck 1987, Riley 1998).
(1952) conclusions that over 50% of icing conditions Simply seeking ice crystals may not be a reliable
in general are found during climb or descent. solution to determining supercooling, however. Clouds
may be composed completely of supercooled liquid
water and still contain no ice crystals, or they may con-
* Personal communication, M. Politovich, National Center for Atmo- tain a concentration of ice crystals that is so small as to
spheric Research, Boulder, Colorado, 1997. be not detectable. The success of using mixed-phase

23
To Contents
clouds as an indicator of supercooling is thus depen- Additional information about the utility of using
dent upon the probability of supercooled liquid clouds mixed-phase clouds for indications of supercooling is
containing detectable ice crystals. Parameterizing nucle- available from various field programs. Tremblay et al.
ation of water droplets in clouds, and the glaciation (1996) observed mixed-phase clouds at 4461 points as
process, is one of the classical problems cloud physics part of CFDE in 1995 off the Newfoundland coast. Plots
has yet to solve. of the proportion of liquid water vs. ice water within
When a cloud is cooled below 0°C, ice crystals could mixed-phase clouds showed a temperature dependence
form. However, because there are relatively few ice between 0°C and –10°C if liquid-water contents larger
nuclei in the atmosphere when compared with conden- than 0.3 g m–3 are ignored. As temperature decreased,
sation nuclei, nucleation often does not begin until drop- the proportion of ice increased. However, the relation-
lets cool to –10°C (Rogers and Yau 1989). Observations ship is also proportional to the cloud liquid-water con-
of 258 clouds by Hobbs et al. (1974, as cited by Rogers tent, with the proportion of cloud water nucleating
and Yau 1989) showed that glaciation typically does not increasing, at a given temperature, with cloud total
begin until cloud-top temperatures cool to about –4°C, liquid-water content. In flights through summer cumu-
after which the percentage of clouds containing ice lus in southern Missouri, Koenig (1963) found glacia-
increases to 100% at a cloud-top temperature of about tion related to drop size, with clouds with large drops
–20°C. Rogers and Yau (1989) state that it is impossi- rapidly forming high concentrations of ice crystals
ble to determine at what cloud-top temperature any indi- regardless of the availability of ice nuclei.
vidual cloud will begin to glaciate or to estimate how Bower et al. (1996) surveyed frontal and maritime
much glaciation will occur. Thus, in general, clouds with convective clouds from the United Kingdom and con-
tops warmer than about –5°C are ice free, and clouds tinental convective clouds from New Mexico and Mon-
with tops colder than –20°C are virtually guaranteed to tana to refine parameterization schemes for global circu-
have ice crystals (Riley 1998). lation models. Detailed analyses were done for all
The first crystals to appear in a cloud must form on clouds that had been measured for glaciation activity
ice nuclei (Rogers and Yau 1989). Additional crystals using aircraft-mounted instrumentation. Continental and
are formed from secondary processes such as the frac- maritime frontal clouds had very rapid glaciation, begin-
ture of ice crystals and the shattering or splintering of ning at 0°C, with total glaciation typically ocurring at
drops as they freeze. These crystal fragments then strike temperatures of –10 to –15°C. Continental and mari-
liquid-water droplets, causing them to freeze though time convective cloud glaciation was much slower and
contact nucleation, or the fragments simply serve as less complete, with glaciation beginning at 0°C, but at
deposition nuclei (Houze 1993). –3°C, typical clouds were only about 40% glaciated.
Overall, glaciation is difficult to predict because it At –15°C, some clouds were still 90% supercooled
depends upon cloud type, cloud age, liquid-water con- liquid water.
tent, and geographical location—especially as related It is not clear from these studies whether glaciation
to air mass type and availability of icing nuclei (Rogers begins in earnest at 0°C. Characteristics of mixed-phase
and Yau 1989). Since the probability of glaciation clouds must be better defined to determine the proba-
increases with decreasing temperature, it would be bility of ice crystals at temperatures below 0°C. The
expected to find a monotonically increasing percent- most current and thorough review of mixed-phase
age of cloud water to be frozen at lower temperatures. clouds and aircraft icing is by Riley (1998).
This does not appear to occur, however. Instead, once
clouds begin to glaciate, freezing occurs rapidly, and 4.4 In-situ instrumentation
the final ice particle concentration is not proportional In-situ measurements of cloud microphysics are
to temperature (Pruppacher and Klett 1997). needed to support and augment the remote sensing of
Mixed-phase clouds are also not necessarily uni- in-flight icing conditions, and in-situ instrumentation is
formly glaciated. On the basis of a large number of needed for improved characterization of icing cloud
soundings of nimbostratus clouds in Russia, Borovik- microphysics. Drop-size distributions are often not cor-
ov et al. (1963) report that three different types of mixed rectly represented by current measurement methods, and
cloud structure can occur: clouds can consist of rela- ice crystals and drops can be confused by the coarse-
tively uniform mixtures of crystals and water through- ness of sensing systems. Two measurements by the same
out, successive layers of water droplets and ice crys- model of instrument often do not agree when used on the
tals, or three or four layers of warm water, supercooled same aircraft, which indicates repeatability or calibration
water, mixed conditions, and ice. The relative frequency problems. The dynamic range and sensitivity of instru-
of each was observed 52%, 28%, and 20% of the time, ments is often insufficient, and gaps in size distributions
respectively, in Russia. synthesized from combinations of instruments often occur.

24
To Contents
 In-situ instrumentation would be useful onboard all and the reverse-flow thermometer were both accurate
aircraft for sensing when icing conditions have been within clouds, and the Rosemount was not (Lawson and
entered and for near-real-time calibration of onboard Cooper 1990).
remote sensors. Instrumentation intended for use on all Marillier et al. (1991) constructed an ultrasonic
aircraft may not require the accuracy of research instru- thermometer that measured temperature ahead of an air-
ments. Ease of use, maintenance, cost, and size are more craft to avoid cloud-wetting effects and to obtain temper-
important factors in these applications. ature at 100 Hz. Measurements were typically accurate
Most current aircraft-mounted instrumentation suited to within a few tenths of a degree Celsius of a colocated
for cloud microphysical measurements is intended for Rosemount probe.
research applications. Very few instruments, notably air- Lawson and Rodi (1992), Friehe and Khelif (1993),
temperature measurement devices, are sufficiently inex- and Haman and Malinowski (1996) have constructed
pensive and robust for general field use. The following very fast response thermometers in an attempt to match
is a brief review of available instrumentation and their the response time of a forward-scattering spectrometer
general applicability. This review does not discuss probe (FSSP). These thermometers are very delicate,
instrumentation for detecting ice on aircraft surfaces, with either platinum wires 12.5 µm in diameter (Law-
either preflight or in-flight. A review of in-flight ice son and Rodi 1992), tungsten wires 2.5 µm in diameter,
detectors is available in the FAA Aircraft Icing Hand- or small thermistors (Friehe and Khelif 1993, Haman
book (FAA 1991) and in an SAE document (SAE 1995). and Malinowski 1996). Both wire probes are accurate
Cloud microphysics measurement instruments are but delicate and broke easily when stressed by high-
typically designed to measure temperature, liquid-water speed airflow and precipitation. All three instruments
content, and elements of the drop-size spectrum. Early are too delicate for general operational use, and the ther-
instrumentation was manually operated, but most mod- mistor instrument had calibration difficulties.
ern instruments are electronic.
4.4.2 Liquid-water content measurement
4.4.1 Temperature measurement The first common instrument for liquid-water meas-
Temperature cannot be measured with a standard urement was the rotating multicylinder, developed at
outside air-temperature probe because, in addition to Mt. Washington Observatory between 1940 and 1945
accuracy and exposure problems, it is wetted by cloud (Lewis et al. 1947, 1953; FAA 1991; Howe 1991). In
water. A wet thermometer measures the wet-bulb tem- the 1930s, a single rotating cylinder was used. The rotat-
perature. Lawson and Cooper (1990) provide a detailed ing multicylinder, still in use at Mt. Washington Observ-
analysis of the problem. Thus, the most important task atory but no longer used on aircraft, is used to deter-
is to protect the thermometer from cloud droplets with- mine cloud liquid-water content and the shape of the
out disturbing the measurement. drop-size spectrum. The multicylinder typically con-
A common temperature measurement instrument is sists of six cylinders, stepped in diameter. The shape of
the Rosemount total temperature probe (Haman et al. the drop-size spectrum is determined by comparing the
1997). It is used by the U.S. Air Force (Glass and Gran- mass of ice on each cylinder to curves of expected ice
tham 1981), by NCAR (Sand et al. 1984) on its King accretion on each cylinder for a given shaped drop spec-
Air, and on the Canadian Convair (Cober et al. 1996b). trum, after theory developed by Langmuir and Blodgett
The instrument measures the resistance of a platinum (1946). Cloud liquid-water content is related to the mass
wire in a bridge circuit. Accuracy is claimed to be from of ice on the cylinder; the smallest cylinder with the
0.5°C (Sand at al. 1984) to ±1°C (Cober et al. 1996b), highest collection efficiency is typically used for the
but wetting in clouds is a problem. calculation.
NCAR and the AES also use reverse-flow tempera- Rotating multicylinders are manually operated and,
ture probes. According to Cober et al. (1996b), they are depending upon cloud liquid-water content and rela-
as accurate as the Rosemount and prevent wetting of tive wind speed, are exposed from a few minutes to
the probe. more than 20 minutes (Kline 1949, Howe 1991).
There has been considerable effort to improve the Jeck (1980) thoroughly reviews problems with the
accuracy, resolution, and response time of thermome- multicylinder method, especially with regard to runoff
ters to measure small-scale cloud features. Lawson and near the Ludlum limit. The advantage of the multi-
Cooper (1990) analyzed the Ophir radiometric thermo- cylinder method is that, with a skilled operator, it is
meter to improve reliability within clouds and to cheap, easy to use, and reasonably accurate in colder
improve response time. However, the sample volume temperatures, smaller drop sizes, and moderate liquid-
is quite large (10 m in depth), and the instrument is water contents.
large, expensive, and complex. The Ophir thermometer The only other early method of measuring liquid-

25
To Contents
water content that received much use was the icing- serviced. King et al. (1978) reported a sensitivity of
rate meter, a horizontal rod with holes facing into the 0.02 g m–3, a response time better than 0.05 s, and an
relative wind (Perkins 1952). Ice plugging the holes accuracy of 5% at 1.0 g m–3. Baumgardner (1983) could
would be sensed with a pressure transducer, triggering draw no conclusions about the King probe, other than
a deicing cycle. Assumptions about ice density and the that it was promising and deserved more study. Cober
meter’s collection efficiency allowed rough estimation et al. (1996b) use the King probe on the Canadian Con-
of liquid-water content. Electronic devices have since vair and claim accuracies of ±0.02 g m–3 for liquid-
replaced multicylinders. water contents of less than 0.2 g m–3. In a range of
Today, the primary electronic devices for measur- liquid-water contents between 0.1 and 1.25 g m–3, and
ing liquid-water content are the Rosemount ice detector with MVDs between 10 and 40 µm, Ide (1996) found
and the King and Johnson–Williams hot wire probes accuracy for the King probe to be within ±0.1 g m–3 of
(FAA 1991, Knollenberg 1981, Glass and Grantham the calibrated NASA Glenn Research Center Icing
1981). The Rosemount ice detector is a standard instru- Research Tunnel (IRT). In general, wind-tunnel testing
ment on most icing research aircraft and on the ground, has shown the King probe to be generally accurate to
and it may also be used to compute liquid-water content 5% at 1.0 g m–3, and it is generally superior to but more
by relating the deicing rate of the detector to relative fragile than the Johnson–Williams probe (FAA 1991).
wind velocity (Brown 1981, FAA 1991, Claffey et al. Another hot-wire-based instrument is the Nezorov
1995). A 6-mm-diameter by 25-mm-long probe vibrates probe developed in Russia (Korolev et al. 1996). Cur-
axially at its resonant frequency of 40,000 Hz. As ice rently being flown on the Canadian NRC Convair and
accretes on the probe, its frequency drops until, at a the NASA Glenn Research Center Twin Otter, it has
preset frequency, a heater deices the probe. Liquid water the unique ability to quantify both the supercooled
may be computed if the mass of ice, the exposure time, liquid-water content and the ice-water component of
the relative wind velocity, and the collection efficiency clouds (Miller et al. 1998). Similarly to the Johnson–
of the probe are known. The detector is reasonably accu- Williams probe, a reference heater corrects for convec-
rate, within the range of conditions found in most moun- tive heat losses. Though details are not available, it
tain and aircraft applications, at moderate liquid-water appears that liquid- and ice-water components are sepa-
contents (Claffey et al. 1995). The typical liquid-water rated by the lag caused by phase changes as water
performance range is 0.05 to 3.0 g m–3 (FAA 1991). vaporizes within the instrument (but ice particles may
The Johnson–Williams probe exposes a hot wire to break away, causing negligible heat loss—and error). The
the droplet-laden air flow, and a second “compensat- exact process is not clear. Comparisons with the King
ing” wire is protected from liquid water but exposed to probe in CFDE flights show less than 10% disagree-
the air flow (Knollenberg 1981, FAA 1991). The second ment, with the Nezorov showing better performance in
wire compensates for variations in air speed, altitude, SLD environments. Wind-tunnel tests demonstrated
and air temperature. The resistance of the wires changes better stability than the King probe at low temperatures
as they warm and cool, and the change of resistance is and the ability to measure the frozen component of
measured through a Wheatstone Bridge circuit. The mixed-phase clouds. Verification in snow has not been
instrument has an absolute liquid-water range from 0.0 possible because of a lack of standards.
to 1.5 g m–3 (Jeck 1980) to 6.0 g m–3 (FAA 1991). Per- Liquid-water content may also be measured opti-
sonne et al. (1982) found undermeasurement of liquid- cally, typically utilizing the interaction of laser-based
water content in large-drop environments for that por- collimated light and droplets. Gerber (1991, 1996) has
tion of the liquid-water content in droplets larger than developed an instrument, the particle volume monitor
30 mm in diameter. A ±20% error limit is often assumed (PVM), that measures cloud liquid-water content, inte-
for the probe, but it can be smaller with wind-tunnel grated particle surface area, and effective cloud droplet
calibration (Baumgardner 1983, Sand et al. 1984). radius. All measurements are made simultaneously in
The CSIRO, or King, hot-wire probe measures liquid a large, 1.25-cm3 sample volume. The instrument oper-
water by maintaining a copper wire coil nominally 1.5 ates by passing droplets through a laser beam, which
mm in diameter exposed to the air stream at a constant then forward-scatter laser light through a lens and a
temperature (King et al. 1978, Knollenberg 1981, FAA variable transmission filter onto a detector. Output from
1991). The electrical energy necessary to maintain a the detector is mathematically inverted to derive liquid-
constant temperature under the cooling influence of the water content and effective drop radius. The instrument
air stream and impinging water droplets is related to resembles a class of instruments called “laser-diffraction
liquid-water content after corrections are made for air particle-sizing instruments” (Gerber 1996). Compari-
temperature and wind speed. The probe requires no sons of the PVM with other instruments in environ-
wind-tunnel calibration, is rugged, and is easily field mental chambers, on mountain tops, and on aircraft have

26
To Contents
 been encouraging. The instrument has been demon- 4.4.3 Drop-size measurement
strated to be reliable in large-drop environments and in Measurement of cloud drop-size spectra has been
large liquid-water contents. It also does not appear to important since early aircraft icing research, primarily
be disturbed by ice crystals, but independent tests must because of their impact on ice shape, ice type, collection
be done before the full accuracy and reliability of the efficiency, and runback. However, the increased inter-
instrument is known. est in SLDs in recent years has placed renewed emphasis
The forward-scattering spectrometer probe (FSSP), on drop-size measurement. Fewer instruments are avail-
developed by Knollenberg (1981) and manufactured by able for measuring drop-size spectra than for measuring
Particle Measuring Systems (PMS), is the most com- liquid-water content, but many of the instruments
monly used optical probe; it is found on nearly all described above for measuring liquid-water content have
research aircraft (Sand et al. 1984, Cober et al. 1996b, dual uses and measure both. In the case of optical instru-
Baumgardner et al. 1993, Thomas and Marwitz 1995). ments, liquid-water content is typically derived from the
Intended for measuring drop sizes from 2 to 47 µm in measured drop-size spectrum and drop concentration.
diameter (94 µm in extended mode), cloud liquid-water Except for oiled or soot-covered slides, the rotating
content can be computed by integrating the spectrum multicylinder described earlier was the first widely used
of drop sizes. However, liquid-water measurements are instrument for obtaining the shape of the drop-size spec-
prone to large error due to over- or undersizing of drop trum and MVD (Lewis et al. 1947, 1953; FAA 1991;
sizes. The instrument operates by forward-scattering Howe 1991). The drop-size spectrum shape is deter-
light through cloud droplets as they pass through a nar- mined by, in effect, fitting the accreted ice weights of
row laser beam to a detector that records a drop size the six cylinders to a series of curves, each represent-
proportional to the flash of light. Drops are classified ing a different droplet spectrum shape. The rotating multi-
into fifteen 3-µm-wide bins. Sources of error in the cylinder method provides only a general indication of
instrument are a small sample volume, false sizes from the breadth of the droplet size distribution, in part
ice crystals, ice accretion and fogging of the optics, because it can be fit to only a finite set of curves and
blockage of airflow through the instrument by ice, and because some clouds have bimodal or multimodal dis-
saturation of the instrument’s electronics at high air- tributions. Howe (1991) states that accuracy in deter-
speeds and large particle concentrations (FAA 1991). mination of liquid-water content and droplet size is
In addition, large drops are typically incorrectly meas- better than ±10% when cloud drop-size distributions
ured. As a result of drop-size measurement errors, are narrow or moderately broad. When drop-size distri-
liquid-water contents typically have up to 34% error butions are extremely broad, accuracy is reduced to
(FAA 1991, Baumgardner 1983). Ide (1996) found about ±20%.
liquid-water contents computed from an FSSP to be Gerber’s PVM (1991, 1996; Gerber et al. 1994) pro-
overestimated by 50% in MVDs up to 60 µm, and by vides the effective drop radius of clouds. Few compari-
100 to 150% in larger MVDs at NASA’s Glenn Research sons have been made with other instruments, but a com-
Center Icing Research Tunnel. Baumgardner et al. (1993) parison with the FSSP (Gerber 1996) shows a linear
and Brenguier (1993), however, have successfully relationship between the two instruments, though not a
modified the FSSP to measure the microstructure of 1:1 relationship. Gerber suggested that the mismatch,
clouds at the centimeter scale. with the FSSP providing smaller drop sizes than the
The Phase Doppler Particle Analyzer (PDPA), devel- PVM, was due to errors in the FSSP.
oped by Aerometrics, Inc. with assistance from NASA The FSSP was described in the liquid-water discus-
Glenn Research Center (NASA 1997), measures drop sion above, where some if its problems of measuring
diameters from 0.7 to 125 µm but can be extended to drop size were also discussed. Overall, the instrument
2000 µm (Aerometrics 1997). Droplet sampling is made tends to broaden the drop-size spectra and, in drops
in a small sample volume at the intersection of two laser larger than about 45 µm, measurements may not be
beams. Droplets passing through the beams create an inter- trustworthy (FAA 1991).
ference fringe pattern that is projected into several detect- Optical array probes (OAPs), manufactured by Par-
ors. The detectors produce a Doppler signal proportional ticle Measuring Systems (PMS), measure drop size by
to the droplet’s velocity and size. Droplet number density, imaging (Knollenberg 1981, Oldenburg and Ide 1990a,
and thus liquid-water content, are also computed. Cali- FAA 1991). OAPs image droplet shadows onto an array
bration is not necessary. Models of the instrument have of photodiodes by allowing drops to flow through a
been developed for both wind-tunnel and aircraft use. laser beam. The loss of light on an individual array ele-
However, the PDPA typically does not appear on equip- ment is detected by a logic circuit that measures the
ment lists of primary cloud research aircraft. shadow size. The drop size is a function of the shadow

27
To Contents
size and the optical elements between the drop and the development for retrieving liquid-water content and
detector array. If the measurement is made along one drop-size spectra.* The second instrument, the CDS,
dimension of the particle, the probe is called a 1-D probe. measures the angular resolution of forward-scattered
Only shadows fully within the array are accepted, so large light from an ensemble of cloud drops. A 256-photo-
droplets—droplets not fully within the array—are diode array measures the scattered light, and liquid-
rejected. The range of drop sizes measured is typically water content is computed from the angular measure-
20 to 300 µm, or 300 to 4500 µm; and other ranges are ments of the forward-scattered light, which expresses
available (Knollenberg 1981, FAA 1991). the collective drop sizes of the drop ensemble. Liquid-
OAP 2-D probes are similar to 1-D probes optically, water content measurements have been successfully
but enhanced signal processing speed allows the photo- compared with other instruments in a wind tunnel, and
diode array to be scanned faster, retrieving, in effect, a drop-size spectra measurements have been tested in the
shadow of particles as they traverse the laser beam. The laboratory, aboard an aircraft, and at Mt. Washington
result is an image of particles indicating the shape of Observatory. No data have been published in the open
ice crystals and the size of drops. Particle sizes are literature demonstrating the CDS’s ability to measure
binned into 64 size classes. Two-D probes have been liquid water or the drop-size spectrum.
configured to detect particles up to 6400 µm in diame-
ter (Marcotte et al. 1996). The accuracy of OAPs has 4.5 Terminology
been scrutinized in recent years with the renewed inter- A remote-sensing system designed to detect and map
est in SLDs. Problems of aircraft speed and location of icing potential within a projected flight path will sense
the drop within the imager depth of field can cause the meteorological conditions that create ice on an air-
OAPs to miss smaller drops and oversize large drops craft: liquid-water content, temperature, and drop size.
(Lawson et al. 1996, Morrison et al. 1997). These prob- Then the remote-sensing system may utilize expert sys-
lems are detected when FSSP and OAP ranges overlap tem or fuzzy logic to create information for a cockpit
and when OAPs are compared with other instruments. display that the pilot can use if icing is entered. Since
New array-processing techniques and algorithms have ice cannot occur on the aircraft until the aircraft enters
been suggested to correct these problems (Hobbs et al. the icing conditions, only an explicit numerical model
1996, Korolev et al. 1996). operated with the remote-sensing system, a set of guide-
A comparison of the PDPA with the FSSP and a PMS lines that relate icing potential to expected performance,
OAP (Oldenburg and Ide 1990a,b) indicates that all would solve the problem of how to relate meteorolog-
three instruments generally agree well. Disagreements ical conditions to the pilot. Bragg et al. (1998) propose
occurred in drop sizes smaller than about 10 µm because a smart icing system that recognizes how aircraft sys-
smaller droplets were suspected of freezing, and each tems should respond to icing and advises the pilot after
instrument treats ice crystals differently because of the the aircraft has entered icing. A similar system could
differing technologies, causing mismatched sample sta- be activated before an aircraft enters icing and act in
tistics. They also disagreed for MVDs larger than 30 response to remotely sensed icing conditions ahead of
µm because of the configuration of the PDPA, which the aircraft.
could be reconfigured to detect larger drop sizes. The Current terminology used to categorize or classify
PDPA has the advantage of being a smaller instrument icing intensity, or severity, is often inconsistent among
than the PMS probes, and it has the capacity of sensing government agencies, difficult to interpret and, at times,
a wide range of drop sizes with one instrument, whereas contradictory (Erickson et al. 1996; Green 1995; Auld
the PMS probes require two instruments, the FSSP and 1989; Newton 1977, 1979). Currently, the National
an OAP. Weather Service defines icing intensity with descrip-
Lawson et al. (1996, 1998) and Lawson and Cor- tions that are related only to aircraft and contain no
mack (1995) describe new optical probes that solve meteorological criteria (Auld 1989, Newton 1979).
problems with the FSSP and OAPs. The new instru- According to Newton (1979), NWS definitions of trace,
ments, which are in near-production stage for both light, moderate, and severe are only reporting defini-
ground and airborne use, are the cloud particle imager tions and contain no meteorological information that
(CPI) and the cloud droplet spectrometer (CDS). The can be used to forecast icing.
CPI creates images of cloud particles at a rate of 30 s–1, Newton (1979) also indicates that the National
but at a potential rate of 240 s–1. Image detail as small Weather Service definitions are not related to FAA icing
as 2 µm is possible, with maximum possible size limited regulations (FAA 1991). Uniformly understood icing
by sample volume. Shadowgraph-type images of ice
crystals can be made of nearly photographic quality. Im- * Personal communication, R. Paul Lawson, Stratton Park Engineer-
ages can be obtained in real-time, and techniques are in ing Corp., 1999.

28
To Contents
 definitions are needed that are defined in standard terms More work is needed to perfect existing information-
(Auld 1989, Newton 1979). A graduated, parametric retrieval methods and to make them more robust. Dop-
method of describing icing severity should be developed pler and polarimetry techniques may be able to provide
that is meaningful to meteorologists, aircraft designers, information about drop-size distributions. Most cloud
regulators, and operators (Erickson et al. 1996). microphysical work occurs in the X, Ka, and W bands,
A remote-sensing system must, at minimum, supply most ground-based precipitation measurement occurs in
information to the pilot with regard to the icing poten- the S and C bands, and in X band from aircraft. A detailed
tial of a given volume of atmosphere in the flight path. sensitivity analysis is needed to find the optimal mix of
A cockpit display must relay to the pilot information that requirements for detecting liquid-water content and
enables a risk-management decision to be made. This drop-size information from ground-based and airborne
means the display must either provide information that systems.
is uniformly applicable to all aircraft, but to which the Radar has shown modest success at measuring atmo-
pilot can reference his aircraft, or the system must directly spheric precipitation and cloud water content—both liquid
indicate the hazard potential to that particular aircraft. and frozen. Rainfall rates may be estimated with S- and
According to Newton (1977, 1979), the most tenable C-band radars. NWS NEXRAD radars are an example
solution with current definitions is to use a form of the of a system with some rainfall rate capabilities that may
old NACA icing intensity curves. The FAA is address- be useful for detecting freezing rain and perhaps freezing
ing terminology problems as part of the FAA Inflight drizzle conditions around airports. Although not practi-
Aircraft Icing Plan (FAA 1997). cal on aircraft because of size, using S band to detect
freezing precipitation, X- and Ka-band differential atten-
uation techniques to detect cloud water, or multiband
5.0 TECHNOLOGY DEVELOPMENT
X-, Ka- and W-band neural network techniques may be
5.1 Summary optimal combinations at airports. In addition, Doppler and
A remote-sensing system, operated either from the polarization techniques may provide some indication of
ground or from an aircraft, must sense the three- the drop-size spectra and the presence of ice crystals.
dimensional fluctuation of atmospheric liquid-water con- A possible problem of radar, especially of the differ-
tent, drop size, and temperature both from outside of icing ential attenuation methods using two or more wave-
conditions and from within icing conditions. In addi- lengths, is the need to know droplet temperature to
tion, the sensing system must have a range, resolution, extract an accurate liquid-water content measurement
and refresh rate sufficient to satisfy pilot needs. The capa- because of the dependency of backscatter on drop tem-
bilities of different technologies are often vastly differ- perature (Rinehart 1997). In addition, in military appli-
ent, depending upon whether scanning is toward space, cations radar could place an aircraft that is sensing icing
toward the earth’s surface, or in the horizontal plane conditions at tactical disadvantage. Radars with stealth-
ahead of an aircraft. In addition, it is likely that no sin- like capabilities should be investigated.
gle sensing technology is capable of providing all sens- In general, for ground-based systems, the focus
ing needs—a fusion of technologies will be necessary. should be on S- or C-band systems for determination of
Radar is the most viable technology for providing precipitation rate and precipitation liquid-water content
range-resolved cloud liquid-water content through large estimations and on X, Ka, and W bands for determin-
expanses of cloud. Its scan rate is rapid, orientation is ation of cloud water content. Doppler and polarization
generally not a limitation (unless observing below the techniques are available and practical in ground-based
horizon where earth reflections may be a problem), and systems, but for airborne systems they may not be use-
there are a variety of bands from which to choose. Radar ful. For general aviation aircraft, (relatively) inexpensive
is a reasonably mature technology, and its basic back- Ka- or W-band radar could be used to simply indicate
scatter capabilities may be augmented by Doppler and the location of cloud bases and tops if pointed to zenith
polarization techniques. and nadir—a useful tool to escape icing.
An important issue that must be addressed in the use Microwave radiometry has the potential to remotely
of radar for remotely detecting icing conditions is to sense icing conditions and, like radar, has been proposed
develop accurate and robust techniques for retrieving and used for several icing studies. Microwave radiome-
liquid water, drop size, and temperature information from ters are capable of integrating or profiling liquid-water
radar backscatter. Developments in this area include the content and water vapor and profiling temperature. These
differential attenuation method of liquid-water retrieval capabilities have been demonstrated for zenith and in-
demonstrated by NOAA-ETL (Martner et al. 1991; termediate angles.
1993a,b) and the neural net retrieval of liquid water and There are several ways radiometers could be used to
drop size by Quadrant Engineering (Mead et al. 1998). detect icing conditions. Ground-based airport systems

29
To Contents
could use scanning radiometers to measure liquid water and some aspects of the drop-size spectra and cloud phase
and temperature profiles with height. Integrated water composition using polarization. However, lidar’s inher-
could be distributed through the atmosphere to deter- ent liability, its inability to penetrate cloud of large opti-
mine volumetric cloud water content if combined with cal depth more than a few hundred meters, is severe.
cloud base, top, and layer information from a Ka radar. Lidar may be a useful, inexpensive technique for allow-
However, this can now be accomplished without radar ing aircraft, especially those flying night VFR (visual flight
by the Radiometrics profiling radiometer. The ability of rules), to avoid icing, but it cannot provide guidance to
radiometry to scan liquid water at airports is nearly a escape icing. Its potentially low cost, high scan-rate capa-
mature technology, with the greatest concerns being the bility, and small size may make it a practical tool for
need to keep sensors free of moisture and improving small helicopters and light aircraft, especially aircraft
scanning rates. Long scan time is one of the most seri- operating in the Far North with no ice protection in often
ous radiometer problems because they are passive limiting weather and few winter hours of daylight. Devel-
devices. At airports, this can be overcome by using multi- opment along these directions is being made in Canada.
ple radiometers, each assigned to a different sector of Canadian needs for this VFR cloud and icing avoidance
sky, but it is a greater problem for airborne systems. capability may be great in the northern territories.
Airborne radiometry of cloud liquid water does not Temperature measurement is needed to determine if
appear to have been tried from aircraft. It is theoretically cloud water or liquid precipitation is supercooled. It is
possible for airborne radiometers to measure integrated also needed for radar retrieval of cloud water content.
liquid water scanning in the horizontal, and that feasi- Two mature technologies are available for sounding
bility is being analyzed using measurements made from temperature from the ground: RASS (radio acoustic
the summit of Mt. Washington at the Mt. Washington sounding systems) and microwave radiometers. RASS
Icing Sensors Project (MWISP). However, it may be provides greater temperature resolution and thus accu-
more feasible at 85 GHz than at the more commonly racy, which is especially needed during the winter when
used 37 GHz because of greater sensitivity to cloud water inversions are common. However, RASS’s range is gen-
at 85 GHz. This needs to be explored. erally limited to altitudes of less than 3 km agl. RASS
As an alternative to horizontal scanning, or in addi- and microwave sounders could be used together at air-
tion to it, radiometers might be able to sense vertically, ports to provide the resolution needed at lower altitudes
at zenith and at nadir, from an aircraft, as is now done and temperatures above the terminal airspace.
from the ground and from satellites, respectively. The At the present time, there is no explicit capability to
amount of liquid water above and below the aircraft range-resolve temperature ahead of an aircraft. Radio-
could then be determined, and radar might be used to metry is the most promising possibility. The NASA Jet
simply determine cloud tops and bases. The slow inte- Propulsion Lab scans air temperature at high altitudes
gration time of radiometers, however—many seconds— ahead of an aircraft, with temperature provided at a rel-
would provide only an averaged or integrated liquid- atively fixed distance. Though not range-resolved, air-
water content. This may be useful information, espe- craft motion effectively range-resolves the temperature
cially in level flight within relatively uniform clouds, and thus may provide an interim solution. The efficacy
but during climb-out and descent, and within fluctuat- of this system needs additional exploration for icing
ing clouds such as cumuliform, this technique may not applications; its performance within clouds is unknown.
be as useful because of the long integration times and RASS has been found not to be practical for sensing tem-
limited spatial resolution of microwave radiometers. In perature from aircraft.
addition, nadir-viewing aircraft radiometers would have There is a possibility of range-resolving air tempera-
the complexity of radiation from the earth’s surface, and ture above and below aircraft, at zenith and nadir, with
varying altitude above the surface, making cloud-water microwave radiometers using the same techniques that
retrieval more difficult. This suggests that although are used by ground-based and satellite-based microwave
ground- and satellite-based radiometers may be able to radiometer temperature sounders. The scan times
measure integrated liquid water successfully now, air- required may cause smearing of temperature due to air-
borne systems are presented with greater difficulties that craft motion, approach and descent altitude changes may
will require more retrieval technique modeling and field- prevent accurate temperature retrieval because of nec-
work. In addition, there is no indication that radiome- essary scan times, and ground radiation may cause prob-
ters can detect characteristics of the drop-size spectra lems when sensing to nadir. A zenith/nadir sensing sys-
or detect ice, though it has been theorized that drizzle- tem does not indicate temperature ahead except through
size drops may be identified by polarization effects. extrapolation, although some scanning capability may
Lidar, especially multiple-field-of-view techniques, be possible. However, it would indicate where warm tem-
has promise for measuring cloud liquid-water content peratures may exist for escape from icing.

30
To Contents
 5.2 Introduction However, capabilities of some technologies are either
A remote-sensing system designed to detect icing enhanced or severely limited by the direction they are
potential ahead of aircraft would consist of three com- sensing, either horizontally or vertically.
ponents: Remote sensing of aircraft icing conditions is devel-
oping along two parallel paths, depending upon the tech-
• A suite of detectors to measure conditions that cause
nology. One requirement is to create an ability to
icing
remotely detect icing conditions ahead of aircraft from
• A processing system to integrate information, as it is
sensors mounted on the aircraft. However, since more
received from the sensors, to assess the icing hazard
aircraft are exposed to icing in the departure and
• An information display system to provide pilots with
approach phases of flight, another requirement is to
timely, useful maps of the icing hazard.
develop ground-based systems that are capable of scan-
The sensing system, which acquires the information ning the airspace around airports. Airport-based sens-
to provide a measure of icing potential with distance ing systems are likely to be developed before airborne
ahead of the aircraft, must accomplish several tasks. systems because greater technological development has
First, it must detect, either directly or through the use occurred with ground-based systems, and because they
of surrogates, cloud and precipitation liquid-water con- present fewer weight, power, and size restrictions.
tent, temperature of the droplets (or the existence of It is currently unlikely that one technology will be
supercooling), and some measure of the breadth of the able to sense liquid water, temperature, and elements
drop-size spectrum. It must map the magnitude of these of the drop-size spectrum, so a remote-sensing system,
conditions ahead of the aircraft for avoidance purposes, whether ground-based or airborne, will probably con-
and it must locate where icing conditions do not exist sist of multiple technologies to obtain all the necessary
ahead of the aircraft for escape purposes if the aircraft information. Technologies under consideration include
is immersed in icing. In addition, the location of each radar, lidar, passive microwave radiometers, and radio
condition must be measured in three-dimensional space acoustic sounding systems (RASS).
often enough to provide a continuously updated image
of conditions to the pilot. The system must also scan a 5.3 Radar
sufficiently large volume of atmosphere, to a great Radar is the most mature technology under consid-
enough distance and with sufficient detail, to provide eration. Initially developed in the 1930s by the British,
pilots with avoidance and escape options. Since remote it was first used for detecting weather phenomena imme-
sensing is a stand-off technique, sensing systems typi- diately after World War II (Battan 1973, Toomay 1982).
cally detect and measure phenomena without being Weather radar operates in the shorter wavelengths of
immersed them. To effectively assist escape from icing, the microwave spectrum and into the centimeter wave-
however, they must also be able to sense that icing con- length spectrum, with longer wavelength (lower fre-
ditions do not exist in a volume of air ahead of the air- quency) radars used to detect larger drops such as rain
craft while sensing from within icing conditions. and shorter wavelengths used to detect cloud droplets.
A variety of technologies are available for remotely Radar wavelengths for atmospheric sensing can be spec-
sensing atmospheric properties (Westwater and Krop- ified by band, frequency, or wavelength (Table 1).
fli 1989). Current technologies are designed to operate As wavelength shortens, the ability to detect smaller
from ground positions to zenith or from satellites to drop sizes improves, but range decreases and attenua-
nadir. Some sensors can scan the atmosphere at inter- tion by precipitation and other atmospheric constitu-
mediate angles, and some can scan in the horizontal. ents such as water vapor increases as well (Houze 1993).

Table 1. Relationships between radar band, frequency,

wavelength, and weather parameter sensed.
Nominal
frequency Nominal Meteorological
Band (GHz) wavelength condition sensed

S 3 GHz 10 cm Precipitation (NEXRAD)

C 6 GHz 5 cm Precipitation (ships)
X 10 GHz 3 cm Precipitation–clouds (aircraft)
Ku 15 GHz 2 cm Precipitation–clouds
K 30 GHz 1 cm Precipitation–clouds
Ka 35 GHz 8.7 mm Cloud droplets
W 94 GHz 3.2 mm Cloud droplets

31
To Contents
The National Weather Service’s WSR-88D Doppler between drop size and backscatter complex and diffi-
radar system is S band; this is the most common wave- cult to predict (Battan 1973, Toomay 1982, Rinehart
length for land-based weather radars because of its abil- 1997). In addition, backscatter from ice particles vs.
ity to detect precipitation-size particles (Crum and liquid drops reverses due to differences in the real part
Alberty 1993, Houze 1993). C-band radar is a com- of the complex index of refraction between water and
mon shipboard radar, and X band is the most common ice, with ice particles providing about 10 times the back-
airborne weather radar. Bands with shorter wavelengths scatter energy of liquid drops of a given size (Battan
than X band suffer from precipitation attenuation, but 1962).
the shorter wavelengths may also allow detail to be Doppler radar measures the motion of drops or ice
retrieved about precipitation. Most cloud microphysi- crystals along the axis of the radar beam (Battan 1973).
cal radar work is currently in the X, Ka, and W bands, The fall speed of water drops is related to their size, and
but disadvantages of the millimeter wavelengths include the fall speed of ice crystals is related to their size and
increased absorption by water vapor and oxygen and shape (Rogers and Yau 1989). A vertically oriented radar
strong extinction by cloud droplets, drizzle, and rain can distinguish drop sizes, and distinguish drops from
(Klugmann and Judaschke 1996). ice crystals, by their fall speed. Shorter-wavelength
Radar detects droplets and ice crystals in the atmo- radars have a greater ability to distinguish between fall
sphere because of signal backscatter from droplets. speeds and thus are better able to resolve drop sizes.
Backscatter occurs as either Rayleigh or Mie scattering. Because of the large volume a radar beam senses, and
Rayleigh scattering occurs when droplet diameters are the large number of drops within a given volume of air,
significantly smaller than the radar wavelength (Battan a spread of the Doppler spectrum typically occurs, which
1962). After some absorption by the droplets, the makes the Doppler signal difficult to interpret. Causes
amount of energy backscattered to the radar is propor- of spread include spread in terminal velocities of the
tional to the sixth power of the drop diameter (Battan drops or ice crystals within a sensing volume, turbu-
1962). Backscatter is also inversely proportional to the lence, and wind shear across the radar beam (Battan
fourth power of the wavelength. As drop diameter 1973). Although Doppler radar is useful for determining
approaches wavelength in size, backscatter increases, drop-size spectra from ground-based radar, it is likely
so the strongest backscatter from small droplets occurs that Doppler techniques will be difficult from aircraft-
from the shortest radar wavelengths (Battan 1962). At mounted radars because most drops will be falling ortho-
wavelengths greater than 3 cm, droplets of 2-mm diam- gonally to the radar beam, making a Doppler shift less
eter and smaller are Rayleigh scatterers, and at wave- detectable. However, turbulence can cause a Doppler
lengths of 10 cm, nearly all drops are Rayleigh scatterers shift because small drops, which are more influenced
(Battan 1973). Small frozen drops and small ice crystals by turbulence than large drops, can be carried toward or
backscatter about 20% as strongly as liquid drops of away from the aircraft, allowing their identification.
the same size (Battan 1962). In all cases, the tempera- The polarization of transmitted and received radar
ture of droplets must be determined to evaluate fully energy can be used to determine the mean values and dis-
the amount of water contributing to the backscatter, tributions of particle size, shape, and spatial orientation
because attenuation from droplets also has a tempera- and to determine their phase (Houze 1993, Zrnic 1996).
ture dependency (Battan 1973, Rinehart 1997). Radar signals may be linearly or circularly polarized.
Attenuation by scattering also increases as wave- Linearly polarized radars transmit and receive energy
length decreases, because more energy is scattered by in horizontal and vertical planes, principally because
intervening precipitation and cloud droplets. These are falling drops typically shorten in the vertical axis and
conflicting factors to consider, because as wavelength lengthen in their horizontal axis (Houze 1993). Differ-
is decreased to detect cloud drops, attenuation increases, ential reflectivity and the linear depolarization ratio are
which limits range in high drop concentrations and high computed from horizontal and vertical polarization.
liquid-water contents (Battan 1962, Rinehart 1997). Differential reflectivity, the ratio of the horizontal trans-
As drop diameter approaches radar wavelength or mitted-to-received energy to the vertical transmitted-to-
becomes larger, Rayleigh scattering no longer applies. received energy, typically indicates the oblateness of
Instead, Mie scattering occurs, which produces a less falling drops—a measure of drop size. Drops smaller
predictable backscatter due to complex interactions than 300 µm are typically spherical, but as size increas-
between energy reflected within the droplet and energy es, oblateness and differential reflectivity both increase.
waves traveling along the droplet surface (Toomay Ice crystals typically show no differential reflectivity.
1982). Depending upon the exact ratio of the drop size The linear depolarization ratio, the ratio of horizontally
to the wavelength, these reflections and traveling waves transmitted to vertically received energy, indicates how
may be additive or subtractive, making the relationship much of the transmitted signal is depolarized. Wet ice

32
To Contents
 particles produce less depolarization than water drops was not a satisfactory proven method for estimating
or dry ice crystals do, so it is a useful method for locat- rain rate from radar. That situation seems to have
ing melting layers (Houze 1993, Rinehart 1997). changed little; NEXRAD needs to utilize rain gauges
Circular depolarizing radars transmit a signal that for correction (Crum and Alberty 1993, Houze 1993,
rotates one complete revolution orthogonal to the beam Rinehart 1997).
axis per radio frequency cycle (Toomay 1982, Houze
5.3.1.2 C-band radar. C-band radars have also been
1993). The circular depolarization ratio is the ratio of
used experimentally to obtain rainfall rates, but again
the parallel (transmitted) component to the orthogonal
they are too large for aircraft use, though they could be
(received) component and indicates the sphericity of
used at airports (Gorgucci and Sarchilli 1996a,b; Tian
particles. Polarization may be useful for determining
and Srivastava 1997). Doviac and Zrnic (1984) argue
some elements of the drop-size spectra, especially from
that this ability to measure rainfall rates, along with
airborne radar that is scanning drops and crystals ortho-
Doppler detection of wind shear, would be a useful tool
gonal to their falling direction and thus maximizing
for predicting freezing rain at the surface and aloft.
shape deformation, or long-axis orientation, to the hori-
zontal. 5.3.1.3 X-band radar. X-band radar is small enough
to be carried aboard aircraft and often is. It is useful for
5.3.1 Detection of liquid water detecting precipitation, although with less accuracy than
longer wavelengths (Rogers and Yau 1989). X band is
5.3.1.1 S-band radar. The ability to detect raindrop-
typically considered a precipitation radar, but it is capa-
sized particles may be needed in a remote ice-detection
ble of detecting cloud liquid water. For example, Paluch
system because freezing rain is a serious aircraft icing
et al. (1996), comparing the use of X- and Ka-band
threat, although it is typically not considered as dan-
radars for detecting cloud liquid-water content, found
gerous as freezing drizzle. Raindrop sizes begin at about
a close, consistent correlation between radar reflectiv-
500 µm diameter and extend to about 5 or 6 mm in
ity and cloud liquid-water content in summer cumulus
thunderstorms (Pruppacher and Klett 1997). It is pos-
in Florida. However, they indicated that Bragg scatter-
sible to experience icing in very large drops near the
ing—the susceptibility of longer wavelengths to detect
tops of towering cumulus in the tropics or in the mid-
“angels” caused by turbulence—was a potentially
latitudes in the summer months. However, freezing rain-
greater problem with X-band radar (White et al. 1996).
drops will usually be found at the smaller end of the
They also indicated that there is typically a strong rela-
size spectrum, typically no larger than 3 mm, because
tionship between drop size and reflectivity, which is a
turbulence is small in most freezing rain (Jeck 1996).
source of error in radar liquid-water measurements. In
According to Battan (1973), S-band 10-cm radar can
the Florida observations, however, there was a strong
successfully estimate rainfall rates, and thus liquid-
reflectivity–liquid-water content relationship because
water content, for long distances. The National Weather
most of the liquid water was concentrated within the
Service’s NEXRAD, for example, is a 10-cm radar with
large end of the drop-size spectrum. This suggests that
a range of about 460 km for reflectivity measurements.
broad drop-size spectra will produce a larger rainfall
NEXRAD does not provide explicit precipitation liquid-
rate estimate error than narrow spectra.
water content at the present time, but it does provide
vertically integrated liquid water from rainfall contained 5.3.1.4 Millimeter-band radar. Millimeter-wave
in a 4 km by 4 km grid (Crum and Alberty 1993). Thus, radars, principally the Ka and W bands, are the current
precipitation liquid water may be estimated near air- choices for detecting cloud microphysical properties and
ports from a ground-based remote detection system. precipitation (Mead et al. 1994, Kropfli and Kelly 1996).
Doviac and Zrnic (1984) indicate that accurate estimates Since the backscattering cross-section of a drop is
of rainfall rates and rainfall liquid water require a knowl- inversely proportional to the fourth power of the wave-
edge of the raindrop size distribution, which, if unknown, length, long-wavelength radars, despite their great range
can cause rainfall rates to differ by a factor of four. In and power, are at a disadvantage for detecting cloud
their review of the extensive work that has been done droplets. They can compensate with larger antennas and
estimating rainfall rates from radar, they indicate that more power, but at high cost (Martner and Kropfli
spatially detailed measurements of precipitation liquid 1993), and, though they may be useful at airports for
water require knowledge of drop sizes unless the drop scanning for icing conditions, they do not fit on air-
sizes are reasonably uniform. Curvature of the earth craft. Kropfli and Kelly (1996) indicate that millimeter-
also causes radar to observe different portions of storms wavelength radars are sensitive to small hydrometeors,
with distance and thus different drop-size distributions. have excellent spatial resolution, minimal ground clut-
Doviak (1983), in a review of rain rate estimation ter problems (which allows observation of weakly
methods using radar, indicated that at that time there reflecting cloud with high resolution), and are easily

33
To Contents
portable. However, because of attenuation in moderate by stating that they are severely attenuated by rainfall,
to heavy rainfall, they are best used in nonprecipitating but they provide fine-scale measurements of nonpreci-
clouds. Ice crystals can also cause non-Rayleigh scatter- pitating and drizzle clouds because the amount of energy
ing, which causes loss of signal. If combined with a reflected by cloud droplets and ice crystals sharply
longer-frequency radar, such as X band, differential increases as wavelength decreases.
attenuation can be used to extract cloud liquid-water There have been many applications of Ka-band radar
content (Martner et al. 1991). According to Mead and to measuring cloud microphysics, it appears to be the
his colleagues, as of 1994, only five universities and most useful of the millimeter wavelengths for aircraft
one government laboratory had operating millimeter- icing. Ka band, when compared with the shorter-wave-
wave radars. Applications include studying internal length W band, can penetrate optically thick clouds with
circulations of cumulus clouds, remotely measuring high liquid-water contents, it can detect clouds above
rainfall drop-size distributions, and studying drizzle in light precipitation and multiple cloud layers, and Mie
stratus clouds (Mead et al. 1994). scattering is relatively uncommon (Kropfli et al. 1995).
Bragg scattering caused by atmospheric refraction, White et al (1996) indicate that Ka-band radar can be
which is a problem for longer-wavelength radars, is used to obtain the reflectivity, Doppler velocity, and
negligible for millimeter-wavelength radars (Kropfli Doppler spectral width of drizzle drops. Directly related
and Kelly 1996). However, the short wavelengths often to reflectivity are drizzle liquid-water content, flux, and
restrict the applicability of Rayleigh scattering. Thus, number concentration. Doppler velocity and spectral
for Ka-band radar, Mie scattering begins for water drops width provide estimates of the magnitudes and shape
at about 2.7 mm diameter, and for W-band radar at about of the drop-size spectrum. Measurements made during
1 mm diameter, which increases the complexity of inter- the Atlantic Stratocumulus Transition Experiment
pretation. Mie scattering is an even more difficult prob- (ASTEX) by Ka-band radar provided liquid-water con-
lem for ice crystals and snowflakes, where Mie theory tents from 0.01to 0.14 g m–3, drop diameters from 20 to
has not been developed (Kropfli and Kelly 1996). High 320 µm, and drop concentrations from 0.3 to 700 L–1.
humidity is also a problem with the shortest wave- Kropfli et al. (1995) further demonstrated in ASTEX
lengths, such as W band, where its attenuation effects that Ka-band radar detected ice masses as low as 0.003 g
may reduce sensitivity to small drops. m–3 in cirrus clouds at 7-km range. They also demon-
Doppler radar in the millimeter wavelengths excels strated retrieval of integrated liquid water at zenith, as
in measuring drop fall speeds because of its high fre- may be applied at airports, and compared it with radio-
quency, which allows high precision. Doppler tech- meter-derived integrated liquid water (Martner and
niques, as indicated earlier, may be helpful for airport- Kropfli 1993). Radar and radiometer-integrated liquid-
based sensing systems, but they may not be useful for water estimates compared well except in drizzle.
aircraft-mounted systems unless they are scanned up The University of Massachusetts has developed a
and down at large angles (Kropfli and Kelly 1996). dual-wavelength, ground-based cloud-profiling radar
Polarization measurements may also be used, especial- system operating at Ka and W bands (Sekelsky and
ly with Ka-band radar, to determine particle shape and McIntosh 1996). A system with a 1-m-diameter anten-
orientation (Kropfli and Kelly 1996). For example, the na is used to make polarimetric and Doppler measure-
circular depolarization ratio has been used to distin- ments of clouds at both wavelengths. The authors state
guish plate-like crystals from aggregates in Colorado that dual-wavelength millimeter-wave particle sizing
clouds using a ground-based Ka-band radar, and future (drops and ice crystals) has the potential for more accu-
capabilities may allow distinguishing between other racy than single-wavelength Doppler methods. In addi-
crystal types. tion, they indicate that MVD and the shape of drop-
Ka-band radar can determine cloud base and top, and size spectra can be determined more accurately with
thus thickness, and cloud structure, which may be use- both wavelengths. They speculate that it also may be
ful for avoiding icing (Politovich et al. 1995) when possible to discriminate glaciated from liquid clouds
combined with a microwave radiometer capable of and to estimate particle sizes in fully glaciated clouds.
measuring vertically integrated liquid water. Vertical W-band radars, operating at about 95 GHz, are
cloud liquid-water profiles may be mapped by adjust- becoming increasingly popular for cloud microphysics
ing drop concentration to force the two signals to fit. work (Mead et al. 1994, Kropfli and Kelly 1996). They
Kropfli and Kelly (1996) state that, though W-band offer even better size, weight, and power advantages
radars are superior for airborne use because of their than Ka-band radars, but they also suffer more severely
smaller size, the Ka band is less attenuated by water from attenuation in large drops, humidity, and precipi-
vapor and cloud water, making it more useful. Kropfli tation. They are used to measure cloud structure and
and Kelly (1996) summarize millimeter-wave radars excel at observing drizzle, which is a subject of intense

34
To Contents
 study because it is believed to regulate the thermody- 5.3.1.5 Liquid-water content retrieval techniques.
namic structure and radiative coupling of the marine
5.3.1.5.1 Single-band retrieval techniques. Frisch et
boundary layer (Frisch et al. 1995). Lhermitte (1987)
al. (1995) develop theory and demonstrate, using the
developed theory for W-band radar and demonstrated
NOAA ETL radar (Martner and Kropfli 1993), the
some of its capabilities as a Doppler radar, but he did not
retrieval of drizzle and cloud droplet parameters with a
measure cloud liquid-water content.
Ka-band radar and a radiometer for measuring integrated
A University of Massachusetts 95-GHz dual-polarized
liquid water. Doppler techniques were used to measure
radar for ground-based and airborne use flies on the
the drizzle drop-size spectra, and drizzle and cloud
University of Wyoming King Air research aircraft (Mead
liquid-water content were measured from reflectivity.
et al. 1994). It has a demonstrated range of 0.1 to 2.9 km
Cloud liquid-water content was computed by using
and has observed 1- to 2-mm graupel with some rimed,
Doppler velocities of drizzle and cloud to parse the two
branched crystals, crystal aggregates up to 4 mm diame-
liquid-water contents. Overall, by using Ka-band radar,
ter, and needles up to 1 mm diameter. In a study with
Doppler features, and a simple drizzle model, the
this radar, a 30-km segment of shallow stratus produc-
authors were able to extract drop number, size distribu-
ing freezing drizzle was flown. The radar beam, point-
tion, liquid-water content, and mean liquid-water flux.
ing vertically, observed detailed cloud structure at 30-m
They indicate that there is a potential for ground-based
resolution, and radar backscatter was compared to cloud
remote sensors to do long-term monitoring of cloud and
parameters measured with in-situ instruments. The radar-
drizzle parameters, such as at airports. This radar sys-
enhanced interpretation of in-situ measurements was not
tem in a scanning mode, together with a RASS for
itself used to measure specific cloud physical parame-
measuring temperature profiles, may be an adequate
ters. A similar radar, built by the University of Massa-
airport-based system.
chusetts and the NASA Jet Propulsion Lab, the Airborne
Liao and Sassen (1994) have also developed a tech-
Cloud Radar, flies on a NASA DC-8. More than 50 hours
nique, although only as modeled theory, that allows
of testing has occurred in cirrus, stratus, and cumulus
extraction of cloud liquid-water content by linking
clouds, and melting layers have been observed (GEWEX
liquid-water content and reflectivity, assuming a drop
1996).
concentration of 100 cm–3. The model applies for esti-
In other W-band applications, Klugmann and Judasch-
mating liquid-water content in nonprecipitating cumulus
ke (1996) have developed a W-band Doppler radar in
and stratocumulus clouds. They also developed theory
Germany to measure vertical velocities within clouds by
for extracting ice water contents from clouds with Ka-
tracking drop speeds. Clothiaux et al. (1995) also
band reflectivity.
explored the use of W-band radar combined with other
remote sensors. They indicate that, when pointed at 5.3.1.5.2 Differential attenuation and dual-band tech-
zenith, W-band radar is valuable for mapping cloud base niques. Combining two radar frequencies and analyz-
and top, but base is often indicated as too low because of ing cloud liquid-water content using differential atten-
precipitation. Though methods were not developed to uation has become a preferred method of measuring
compute cloud liquid-water content from the radar, in-situ range-resolved cloud liquid water. According to Mart-
measurements of liquid-water content were compared ner et al. (1993a), using two radar wavelengths with
with the radar calibration, and calibrations were consis- significantly different liquid-water attenuation coeffi-
tent with in-situ measurements. cients to measure cloud water and ice content was first
Sassen and Liao (1996) developed theory for meas- theorized by Atlas (1954). Martner et al. (1991) and
uring the contents of ice and water clouds from W-band Gosset and Sauvageot (1992) independently developed
radars. They indicate that for most cloud drops, Rayleigh field tests and additional theory, concluding that the best
scattering applies in W band, and they provide algorithms radar wavelengths were in the X and Ka bands. Sand
for computing cloud liquid and ice content from reflec- and Kropfli (1991) patented the concept. According to
tivities. Fox and Illingworth (1997a,b) computed W-band Martner et al. (1993a), as the radar beams enter a cloud,
reflectivity and cloud liquid-water content from more the Ka band is attenuated more rapidly than the X band.
than 4000 km of flight in-situ drop spectra measurements Assuming Rayleigh attenuation, the range derivative
by aircraft. They computed the probability of detecting of the difference of the reflectivities is proportional to
various values of liquid-water content as a function of the liquid-water content. Both water and ice contribute
radar sensitivity. Computations were complicated by the to the reflectivity. However, according to Martner et al.
predominance of drizzle drops in marine stratocumulus. (1993b), only the liquid water generates the differen-
They concluded that a highly sensitive space-based ra- tial attenuation needed to compute liquid-water content.
dar could detect 100% of all marine and stratocumulus Thus, the attenuation due to ice is the same for both
clouds, but it was not clear if cloud liquid-water content bands—very small.
could be directly measured. According to Gosset and Sauvageot (1992), the dual-

35
To Contents
wavelength differential attenuation method allows dis- radiometer or aircraft support. The February and March
crimination between the solid and liquid phases in a cases gave highly variable results (Martner et al. 1993b).
nonprecipitating cloud and an estimation of water and Measurements were made within upslope clouds in one
ice contents. The attenuation of ice is negligible com- March case with aircraft-measured liquid-water contents
pared with the attenuation of water. Thus, after correc- of up to 0.4 g m–3, a cloud droplet MVD of about 15
tion for the temperature of the ice and water particles µm, and no ice crystals. However, the radar measured
and the coefficients of attenuation of water and ice, liquid no liquid-water content. In another case with drizzle, liq-
water can be computed from the difference between uid-water contents were occasionally negative, perhaps
the reflectivities. They claim that theory indicates that because drops fell outside of the Rayleigh regime. In the
the technique allows the quantity and location of super- April tests, liquid-water contents of 0.2 to 0.6 g m–3 were
cooled water to be estimated precisely and easily. An measured, but no in-situ verification was available.
analysis of paired X and Ka bands, and Ka and W bands, Martner et al. (1993a,b) have evaluated their meas-
concluded that the X- and Ka-band pair was best for urements extensively. They identify several meteorologi-
both air- and ground-based systems because attenua- cal conditions that cause problems and radar deficiencies
tion of the W band is too high. There are problems with that may be responsible for problems. If non-Rayleigh
Mie scattering, however, if water drops are large. Gosset scattering occurs due to large water drops or ice crys-
and Sauvageot (1992) indicated that, with a knowledge tals, there is a reversal of the X– Ka-band reflectivity
of drop temperature, the technique is useful in mixed difference trends, producing negative liquid-water con-
clouds and light precipitation. Since liquid-water con- tents. Measurements of liquid-water content may also
tent at any range gate is proportional to the local range be particularly poor in regions of large aggregate snow-
derivative of the reflectivity difference of the two radars, flakes and just below the melting layer where liquid-
precise absolute calibrations of the radar reflections are covered ice crystals occur. In addition, false positive and
not needed because only a change in their relative val- negative liquid-water contents can occur at locations
ues with range is important (Martner et al. 1991). within clouds where large changes in drop size occur—
Martner et al. (1991; 1993a,b) have done the most the boundaries between smaller drops and larger drops.
definitive field testing of dual-wavelength X-band and For example, if droplets change from small to large, X-
Ka-band radars for measuring supercooled liquid water. band reflectivity may increase rapidly and Ka band does
During the 1991 NCAR Winter Icing Storms Project not—producing a large false positive liquid-water con-
(WISP), radars were installed northeast of Boulder, tent. The reverse can occur as the beams move from large
Colorado, and liquid water was measured intermittently to small droplets, causing a negative liquid-water con-
from February to April. A steerable microwave radio- tent.
meter to measure integrated liquid water and occasional Clouds composed of small droplets, typically those
research flights through the radar beams were avail- of continental origin, are difficult to detect, especially at
able to verify radar measurements. The radars and radio- long range and if liquid-water content is low. Ice crys-
meter were scanned at an angle of 7.5° above the tals within the clouds can help make them visible, but
horizon, and the research aircraft flew up the radar then Rayleigh problems may occur (Martner et al.
beam. Liquid-water content was analyzed with a least- 1993a,b). Higher-power radars may solve the problem.
squares fit between the X- and Ka-band reflectivities Overall, Martner et al. (1993a,b) were encouraged by
for a 4-km window of gates (53 gates of 75 m each). the promise of dual-wavelength differential attenuation
Computed liquid-water content was assigned to the to obtain cloud liquid-water content. Most of the prob-
center range gate of the 53 gates and then was shifted lems can be solved, and they indicate that more field
out one gate, and the computation was repeated. This trials should be performed with improved hardware and
was repeated for each gate along the beam from 2.0 to in more favorable weather conditions (Martner et al.
22.6 km from the radar (Martner et al. 1993a). Cloud 1993a).
temperature, necessary for liquid-water content com- Fournier (1993) also proposed a dual X- and Ka-band
putations, was either estimated or available from the radar as a terminal aviation weather-sensing system to
research aircraft. estimate cloud parameters at distances from 20 to 30
Test results during WISP were mixed because of km. He indicates that the radar would be capable of esti-
radar design deficiencies due to the low-budget nature mating, through differential attenuation, cloud type, visi-
of the tests, weather conditions, and inability to obtain bility, wind fields, median drop diameters, ice vs. liquid-
complete in-situ measurements of the radar-measured water content, and light and moderate precipitation.
cloud conditions. Seven cases were analyzed from Development of this system did not continue beyond the
WISP91, five cases in February and March with radio- proposal stage, however, because of funding cuts at
meter and aircraft support and two in April without Transport Canada.

36
To Contents
 5.3.1.5.3 Multiple-band retrieval of liquid-water con- cussed earlier, Doppler techniques may be used to deter-
tent. Considerations for multiple-wavelength radars mine droplet size from the fall speeds, but the radar must
include the use of more than two wavelengths and rela- attain a high angle scan, near zenith or nadir, to obtain a
tionships beyond differential attenuation. For example, reliable velocity measurement. Since the range of drop
Jameson (1994) develops theory necessary to measure speeds in the drop sizes of interest is small, use of Dop-
rainfall rate, rain water content, and mass-weighted mean pler techniques may be difficult. Cloud drops all fall at
drop diameter from satellite and airborne radars using less than 27 cm s–1, and drizzle falls from about 27 cm
multiple-wavelength radars. He claims that simple differ- s–1 to 2.06 m s–1 (Rogers and Yau 1989). Larger drops
ential attenuation can be used to determine water content. fall faster.
However, three wavelengths—38, 25, and 13 GHz—are Polarization relies upon the oblateness of larger drops
necessary to measure rain rate, rain water content, and as they fall and distort in shape due to aerodynamic drag.
mass-weighted mean drop diameter. In his summary, Since drops smaller than 300 µm in diameter are typi-
Jameson states that two frequencies are needed to meas- cally spherical, drop shape cannot be used to determine
ure one parameter, and the measurement of two or three droplet size in the cloud drop range (Pruppacher and
parameters requires at least three, and ideally four, fre- Klett 1997). Depending upon the wavelengths, especially
quencies. Some of these concepts might be applied to for multiwavelength radars, it may also be possible to
measuring icing potential. In a somewhat different use Rayleigh vs. Mie scattering to sort drops by size.
approach, Srivastava and Tian (1996) theorize that two
5.3.2.1 Doppler radar techniques. Thomson and List
radars of the same or nearly the same wavelength, but
(1996) developed a new method to determine rainfall
physically located apart, could be used to improve com-
drop-size spectra with a vertically pointing X-band Dop-
putations of rainfall through simultaneous use of each
pler radar. Errors in raindrop fall speeds with Doppler
radar’s attenuation and reflectivity.
radar occur from vertical wind, turbulence, pressure
A neural-network-based retrieval technique has been
dependence of terminal fall speed, and deviations from
developed at Quadrant Engineering and at the Univer-
Rayleigh scattering. Disdrometer measurements of drop
sity of Massachusetts based upon backscatter informa-
size during radar measurements in the Canadian Atlantic
tion from three radar bands: X, Ka, and W (Mead at al.
Storms Project indicated that the power spectrum of the
1998, Koenig et al. 1999). A neural network was trained
raindrop velocities was related to the drop-size spectra.
to estimate cloud temperature, liquid-water content, and
Vertical wind effects were removed, and the drop spec-
drop mean volume diameters (MeanVD) and mean radar
trum was calculated from reflectivity. Good agreement
reflectivity diameters (MZD) from the backscatter power
between measurements and radar calculations of drop
measurements from one, two, and three bands. Range
spectra were found in two test cases.
resolution was 2 km, and cloud parameters were syn-
Gossard (1994) proposed a method for extracting
thetically created for a wide range of conditions found
cloud droplet-size spectra information using Doppler
in precipitating and nonprecipitating stratiform and
radar in the Ka band. Doppler radar typically cannot
cumiliform clouds. The neural net was trained with
detect cloud droplet spectra from fall speed because the
10,000 cases and tested with 200 cases. Temperature
settling velocity of cloud droplets allows them to be car-
retrievals were not theoretically possible because of radar
ried by updrafts and downdrafts. Gossard developed a
noise. However, with three radar bands, liquid water was
technique, tested on a long-wavelength wind profiler
retrieved with less than 0.17 g m–3 error, and MeanVD
Doppler radar, that allows extraction of the cloud drop-
and MZD with less than 16% error. The neural net is
size spectrum by measuring the spectrum shape parame-
currently being evaluated with actual cloud information
ter independently of updrafts and downdrafts. He pro-
to determine prediction accuracy.
poses that the method is effective in detecting drop
growth to the drizzle-size range in stratus clouds. The
5.3.2 Detection of mixed-phase conditions
method cannot be used in precipitation because the rain-
and drop-size spectra
drops overwhelm the cloud signal, and there is too much
In addition to cloud liquid-water content, detection
error in the technique to determine cloud liquid-water
of icing conditions requires information about the loca-
content, even with no precipitation. The technique may
tion of liquid water vs. ice particles, and about the size
be applicable at airports with ground-based radars, and
distribution of water drops. The latter is particularly
it may be useful for airborne radars if high-elevation
important for detecting whether drops are within the
scanning is performed on clouds above and below the
drizzle or raindrop size ranges.
aircraft. Gossard et al. (1997) expand on the technique
The technique most useful for detecting some of these
and indicate that errors in liquid-water content are pos-
parameters, but especially drop size, depends upon
sible within a factor of two and that relative liquid water
whether the radar is ground-based or airborne. As dis-
and rainfall flux through clouds should be accurate.

37
To Contents
 Lhermitte (1987) uses W-band Doppler radar to raindrops. They conclude that the combination of fall
extract information about the rainfall drop-size spectra speed and the circular depolarization ratio will yield more
in clouds. Using Doppler velocity measurements of drop information than either can individually.
fall speeds, and Rayleigh and Mie scattering, drop sizes
were detected in field tests in Florida and Colorado. He 5.3.2.3 Polarization radar techniques. Reinking et
proposes that combining several radar wavelengths, such al. (1996) proposed and tested a method of differentiat-
as W, Ka, and X bands, would provide the capability of ing drizzle from rain and ice crystals using elliptical
detecting the full range of raindrop sizes using Doppler depolarization ratios (EDRs) and linear depolarization
techniques. Lhermitte (1988) also presents a method of ratios (LDRs) with the NOAA ETL scanning Ka-band
backing out air vertical velocities from raindrop fall radar. Drizzle drops are spherical and do not polarize
speeds to improve measurement of drop-size spectra. the signals, raindrops are nonspherical and will depolar-
The technique relies upon Mie backscattering oscilla- ize the radar signal, and depolarization by ice crystals
tions around Mie scattering maxima and minima caused depends upon their shape and orientation. Scattering cal-
by raindrops of various sizes (Lhermitte 1987). The tech- culations, and measurements during WISP in 1993, indi-
nique would be useful in providing vertical profiles of cate that EDR provides a good capability to distinguish
drop-size spectra within clouds and could be applied to between ice crystals of various habits, drizzle, and rain.
S-band radars such as NEXRAD. If the requirement is simply to distinguish drizzle from
ice, then LDR is better and could be applied to NWS
5.3.2.2 Doppler and polarization radar techniques. NEXRAD radars for use at airports. Matrosov et al.
Wilson et al. (1997) developed a method of determining (1996) and Reinking et al. (1997) have further differen-
parameters of the drop-size distribution of rainfall to tiated ice crystal types using EDR and LDR with drizzle
estimate rainfall rates. This is accomplished, using S- drops as a reference. They are able to discriminate hydro-
band radar, by measuring differential Doppler velocity meteor types within cloud systems, which will help deter-
(DDV), the difference between Doppler velocities at mine the presence of cloud ice and drizzle.
vertical and horizontal polarization. DDV is indepen- Pazmany et al. (1994) described the development of
dent of turbulence and most shear. It is used with reflec- a new W-band dual-polarized Doppler radar at the Uni-
tivity and differential reflectivity to produce a three-para- versity of Massachusetts. The radar flies on the Univer-
meter gamma fit to the drop-size distribution. The radar sity of Wyoming King Air, as described earlier, and oper-
beams must have elevation angles of 10° to 40° for accu- ates pointing either horizontally, along the flight path,
rate measurements. The technique is not affected by drop or vertically. Early tests indicated the ability to detect
oscillations and can detect ice crystals and determine melting bands and hydrometeor type. Flights with the
the location of melting layers. Because of the radar wave- radar in 1992 and in WISP94 provided airborne in-situ
length, the technique would be usable only with ground- and radar measurements of snowstorms, needle ice crys-
based radars, such as the NEXRAD. tals, and melting layers. Most observations were made
Takahashi et al. (1996) utilized a Doppler and dual- at vertical incidence to clouds above the aircraft. The
polarized X-band radar system to detect the drop-size paired radar and in-situ measurements will be used to
distribution of rainfall in isolated cumulus clouds in develop relationships between the two.
Japan. Drop-size distribution was computed from ter-
minal fall velocities of the drops and the differential 5.3.2.4 Neural net and other radar techniques.
reflectivity and horizontal polarization. Drop sizes larger Mead and Pazmany (Mead at al. 1998, Koenig et al.
than 140 µm were detectable. The linear depolarization 1999) proposed using a three-band radar consisting of
ratio and the correlation between the vertical and hori- X, Ka, and W to detect liquid-water content and elements
zontal polarized waves were used to determine mixed- of the drop-size spectrum, using a neural net for post-
phase layers, such as the bright band. Two parameters processing the radar returns. This technique, described
of an assumed exponential drop-size distribution were earlier, shows excellent potential for estimating elements
returned by the radar. Measurements made of rainfall in of the drop-size spectrum, but it cannot detect mixed-
isolated cumulus clouds were plausible, but no in-situ phase conditions.
verification measurements were made. Using C-band radar, Huggel et al. (1996) improved
In one of the first modern uses of a Ka-band radar, estimates of rainfall rates by more accurately estimating
Kropfli et al. (1982) used a Doppler radar scanning at a drop-size distribution. They argue that most liquid pre-
high elevation angle, and circular depolarization, to dis- cipitation is formed in a bright band where falling ice
tinguish falling precipitation forms. Fall speed distin- crystals melt and coalesce into raindrops. By develop-
guished graupels, aggregates, and dendrites into five cat- ing a relationship between reflectivity within the “bright
egories within a squall line, and the circular depolari- band” and measured drop sizes with disdrometers, they
zation ratio was able to distinguish frozen particles from were able to predict drop-size distributions below the

38
To Contents
 melting layer in moderate intensity rainfall, from 1 to spheric paths over a frequency range of 1 to 300 GHz
10 mm hr–1, in Switzerland. is performed with RADTRAN or similar radiative trans-
fer models* (Falcone et al. 1982). RADTRAN is a
5.4 Passive microwave radiometers design tool to assess potential environmental impacts
Microwave radiometers operate by receiving thermal on microwave sensors. In effect, it allows reasonably
energy emitted and scattered by the earth’s atmospheric complete modeling of the atmospheric radiation envi-
constituents (Grody 1997). They are passive instruments, ronment due to gases and clouds, including polariza-
receiving natural radiation and actively emitting none tion, prior to actually building a radiometer.
of their own. Radiative energy is emitted, scattered, and
5.4.1 Detecting liquid-water content
absorbed by the atmosphere, and radiometers act as ther-
Cloud liquid water and rainfall rates have been
mometers to measure a narrow spectral portion of this
observed with radiometers over oceans and over land
energy. Passive microwave radiometers are used to
from the ground, from the air, and from satellites.
measure many atmospheric and surface characteristics,
Observing from the ground up to clouds, or from aloft
but their ability to measure atmospheric temperature,
down to clouds over water, is least difficult because
cloud liquid-water content, and attributes of cloud and
clouds contrast well with the thermally cold background
precipitation constituents such as phase is needed for
of space or water surfaces. Sensing cloud water from
estimating icing hazards. Microwave measurement is
aloft toward the ground is more difficult, as indicated
based on the brightness temperature of atmospheric con-
above, because of the radiative diversity of land sur-
stituents. The radiation intensity observed by a radio-
faces. Airport-based radiometers scanning for icing
meter is a function of the temperature, reflectivity, trans-
conditions will scan upward toward the cold of space.
missivity, and emissivity of the emitter and attenuation
The situation may be more difficult for aircraft-mounted
by constituents between the emitter and the radiometer
radiometers because they will scan ahead, above, and
at the specific wavelength of interest.
below the aircraft flight path and, depending upon the
Each atmospheric constituent—gas, liquid, and solid-
aircraft’s flight altitude and attitude, they may be look-
—has a unique absorption spectrum. The atmosphere
ing at the horizon, toward space, or toward the earth’s
in general absorbs in several narrow wavelength bands
surface, which could be either land- or water-covered.
and allows radiation to be transmitted through several
Thus, sensing cloud water along flight paths may be
broad windows due to the atmosphere’s gaseous com-
considerably more complex and difficult than most
position. The primary absorbers are oxygen and water
radiometer sensing to date. In general, cloud liquid water
vapor. Oxygen absorbs and re-emits in the 50–60-GHz
can be sensed with microwave radiometers, but ice is
region and at 118 GHz and is used for temperature pro-
difficult to detect. Thus, cloud-water sensing is not com-
filing. Water vapor has peak absorption and re-emis-
plicated by the presence of ice, but the presence of ice
sion at 22, 37, and 183 GHz (Grody 1997). Liquid-wa-
cannot be used to determine whether cloud water is
ter peak absorption and emission occurs near 37 GHz
supercooled.
and 89 GHz.
Passive microwave detection of cloud water and pre- 5.4.1.1 Sensing upward from ground. Westwater
cipitation is practical from the ground because the back- (1978) reviewed the theory and assessed the accuracy
ground of space has a low brightness temperature, pro- of determining cloud liquid water from upward-look-
viding high radiative contrast with clouds. Satellites can ing radiometers. His analysis yielded a two-wavelength
detect cloud water and rainfall rates over water bodies system, operating at 30 GHz (1.0-cm- wavelength) and
because water also provides a low brightness tempera- 21 GHz (1.4-cm wavelength) to detect liquid water and
ture, but land masses are warmer and more radiatively water vapor, respectively. Two wavelengths are needed
complex, making interpretation difficult. Thus, obser- because, if a single-frequency radiometer is used, varia-
vation direction is more critical for passive radiometry tions in water vapor cause apparent changes in cloud
than for radar, because the latter creates the energy that liquid water (Hill 1991b), though Hill disputes the need
it observes and so contrast is typically more adequate. for water vapor measurements during the winter (Hill
Radiometers are also capable of measuring polar- 1991a). Westwater indicates that knowing cloud temper-
ized energy. Polarization describes brightness tempera- ature would improve the accuracy of water vapor and
ture (radiance) along either a horizontal plane between cloud water estimates.
the emitter and the radiometer—vertical polarization, Hogg et al. (1983a,b) describe a dual-frequency
or along a plane orthogonal to that path—horizontal
polarization. Cloud and precipitation drops are usually * Statement made by T. Lines, Raytheon Corp., Denver, Colorado, at
vertically polarized. Inflight Remote Sensing Icing Avoidance Workshop, Meteorologi-
Analysis of microwave attenuation for typical atmo- cal Panel, 2 April 1997.

39
To Contents
radiometer at NOAA ETL that measures both integrated icing pilot reports. Westwater and Kropfli (1989) cite
water vapor (20.6 GHz) and cloud liquid water (31.6 Fotino’s work as one among several demonstrating the
GHz). Accuracy was determined by comparing with utility of scanning microwave radiometers at airports
liquid-water estimates derived by microwave transmis- for detecting aircraft icing conditions. They stress the
sions from a COMSTAR satellite through clouds. A importance of the 21-GHz and 31-GHz frequencies for
scatter plot comparing the two methods showed a nearly measuring integrated liquid water. They demonstrate
1:1 relationship, with scatter in the 1:1 relationship how the radiometers can be used to continuously map
increasing as liquid water increased. They also demon- integrated liquid water, and they indicate that 90.0 GHz
strate azimuth scans of the radiometer at an elevation is also a frequency that can be used to detect liquid
angle of 12.5°, showing liquid-bearing cloud fluctua- water. It is six times more sensitive than 31 GHz, and
tions depending upon the antenna direction. In heavy can measure integrated liquid per unit area ranging in
rains, such as 80 mm hr–1, the radiometer saturates due magnitude from 0.03 to 5.0 mm, which translates to
to excessive radiation. Signals were not affected by water 0.06 to 10.0 g m–3 of cloud water. This higher sensitiv-
or snow on the antenna. The authors suggest that an ity is confirmed by Grody (1997), who indicates, how-
airborne system could scan from the zenith to any angle ever, that scattering from precipitation droplets is a seri-
forward along the line of flight, suggesting that hori- ous problem at 85–90 GHz.
zontal sensing may be possible. Hill (1991a, 1992) made some of the first measure-
Gary (1983) describes a microwave system designed ments comparing radiometer-measured supercooled
to monitor aircraft icing conditions that was demon- liquid water with in-situ aircraft measurements. The
strated at Buffalo International Airport. Temperature pro- Utah State University radiometer, a copy of the NOAA
files, water vapor, and cloud liquid water were meas- ETL scanning radiometer discussed above, operates at
ured with radiometers. Water vapor and liquid-water 20.6 and 31.65 GHz. Field tests were done at Sodus
measurements were made from radiometers operated Point, N.Y. A research aircraft carrying a Rosemount
at 22.23 and 31.4 GHz, respectively. Water vapor and ice detector to measure supercooled liquid water was
temperature were verified with radiosondes, but intended flown in ascending and descending spirals centered over
overflights to verify liquid water did not occur. How- the radiometer. Measured liquid-water contents were
ever, forecasts of aircraft icing and pilot reports of icing low, near the limits of the radiometer resolutions, so
did compare well, suggesting that the liquid-water meas- completely valid comparisons could not be made. Seven
urements were reasonable. This study is the first example validation tests were made, with two producing spuri-
of a system explicitly designed and tested, using remote- ously high readings by the radiometer, potentially attrib-
sensing devices, for detecting aircraft icing conditions. utable to cloud-entrained snowfall that did not adhere
A workshop about remote detection of aircraft icing, to the ice detector and a melt layer along an inversion.
sponsored by the University of North Dakota (Smith Hill (1991b) also compared the ability of a dual-
1985), concluded that passive microwave radiometers frequency radiometer, the Utah State instrument cited
would be useful for determining cloud liquid water. above, and a single-frequency radiometer operating at
However, since radiometers only provide integrated 31.65 GHz to measure cloud water. A dual-frequency
liquid water, cloud top and base would also have to be unit corrects cloud water content by accounting for
measured to provide an estimate of cloud liquid-water changes in water vapor. However, according to Hill,
content in mass/volume units. They recommend that these changes are very small during the winter and intro-
radiometers be used in scanning mode but indicate that duce little error if ignored. A comparison between the
scanning is slow and would take about 5 min per 360° two radiometers during winter tests indicated a nearly
scan at one elevation angle. A solution to this may be to 1:1 relationship, suggesting that the winter water vapor
use multiple radiometers, each scanning assigned sec- correction is not needed. This would simplify cloud-
tors and elevation angles. water measurements from aircraft, reducing hardware
Fotino et al. (1986) related radiometer-derived zenith and computational requirements.
measurements of temperature and liquid water near Stankov et al. (1992) describe the use of four micro-
Denver to pilot reports of icing. Frequencies of 20.6 wave radiometers to measure supercooled liquid water
GHz and 31.65 GHz were use to measure water vapor in the Denver and Boulder area during WISP91. Though
and liquid water, respectively. Measurements were inte- the radiometer estimates of liquid water were not explic-
grated over 2-min periods. Comparisons with pilot itly compared with aircraft measurements, they con-
reports were difficult because they are often inaccurate clude that the radiometer’s ability to measure liquid
in time and location, and pilots avoid icing upon hear- water was excellent. The only reported problem with
ing reports of the conditions. They found overall strong the radiometers involved keeping snow from adhering
correlations between the radiometer measurements and to the sensor windows.

40
To Contents
 Martner et al. (1993a), as part of the Lake Ontario the production of ice crystals and water in rising air.
Winter Storms project, installed Doppler radar, wind The process operates only in clouds with updrafts. Air
profilers, and microwave radiometers to measure inte- temperature and Doppler radar measurements of
grated liquid water on the eastern shore of Lake Ontario. updrafts are used to describe the rate of production of
A freezing rainstorm on 15 February 1990 was contin- water vapor in excess of saturation with respect to ice.
uously monitored by the microwave radiometers at 20.6, The radar reflectivity is inverted to determine the profile
31.65, and 90.0 GHz. Vertically pointing scans and of ice-particle size distribution. From the ice-particle
scans at a 7.5° elevation angle were made. The radio- size distribution, the rate that water vapor can be con-
meter could not observe snow and ice, but it immedi- sumed by pure deposition is determined. The differ-
ately signaled the onset of melting aloft at the begin- ence between the rate of water vapor production and
ning of liquid precipitation. Some overestimates of deposition on ice crystals is the production of super-
liquid water may have occurred because water-coated cooled liquid water at each level within the cloud. This
snowflakes appeared to the radiometer to be completely technique was simulated, but no field studies have been
liquid water as they melted. In addition, high emission conducted to demonstrate its viability.
during rainfall occasionally saturated the radiometer Solheim and Godwin (1998) describe the develop-
signal. The radiometers were valuable for their ability ment of a passive microwave radiometer that profiles
to observe the immediate onset of freezing precipita- atmospheric temperature, water vapor, and liquid cloud
tion aloft. water. Operating at two frequencies, the radiometer pro-
Huggins (1995) describes the use of a dual-wave- files water vapor by frequency scanning near 22 GHz
length radiometer system (20.6 and 31.65 GHz), similar and temperature and liquid-water profiles by scanning
to the ETL system but mounted on a truck for mobile near 60 GHz. Initial use demonstrates radiometer pro-
use. The system was used to monitor the spatial distri- files using a variety of retrieval algorithms compared
bution of liquid water over the Wasatch Plateau of Utah with radiosonde profiles. The instrument, although it
for cloud seeding. A unique spinning mirror system was needs additional testing, holds promise for providing
used to keep snow and rain off the radiometer mirror. temperature and liquid-water profiles at airports to
Most microwave radiometers measure integrated assess aircraft icing conditions.
liquid water; they provide no indication of the distribu- The effects of rainfall on integrated cloud liquid-
tion of liquid water vertically through a cloud. The dis- water measurements also demand attention because
tribution of liquid water within a cloud layer, and among rainfall causes cloud liquid water to be underestimated
layers if there is more than one layer, is needed for accu- in the 20–90-GHz range. Sheppard (1996) examined
rate aircraft icing estimates. Stankov et al. (1995) melded the effect of rainfall rate on vertically pointing micro-
climatological and iterative techniques that had been wave radiometer measurements of integrated liquid
used unsuccessfully in the past into a new technique water. Using a radiative transfer model, the amount of
using Ka-band radar, wind profilers, and ceilometers. error in estimated cloud liquid water increased with
Their goals were to determine cloud base height, cloud rainfall rate. The theory was tested against measure-
top, and cloud layers. Then, using temperature profiles ments made in CASP. Sheppard indicates that a future
and three different assumptions about the distribution study should investigate the effects of solid precipita-
of water within clouds, they parsed integrated water tion.
measured by the radiometers through the clouds. These 5.4.1.2 Sensing from satellites or aircraft over
experimental techniques were compared with aircraft water or land. Passive microwave sensing of cloud
measurements as part of WISP. Results were mixed: liquid water from a satellite or aircraft over land or water
some profiles compared well with aircraft measure- is more difficult than viewing toward zenith from be-
ments, and others compared poorly within the same low clouds. When observing toward the earth’s surface,
storm. Stankov and his colleagues indicated that more as explained earlier, the brightness temperature of water
comparisons are needed. This work demonstrates a fun- and land surfaces must be considered. When viewing
damental need for methods of distributing liquid water toward space, background temperature is near absolute
through clouds after total integrated values are meas- zero (~3°K) and clouds emit strongly in contrast. When
ured. This is important for determining the concen- viewing toward water bodies from above clouds, the
trations of supercooled liquid water in each cloud profile. water bodies appear cold, with brightness temperatures
Sauvageot (1996) has developed a method for con- of about 140°K at 37 GHz (this can vary by 15°K or
structing vertical profiles of liquid and ice water in more), because the emissivity of water bodies is about
mixed-phase clouds using a microwave radiometer to 0.5 (Jones and Vonder Haar 1990, Greenwald et al.
determine total integrated liquid water, a Doppler radar 1993). Land surfaces typically have larger brightness
to determine updraft velocities, and theory to explain temperatures than water because land surfaces are often

41
To Contents
warmer, their emissivity is close to 1, surface rough- effects prior to sensing clouds. A two-stage algorithm,
ness and vegetation contribute to variability, and, most developed by Jones and Vonder Haar (1990), estimates
importantly, soil moisture varies widely and radiates ground transmittance on clear days, and then, after clouds
strongly (Jones and Vonder Haar 1990). This larger and move over, removes the effects of ground transmittance.
more variable emmitance over land requires different Greenwald et al. (1997b) further advanced the technique
retrieval techniques than over water. The above expla- by using the polarization differences of the brightness
nation also suggests why, in principle, sensing cloud temperature effective over some land surfaces. The ad-
liquid water from the ground should be, and is, more vantage of polarization is that it improves the ability to
accurate, with values typically being correct within estimate the liquid-water content of low-lying clouds.
15%, as demonstrated by Hill (1992). Comparisons with ground-based radiometer measure-
5.4.1.2.1 Liquid-water retrieval over water. Retrieval ments in Colorado were generally good. This method
over water is relatively direct, and techniques for retriev- would be difficult to apply to in-flight detection of cloud
ing cloud liquid water over the oceans have been in use liquid water because a priori ground radiance informa-
for about 20 years with some of the original techniques tion would be difficult to obtain.
developed by Grody (1997). Schemes to retrieve liquid 5.4.1.2.3 Liquid-water retrieval from the horizontal.
water from microwave emissions are statistical, semi- Savage et al. (1999) have developed a technique for
statistical, and semiphysical (Greenwald et al. 1993). locating and estimating cloud liquid-water content using
Greenwald and his colleagues developed a simple physi- two frequencies, 37 and 89 GHz, and three viewing
cal technique that is accurate and has been rigorously angles. A radiometer placed on the nose of an aircraft
verified, unlike many other methods. Verification was would scan horizontally ahead of the aircraft and 2°
against ground-based microwave radiometers looking above and below the flight path. In a clear-sky condi-
skyward detecting cloud liquid water at four oceanic tion, the +2° beam sees colder temperatures, observing
locations. Relative errors in the algorithm range from toward cold space, than does the –2° beam observing
20 to 40%, and occasionally to 50%, with the largest toward the warmer surface of the earth. As the aircraft
errors in areas with low liquid-water content and thin approaches a cloud, the temperature of both beams con-
clouds. The algorithm is also only valid for nonprecipi- verges toward that of the horizontal beam. During this
tating clouds, because liquid precipitation radiates process, the horizontal beam provides an estimate of
strongly. Frozen precipitation does not cause problems the cloud temperature. An estimate of liquid-water con-
because ice is typically not visible at microwave wave- tent magnitude is obtained by comparing the bright-
lengths. ness temperatures of the 37- and 89-GHz beams in the
Lee et al. (1994) applied the Defense Meteorological +2° orientation. Since the 37-GHz beam penetrates far-
Satellite special sensor microwave/imager (DMSP ther than the 89-GHz beam, it will be colder than the
SSM/I) in an attempt to predict aircraft icing over ocean 89-GHz beam if there is little liquid water, because it
areas. They utilized a statistical retrieval algorithm to can detect the cold of space through the water. As liquid-
convert transmittance to cloud water using the 37-GHz water content increases, the +2° 89- and 37-GHz bright-
and 85-GHz bands. Though heavy precipitation con- ness temperatures converge as cold space is obscured.
taminates 37-GHz retrievals, according to Lee and Clar- Savage et al. (1999) also believe that the presence of
ke supercooled water and significant precipitation are drizzle-size drops can be detected by sensing polarized
mutually exclusive and thus reduce the magnitude of radiation scattered from the earth’s surface by large
the problem. Most analyses, however, were performed drops. The Savage techniques are being evaluated using
with the 85-GHz band because it has higher resolution. information gathered during the MWISP field project.
Aircraft icing was expected when liquid water was These examples of microwave radiometer capabili-
greater than 0.2 kg m–2 and temperatures were between ties over land and water surfaces represent only a small
0°C and –20°C. portion of all work accomplished, but cloud water
5.4.1.2.2 Liquid water retrieval over land. Microwave retrieval over land is new and is not yet available opera-
retrievals over land are possible if emittance from the tionally.* Precipitation retrieval from satellites over
land surface can be accounted for. In addition, the water and land has been developed even more than
retrieval wavelength is changed over land to 85.5 GHz, cloud water retrieval; it is presented in papers by Spen-
which is more sensitive to cloud liquid water than are cer et al. (1989), Petty and Katsaros (1992), Vivekanan-
other microwave channels, and surface effects become dan et al. (1993), and Ferraro and Marks (1996).
less important as cloud liquid water increases and atmo-
spheric attenuation obscures the ground at this frequency.
Jones and Vonder Haar (1990) and Greenwald et al. * Personal communication, J. Vivekanandan, National Center for
(1997a) developed methods of subtracting surface Atmospheric Research, Boulder, Colorado, 1997.

42
To Contents
 5.4.2 Drop-size spectra and cloud phase to icing, they may be penetrated, but the hazard is that
The only research found that described detection of clouds beyond cannot be sensed if there is no cloud-
cloud drop size with microwave radiometers was that free space ahead of them, so aircraft flying IFR cannot
of Savage et al. (1999), described above. Generally, use lidar to maximum advantage. Lidar could indicate
cloud ice particles cannot be detected by microwave to night VFR aircraft whether cloud lies ahead, and it
radiometers because ice is transparent to microwaves. could indicate whether liquid precipitation that could
However, Wu (1987) applied four channels of NASA’s be freezing also lies ahead below clouds, giving the
Advanced Microwave Moisture Sounder, flown on aircraft advance warning. Thus, lidar appears to have
high-altitude aircraft, to the problem of detecting the the greatest utility to night VFR pilots who have multi-
ice-water content of clouds in the microwave frequen- ple reasons to avoid clouds and may need to avoid pre-
cies of 92-, 183- (±2), 183- (±5), and 183- (±9) GHz cipitation. Lidar is of greatest potential utility for avoid-
bands. A microwave radiativex transfer routine was ing icing and of least potential utility for escaping icing.
developed that allows detection of the ice-water con- Lidar can operate over a range of wavelengths, it
centration through mixed-phase clouds by observing can be polarized, and it can be used in single-scattering
the changes in brightness temperature of each frequency. and multiple-scattering modes where multiple-field-of-
Some success was claimed by comparing computed ice- view lidars can utilize the information. Its most typical
water contents with observations made in the near- cloud uses are for determining liquid-water content,
infrared during the Cooperative Convection Precipita- phase, drop number, mean drop size, and optical thick-
tion Experiment. The study also showed that cloud ness. Lidar is also widely used to determine ceiling
brightness temperature at each frequency depended not height, though there is often difficulty with optically
only on total ice-water content of a cloud, but also on thin clouds, virga, and precipitation.
its distribution within the cloud. Further developments Using an infrared-wavelength CO2 lidar operating
of Wu’s technique have not been published in the last at 10.6 µm, Eberhard (1993) developed theory and
decade, so it is not clear whether the technique is fully demonstrated retrieval of the mean radius of cloud drop-
viable. size distributions. The lidar determines the extinction-
to-backscatter ratio, which is then fitted to a variety of
5.5 Lidar expected drop-size distributions until a fit is obtained.
Lidar, or light detection and ranging, is the optical The method is valid for distributions with drop sizes
equivalent of radar, operating in the visible and infra- falling between 1 and 17 µm. Data from fair-weather
red wavelengths. Unlike radar, however, wavelengths cumulus clouds at Cape Kennedy provided reasonable
used for lidar suffer rapid extinction in optically thick results, although no in-situ measurements were avail-
clouds, so their use for sensing cloud properties is lim- able for validation. Eberhard indicated that an 11-µm
ited. Cloud scattering rapidly attenuates the signal, pre- wavelength may provide better results.
venting most lidars from penetrating dense clouds for Bissonnette and Hutt (Hutt et al. 1994; Bissonnette
more than a few hundred meters. However, multiple and Hutt 1995a,b) at the Defence Research Establish-
scattering of lidar returns from clouds can be used to ment at Valcartier, Quebec, Canada, have used the back-
advantage for interpreting elements of cloud composi- scattered power from a 1.06-µm multiple-field-of-view
tion. (MFOV) polarized lidar to characterize cloud, fog, and
Overall, lidar may be able to contribute to aircraft aerosols. The system measures the backscatter from a
icing avoidance by remotely sensing cloud conditions, central beam with 1.5-m-long pulses and multiscattered
but only in very specific and limited ways because of return signal intensity at three or more coaxial fields of
the extinction problem. An ideal supercooled liquid- view with a maximum of ten possible fields of view.
water sensor will range-resolve liquid water and drop The amount of scatter returned is proportional to the
size many kilometers ahead of an aircraft, even if the number density of drops in the cloud. Fitted to a multiple
aircraft is flying within clouds. A lidar operates effec- scattering lidar equation, the measurements provide a
tively only when the aircraft is flying within a nearly scattering coefficient and a droplet effective radius.
cloud-free atmosphere. Lidar can sense through cloud- From this, and an assumed gamma drop-size distribu-
free atmosphere to the nearest clouds and determine tion, liquid-water content and extinction coefficients
the properties of the first few hundred meters of those are computed. Range-resolved droplet size distribution
clouds, but it cannot penetrate them to determine what (1 to ~100 µm), liquid-water content to 1.0 g m–3, and
lies beyond. Therefore, lidar can only help aircraft avoid extinction coefficient measurements have been made
icing conditions by determining whether there are at ranges to 1000 m and verified in situ (Bissonnette et
clouds in the immediate flight path and if they are con- al. 1998). The linear polarization ratio, between paral-
ducive to airframe icing. If the clouds are not conducive lel and perpendicular polarization, is about 35 to 40%

43
To Contents
when clouds are composed primarily of ice crystals. liquid-water content and drop-size distribution. How-
The 1.06-µm-wavelength is not eye-safe, but an eye- ever, because Raman systems are large and expensive
safe 1.54- or 2.0-µm lidar could be built with 1.5- to and can only be used at night, their application to air-
3.0-m resolution and a 100-Hz pulse repetition rate. craft icing is unlikely.
Temperature at the cloud could also be determined by In general, lidars have limited use for remotely
ranging with the lidar and sensing temperature with an detecting aircraft icing conditions because of their lim-
IR radiometer. Evidence of supercooling could also be ited ability to penetrate clouds. Many ground-based and
assessed using polarization techniques to determine if airborne lidars are currently used for boundary-layer
the cloud is mixed phase. research and wind-shear monitoring (Targ et al. 1991,
Benayahu et al. (1995) developed a method for Hannon and Henderson 1995). Canadian organizations
retrieving drop number and drop-size distribution from interested in aircraft icing have proposed a remote-
clouds, and potentially liquid-water content, assuming sensing system for detecting icing conditions that may
multiple scattering occurs at all times. A total scatter use lidar as a principal component (EWA 1996).
signal and multiple scattering signal, simultaneously Therefore, though not as promising for penetration of
received from two separated receivers, contained infor- clouds, lidar does have demonstrated capability and may
mation about the shape of the drop-size distribution and be a candidate technology for limited and specialized
the mean droplet radius. A field test was conducted on applications.
a marine stratus cloud off the coast of Israel with concur-
rent lidar and in-situ aircraft measurements, with good 5.6 Temperature measurement
agreement between the measurements. Drop temperature, or a surrogate for drop tempera-
Eberhard (1995), working with a CO2 laser, argues ture, is necessary to determine if liquid cloud water is
that depolarization in longer wavelengths cannot be supercooled. In most cases, drop temperatures will be
used to discriminate ice from water in clouds. However, nearly the same as the air temperature, neglecting radi-
the ratio of backscatter between two different wave- ative exchanges, but droplets can be considerably dif-
lengths can indicate the presence of ice. Seven wave- ferent from air temperature in several situations. Snow
lengths between 10.4 and 11.5 µm were selected and falling into warm air, such as in overrunning, will melt.
paired to test the theory, and the backscatter ratio The temperature of snow and ice crystals will rise to
between water and ice was shown to range from 2 to 5 0°C until melting is complete, and then they will con-
depending upon the frequency pairs chosen. The method tinue to warm. These drops may then fall into colder
demonstrated the feasibility of detecting ice vs. water air below and remain warmer than the air until conduc-
without polarization and estimated an approximate pro- tion, convection, radiative exchange, and evaporation
portion of ice vs. water. No field tests were conducted cools them to the dew-point temperature.
to verify the method, though they were planned. The There likely will be little choice whether to measure
technique is simple and eye safe. drop temperature or air temperature, because some sens-
When a laser beam is incident upon water droplets, ing technologies may preferentially sense water drops
most of the energy is scattered away from the drop with- or ice crystals and others may preferentially sense air
out change but scatters a small portion of the light at temperature. A surrogate for air-temperature measure-
different wavelengths. The scattering of light at differ- ments is the detection of ice crystals. If a cloud is mixed
ent wavelengths is Raman scattering, actually an phase, then it is likely that any liquid water is super-
exchange of energy between a photon and a molecule cooled. Ice crystals may be detected by radar and lidar.
with a resulting change in the energy—and thus wave- Ice detection methods were discussed above, so they
length—of the photon (Carey 1987). Raman scattering will not be dealt with here.
has become a useful technique for atmospheric sensing There are at least four types of sensors for measur-
of nitrogen, oxygen, and water vapor and for tempera- ing temperature profiles:
ture profiling. Melfi et al. (1997) report on a potential
• Microwave radiometers
use of Raman scattering for measuring the liquid-water
• Infrared radiometers
content of clouds. During water-vapor measurements
• Radio acoustic sounding (RASS)
with a XeF laser centered at 0.35 µm, they detected
• Raman lidar.
two thin cloud layers as they passed over the lidar. Melfi
et al. (1997) indicate that Raman lidar techniques hold Radiative techniques are infrared and millimeter
promise as a new method for remotely measuring cloud wave, acoustic systems are based on tracking the speed
of sound by radar, and lidars detect Raman scattering.
* Personal communication, L. Bissonnette, Defence Research Estab- The purpose of this review is to indicate how tempera-
lishment, Valcartier, Quebec, Canada, 1997. ture is measured remotely in the atmosphere and to indi-

44
To Contents
 cate which methods may be most viable for remote sens- square error (rms) in the lowest 3000 m of the atmo-
ing of aircraft icing conditions. sphere. Radiometers still do not have sufficient accu-
racy or resolution with height to replace radiosondes.
5.6.1 Infrared radiometers When comparing the accuracy of radiosondes, RASS,
Infrared radiometers are not sounders; temperatures and radiometers for temperature profiling, Schroeder
obtained are an integration of the cloud boundary layer, (1990) determined that the three devices compared well
depending upon cloud optical depth, and the depth from in the summer when temperature changes with height were
which contributing radiating energy originates. Exam- not rapid, but during the winter the RASS clearly pro-
ples of infrared thermal imagers are 10.8- and 11.8-µm vided better resolution.
sensors of the NOAA Advanced Very High Resolution Decker et al. (1978) constructed a radiometer system
Radiometer used to determine cloud-top temperatures operating around 60 GHz that scanned to an elevation
(Giraud et al. 1997, Lee 1997). Bissonnette* suggests angle of 45° from zenith. They indicated that temper-
that cloud temperature could be determined in a stand- ature retrievals are possible from elevation angles other
off situation using an infrared radiometer and range than zenith, though they chose not to report on them.
determined by lidar. This method would not be viable The Solheim and Godwin (1998) microwave radiometer
within cloud, however, because the temperature pro- discussed earlier profiles temperature, water vapor, and
vided would represent an integration of the cloud mass liquid water at zenith and at low elevation angles.
only a few hundred meters ahead of the aircraft, depend- Gary et al. (1992) report on an automatic tempera-
ing upon cloud optical depth. ture profiler operated on the NASA ER-2 high-altitude
aircraft. The radiometer, operating at 60 GHz, scans in
5.6.2 Microwave radiometers 10 angular steps from 50° below the flight path to 60°
Microwave temperature sounders detect temperature above the flight path ahead of the aircraft. Brightness
changes with altitude by either sensing from the ground temperatures from 15 distinct altitudes span from 2 km
to zenith or from satellites to nadir. Oxygen absorbs and below the aircraft to 3 km above it. Temperature retrieval
thus re-emits in the 60-GHz region (±10 GHz) and at accuracy is a function of distance from the aircraft, with
118.75 GHz. About 45 absorption lines centered on 60 greater distance causing greater uncertainty. Measure-
GHz are used to determine temperature with height in ments are not range gated. However, as the aircraft flies,
the atmosphere by pressure broadening (Grody 1997). if scanning is rapid enough, temperatures measured sev-
The absorption lines found in the ±10 GHz region around eral kilometers ahead of the aircraft can be assembled
60 GHz result from decreasing atmospheric pressure to create a composite temperature map. Such maps have
with altitude (Elachi 1987). Oxygen molecular collisions been assembled by the authors to create altitude tem-
are frequent enough that at given pressures they reach a perature profiles and horizontal profiles of temperature.
local thermodynamic equilibrium. This results in a shift This profiling technique may be promising for detect-
in the wavelength of emission around 60 GHz as pressure ing temperature ahead of aircraft in an icing environ-
changes with altitude, with lines of maximum emission ment. It deserves further exploration, but the instrument’s
shifting closer to 60 GHz with altitude. That is, the spec- capabilities within clouds are unknown.
trum of wavelength from which oxygen emits radiation
near 60 Hz narrows as pressure decreases with altitude. 5.6.3 Radio acoustic sounding systems (RASS)
As a result, temperature can be retrieved with altitude by RASS is used operationally by NOAA; it operates
using the expected wavelengths of emission of oxygen by directing acoustic waves, typically at 900 Hz, verti-
at given altitudes. Measurements are typically made on cally into the atmosphere (Schroeder 1990, Matuura et al.
either side of 60 GHz. 1986, May et al. 1989). Compression and rarefaction by
Microwave sounders routinely retrieve temperatures the sound wave alters the air’s dielectric constant, allow-
with height (Westwater and Grody 1980, Westwater et ing radar reflection. A strong reflection is obtained when
al. 1983, Askne 1987, Gary 1989, Solheim and Godwin the acoustic signal is matched to half of the radar wave-
1998). The primary problem with satellite and ground- length, creating Bragg scattering (May et al. 1988).
based radiometers is their inability to detect rapid NOAA uses 404-MHz Doppler radar to track the acous-
changes in temperature with altitude, such as inversions, tic wave, the same radars that are used for wind profiling
in the lower few kilometers of the atmosphere (West- (Westwater 1997).
water 1997). However, overall accuracy of ground-based RASS is generally immune to cloud effects, but there
radiometers is typically better than 1.2°C root mean are other sources of error. Humidity changes the speed
of sound and, if not considered, can cause errors of up
* Personal communication, L. Bissonnette, Defence Research Estab- to 2.2°C, and vertical wind velocities of 3 m s–1 can
lishment, Valcartier, Quebec, Canada, 1997. produce a nearly 7°C temperature error (North et al.

45
To Contents
1973). The NOAA wind profilers are capable of meas- Though research and development may be accom-
uring temperature accurately in horizontal and vertical plished through partnering, purchasing, and in-house
wind by averaging over a long time period, about 6 work, managers must acquire expertise to understand
min, and by using multiple acoustic sources. Vertical and properly direct activities. They must have the under-
wind errors have been experimentally reduced to as low standing, focus, and comprehensive vision to complete
as 0.1°C using 6-min averaging times (Angevine and the project.
Ecklund 1994). Overall RASS accuracies are compar- Adapt means to take technology and skills and adapt
able to radiosondes, with overall error consistently about it to needs (Deffeyes 1996). A full understanding of the
1°C rms, with altitudes of 3.5 km agl reached 50% of needs, operations, environment, and technologies avail-
the time (Westwater 1997). Experimentally, RASS able allows developers to adapt existing or developing
measurements have been made to 15 km altitude and technologies and techniques to the requirements of the
more (Matuura et al. 1986). product. This means, for example, adapting microwave
RASS is certainly a viable technique for sensing tem- radiometers or differential attenuation radar to operate
perature from the ground during icing conditions around on aircraft with other sensors as a system to satisfy the
airports. Its only limitation at airports may be an occa- icing information needs of operators and pilots.
sional inability to reach needed altitudes. Use of RASS Adopt means to acquire new ways of thinking and
in airborne applications, especially sensing ahead of of doing business (Deffeyes 1996). This will be neces-
the aircraft, has been assessed to be impractical because sary for system developers, as well as for regulatory
of problems with aircraft pitch and yaw and cross winds bodies, manufacturers, operators, and pilots. Adopting
causing loss of signal (Mead et al. 1998). new techniques may improve efficiency and safety, but
it requires willingness to change. For example, FAR
5.6.4 Raman lidar 25, Appendix C, has been the standard for aircraft design
Raman lidar techniques, discussed above, are also criteria in icing conditions. As new information is
used to measure atmospheric temperature profiles (Gill acquired characterizing the icing environment, users of
et al. 1979, Evans et al. 1997). Raman lidar uses a variety Appendix C may be required to change the range of
of wavelengths, with examples at 0.55 and 0.35 µm conditions for certifying aircraft for flight in icing con-
(Evans et al. 1997, Vaughan et al. 1993). Raman lidar ditions.
is typically operated only at night because the signal is A logical initial location to provide icing protection
overwhelmed by solar radiation. Long integration times using remote-sensing technology is at airports. The need
of 10 min are often necessary to obtain accurate temper- for ice protection is greatest in the approach and depar-
atures (Evans et al. 1997). In addition, it suffers from ture phases of flight because aircraft are operating at
the typical extinction problems suffered by all lidars in lower altitudes, lingering in conditions for longer
clouds. As a result, Raman lidar would not be practical periods at slower speeds, and operating closer to maneu-
for ground or aircraft-mounted remote sensing in icing vering limits. Protecting airports would also provide
conditions. the greatest benefit for the least cost.
There are few known remote temperature-measuring The ability to sense cloud microphysical properties
methods suitable for operation from airborne platforms remotely from the earth’s surface is mature in some
in icing conditions. The best possibilities lie with scan- technology areas. For example, RASS is a mature tech-
ning microwave radiometers, but they scan slowly and nology for measuring temperature profiles in the lower
may be difficult to use from a moving platform. Lidar atmosphere. Passive microwave radiometers are mature,
methods are not practical, and infrared radiometers but have less resolution than RASS for measuring tem-
operated in the 3.8- to 16.8-µm region have a short range perature profiles from the surface to midtroposphere
in clouds because of reduced optical depth. For ground- altitudes. Integrated liquid-water measurements may be
based systems, RASS and radiometers are proven and made from the earth’s surface using microwave radio-
thus offer the best prospects of success. meters, and the technology is mature. Liquid water may
be distributed among clouds with lidar ceilometers used
to determine cloud base and Ka- or W-band radar to
6.0 RECOMMENDATIONS
determine cloud base and top locations, the locations
Development of a remote-sensing system to detect of multiple cloud layers, and, if needed, cloud phase,
icing potential requires developers and users to be adept, or the new radiometer that profiles temperature, water
to adapt, and to adopt (Deffeyes 1996). Adept means that vapor, and cloud liquid water may be a viable option. A
developers have a sufficient understanding of the oper- system composed of this hardware, driven by expert
ational, meteorological, and sensor technology issues system logic, and perhaps supplemented with satellite
to develop a coherent product that addresses user needs. information, would make a usable prototype airport-

46
To Contents
 based remote-sensing system for mapping icing poten- safety in six areas, including weather. The weather
tial. A prototype ground-based demonstration system requirements include the need to detect icing condi-
would allow operational and technological problems tions remotely.
to be identified and resolved before airborne systems
reach a similar stage of development. Airborne systems 7.2 FAA
are much further from prototype system demonstration The FAA is the U.S. government agency responsible
because little promising technology is near maturity. for aviation operations and safety. It implemented the
The following recommendations for research are FAA Inflight Aircraft Icing Plan in April 1997 after three
presented for operations, meteorological sensing needs, meetings examining causes, and solutions to the causes,
and technology. of the October 1994 ATR-72 crash.
In-flight icing is recognized as a significant hazard
6.1 Operational needs research to both civilian and military aviation. In May 1996, the
• Assess human factors issues of remote-sensing FAA held an International Conference on Aircraft Icing
systems, assess cockpit and aircraft integration in Springfield, Virginia, attended by over 400 civilian
issues, and develop avoid-and-exit protocol and and military participants from 20 countries. As a result
training. of the conference, the development of in-flight ice detec-
• Assess integration into the weather system infra- tion emerged as a goal to “accelerate development of
structure for other aircraft, air traffic controllers, airborne technologies that remotely assess icing condi-
and meteorologists. tions by working with groups that already are support-
• Identify aircraft flight envelopes in icing condi- ing research in this area.” In response, the NASA Glenn
tions and characterize the icing hazard. Research Center, the FAA Technical Center, and
CRREL organized a cooperative research program to
6.2 Meteorological needs research accelerate development of systems for remotely detect-
• Characterize absolute magnitudes of the cloud ing icing conditions in the flight path.
microphysical conditions that produce icing.
• Assess the spatial and temporal variability of icing 7.3 NCAR
weather conditions at multiple scales. The National Center for Atmospheric Research,
• Develop an icing metric algorithm to convert funded by UCAR, which is in turn funded by the
liquid-water content, temperature, and elements of National Science Foundation, is the nation’s preemi-
the drop spectra into a measure of icing potential. nent meteorological research organization. With NASA,
the FAA, and Canada’s Atmospheric Environment Ser-
6.3 Technology needs research vice, NCAR has led the nation in aircraft icing weather
• Assess the feasibility of remote-sensing technol- research for several decades and will continue in that
ogies to provide liquid-water content, drop size, role. NCAR excels in developing forecast ability, instru-
and temperature information. mentation, and atmospheric characterization. In addi-
• Develop methods of assessing feasibility studies tion, NCAR operates airborne research platforms well-
through measurements with hardware. suited for proving instrumentation concepts.
• Develop prototype technologies for field testing.
7.4 NOAA ETL
7.0 GOVERNMENT ROLES, MISSIONS AND The National Oceanographic and Atmospheric
COLLABORATIVE ACTIVITIES Administration’s Environmental Technology Labora-
tory in Boulder, Colorado, is a pioneer developer of
7.1 NASA remote-sensing systems for detecting atmospheric phen-
NASA is the U.S. government agency primarily omena using radar, microwave radiometers, and RASS.
responsible for aviation research. NACA, NASA’s pre- ETL has developed some of the finest research radars
decessor, conducted pioneering icing research in the and radiometers available and has played an important
1940s that continues today, principally at the Glenn role in several icing research programs, such as the
Research Center at Lewis Field in Cleveland, Ohio. In 1989–1994 Winter Icing and Storms Project in Colo-
February 1997, President Clinton released the recom- rado, and the 1999 Mt. Washington Icing Sensors
mendations of the White House Commission on Avia- Project in New Hampshire. ETL often partners with
tion Safety and Security for improving aviation safety. NCAR in atmospheric research and will play an impor-
In response, NASA implemented the Aviation Safety tant role in atmospheric characterization and remote-
Program, a $500-million program to improve aviation sensing technology development.

47
To Contents
7.5 Department of Defense 9.0 LITERATURE CITED
Aviation has been the nation’s first line of defense
Aerometrics (1997) ADA-100 Icing Probe. Product bro-
for 50 years. Each of the military services operates a
chure.
fleet of aircraft best suited to their mission requirements.
Ahmed, M., and R. Brown (1995) Results from the
Only the Army, however, directly addresses cold-
freezing rain global climatology. Unpublished internal
weather problems because of its close association with
report, U.K. Meteorological Office.
winter weather conditions on and near the ground. The
AHS (1995) Tape transcription of International Icing
Army’s Cold Regions Research and Engineering Lab-
Symposium ’95 Workshop, American Helicopter Soci-
oratory has taken the DoD lead in developing remote-
ety, Montreal.
sensing systems for avoiding aircraft icing conditions.
Anderson, C., and D. Carbaugh (1993) Flight crew
Working hand in hand with NASA and the FAA,
interface aspects of forward-looking airborne windshear
CRREL is assisting in the management and develop-
detection systems. NASA Contractor Report 191437.
ment of icing avoidance capabilities to improve mili-
Angevine, W., and W. Ecklund (1994) Errors in radio
tary readiness.
acoustic sounding of temperature. Journal of Atmo-
spheric and Oceanic Technology, 11: 837–842.
8.0 IMPLEMENTATION Askne, J., G. Elgered, H. Nordius, G. Skoog, E. Win-
berg, A. Hagard, E. Andersson, N. Gustafsson, J. Svens-
Two tasks are involved in the implementation of a
son, and I. Carlsson (1987) The ONSAM Experiment:
remote-sensing ice-avoidance capability for aviation.
Remote sensing techniques for vertical sounding of the
The first is the development of a coherent and cooper-
atmosphere. Journal of Atmospheric and Oceanic Tech-
ative research and development plan, led by the federal
nology, 4: 180–190.
government but coordinated with industry and univer-
Atlas, D. (1954) The estimation of cloud parameters by
sities. The second is development of requirements,
radar. Journal of Meteorology, 11: 309–317.
either voluntary or mandated, by the FAA to place
Auld, H. (1989) Airframe icing: Some classification
remote-sensing systems at airports or aboard aircraft,
problems. Canadian Aeronautics and Space Journal,
as is now common with weather-avoidance radar and
35(3): 152–154.
wind shear alert, collision-avoidance, and terrain-avoid-
AVEMCO (1983) Ice: As unwelcome a sight as there
ance systems.
is. AVEMCO Pilot Bulletin, Flying Safety Update No.
Research and development must be led by the federal
60.
government. The government is mandated the respon-
Battan, L. (1962) Radar Meteorology. Chicago, Illinois:
sibility of enforcing aviation safety, and it has the pub-
University of Chicago Press.
lic trust. Development of icing avoidance capability
Battan, L. (1973) Radar Observation of the Atmosphere.
with onboard remote sensors is a high-risk venture that
Chicago, Illinois: University of Chicago Press.
is not likely to be undertaken by industry. In addition,
Baum, R., and T. Seymour (1980) In-flight evaluation
the White House Commission on Aviation Safety and
of a severe weather avoidance system for aircraft. Flight
Security requires that the government take a leadership
Dynamics Laboratory, Air Force Wright Aeronautical
role in improving aviation safety. Weather, including
Labs, Wright–Patterson Air Force Base, AFWAL-TR-
icing, is related to roughly 27% of all general aviation
80-3022.
accidents and 33% of all commercial aviation accidents.
Baumgardner, D. (1983) An analysis and comparison
In response to accidents, the FAA Inflight Aircraft Icing
of five water droplet measuring instruments. Journal of
Plan was developed. The NASA Aviation Safety Pro-
Climate and Applied Meteorology, 22: 891–910.
gram was also implemented in response to the report
Baumgardner, D., B. Baker, and K. Weaver (1993) A
of the White House Commission on Aviation Safety.
technique for the measurement of cloud structure on
The Department of Defense is responding to a need to
centimeter scales. Journal of Atmospheric and Oceanic
improve military safety and readiness.
Technology, August, 10: 557-565.
The result is a need for federal leadership and coor-
Beheim, M. (1979) Executive summary of Aircraft Icing
dination in the development and implementation of a
Specialists Workshop. In Proceedings of Aircraft Icing,
capability for remotely detecting and avoiding in-flight
Cleveland, Ohio, 19-21 July 1978, NASA Conference
icing conditions. In response, NASA, the FAA, and
Publication 2086, FAA-RD-78-109, p. 1–16.
CRREL have signed cooperative agreements to assess
Benayahu, Y., A. Ben-David, S. Fastig, and A. Cohen
the need and encourage the development of a remote-
(1995) Cloud-droplet-size distribution from lidar multiple-
sensing capability. This report supports that goal.
scattering measurements. Applied Optics, 34(9): 1569–
1578.

48
To Contents
 Bennett, I. (1959) Glaze, its meteorology and climatol- Royal Meteorological Society, 122: 1815–1844.
ogy, geographical distribution, and economic effects. Boyer, B. (1994) 3-D weather displays for aircraft cock-
Quartermaster Research and Engineering Center, Tech- pits. Department of the Air Force, DTIC ADA289759.
nical Report EP-105. Bracken, J., S. Green, J. Bettcher, and S. Erickson
Bernstein, B.C. (1996) The stovepipe algorithm: Identi- (1996) An operational perspective on ice detection and
fying locations where supercooled large droplets are protection systems research and development. In Pro-
likely to exist. In Proceedings of the FAA International ceedings of the FAA International Conference on Air-
Conference on Aircraft Inflight Icing, vol. II, Working craft Inflight Icing, Springfield, Virginia. vol. I, Work-
Group Papers, Springfield, Virginia, p. 353–364. ing Group Papers, p. 137–148.
Bernstein, B.C., and B.G. Brown (1997) A climatolo- Bragg, M. (1996) Aircraft aerodynamic effects due to
gy of supercooled large drop conditions based upon large droplet ice accretions. In Proceedings of 34th
surface observations and pilot reports of icing. 7th Aerospace Sciences Meeting & Exhibit, 15–18 Janu-
Conference on Aviation, Range and Aerospace Meteor- ary, Reno, Nevada. AIAA-98-0932. Reston, Virginia:
ology, Long Beach, California, 2–7 February, p. 82-87. American Institute of Aeronautics and Astronautics.
Berry, F., Jr., E. Bollay, and N. Beers (1945) Hand- Bragg, M, W. Perkins, N. Sarter, T. Baser, P. Voul-
book of Meteorology. New York: McGraw-Hill Book garis, H. Gurbacki, J. Melody, and S. McCray (1998)
Company. An interdisciplinary approach to inflight aircraft icing
Bertorelli, P. (1992) Icing: Assessing the risk. IFR, safety. In Proceedings of 34th Aerospace Sciences Meet-
November, p. 6–9. ing & Exhibit, 15–18 January, 1996, Reno, Nevada.
Bettcher, J., S. Green, S. Erickson, and J. Bracken AIAA-98-0095. Reston, Virginia: American Institute
(1996) A review of certification for flight in icing con- of Aeronautics and Astronautics.
ditions and recommendations for necessary improve- Brayton, R., and K. Hakala (1996) Working Group
ments. In Proceedings of the FAA International Confer- on Operational Regulations and Training Requirements.
ence on Aircraft Inflight Icing, Springfield, Virginia, vol. In Proceedings of the FAA International Conference on
I, Working Group Papers, p. 437–446. Aircraft Inflight Icing, Springfield, Virginia. vol. I, Plen-
Bissonnette, L., and D. Hutt (1995a) Lidar remote ary Sessions, p. 217–222.
sensing of cloud and fog properties. In Proceedings of Brenguier, J. (1993) Observations of cloud microstruc-
European Symposium on Optics for Environmental and ture at the centimeter scale. Journal of Applied Meteor-
Public Safety, Munich, June 19–23 1995, Proceedings ology, 32: 783–793.
no. 2506, paper no. 62. Sponsored by Commission of Broderick, A. (1996) Conference charge. In Proceed-
the European Communities, Directorate General for ings of the FAA International Conference on Aircraft
Science, Research, and Development; Air and Waste Inflight Icing, Springfield, Virginia. vol. I, Plenary Ses-
Management Association; Deutsche Forschungsanstalt sions, p. 3–5.
für Luft- und Raumfahrt eV; European Optical Society; Brown, D., and L. Dorr (1997) NASA, FAA and DoD
and International Society for Optical Engineering. challenged to achieve White House Commission goal
Bissonnette, L., and D. Hutt (1995b) Multiple scat- to reduce aircraft accidents. NASA News Release 97-
tered aerosol lidar returns: Inversion method and com- 26, 12 February.
parison with in situ measurements. Applied Optics, 34: Brown, E. (1981) An evaluation of the Rosemount ice
6959–6975. detector for aircraft hazard warning and for undercooled
Bissonnette, L., G. Roy, and J. Thériault (1998) Lidar cloud water content measurements. National Center for
monitoring of adverse weather. Defense Research Estab- Atmospheric Research, Technical Note, October.
lishment, Valcartier, Québec, DREV-R-9710. Buck, R. (1988) Weather Flying. New York: Macmill-
Bond, T., A. Reehorst, and C. Ryerson (Ed.) (1997) an Publishing Company.
NASA–LeRC/CRREL/ FAA Inflight Remote Sensing Carle, W. (1997) Comments in meteorological panel
Icing Avoidance Workshop, Summaries and Presenta- summary. In NASA–LeRC/CRREL Inflight Remote Sens-
tions, 1–2 April, Ohio Aerospace Institute, Cleveland, ing Icing Avoidance Workshop, Summaries and Presen-
Ohio. tations, 1–2 April, Ohio Aerospace Institute, Cleveland,
Borovikov, A., I. Gaivoronskii, E. Zak, V. Kostarev, Ohio.
I. Mazin, V. Minervin, A. Khrgian, and S. Shmeter Carey, P. (1987) Raman spectroscopy. In Encyclopedia
(1963) Cloud Physics. Jerusalem: Israel Program for of Physical Science and Technology. Orlando, Florida:
Scientific Translations. Academic Press, vol. 2, p. 1–17.
Bower, K., S. Moss, D. Johnson, T. Choularton, J. Choularton, T.W., and N. Bower (1993) A model of
Latham, P. Brown, A. Blyth, and J. Cardwell (1996) the development of droplet effective radius in convec-
A parameterization of the ice water content observed in tive cloud. Quarterly Journal of the Royal Meteoro-
frontal and convective clouds. Quarterly Journal of the logical Society, 119: 443–456.

49
To Contents
Claffey, K, K. Jones, and C. Ryerson (1995) Use and Cooper, W., M. Politovich, and W. Sand (1982) Nature
calibration of Rosemount ice detectors for meteorologi- of icing conditions encountered during meteorological
cal research. Atmospheric Research, 36: 277–286. research. Unpublished contract report. Martin, Pringle,
Clark, J. (1997) Introduction to the Operations Panel Oliver, Triplett and Wallace, Wichita, Kansas.
discussions. NASA–LeRC/ CRREL/FAA Inflight Remote Cotton, W., and R. Anthes (1989) Storm and Cloud
Sensing Icing Avoidance Workshop, Summaries and Dynamics. San Diego, California: Academic Press.
Presentations, 1–2 April, Ohio Aerospace Institute, Creley, J. (1990) Ice capade. Flying, June, p. 134–135.
Cleveland, Ohio. . Crum, T., and R. Alberty (1993) The WSR-88D and
Clothiaux, E., M. Miller, B. Albrecht, T. Ackerman, the WSR-88D Operational Support Facility. Bulletin of
J. Verlinde, D. Babb, R. Peters, and W. Syrett (1995) the American Meteorology Society, 74(9): 1669–1687.
An evaluation of a 94-GHz radar for remote sensing of Curry, J., and G. Liu (1992) Assessment of aircraft
cloud properties. Journal of Atmospheric and Oceanic icing potential using satellite data. Journal of Applied
Technology, 12(2): 201–229. Meteorology, 31: 605–621.
Cloutier, C. (1997) Presentation. NASA–LeRC/CRREL/ Decker, M., E. Westwater, and F. Guiraud (1978)
FAA Inflight Remote Sensing Icing Avoidance Work- Experimental evaluation of ground-based microwave
shop, Summaries and Presentations, 1–2 April, Ohio radiometric sensing of atmospheric temperature and
Aerospace Institute, Cleveland, Ohio. water vapor profiles. Journal of Applied Meteorology,
Cober, S., G. Isaac, and J. Strapp (1995) Aircraft icing 17: 1788–1795.
measurements in east coast winter storms. Journal of Decker, M., P. Fotino, and J. Schroeder (1986) Remote
Applied Meteorology, 34: 88–100. detection of aircraft icing conditions using microwave
Cober, S., G. Isaac, and J. Strapp (1996a) Aircraft radiometers. NOAA Technical Memorandum ERL
icing from supercooled drizzle. In Proceedings of the WPL-137.
12th International Conference on Clouds and Precipi- Deffeyes, R. (1996) Adept, adapt, adopt: Three hurdles
tation, 19–23 August, Zurich, Switzerland. International for new products. IEEE AES Systems Magazine, Octo-
Commission on Clouds and Precipitation (ICCP) of ber, p. 3–5.
International Association of Meteorology and Atmo- Doviak, R. (1983) A survey of radar rain measurement
spheric Sciences (IAMAS) of International Union of techniques. Journal of Climate and Applied Meteorol-
Geodesy and Geophysics (IUGG), p. 1318–1321. ogy, 22(5): 832–849.
Cober, S., G. Isaac, J. Strapp, and D. Marcotte Doviak, R., and D. Zrnic (1984) Doppler Radar and
(1996b) Analysis of aircraft icing environments asso- Weather Observations. San Diego, California: Academic
ciated with supercooled drizzle. In Proceedings of the Press, Inc.
FAA International Conference on Aircraft Inflight Icing, Eberhard, W. (1993) CO2 lidar technique for observ-
Springfield, Virginia. vol. II, Working Group Papers, ing characteristic drop size in water clouds. IEEE Trans-
p. 1–12. actions on Geoscience and Remote Sensing, 31: 56–63.
Cober, S., J. Strapp, and G. Isaac (1996c) An example Eberhard, W. (1995) Theory for dual-wavelength CO2
of supercooled drizzle drops formed through a collision– lidar method to distinguish ice, mixed-phase, and water
coalescence process. Journal of Applied Meteorology, clouds. Journal of Atmospheric and Oceanic Technol-
December, 35: 2250–2260. ogy, 12: 130–140.
Coffey, R. (1995) Big fat drops of ice. IFR, October, p. Elachi, C. (1987) Introduction to the Physics and Tech-
6–8. niques of Remote Sensing. New York: John Wiley and
Coleman, W. (1996) Presentation by the Regional Air- Sons.
line Association. In Proceedings of the FAA Interna- Engelberg, S., and A. Bryant (1995) Lost chances in
tional Conference on Aircraft Inflight Icing, Springfield, making a commuter plane safer. The New York Times,
Virginia. vol. I, Plenary Sessions, p. 217–222. 26 February, p. 1, 24–26.
Collins, R. (1989) Ice. AOPA Pilot, December, p. 43– Erickson, S. (1997) Air line pilots. NASA–LeRC/
47. CRREL/FAA Inflight Remote Sensing Icing Avoidance
Considine, G.D. (1997) Modeling the diurnal variabil- Workshop, Summaries and Presentations, 1–2 April,
ity in cloud microphysics in boundary layer clouds. Ohio Aerospace Institute, Cleveland, Ohio.
Journal of Geophysical Research, January, 102(D2): Erickson, S., S. Green, J. Bettcher, and J. Bracken
1717–1726. (1996) Difficulties in the operation of aircraft in icing
Cooper, W., and W. Sand (1997) Threshold conditions conditions In Proceedings of the FAA International Con-
for ice accretion. In 35th AIAA Aerospace Sciences ference on Aircraft Inflight Icing, Springfield, Virginia.
Meeting, 6–10 January, Reno, Nevada. AIAA-97-0307. vol. I, Working Group Papers, p. 575–584.
Reston, Virginia: American Institute of Aeronautics and Evans, K.D., S.H. Melfi, R.A. Ferrare, and D.N.
Astronautics. Whiteman (1997) Upper tropospheric temperature

50
To Contents
 measurements with the use of a Raman lidar. Applied wave temperature profiler during the airborne Antarctic
Optics, 36(12): 2594-2602. ozone experiment. Journal of Geophysical Research,
EWA (1996) Review of in-flight ice detection and meas- 94(D9): 11,223–11,231.
urement technologies. Document No. 1185-000-D000, Gary, B., R. Denning, and S. Walter (1992) Airborne
Electronic Warfare Associates–Canada, Ltd., Ottawa. radiometric temperature profiling. In Proceedings of the
FAA (1991) Aircraft Icing Handbook. U.S. Department Specialists Meeting on Microwave Radiometry and
of Transportation, vol. 1, DOT/FAA/CT-88/8-1, DTIC Remote Sensing, 14–16 January, Boulder, Colorado.
ADA 238 039. Gerber, H. (1991) Direct measurement of suspended
FAA (1997) FAA inflight aircraft icing plan. United particulate volume concentration and far-infrared extinc-
States Department of Transportation, April. tion coefficient with a laser-diffraction instrument.
Falcone, V., L. Abreu, and E. Shettle (1982) Atmo- Applied Optics, 30(33): 4824-4831.
spheric attenuation in the 30- to 300-GHz region using Gerber, H. (1996) Microphysics of marine stratocumu-
RADTRAN and MWTRAN. Proceedings of the Soci- lus clouds with two drizzle modes. Journal of the Atmo-
ety of Photogrammetric Optical Instrumentation Engi- spheric Sciences, 32(12): 1649–1662.
neering, 337: 62–66. Gerber, H., B. Arends and A. Ackerman (1994) New
Feingold, G., B. Stevens, W.R. Cotton, and A.S. Frisch microphysics sensor for aircraft use. Atmospheric
(1996) The relationship between drop in-cloud residence Research, 31: 235–252.
time and drizzle production in numerically simulated GEWEX (1996) New airborne cloud radar data.
stratocumulus clouds. Journal of the Atmospheric Sci- GEWEX News, 6(3): 16.
ences, 53(8): 1108–1122. Gill, R., K. Geller, J. Farina, and J. Cooney (1979) Meas-
Ferraro, R., and G.F. Marks (1995) The development urement of atmospheric temperature profiles using Raman
of SSM/I rain-rate retrieval algorithms using ground- lidar. Journal of Applied Meteorology, 18: 225–227.
based radar measurements. Journal of Atmospheric and Giraud, V., J.C. Buriez, Y. Fouquart, and F. Parol
Oceanic Technology, 12: 755–770. (1997) Large-scale analysis of cirrus clouds from
Fletcher, N. (1962) The Physics of Rainclouds. Cam- AVHRR data: Assessment of both a microphysical index
bridge: University Press. and the cloud-top temperature. Journal of Applied
Fotino, P., J. Schroeder, and M. Decker 1986) Ground- Meteorology, 36: 664–674.
based detection of aircraft icing conditions using mi- Glass, M., and D. Grantham (1981) Response of cloud
crowave radiometers. IEEE Transactions in Geoscience microphysical instruments to aircraft ice conditions. Air
and Remote Sensing, GE-24(6): 975–982. Force Geophysics Laboratory, Report AFGL-TR-81-
Fournier, G. (1993) Transport Canada proposed R&D 0192.
program for the development of a multi-parameter dual Goe, D. (1997) Comments in meteorological panel sum-
X–Ka band doppler radar for aviation meteorology appli- mary. NASA–LeRC/CRREL/ FAA Inflight Remote Sens-
cations. In Proceedings of the 5th International Confer- ing Icing Avoidance Workshop, Summaries and Presen-
ence on Aviation Weather Systems, Vienna, Virginia. tations, 1–2 April, Ohio Aerospace Institute, Cleveland,
American Meteorological Society, p. 146–147. Ohio.
Fox, N., and A. Illingworth (1997a) The retrieval of Gorgucci, E. and G. Scarchilli (1996a) Error structure
stratocumulus cloud properties by ground-based cloud of radar rainfall measurement at C-band frequencies with
radar. Journal of Applied Meteorology, 36: 485-492. dual polarization algorithm for attenuation correction.
Fox, N., and A. Illingworth (1997b) The potential of a Journal of Geophysical Research, 101(D21): 26,461–
spaceborne cloud radar for the detection of stratocumulus 26,471.
clouds. Journal of Applied Meteorology, 36: 676–687. Gorgucci, E., and G. Scarchilli (1996b) Operational
Friehe, C.A., and D. Khelif (1993) Fast-response air- monitoring of rainfall over the Arno River Basin using
craft temperature sensors. Journal of Atmospheric and dual-polarized radar and rain gauges. Journal of Applied
Oceanic Technology, 9: 784–795. Meteorology, 35: 1221–1230.
Frisch, A., C. Fairall, and J. Snider (1995) Measure- Gossard, E. (1994) Measurement of cloud droplet size
ment of stratus cloud and drizzle parameters in ASTEX spectra by Doppler radar. Journal of Atmospheric and
with a Ka-band Doppler radar and a microwave radio- Oceanic Technology, 11: 712–726.
meter. Journal of the Atmospheric Sciences, 52: 2788– Gossard, E., J. Snider, E. Clothiaux, B. Martner, J.
2799. Gibson, R. Kropfli, and A. Frisch (1997) The poten-
Gary, B. (1983) Microwave monitoring of aviation icing tial of 8-mm radars for remotely sensing cloud drop size
clouds. Report AFGL-TR-83-0271. Air Force Geophysics distribution. Journal of Atmospheric and Oceanic Tech-
Laboratory, Hanscom Field, Massachusetts. nology, 14: 76–87.
Gary, B.L. (1989) Observational results using the micro- Gosset, M., and H. Sauvageot (1992) A dual-wave-

51
To Contents
length radar method for ice-water characterization in Hill, G. (1991a) Comparison of simultaneous airborne
mixed-phase clouds. Journal of Atmospheric and and radiometric measurements of supercooled liquid
Oceanic Technology, 9: 538–547. water. Journal of Applied Meteorology, 30: 1043–1046.
Green, S. (1995) Inflight certification working paper. Hill, G. (1991b) Measurement of atmospheric liquid
In Proceedings of the International Icing Symposium, water by a ground-based single-frequency microwave
18–21 September, Montreal, p. 401–411. radiometer. Journal of Atmospheric and Oceanic Tech-
Green, S., J. Bettcher, and S. Erickson (1996) Inflight nology, 8: 685–690.
structural icing: An operational analysis and global Hill, G. (1992) Further comparisons of simultaneous
approach. In Proceedings of the FAA International airborne and radiometric measurements of supercooled
Conference on Aircraft Inflight Icing, Springfield, Vir- liquid water. Journal of Applied Meteorology, 31: 397–
ginia. vol. I, Plenary Sessions, p.129–164. 401.
Greenwald, T., G. Stephens, T. Vonder Haar and D. Hill, G.E. (1994) Analysis of supercooled liquid water
Jackson (1993) A physical retrieval of cloud liquid wa- measurements using microwave radiometer and vibrat-
ter over the global oceans using special sensor micro- ing wire devices, Journal of Atmospheric and Oceanic
wave/imager (SSM/I) observations. Journal of Geo- Technology, 11: 1242–1252.
physical Research, 98(D10): 18,471–18,488. Hill, G. (1997) Regulatory and national air space issues.
Greenwald, T, C. Combs, A. Jones, D. Randal, and NASA–LeRC/CRREL/FAA Inflight Remote Sensing Ic-
T. Vonder Haar (1997a) Further developments in esti- ing Avoidance Workshop, Summaries and Presentations,
mating cloud liquid water over land using microwave 1–2 April, Ohio Aerospace Institute, Cleveland, Ohio.
and infrared satellite measurements. Journal of Applied Hobbs, P., and A. Deepak (1981) Clouds, Their For-
Meteorology, 36: 389–405. mation, Optical Properties, and Effects. New York:
Greenwald, T., S. Christopher, and J. Chou (1997b) Academic Press.
Cloud liquid water path comparisons from passive Hobbs, P., and A. Rangno (1996) Precipitation from a
microwave and solar reflectance satellite measurements: maritime cloud layer with very low droplet concentra-
Assessment of sub-field-of-view cloud effects in micro- tions, Atmospheric Research, 40: 99–107.
wave retrievals. Journal of Geophysical Research, Hobbs, R., B. Morrison, R. Ashenden, and R. Ide
102(D16): 19,585–19,596. (1996) Comparison of two data processing techniques
Grody, N. (1997) Remote Sensing from Satellites Using for optical array probes In Proceedings of the FAA Inter-
Microwave Radiometry. NOAA, microwave radiome- national Conference on Aircraft Inflight Icing, Spring-
try short course text at 1997 American Meteorological field, Virginia. vol. II, Working Group Papers, p. 69–
Society Meeting, Long Beach, California. 90.
Guttman, N.B., and R.K. Jeck (1987) Aircraft icing Hobbs, P.V., L.F. Radke, R.R. Weiss, D.G. Atkinson,
environment in low ceiling conditions near Washing- J.D. Locatelli, K.R. Biswas, F.M. Turner, and C.E.
ton, D.C. Weather and Forecasting, 2: 114–126. Robertson (1974) Structure of clouds and precipita-
Haman, K., and S. Malinowski (1996) Temperature tion over the Cascade Mountains and their modifica-
measurements in clouds on a centimetre scale prelimi- tion by artificial seeding (1972–73). Research report 8,
nary results. Atmospheric Research, 41: 161–175. Department of Atmospheric Sciences, University of
Haman, K., A. Makulski, and S. Malinowski (1997) Washington, Seattle.
A new ultrafast thermometer for airborne measurements Hoffman, H. (1984) On the state of knowledge on
in clouds. Journal of Atmospheric and Oceanic Technol- meteorologically caused icing of aircraft in clouds.
ogy, 14: 217-227. Translation of Deutsche Forschungs- und Versuchsan-
Hannon, S., and S. Henderson (1995) Wind measure- stalt für Luft- und Raumfahrt—Mitt. 83-12, European
ment applications of coherent lidar. Review of Laser Space Agency Technical Translation ESA-TT-861, June
Engineering, 23(2): 124–130. 1984.
Hansman, R.J., Jr. (1985) Droplet size distribution ef- Hoffman, H., R. Roth, and J. Demmel (1986) Results
fects on aircraft ice accretion. Journal of Aircraft, 22(6): of icing flight in winter 1983/84: Dependence of ice
503–508. accretion thickness on cloud physics parameters as well
Hansman, R.J., Jr. (1997) Overview of human cen- as object and cloud parameters. Translation of Deutsche
tered design considerations for inflight remote sensing Forschungs- und Versuchsanstalt für Luft- und Raum-
icing avoidance. NASA–LeRC/CRREL/FAA Inflight Re- fahrt—FN-85-39, European Space Agency Technical
mote Sensing Icing Avoidance Workshop, Summaries Translation ESA-TT-969, April 1986.
and Presentations, 1–2 April, Ohio Aerospace Institute, Hogg, D.C., M.T. Decker, F.O. Guiraud, K.B. Earn-
Cleveland, Ohio. shaw, D.A. Merrett, K.P. Moran, W.B. Sweezy, R.G.
Helmreich, R. (1997) Managing human error in avia- Strauch, E.R. Westwater, and C.G. Little (1983a) An
tion. Scientific American, May, p. 62–67. automatic profiler of the temperature, wind and humid-

52
To Contents
 ity in the troposphere. Journal of Climate and Applied Physics, 15–18 November, Chicago. American Meteor-
Meteorology, 22: 807–831. ological Society, p. 408–411.
Hogg, D., F. Guiraud, J. Snider, M. Decker, and E. Jeck, R. (1983) A new data base of supercooled cloud
Westwater (1983b) A steerable dual-channel micro- variables for altitudes up to 10,000 feet AGL and the
wave radiometer for measurement of water vapor and implications for low-altitude aircraft icing. Department
liquid in the troposphere. Journal of Climate and of Transportation, FAA/CT-83-21, NRL Report 8738.
Applied Meteorology, 22: 789–806. Jeck, R. (1989) Selecting representative test points for
Horne, T. (1994) Tailplane icing. AOPA Pilot, March, icing conditions using the new FAA/NRL database.
p. 96–100. Presented at the 27th American Institute of Aeronau-
Houze, R. Jr. (1993) Cloud Dynamics. San Diego, tics and Astronautics Aerospace Sciences Meeting,
California: Academic Press. Reno, Nevada. Reston, Virginia: American Institute of
Howe, J. (1991) Rotating multicylinder method for the Aeronautics and Astronautics.
measurement of cloud liquid-water content and drop- Jeck, R. (1996) Representative values of icing-related
let size. USA Cold Regions Research and Engineering variables aloft in freezing rain and freezing drizzle. In
Laboratory, CRREL Report 91-2. Proceedings of the FAA International Conference on
Hudson, J.G., and G. Svensson (1995) Cloud micro- Aircraft Inflight Icing, Springfield, Virginia. vol. I,
physical relationships in California marine stratus. Jour- Working Group Papers, p. 57–68.
nal of Applied Meteorology, 34: 2655–2666. Jeck, R. (1998) A workable, aircraft-specific icing
Huggel, A., W. Schmid, and A. Waldvogel (1996) severity scheme. 36th American Institute of Aeronau-
Raindrop size distribution and the radar bright band. tics and Astronautics Aerospace Sciences Meeting, 12–
Journal of Applied Meteorology, 35: 1688–1701. 15 January, Reno, Nevada, AIAA-98-0094. Reston, Vir-
Huggins, A.W. (1995) Mobile microwave radiometer ginia: American Institute of Aeronautics and Astronau-
observations: Spatial characteristics of supercooled tics.
coud water and cloud seeding implications. Journal of Jones, A., and Vonder Haar, T. (1990) Passive micro-
Applied Meteorology, 34: 432–446. wave remote sensing of cloud liquid water over land
Hutt, D., L. Bissonnette, and L. Durand (1994) Mul- regions. Journal of Geophysical Research, 95(D10):
tiple field of view lidar returns from atmospheric aero- 16,673–16,683.
sols. Applied Optics, 33(12): 2338–2348. King, W., D. Parkin, and R. Handsworth (1978) A
Ide, R. (1996) Comparison of liquid water content hot-wire liquid device having fully calculable response
measurement techniques in an icing wind tunnel. In characteristics. Journal of Applied Meteorology, 17:
Proceedings of the FAA International Conference on 1809–1813.
Aircraft Inflight Icing, Springfield, Virginia. vol. I, Kirkpatrick, G.M. (1970) General aviation weather
Working Group Papers, p. 91–103. avoidance sensor study. NASA Report NAS 12-2032.
Isaac, G., A. Korolev, J. Strapp, S. Cober, A. Trem- NTIS No. N70-36981. Electronics Research Center,
blay, and R.A. Stuart (1996) Freezing drizzle forma- Cambridge, Masssachusetts.
tion mechanisms. In Proceedings of the 12th Interna- Kline, D.B. (1949) Investigation of meteorological
tional Conference on Clouds and Precipitation, 19–23 conditions associated with aircraft icing in layer-type
August, Zurich, Switzerland. International Commission clouds for 1947–1948 winter. Technical Note No. 1793,
on Clouds and Precipitation (ICCP) of Internatinal National Advisory Committee for Aeronautics.
Association of Meteorology and Atmospheric Sciences Kline, D.B., and J.A. Walker (1951) Meteorological
(IAMAS) of International Union of Geodesy and Geo- analysis of icing conditions encountered in low-altitude
physics (IUGG), p. 11–14. stratiform clouds. Technical Note No. 2306, National
Isaac, G.A., and R.S. Schemenauer (1979) Large Advisory Committee for Aeronautics.
particles in supercooled regions of northern Canadian Klugmann, D., and R. Judaschke (1996) A 94-GHz
cumulus clouds. Journal of Applied Meteorology, 18: FM–CW doppler radar profiler using a solid state trans-
1056–1065. mitter. Beitr. Physica Atmospherica, 69(1): 239–246.
Jameson, A. (1994) The estimation of rainfall para- Knollenberg, R. (1981) Techniques for probing cloud
meters using spaceborne and airborne nadir-pointing microstructure. In Clouds, Their Formation, Optical
radars. Journal of Applied Meteorology, 33: 230–244. Properties, and Effects (P. Hobbs and A. Deepak, Ed.).
Jeck, R. (1980) Icing characteristics of low altitude, New York: Academic Press, p. 15–91.
supercooled layer clouds. U.S. Department of Trans- Koenig, L.R. (1963) The glaciating behavior of small
portation, Interim Report No. FAA–RD-80-24. cumulonimbus clouds. Journal of the Atmospheric Sci-
Jeck, R. (1982) 5500 miles of liquid water and drop- ences, 20: 29–47.
size measurements in supercooled clouds below 10,000 Koenig G., C. Ryerson, J. Mead, and A. Pazmany
feet AGL. In Proceedings of the Conference on Cloud (1999) Multi-frequency radar technique for detecting

53
To Contents
cloud parameters for predicting aircraft icing potential. icing. In Compendium of Meteorology (T. Malone, Ed.).
In Proceedings of the American Meteorological Soci- Boston: American Meteorological Society, p. 1197–
ety 8th Conference on Aviation, Range and Aerospace 1203.
Meteorology, 12–15 January, Dallas, Texas. Paper Lewis, W., and W.H. Hoecker (1948) Observations
P13.13, p. 502–506. of icing conditions encountered in flight during 1948.
Korolev, A., J. Strapp, and G. Isaac (1996) On the Technical Note 1904, National Advisory Committee for
accuracy of PMS optical array probes. In Proceedings Aeronautics.
of the FAA International Conference on Aircraft Inflight Lewis, W., D. Kline, and C. Steinmetz (1947) A fur-
Icing, Springfield, Virginia. vol. I, Working Group ther investigation of the meteorological conditions
Papers, p. 105–113. conducive to arcraft icing. Technical Note No. 1424,
Kropfli, R., and R. Kelly (1996) Meteorological National Advisory Committee for Aeronautics.
research applications of mm-wave radar. Meteorologi- Lewis, W., P. Perkins, and R. Brunn (1953) Proce-
cal and Atmospheric Physics, 59: 105–121. dure for measuring liquid-water content and droplet
Kropfli, R., W. Moninger, and E. Pasqualucci (1982) sizes in supercooled clouds by rotating multicylinder
Cloud particle identification near the meeting layer with method. Research memorandum, NACA RM E53D23,
dual polarization K-band doppler radar. In Proceedings National Advisory Committee for Aeronautics.
of American Meteorological Society Conference on Lhermitte, R. (1987) A 94-GHz Doppler radar for
Cloud Physics, 15–18 November, Chicago, Illinois, p. cloud observations. Journal of Atmospheric and
412–415. Oceanic Technology, 4: 36–48.
Kropfli, R., S. Matrosov, T. Uttal, B. Orr, A. Frisch, Lhermitte, R. (1988) Observation of rain at vertical
K. Clark, B. Bartram, R. Reinking, J. Snider, and incidence with a 94-GHz Doppler radar: An insight on
B. Martner (1995) Cloud physics studies with 8-mm Mie scattering. Geophysical Research Letters, 15:
wavelength radar. Atmospheric Research. 35: 299–313. 1125–1128.
Langmuir, I., and K. Blodgett (1946) A mathemati- Liao, L., and K. Sassen (1994) Investigation of rela-
cal investigation of water droplet trajectories. In Col- tionships between Ka-band radar reflectivity and ice and
lected Works of Irving Langmuir. Elmsford, New York: liquid water contents. Atmospheric Research, 34: 231–
Pergamon Press, vol. 10, p. 335–393. 248.
Lawson, R., and W. Cooper (1990) Performance of Manningham, D. (1991) Rime and reason. Business
some airborne thermometers in clouds. Journal of Atmo- and Commercial Aviation, p. 82–85.
spheric and Oceanic Technology, 7: 480–494. Marcotte, D., J. Strapp, S. Cober, and G. Isaac (1996)
Lawson, R., and R. Cormack (1995) Theoretical The NRC Convair-580 research aircraft: Configuration
design and preliminary tests of two new particle spectro- in the Canadian freezing drizzle experiment. In Proceed-
meters for cloud microphysics. Atmospheric Research, ings of the FAA International Conference on Aircraft
35: 315–348. Inflight Icing, Springfield, Virginia, vol. I, Working
Lawson, R., and A. Rodi (1992) A new airborne ther- Group Papers, p. 125–135.
mometer for atmospheric and cloud physics research. Marillier, A., M. Cabane, and D. Cruette (1991) Pre-
Part I: Design and preliminary flight tests. Journal of liminary tests of an ultrasonic thermoanemometer for
Atmospheric and Oceanic Technology, 9: 556–574. aircraft measurements. Notes and Correspondence, 8:
Lawson, R., A. Korolev, T. Huang, L. Angus, K. 597–605.
Weaver, J. Strapp, G. Isaac, and S. Cober (1996) Martner, B., and R. Kropfli (1993) Observations of
Some instrumentation effects on airborne measurements multi-layered clouds using Ka-band radar. Presented at
of the drop size distribution in freezing drizzle. In Pro- American Institute of Aeronautics and Astronautics 31st
ceedings of the FAA International Conference on Air- Aerospace Sciences Meeting, 11–14 January, Reno,
craft Inflight Icing, Springfield, Virginia. vol. I, Work- Nevada.
ing Group Papers, p. 115–123. Martner, B., R. Kropfli, L. Ash, and J. Snider (1991)
Lawson, R.P., A.V. Korolev, S.G. Cober, T. Huang, Progress report on analysis of differential attenuation
J.W. Strapp, and G. Isaac (1998) Improved measure- radar obtained during WISP-91. NOAA Technical
ments of the drop size distribution of a freezing drizzle Memorandum ERL WPL-215.
event. Atmospheric Research, 47–48: 181–191. Martner, B., R. Kropfli, L. Ash, and J. Snider (1993a)
Lef, T.F., J.R. Clark, and S.D. Swadley (1994) Poten- Dual-wavelength differential attenuation radar measure-
tial applications of the SSM/I cloud liquid water para- ments of cloud liquid water content. In Proceedings of
meter to the estimation of marine aircraft icing. Weather the 26th International Conference on Radar Meteorol-
and Forecasting, 9: 173–182. ogy, Norman, Oklahoma, Preprints. American Meteo-
Lewis, W. (1951) Meteorological aspects of aircraft rological Society, p. 596–598.

54
To Contents
 Martner, B., R. Kropfli, L. Ash, and J. Snider (1993b) NASA (1997) Advanced probe fights icing and corro-
Cloud liquid water content measurement tests using dual- sion. NASA Tech Briefs.
wavelength radar. NOAA Technical Memorandum ERL NCAR (1990) 1990 winter icing storms project (WISP90)
ETL-235. data catalog. National Center for Atmospheric Research,
Masters, C.O. (1983) A new characterization of super- Research Applications Program, Boulder, Colorado.
cooled clouds below 10,000 feet AGL. Department of Newton, D.W. (1977) An integrated approach to the
Transportation, Final Report, FAA/CT-83-22. problem of aircraft icing. In Proceedings, AIAA Aircraft
Matrosov, S., R. Reinking, R. Kropfli, and B. Bar- Systems and Technology Meeting, Seattle, Washington,
tram (1996) Estimation of ice hydrometeor types and 22–24 August, New York. Reston, Virginia: American
shapes from radar polarization measurements. Journal Institute of Aeronautics and Astronautics.
of Atmospheric and Oceanic Technology, 13: 85–96. Newton, D.W. (1979) A review of the icing situation
Matuura, N., Y. Masuda, H. Inuki, S. Kato, S. Fukao, from the standpoint of general aviation. In Proceedings
T. Sato, and T. Tsuda (1986) Radio acoustic measure- of Aircraft Icing, Cleveland, Ohio, 19–21 July 1978.
ment of temperature profile in the troposphere and strato- NASA Conference Publication 2086, FAA-RD-78-109,
sphere. Nature, 323: 426–428. p. 31–38.
May, P., R. Strauch, and K. Moran (1988) The alti- North, E., A. Peterson, and H. Parry (1973) RASS, a
tude coverage of temperature measurements using RASS remote sensing system for measuring low-level temper-
with wind profiler radars. Geophysical Research Let- ature profiles. Bulletin, American Meteorological Soci-
ters, 15: 1381–1384. ety, 54(9): 912–919.
May, P., K. Moran, and R. Strauch (1989) The accu- NTSB (1996) In-flight icing encounter and loss of con-
racy of RASS temperature measurements. Journal of trol: Simmons Airlines, d.b.a. American Eagle Flight
Applied Meteorology, 28: 1329–1335. 4184. Avions de Transport Regional (ATR) Model 72-
Mayer, P., R. Jeck, and P. Drake (1984) An analysis of 212, N401AM, Roselawn, Indiana, October 31 1994.
icing-related mishaps on U.S. Navy aircraft from 1964 National Transportation Safety Board Aircraft Accident
to 1984. Navy Research Lab Memorandum, Report 5418. Report, Volume 1: Safety Board Report, NTSB/AAR-
McClean, P. (1992) Ice over Lake Erie. AOPA Pilot, 96/01.
January, p. 83–84. Oldenburg, J., and R. Ide (1990a) Comparison of drop
Mead, J., A. Pazmany, S. Sekelsky, and R. McIntosh size distributions from two droplet sizing systems.
(1994) Millimeter-wave radars for remotely sensing NASA Technical Memorandum 102520.
clouds and precipitation. Proceedings IEEE, 82(23): p. Oldenburg, J., and R. Ide (1990b) Comparison of two
1891–1906. droplet sizing systems in an icing wind tunnel. NASA
Mead, J., A. Pazmany, and M. Goodberlet (1998) Technical Memorandum 102456.
Evaluation of technologies for the design of a prototype Owen, T. (1997) A regulating agency’s operational per-
in-flight remote aircraft icing potential detection system. spective. NASA–LeRC/ CRREL/FAA Inflight Remote
DOT/FAA/AR-98/72 report. Sensing Icing Avoidance Workshop, Summaries and
Melfi, S.H., K.D. Evans, J. Li, D. Whiteman, R. Fer- Presentations, 1–2 April, Ohio Aerospace Institute,
rare, and G. Schwemmer (1997) Observation of Raman Cleveland, Ohio.
scattering by cloud droplets in the atmosphere. Applied Paluch, I., C. Knight, and L. Miller (1996) Cloud
Optics, 36(15): 3551–3559. liquid water and radar reflectivity of nonprecipitating
Miller, A., and R. Anthes (1980) Meteorology. Colum- cumulus clouds. Journal of Atmospheric Sciences,
bus, Ohio: Charles E. Merrill Publishing Company, 4th 53(11): 1587–1603.
edition. Parelon, E. (1996) JAA Certification Policy NPA 25F-
Miller, D., T. Ratvasky, B. Bernstein, F. McDonough, 219, Flight in icing conditions: Acceptable handling
and W. Strapp (1998) NASA/FAA/NCAR supercooled characteristics and performance effects. In Proceedings
large droplet icing flight research: Summary of winter of the FAA International Conference on Aircraft Inflight
96–97 flight operations. In 36th AIAA Aerospace Sci- Icing, vol. I, Working Group Papers, Springfield, Vir-
ences Meeting, 12–15 January, Reno, Nevada. AIAA- ginia, p. 413–436.
98-0577. Reston, Virginia: American Institute of Aero- Pazmany, A.L., R. McIntosh, R. Kelly, and G. Vali
nautics and Astronautics. (1994) An airborne 95-GHz dual-polarized radar for
Morrison, B.J., J. Jung, and R.L. Rose (1997) Mobile cloud studies. IEEE Transactions on Geoscience and
aircraft icing forecast system. In Proceedings of the Remote Sensing, 32(4): 731–739.
Conference on Cloud Impacts on DoD Operations and Peer, R. (1986) Tips on avoiding low-level icing. 5th
Systems, 23–25 September, Newport, Rhode Island, p. 235– Weather Wing Forecaster Memo.
238. Perkins, P.J. (1952) Preliminary survey of icing condi-

55
To Contents
tions measured during routine transcontinental arline Meteorological Society Conference on Weather Analy-
operation. National Advisory Committee for Aeronaut- sis and Forecasting, 19–23 August, Norfolk, Virginia.
ics. Research Memorandum. Rodert, L. (1951) Physical and operational aspects of
Personne, P., J. Brenguier, J. Pinty, and Y. Pointin aircraft icing. In Compendium of Meteorology (T. Mal-
(1982) Comparative study and calibration of sensors for one, Ed.). American Meteorological Society, Boston, p.
the measurement of the liquid water content of clouds 1190–1196.
with small droplets. Journal of Applied Meteorology, 21: Roebber, P. (1988) Helicopter and airframe icing poten-
189–196. tial off the Canadian east coast. Atmospheric Environ-
Petty, G., and K. Katsaros (1992) Nimbus-7 SMMR ment Service, Downsview, Ontario, Canadian Climate
precipitation observations calibrated against surface Center Report No. 88-5.
radar during TAMEX. Journal of Applied Meteorology, Rogers, R., and M. Yau (1989) A Short Course in Cloud
31: 489–505. Physics. Oxford: Pergamon Press.
Politovich, M. (1982) Characteristics of icing conditions Ryerson, C. (1990) Atmospheric icing rates with eleva-
in wintertime stratiform clouds. In Proceedings of Con- tion on northern New England mountains, U.S.A. Arc-
ference on Cloud Physics, 15–18 November, Chicago, Illin- tic and Alpine Research, 22(1): 90–97.
ois. American Meteorological Society, p. 404–407. Ryerson, C. (1996) Remote detection and avoidance of
Politovich, M. (1989) Aircraft icing caused by large inflight icing. In Proceedings of the FAA International
supercooled droplets. Journal of Applied Meteorology, Conference on Aircraft Inflight Icing, vol. II, Working
28: 856–868. Group Papers, Springfield, Virginia, p. 179–190.
Politovich, M., and B.C. Bernstein (1995) Production Ryerson, C. (1997) Prospects and problems of remotely
of depletion of supercooled liquid water in a Colorado detecting inflight icing potential. In Proceedings of the
winter storm. Journal of Applied Meteorology, 34: 2631– American Meteorological Society 7th Conference on
2648. Aviation, Range and Aerospace Meteorology, 2–7 Feb-
Politovich, M., and G. Vali (1983) Observations of liquid ruary, Long Beach, California, p. 138–143.
water in orographic clouds over Elk Mountain. Journal Ryerson, C. (1998) Aircraft icing remote sensing. In
of the Atmospheric Sciences, 40(5): 1300–1312. 36th AIAA Aerospace Sciences Meeting & Exhibit, 12–
Politovich, M., B. Stankov, and B. Martner (1995) 15 January 1998, Reno, Nevada. AIAA-98-0578. Res-
Determination of liquid water altitudes using combined ton, Virginia: American Institute of Aeronautics and
remote sensors. Journal of Applied Meteorology, 34: Astronautics.
2062–2075. Ryerson, C., M. Politovich, K. Rancourt, G. Koenig,
Pruppacher, H., and J. Klett (1997) Microphysics of R. Reinking, and D. Miller (2000) Overview of Mt.
Clouds and Precipitation. Dordrecht : Kluwer Academic Washington Icing Sensors Project. 38th AIAA Aerospace
Publishers. Sciences Meeting, 10–13 January, Reno, Nevada. AIAA-
Ramsay, A. (1997) Freezing rain detection and report- 2000-0488. Reston, Virginia: American Institute of Aero-
ing by the Automated Surface Observing System nautics and Astronautics.
(ASOS). In Preprints of the 1st Symposium on Integrated SAE (1995) Aircraft Ice Detectors and Icing Rate Meas-
Observing Systems. American Meteorological Society, uring Instruments. AIR 4367. Warrendale, Pennsylva-
paper J10.13, p. J65–J69. nia: Society of Automotive Engineers.
Reinking, R., S. Matrosov, B. Martner, and R. Kropfli Sand, W., and R. Kropfli (1991) Icing Hazard Detec-
(1996) Differentiation of freezing drizzle from ice hydro- tion for Aircraft. United States Patent 5,028,929, July 2,
meteors and freezing rain with dual-polarization radar. 1991.
In Proceedings of the FAA International Conference on Sand, W., W. Cooper, M. Politovich, and D. Veal
Aircraft Inflight Icing, vol. II, Working Group Papers, (1984) Icing conditions encountered by a research air-
Springfield, Virginia, p. 331–338. craft. Journal of Climate and Applied Meteorology,
Reinking, R., S. Matrosov, R. Bruintjes, and B. Mart- 23(10): 1427–1440.
ner (1997) Identification of hydrometeors with ellipti- Sassen, K., and L. Liao (1996) Estimation of cloud
cal and linear polarization Ka-band radar. Journal of content by W-band radar. Journal of Applied Meteorol-
Applied Meteorology, 36: 322–339. ogy, 35: 932–938.
Riley, J. (1998) Mixed-phase icing conditions: A review. Sauvageot, H. (1996) Retrieval of vertical profiles of
DOT/FAA/AR-98/76 report. liquid water and ice content in mixed clouds from Dop-
Rinehart, R. (1997) Radar for Meteorologists. Grand pler radar and microwave radiometer measurements.
Forks, North Dakota: Rinehart Publications. Journal of Applied Meteorology, 35: 14–23.
Robbins, C., and J. Cortinas, Jr. (1996) A climatology Savage. R., R. Lines, J. Cole, and G. Koenig (1999) A
of freezing rain in the contiguous United States: Pre- passive microwave icing avoidance system (MIAS). In
liminary results. In Preprints of the 15th American Proceedings of the American Meteorological Society 8th

56
To Contents
 Conference on Aviation, Range and Aerospace Meteo- and Oceanic Technology, 12: 488–510.
rology, 12–15 January, Dallas, Texas. Paper P12.9, p. Stewart, R., R. Crawford, K. Szeto, and D. Hudak
457–461. (1996) Horizontal aircraft passes across 0°C regions
Schroeder, J. (1990) A comparison of temperature within winter storms. Atmosphere–Ocean, 34(1): 133–
soundings obtained from simultaneous radiometric, ra- 159.
dio-acoustic, and rawinsonde measurements. Journal of Strapp, J., R. Stuart, and G. Isaac (1996) A Canadi-
Atmospheric and Oceanographic Technology, 7: 495– an climatology of freezing precipitation, and a detailed
503. study using data from St. John’s, Newfoundland. In
Schuyler, N. (1989) The iceman cometh. Private Pilot, Proceedings of the FAA International Conference on
November, p. 31–37. Aircraft Inflight Icing, vol. II. Working Group Papers,
Sekelsky, S., and R. McIntosh (1996) Cloud observa- Final Report, p. 45–55.
tions with a polarimetric 33-GHz and 95-GHz radar. Svensson, E., M. Angelborg-Thanderz, L. Sjöberg,
Meteorological and Atmospheric Physics, 59: 123–140. and S. Olsson (1997) Information complexity—Mental
Shah, A., M. Patnoe, and E. Berg (1998) Droplet size workload and performance in combat aircraft. Ergo-
distribution and ice shapes. In 36th Aerospace Sciences nomics, 40(3): 362–380.
Meeting, 12–15 January, Reno, Nevada. AIAA-98-0487. Takahashi, N, H. Uyeda and K. Kikuchi (1996) Evo-
Reston, Virginia: American Institute of Aeronautics and lution process and precipitation particles of an isolated
Astronautics. echo observed with dual-polarization Doppler radar near
Sheppard, B. (1996) Effect of rain on ground-based Sapporo on July 9 1992. Journal of Facilities Science,
microwave radiometric measurements in the 20- to 90- Hokkaido University, Ser. VII (Geophysics), 10(1):
GHz range. Journal of Atmospheric and Oceanograph- 135–153.
ic Technology, 13: 1139–1151. Targ, R., M. Kavaya, R. Huffaker, and R. Bowles
Siquig, R. (1993) Weather and unmanned aerial vehi- (1991) Coherent lidar airborne windshear sensor: Per-
cles. Report NRL/MR/75432-93-7206. Naval Research formance evaluation. Applied Optics, 30(15): 2013–
Laboratory. 2026.
Smith, P. (1985) Summary of Workshop on Remote Taylor, R. (1991) Aviation Weather. Greenwich, Con-
Detection of Potential Aircraft Icing Conditions. Pre- necticut: Belvoir Publications.
pared for Center for Aerospace Sciences, SDSMT/IAS/ Telford, J.W. (1988) An example of the behavior of an
R-85-06. aircraft with accumulated ice: Latent instability. Jour-
Solheim, F., and J. Godwin (1998) Passive ground- nal of Applied Meteorology, 27: 1093–1109.
based remote sensing of atmospheric temperature, water Telford, J.W. (1996) Clouds with turbulence: The role
vapor, and cloud liquid water profiles by a frequency of entrainment. Atmospheric Research, 40: 261–282.
synthesized microwave radiometer. Meteorologische Thomas, M., and J. Marwitz (1995) Droplet spectra
Zeitschrift., N.F. 7, p. 370–376. in freezing rain. In Proceedings of the American Meteor-
Spencer, R., H. Goodman, and R. Hood (1989) Pre- ological Society 6th Conference on Aviation Weather
cipitation retrieval over land and ocean with the SSM/I: Systems, 15–20 January, Dallas, Texas, p. 253–256.
Identification and characteristics of the scattering signal. Thompson, G., R. Bullock, and T. Lee (1997) Using
Journal of Atmospheric and Oceanic Technology, 6: satellite data to reduce spatial extent of diagnosed icing.
254–273. Weather and Forecasting, 12: 185–190.
Srivastava, R., and L. Tian (1996) Measurement of Thomson, A., and R. List (1996) Raindrop spectra and
attenuation by a dual-radar method: Concept and error updraft determination by combining Doppler radar and
analysis. Journal of Atmospheric and Oceanic Technol- disdrometer. Journal of Atmospheric and Oceanic Tech-
ogy, 13: 937–947. nology, 13: 465–476.
Stack, D. (1996) Inflight icing—The critical need for Tian, L., and R. Srivastava (1997) Measurement of
improved forecasts and indexing of the weather. In Pro- attenuation at C band in a convective storm by a dual-
ceedings of the FAA International Conference on Air- radar method. Journal of Atmospheric and Oceanic
craft Inflight Icing, vol. II, Working Group Papers, Technology, 14: 184–196.
Springfield, Virginia, p. 575–584. Toomay, J. (1982) Radar Principles for the Non-
Stankov, B., E. Westwater, J. Snider, and R. Weber Specialist. London: Lifetime Learning Publications.
(1992) Remote measurements of supercooled integrated Transport Canada (1996) Human Factors for Avia-
liquid water during WISP/FAA aircraft icing program. tion: Advanced Handbook. Transport Canada Safety and
Journal of Aircraft, 29(4): 604–715. Security, TP 12864 (E).
Stankov, B., B. Martner, and M. Politovich (1995) Tremblay, A., S. Cober, A. Glazer, and G. Isaac
Moisture profiling of the cloudy winter atmosphere us- (1996) Observations and parameterization of mixed-
ing combined remote sensors. Journal of Atmospheric phase clouds during the Canadian freezing drizzle

57
To Contents
experiment. In Proceedings of the 12th International 685.
Conference on Clouds and Precipitation, Zurich, Westwater, E. (1997) Remote sensing of tropospheric
Switzerland. temperature and water vapor by integrated observing
Tucker, W. (1983) Current procedures for forecasting systems. Bulletin of the American Meteorological Soci-
aviation icing: A review. USA Cold Regions Research ety, 78(9): 1991–2006.
and Engineering Laboratory, Special Report 83-24. Westwater, E.R., and N.C. Grody (1980) Combined
Vaughan, G., D.P. Wareing, S.J. Pepler, L. Thomas, surface- and satellite-based microwave temperature pro-
and V. Mitev (1993) Atmospheric temperature meas- file retrieval. Journal of Applied Meteorology, 19: 1438–
urements made by rotational Raman scattering. Applied 1444.
Optics, 32(15): 2758–2764. Westwater, E., and R. Kropfli (1989) Remote sens-
Vigeant-Langlois, L., and R. Hansman (1999) Human- ing techniques of the wave propagation laboratory for
centered design considerations for in-flight remote sens- the measurement of supercooled liquid water: Applica-
ing icing avoidance. Presented at 4th International Remote tions to aircraft icing. NOAA Technical Memorandum
Sensing Conference and Exhibition/21st Canadian Sym- ERL WPL-163, NOAA Environmental Research Labo-
posium on Remote Sensing, 21–24 June, Ottawa, Ontario, ratories.
Canada. Westwater, E.R., M.T. Decker, A. Zachs, and K.S.
Vivekanandan, J., J. Turk, and V. Bringi (1993) Com- Gage (1983) Ground-based remote sensing of temper-
parisons of precipitation measurements by the advanced ature profiles by a combination of microwave radiom-
microwave precipitation radiometer and multiparame- etry and radar. Journal of Climate and Applied Meteor-
ter radar. IEEE Transactions on Geoscience and Remote ology, 22: 126–133.
Sensing, 31(4): p. 860–870. White, A., C. Fairall, A. Frisch, B. Orrm, and J.
Vivekanandan, J., G. Thompson, and T. Lee (1996) Snider (1996) Recent radar measurements of turbulence
Aircraft icing detection using satellite data and weather and microphysical parameters in marine boundary layer
forecast model results. In Proceedings of the FAA Inter- clouds. Atmospheric Research, 40: 77–221
national Conference on Aircraft Inflight Icing, vol. I, Willis, P., and P. Tattelman (1989) Drop-size distri-
Plenary Sessions, Springfield, Virginia, p. 365–374. butions associated with intense rainfall. Journal of
Walter, S., and P. Moynihan (1997) Mobile aircraft Applied Meteorology, 28: 3–15.
icing forecast system. In Proceedings of the Confer- Wilson, D., A. Illingworth, and T. Blackman (1997)
ence on Cloud Impacts on DoD Operations and Sys- Differential Doppler velocity: A radar parameter for
tems, 23–25 September, Newport, Rhode Island. Phil- characterizing hydrometeor size distributions. Journal
lips Laboratory, Geophysics Directorate; Office of the of Applied Meteorology, 36: 649–662.
Oceanographer of the Navy; and Army Research Lab- Wright, W. (1995) Capabilities of LEWICE 1.6 and
oratory. Hampton, Virginia: Science and Technology comparison with experimental data. In Proceedings,
Corporation, p. 213–216. International Icing Symposium ’95, American Helicop-
Wanke, C., and R. Hansman, Jr. (1991) Alert genera- ter Society, Montreal, p. 349–361.
tion and cockpit presentation for an integrated cockpit Wu, M-L (1987) Determination of cloud ice water con-
microburst alert system. Joint University Program for tent and geometrical thickness using microwave and
Air Transportation Research 1990–1991, Langley infrared radiometric measurements. Journal of Climate
Research Center, p. 29–39. and Applied Meteorology, 26: 878–884.
Warner, J. (1969) The microstructure of cumulus Young, K. (1993) Microphysical Processes in Clouds.
cloud. Part I. General features of the droplet spectrum. New York: Oxford University Press.
Journal of the Atmospheric Sciences, 26: 1049–1059. Zrnic, D. (1996) Weather radar polarimetry—Trends
Westwater, E. (1978) The accuracy of water vapor and toward operational applications. Bulletin of the Ameri-
cloud liquid determination by dual-frequency ground- can Meteorological Society, 11(7): 1529–1534.
based microwave radiometry. Radio Science, 13(4): 677–

58
To Contents
 APPENDIX A: SYNOPSIS OF OPERATIONAL INFORMATION NEEDS,
STATE OF KNOWLEDGE, STRENGTHS, AND WEAKNESSES

INFORMATION NEEDS
Pilot needs and human factors 10. Create regulatory policy development plan.
1. Determine pilot information needs. 11. Consider system integration, protocol compati-
2. Design information displays conducive to effec- bility, and certification.
tive risk assessment and cockpit resource man- 12. Determine impact on NAS, ATC, and Free Flight.
agement. 13. Develop standards to integrate ground, in-situ,
3. Understand pilot decision-making process for and remotely sensed information.
accurate hazard/risk assessment. 14. Develop auto-reporting standards for ground
4. Develop avoid/escape strategies. (ATC and weather) and other-aircraft linkages.
5. Determine effect of aircraft performance on 15. Develop testing standards for remote-sensing sys-
avoid/exit strategies vs. information needed. tems.
6. Determine an acceptable warning lead time.
7. Determine optimal training requirements. Test beds and platforms
8. Develop unambiguous indications of icing, and 1. Develop training and simulation protocols.
an ability to sense through icing. 2. Locate test beds and platforms for sensor, infor-
9. Improve operational guidelines for flight in icing. mation, avoidance, and escape procedure testing.
3. Develop testing standards.
Operators and manufacturers
1. Minimize remote-sensing system cost, size, STATE OF KNOWLEDGE
power, and weight and maximize maintainability, Pilot needs and human factors
simplicity, and reliability. 1. Pilots have no indication when they are in icing
2. Establish aircraft space, weight, and power equip- conditions that exceed aircraft limitations.
ment constraints. 2. Pilots often do not know how to avoid or escape
3. Develop incentives to use in-flight remote-sensing icing.
icing-avoidance systems. 3. Pilots want to know if icing conditions lie ahead.
4. Establish aircraft operational limits in icing. 4. The stated icing information needs of a few pilots,
and of ALPA pilots as a union, are known.
Regulatory issues, weather forecasting, 5. Pilot reports are weak; spatial, temporal accuracy,
and traffic management and terminology standardization are lacking.
1. Establish aircraft icing performance limits. 6. It is known how pilots use weather radar and wind
2. Develop simple, scientific, standardized scale for shear alert systems, which may serve as analogs.
reporting icing potential. 7. It is known how information displayed affects
3. Develop regulatory operational concept and func- pilot reaction to terrain and wind shear alerts.
tional requirements. 8. An effective icing display needs to be designed.
4. Develop training standards and requirements. 9. Pilots want a simple icing display, but other infor-
5. Develop clear, unambiguous guidelines of when mation may be needed because of airframe and
severe icing conditions are entered. mission differences.
6. Consider status of remote-sensing information— 10. Avoid/exit strategies should be identified and
whether it should be an advisory or warning sys- tested—perhaps in a simulator.
tem. 11. Effects of aircraft performance on avoid/exit strat-
7. Create incentives for operators and manufactur- egies and on information needs by pilots are not
ers to install remote-sensing systems. known.
8. Expand FAR 25, Appendix C, to include freez- 12. Limitations of aircraft performance in avoid/es-
ing drizzle. cape strategies are known through experience.
9. Develop ground-based icing remote-sensing sys- 13. Pilots do not know aircraft limits in icing.
tems for terminal areas.

59
To Contents
Operators and manufacturers aircraft may require processing through a ground
1. Reliability, cost, size, power, and weight estimates station. This would be more difficult for military
for remote-sensing systems can be estimated, aircraft, which may not be able to benefit from
though poorly, from existing onboard systems. such services.
2. Room, power available, and weight allowable for
remote-sensing systems are unknown for most Experimental test beds
aircraft, except when newly delivered from manu- 1. Experimental test beds for training and develop-
facturers. After aircraft delivery, operators may ment of avoid/exit procedures could be developed
provide information. using flight simulators.
3. Simplicity, maintainability, and reliability of icing 2. Test beds for air traffic control and pilot training
remote-sensing systems are currently unknown are needed.
except for extrapolation of records from other
weather-avoidance sensors.
OPERATIONAL INFORMATION KNOWLEDGE
Regulatory issues, weather forecasting, Strengths
and traffic management 1. General operational concept of icing remote-
1. Operating limits of aircraft in icing conditions out- sensing system.
side of FAR 25, Appendix C, are not known. 2. General ideas of pilot information needs.
2. Information needed to create a standardized icing 3. Use of current onboard radar and wind shear alert
potential scale is available. as analogs.
3. An operational concept of a remote-sensing icing- 4. Use of remote-sensing systems in regulated air
avoidance system needs further development. space.
4. Functional requirements of remote-sensing icing- 5. Information display problems from radar and
avoidance system need development. wind shear alert systems.
5. Regulations currently do not address icing condi- 6. Weather information needed by meteorologists.
tions beyond conditions defined in FAR 25,
Appendix C. Weaknesses
6. There is little incentive for operators to place an 1. Lack of standardized, objective icing rating ter-
icing remote-sensing system on aircraft. minology.
7. There are no ground-based icing remote-sensing 2. Unknown limits of aircraft capabilities in icing
systems, though they are under development. conditions outside Appendix C.
8. There is no regulatory policy with regard to 3. Poor knowledge of how to avoid and escape icing.
remote-sensing icing-avoidance systems. How- 4. Lack of incentives for manufacturers and opera-
ever, onboard weather radar and wind shear alert tors to use remote-sensing systems.
systems are analogs for aircraft-based systems, 5. Definition of beyond FAR 25, Appendix C, con-
and ground-based wind shear alert systems are ditions.
analogs for ground-based icing-avoidance sys- 6. Undeveloped functional requirements (specifica-
tems. tions) of a remote-sensing icing detection system.
9. Remote-sensing icing-avoidance systems may 7. Use of remote-sensing systems in Free Flight.
complicate Free Flight and ATC operations in ter- 8. Development of effective display for icing con-
minal areas. ditions.
10. Weather forecasters do not have reliable indica- 9. Effects of aircraft performance on avoid/exit strat-
tions of icing conditions. Only pireps provide egies and information needs.
information, which is often inaccurate as to posi- 10. Location of effective experimental test beds for
tion, time, and intensity. pilots, air traffic controllers, and meteorologists.
11. Weather forecasters know the information they 11. Reliability, cost, size, power, and integration into
want to improve icing forecasts. airframe.
12. Integrating aircraft and ground information is not 12. Quality and type of information that will be pro-
well understood, though some work is being done vided to meteorologists and ATC from remote-
in the AGATE program. sensing systems.
13. Integration of in-situ, remote-sensing, and ground 13. Regulatory hurdles.
information may be difficult because of spatial 14. Integration of ground, in-situ, and remotely
and temporal inconsistencies. sensed information.
14. Reporting remotely sensed information to other 15. Training standards.

60
To Contents
 16. Establishing pilot needs. 3. Investigate concerns and limitations of airframe
17. Establishing remote-sensing system range, scan manufacturers.
rate, resolution, and warning time to meet pilots’ 4. Investigate operator needs and limitations.
needs. 5. Develop specifications for remote-sensing sys-
tem.
6. Investigate information dissemination prob-
GENERAL GOALS
lems—to ground, to other aircraft, information
1. Assess the needs of a greater variety of pilots. needed by ground, information aircraft can pro-
Study information needs of pilots in greater cross- vide (remote and in situ).
platform/cross-mission variety. For example, 7. Investigate feasibility of using simulation to de-
ALPA, military, helicopter vs. turboprop vs. jet velop pilot interfaces, and avoid-and-exit strate-
vs. general aviation. gies.
2. Investigate pilot decision-making process. Inves- 8. Investigate integration of remotely sensed and in-
tigate what information, to what distance, reso- situ measurements for assessing proximity to air-
lution, and detail and types of displays pilots need. craft operational limits.
9. Develop simulation and training aids.

61
To Contents
 APPENDIX B: SYNOPSIS OF SENSING NEEDS, STATE OF KNOWLEDGE,
STRENGTHS, AND WEAKNESSES

SENSING NEEDS
General
Characterization 1. Determine what characteristics of clouds are criti-
1. Good climatologies of icing conditions in all syn- cal to flight from flight tests, tunnel tests, and nu-
optic situations. merical models, in a spectrum of meteorological
2. Synoptic, continental, and global icing patterns conditions from a wide variety of aircraft.
to determine system utility. 2. A meteorology-based icing intensity standard.
3. Fully characterize supercooled large droplet cli- 3. Determine the critical technical capabilities for a
matology. remote-sensing system, such as range needed to
4. Characterize liquid-water content, drop size, and observe through most icing conditions, scanning
temperature conditions in all icing synoptic situa- rate/resolution, accuracy. Information needed is
tions. primarily meteorological, but also operational.
5. Determine the variability of cloud characteristics 4. Weather forecasters and numerical models need
(such as liquid-water content, droplet size, and downlinked objective, timely temperature, liquid-
temperature) within 3-D space (vertical and hori- water content, and drop-size information that is
zontal). accurate in position.
6. MVD or equivalent may not be acceptable
because they poorly represent “nonstandard” (i.e.,
STATE OF KNOWLEDGE
non-Gaussian) distributions of drop sizes
observed in clouds with drizzle drops. Characterization
7. Drop-size distributions are often not correctly 1. 3-D organization of icing patches is poorly under-
represented by current instrumentation, and ice stood with regard to usefulness of a remote-
crystals or drops can confuse sensing systems. sensing system.
Better instrumentation is needed. 2. Range and scale of liquid-water content, drop size,
8. More research flights specifically planned to and temperature variability is not well understood
measure information needed to characterize the in 3-D space, especially at the submesoscale, and
icing environment with regard to remote-sensing especially for supercooled large drops.
systems—with better instrumentation. 3. Climatology of supercooled large drops is poorly
9. Rework existing flight data. understood.
10. Measure cloud microphysical properties and 4. Submesoscale, continental, and global scales of
resulting ice on aircraft with in-situ sensing sys- aircraft icing are poorly understood.
tems to calibrate remote-sensing system dynam- 5. Drop-size distributions are not well characterized
ically. by current instrumentation—especially for larger
11. Improve characterization of freezing rain aloft. drops.
12. Characterize droplet temperature variations. 6. Ranges of liquid water content are reasonably well
13. Characterize mixed-phase clouds with tempera- understood, but not in relation to drop size and
ture. temperature.
7. Synoptic scale of icing is reasonably well under-
In-situ instrumentation stood.
1. In-situ instruments with better dynamic range and 8. General climatologies of aircraft icing exist and
sensitivity. are reasonably well understood.
2. Small, accurate, and inexpensive in-situ technol- 9. Research flight information exists that can be reanal-
ogies. yzed at little cost for better characterization.
3. Improved SLD measurement instrumentation. 10. Characterization of freezing rain and freezing driz-
zle aloft poorly understood.
11. Characterization of temperature poorly understood.

63
To Contents
 In-situ instrumentation Weaknesses
1. Accuracy and dynamic range of in-situ instru- 1. Characterization of the 3-D scale of cloud physi-
ments need improvement, especially for SLD. cal properties at the submesoscale, continental,
2. Small, accurate, and inexpensive in-situ instru- and global scales, including temperature, liquid-
ments are unavailable. water content, and drop sizes.
3. 3-D remote-sensing resolution, range, and angu- 2. Characterization of SLD.
lar scanning area need specification. 3. Accurate, reliable, inexpensive in-situ instrumen-
4. Characteristics of clouds critical to flight in icing tation.
on many airframes in many conditions are poorly 4. There is no objective, weather-based icing index.
understood. 5. Further testing of airfoils needed under a variety
of weather and operating conditions.
General 6. Specifications must be developed for remote-
1. We can currently only speculate about the ideal sensing systems.
remote sensor scanning range needed—distance 7. Characterization of test beds needed.
and angular, accuracy, and 3-D sensing resolu- 8. FAR 25, Appendix C, must be extended.
tion needed.
2. There is no index of icing conditions by intensity General goals
as a function of weather condition alone, inde- 1. Define conditions within which clouds are mixed
pendent of aircraft type. phase, that is, have ice crystals, because mixed-
3. Characteristics of clouds critical to flight from phase situations indicate that supercooled liquid
flight tests, tunnel tests, and numerical models, water exists.
in a spectrum of meteorological conditions from 2. Characterize 3-D spatial scales of icing by icing
wide variety of aircraft, are not available. potential and microphysical properties.
4. There is no objective system for reporting icing 3. Assess continental-scale and global-scale icing
potential independent of aircraft type. patterns, frequencies, and intensities to determine
5. There is currently no objective system of report- needs for commercial aviation and military avia-
ing areas of icing or no icing that is accurate in tion in potential operational theaters.
intensity, position, and time. 4. Characterize SLD conditions with regard to fre-
6. Weather forecasters and numerical models cur- quency, range of conditions, synoptic situations,
rently do not have available objective, timely tem- and continental and global patterns.
perature, liquid-water content, and drop-size 5. Reanalyze old flight data with improved instru-
information that is accurate in position. ment-correction algorithms.
7. A meteorology-based icing intensity standard is 6. Fly new research flights to develop specifications
needed. for remote-sensing systems.
8. Need to expand FAR 25, Appendix C to include 7. Develop small, turnkey, inexpensive in-situ
SLD. instruments for aircraft.
8. Investigate feasibility of dynamically calibrating
SENSING REQUIREMENTS KNOWLEDGE remote-sensing systems from in-situ sensors.
Strengths 9. Develop sensing specifications for remote-sensing
1. General climatologies of icing are available. systems.
2. Information desired to be downlinked to forecast- 10. Develop meteorologically indexed icing intensi-
ers is known. ty scale.
3. There is considerable data and confidence in FAR 11. Assess synoptic meteorological conditions within
25, Appendix C, icing-condition characterization. which remote sensing would be most practical.
4. General research instrumentation is generally ade- 12. Identify appropriate test beds.
quate.
5. Appropriate test platforms are available.

64
To Contents
 APPENDIX C: SYNOPSIS OF SENSOR TECHNOLOGY NEEDS,
STATE OF KNOWLEDGE, STRENGTHS, AND WEAKNESSES

SENSOR TECHNOLOGY NEEDS

wavelengths allows liquid-water content to be
1. Establish sensing needs for phenomena to be retrieved.
sensed, including range and resolution. 8. Millimeter-wavelength radars most suited for
2. Develop inversion theory to determine most cloud-water sensing.
appropriate technologies for sensing 3-D spatial 9. Wider ranges of drop sizes in precipitating clouds
structure of liquid-water content, drop-size spec- increase the need for a multiple-wavelength (more
tra, and temperature ahead of aircraft. than two) radar system.
3. Develop prototype hardware for testing theory 10. Differential attenuation techniques may require
for each selected component of remote-sensing droplet temperature for accurate assessment of
system. liquid-water content.
4. Select test bed for proving concepts developed 11. Gossett and Sauvageot (1992) theoretically
in theory, develop test plans, and test. demonstrated that 3.2-cm (X) and 0.87-cm (Ka)
5. Develop airborne prototype system component bands are the best for detecting water in clouds,
prototypes. but that other wavelength pairs are possible,
6. Test airborne prototypes on flight platform and depending upon desired range and the existence
verify capabilities. of hydrometeors.
7. Develop integrated system comprising all compo- 12. Millimeter-wavelength radars are suited to air-
nents, with processing and display system com- borne cloud studies because of their small size,
patible with onboard systems and protocol. low ground-clutter susceptibility, high resolution,
8. Test integrated prototype system on civilian and and sensitivity to small hydrometeors.
military flight platforms. 13. It may not be possible to uniquely define cloud
9. Certify system. drop-size distributions with radar. Only parame-
ters of a distribution may be available.
STATE OF KNOWLEDGE
Passive radiometers
Radar 1. Operationally used to measure zenith and nadir
1. Ground (vertical scanning) and airborne (hori- temperature profiles, with thermal and spatial reso-
zontal scanning) systems are technically possible. lution decreasing with distance from radiometer.
2. Range-resolved liquid water measurements have 2. Integrated liquid-water path sensed in vertical at
been acquired from clouds. 31.6 GHz with scanning possible.
3. Polarization techniques can be used to detect ice 3. Integrated liquid water might be sensed in hori-
vs. water. zontal to provide integrated water that an aircraft
4. Doppler techniques to detect droplet sizes may could intercept.
not be possible with horizontally scanning sys- 4. Modeling and experiments needed at test beds.
tems. 5. Precipitation can cause difficulties—modeling is
5. Longer wavelengths (lower frequencies) cannot needed.
detect smaller droplets but have longer range and 6. Cloud phase may be possible with polarimetry.
can operate within Rayleigh regime to larger drop 7. Scanning liquid-water radiometer under devel-
sizes. Useful for detecting precipitation. opment provides range resolution.
6. Shorter wavelengths (higher frequencies) detect 8. Passive technology advantage for cost, size,
smaller drop sizes but have shorter range and weight, power, general aviation, and military ap-
cannot operate within Rayleigh regime in larger plications.
drops. 9. Model with RADTRAN or its successors.
7. Differential attenuation of two or more radar 10. Radiometer scanning is slow.

65
To Contents
Lidar 4. A scanning liquid-water radiometer under devel-
1. Scatter proportional to number density of drops. opment provides range resolution.
2. High spatial resolution. 5. Passive technology has an advantage for cost,
3. Multiple-field-of-view lidars can indicate effec- size, weight, power, general aviation, and mili-
tive drop diameter, but distribution must be assumed. tary applications.
4. Liquid-water content can be measured. 6. Model with RADTRAN or its successors.
5. Can estimate relative amount of ice crystals from 7. Radiometers can be small and use little power.
amount of polarization.
6. Current MFOV lidars not eye-safe, but could be Lidar
at 1.54 or 2.0 µm with 1.5- to 3-m resolution. 1. Pulses can be only 2 to 3 m long, yielding very
7. Could be placed on an aircraft. high spatial resolution.
8. Detect to clouds through clear air, but cloud extinc- 2. Multiple field-of-view lidars can indicate effec-
tion allows only few hundred meters penetration. tive drop diameter and liquid water content.
3. Can retrieve relative amount of ice crystals from
Temperature measurements amount of polarization.
1. RASS profiles temperature to radiosonde accura- 4. Multiple field-of-view lidars could be eye safe at
cy to over 3.5 km, even inversions. 1.54 or 2.0 mm with 1.5- to 3-m resolution.
2. RASS has not been used on moving vehicles or in 5. Five-watt power demand, 6- to 8-in. receiving lens,
the horizontal. 100 pulses s–1, 1-m3 volume, could be placed on
3. Pitch, roll, and yaw and aircraft speeds prevent an aircraft.
RASS use on aircraft. 6. Lidar currently used for operational onboard wind
4. RASS acoustic source could be aircraft engine shear alert.
noise, but turbulence and relative wind could cause 7. Small and inexpensive.
loss of signal for up to 1 min depending upon air- 8. Rapid scanning possible.
craft speed and heading.
5. Radiometers may sense temperature in horizon- Temperature measurement
tal. Might try tunable system operating in the vicin- 1. RASS sounds temperature with radiosonde accu-
ity of the oxygen absorption band, with tuning racy, even through inversions.
providing range resolution. 2. Radiometers are used operationally to create ver-
6. Radiometers provide lower-resolution vertical tical temperature profiles.
temperature profiles. 3. Radiometers may sense temperature in horizon-
7. Lidar not applicable. tal. Might try tunable system operating in the
vicinity of the oxygen absorption band, with tun-
ing providing range resolution.
TECHNOLOGY KNOWLEDGE STRENGTHS
Radar
TECHNOLOGY KNOWLEDGE WEAKNESSES
1. Ground (vertical scanning) and airborne (horizon-
tal scanning) systems are possible. Radar
2. Successful attempts have been made to acquire 1. Doppler techniques to detect droplet sizes not
range-resolved liquid water from clouds. possible with horizontally scanning systems.
3. Polarization techniques can be used to detect ice 2. Wider ranges of drop sizes in precipitating clouds,
vs. water. for example, increases the need for a multiple-
4. Millimeter-wavelength radars are suited to air- wavelength (more than two) radar system.
borne cloud studies because of their small size, 3. Dual-wavelength differential attenuation tech-
low ground-clutter susceptibility, high resolution, niques require temperature for accurate liquid-
and sensitivity to small hydrometeors. water content.
5. Radar scans rapidly. 4. Gossett and Sauvageot (1992) theoretically dem-
onstrated that X and Ka bands are the best for
Passive radiometers detecting water in clouds, but that other wave-
1. Operationally used to measure zenith and nadir length pairs are possible. Modeling is needed.
temperature profiles, and possibility in horizontal. 5. It may not be possible to uniquely define cloud
2. Integrated liquid-water path sensed in vertical with drop-size distributions with radar.
scanning possible. 6. Dual-frequency radar at X and Ka bands cannot
3. Cloud phase may be possible with polarimetry. detect small drop diameters to a long range, nor

66
To Contents
 liquid-water content smaller than 0.2 g m–3 (Mart- Temperature measurement
ner at al. 1993) 1. RASS is operational from the ground and has not
7. W-band radar (95 GHz) may be suitable but with been used on vehicles or in the horizontal.
short range and great attenuation by large liquid- 2. Pitch, roll, and yaw and aircraft speed prevent
water content, large drops, and water vapor. Mod- RASS use on aircraft.
eling is needed. 3. Radiometers cannot sense temperature in hori-
zontal using traditional inversion methods.
Passive radiometers 4. Radiometers scan slowly.
1. Liquid water might be sensed in horizontal to pro- 5. Lidar not applicable.
vide integrated water that an aircraft might inter-
cept. General goals
2. Temperature cannot be sensed in horizontal using 1. Perform feasibility studies of ability of technolo-
conventional inversion methods. gies to provide liquid-water content, drop size,
3. Modeling and experiments needed at test beds. and temperature information needed.
4. Precipitation can cause difficulties at some wave- 2. Develop methods of assessing feasibility studies
lengths; modeling is needed. through measurements with hardware.
5. Cloud phase cannot be detected with polarimetry. 3. Develop prototype technologies for field testing.

Lidar
1. High extinction in clouds.
2. Eye safety.

67
To Contents
 Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704-0188
Public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and maintaining the
data needed, and completing and reviewing this collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including suggestions for reducing
this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704-0188), 1215 Jefferson Davis Highway, Suite 1204, Arlington, VA 22202-4302.
Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection of information if it does not display a currently valid
OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD-MM-YY) 2. REPORT TYPE 3. DATES COVERED (From - To)
January 2000 Monograph
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER

5b. GRANT NUMBER

Remote Sensing of In-Flight Icing Conditions:
Operational, Meteorological, and Technological Considerations
5c. PROGRAM ELEMENT NUMBER

6. AUTHOR(S) 5d. PROJECT NUMBER

Charles C. Ryerson
5e. TASK NUMBER

5f. WORK UNIT NUMBER

7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT

NUMBER
Engineer Research and Development Center
U.S. Army Cold Regions Research and Engineering Laboratory
72 Lyme Road ERDC/CRREL M-00-1
Hanover, New Hampshire 03755-1290

9. SPONSORING/MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR / MONITOR’S ACRONYM(S)

National Aeronautics and Space Administration

11. SPONSOR / MONITOR’S REPORT
Washiington, D.C. NUMBER(S)

12. DISTRIBUTION / AVAILABILITY STATEMENT

Approved for public release; distribution is unlimited.

Available from NTIS, Springfield, Virginia 22161.

13. SUPPLEMENTARY NOTES

14. ABSTRACT
Remote-sensing systems that map aircraft icing conditions in the flight path from airports or aircraft would allow icing to be avoided and
exited. Icing remote-sensing system development requires consideration of the operational environment, the meteorological environment,
and the technology available.
Operationally, pilots need unambiguous cockpit icing displays for risk management decision-making. Human factors, aircraft integra-
tion, integration of remotely sensed icing information into the weather system infrastructures, and avoid-and-exit issues need resolution.
Cost, maintenance, power, weight, and space concern manufacturers, operators, and regulators.
An icing remote-sensing system detects cloud and precipitation liquid water, drop size, and temperature. An algorithm is needed to
convert these conditions into icing potential estimates for cockpit display. Specification development requires that magnitudes of cloud
microphysical conditions and their spatial and temporal variability be understood at multiple scales.
The core of an icing remote-sensing system is the technology that senses icing microphysical conditions. Radar and microwave radiom-
eters penetrate clouds and can estimate liquid water and drop size. Retrieval development is needed; differential attenuation and neural
network assessment of multiple-band radar returns are most promising to date. Airport-based radar or radiometers are the most viable near-
term technologies. A radiometer that profiles cloud liquid water, and experimental techniques to use radiometers horizontally, are promising.
15. SUBJECT TERMS
Aircraft icing Drop size In-flight icing Lidar NASA Radar Temperatures
DoD FAA Human factors Liquid water content Operations Remote sensing
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
ABSTRACT OF PAGES

a. REPORT b. ABSTRACT c. THIS PAGE 19b. TELEPHONE NUMBER (include area code)

U U U U 75

To Contents Standard Form 298 (Rev. 8-98)

Prescribed by ANSI Std. 239.18
14. ABSTRACT (cont’d)

The most critical operational research needs are to assess cockpit and aircraft system integration, devel-
op avoid-and-exit protocols, assess human factors, and integrate remote-sensing information into weather
and air traffic control infrastructures. Improved spatial characterization of cloud and precipitation liquid-
water content, drop-size spectra, and temperature are needed, as well as an algorithm to convert sensed
conditions into a measure of icing potential. Technology development also requires refinement of inversion
techniques. These goals can be accomplished with collaboration among federal agencies including NASA,
the FAA, the National Center for Atmospheric Research, NOAA, and the Department of Defense. This
report reviews operational, meteorological, and technological considerations in developing the capability to
remotely map in-flight icing conditions from the ground and from the air.

To Contents