Global Warming: Limitations of General Circulation Models and Costs of
Modeling Efforts (Letter Report, 07/13/95, GAO/RCED-95-164).

Pursuant to congressional request, GAO reviewed the accuracy of general
circulation models (GCM) in forecasting global warming trends, focusing
on the: (1) factors limiting the accuracy of GCM estimates of future
climatic changes; and (2) federal expenditures for GCM for fiscal years
(FY) 1992 through 1994.

GAO found that: (1) although GCM have improved their ability to predict
future climatic changes over the last decade, their estimates are still
limited by their incomplete or inaccurate representations of
climate-affecting processes and by insufficient computer power; (2)
scientists do not fully understand how the climate system responds to
potentially important physical, chemical, and biological processes; (3)
the lack of computer power requires scientists to use simplified
assumptions and structures that increase the uncertainty of the models'
predictions; (4) scientists are conducting research to overcome the
limitations of the computer models; and (5) five federal agencies spent
about $122.6 million for various global modeling projects, which
represented about 3 percent of the global change research program's
budget for FY 1992 through 1994.

--------------------------- Indexing Terms -----------------------------

 REPORTNUM:  RCED-95-164
     TITLE:  Global Warming: Limitations of General Circulation Models 
             and Costs of Modeling Efforts
      DATE:  07/13/95
   SUBJECT:  Scientific research
             Computer modeling
             Budget outlays
             Atmospheric research
             Air pollution control
             Climate statistics
             Oceanographic research
             Data integrity
IDENTIFIER:  U.S. Global Change Research Program
             ICSU/WMO World Climate Research Programme
             Global Climate Observing System
             
**************************************************************************
* This file contains an ASCII representation of the text of a GAO        *
* report.  Delineations within the text indicating chapter titles,       *
* headings, and bullets are preserved.  Major divisions and subdivisions *
* of the text, such as Chapters, Sections, and Appendixes, are           *
* identified by double and single lines.  The numbers on the right end   *
* of these lines indicate the position of each of the subsections in the *
* document outline.  These numbers do NOT correspond with the page       *
* numbers of the printed product.                                        *
*                                                                        *
* No attempt has been made to display graphic images, although figure    *
* captions are reproduced. Tables are included, but may not resemble     *
* those in the printed version.                                          *
*                                                                        *
* A printed copy of this report may be obtained from the GAO Document    *
* Distribution Facility by calling (202) 512-6000, by faxing your        *
* request to (301) 258-4066, or by writing to P.O. Box 6015,             *
* Gaithersburg, MD 20884-6015. We are unable to accept electronic orders *
* for printed documents at this time.                                    *
**************************************************************************


Cover
================================================================ COVER


Report to the Ranking Minority Member, Committee on Commerce,
House of Representatives

July 1995

GLOBAL WARMING - LIMITATIONS OF
GENERAL CIRCULATION MODELS AND
COSTS OF MODELING EFFORTS

GAO/RCED-95-164

Global Warming


Abbreviations
=============================================================== ABBREV

  DOE - Department of Energy
  EPA - Environmental Protection Agency
  GAO - General Accounting Office
  GCM - general circulation model
  NASA - National Aeronautics and Space Administration
  NOAA - National Oceanic and Atmospheric Administration
  NSF - National Science Foundation
  USGCRP - U.S.  Global Change Research Program

Letter
=============================================================== LETTER


B-261119

July 13, 1995

The Honorable John D.  Dingell
Ranking Minority Member
Committee on Commerce
House of Representatives

Dear Mr.  Dingell: 

Increasingly, emissions of heat-trapping greenhouse gases from energy
production, industry, transportation, agriculture, and other human
activities are being concentrated in the earth's atmosphere.  During
the Forum on Global Climate Change Modeling, held in October 1994,
scientists agreed that the buildup of these gases is creating an
enhanced greenhouse effect that will lead to global warming. 
Specifically, they estimate that the surface temperature of the earth
will rise by 1 to 4 degrees Fahrenheit from 1990 to 2050 if emissions
continue to grow without restriction.  An increase of such magnitude
could begin to melt the polar ice caps and, in turn, raise the sea
level and alter weather patterns (particularly patterns of
precipitation).  These changes could bring about the extinction of
certain plant and animal species as they shift to adapt to a warmer
climate. 

Anticipating that costly actions may be needed to avoid or respond to
global warming, policymakers are seeking accurate information about
possible future climatic changes.  The most highly developed tools
now available to project climatic changes are complex computer models
called general circulation models (GCM).  These models, whose
development is supported through a coordinated effort of the U.S. 
Global Change Research Program and five federal agencies, have become
more accurate during the last few decades, but important
uncertainties still limit their predictive capabilities.  Given the
pivotal role that GCMs could play in shaping the response of U.S. 
policymakers to possible future climatic changes, you asked us to
identify the (1) factors limiting the accuracy of GCMs' estimates of
future climatic changes and (2) federal expenditures for GCMs from
fiscal years 1992 through 1994. 


   RESULTS IN BRIEF
------------------------------------------------------------ Letter :1

General circulation models are considerably better now than they were
a decade ago at predicting future climatic changes.  Nevertheless,
the accuracy of their estimates is still limited, primarily by their
incomplete or inaccurate representation of the processes affecting
climate and by insufficient computer power.  For general circulation
models, as for other computer models, the quality of the output
depends upon the quality of the input--the models are only as good as
the data and the scientists' understanding of how the climate system
works.  When scientists do not fully understand how the climate
system responds to potentially important physical, chemical, and
biological processes, they can omit or poorly represent the operation
of these processes in the models.  Such omissions or poor
representations produce inaccuracies in the models' projections of
future climatic conditions.  Limitations on computer power have led
to the use of (1) simplifying assumptions that increase the
uncertainty of the models' predictions and (2) simplifying structures
that preclude the incorporation of the detailed data needed to
accurately project regional and local conditions.  Efforts to
overcome these limitations and improve the accuracy of the models'
estimates are ongoing. 

For fiscal years 1992 through 1994, federal agencies reported
spending an estimated $122.6 million to fund various projects for
modeling global climate change; these expenditures represented
approximately 3 percent of the U.S.  Global Change Research Program's
multiagency budget over this period.\1 Five U.S.  agencies operate
and/or fund such projects:  the Department of Energy (DOE), the
National Aeronautics and Space Administration (NASA), the National
Science Foundation (NSF), the National Oceanic and Atmospheric
Administration (NOAA), and the Environmental Protection Agency (EPA). 
DOE and NASA operated the largest modeling programs and reported
spending approximately 64 percent of the total federal funds.  The
modeling projects were conducted under contracts and grants with
various research laboratories and universities throughout the United
States. 


--------------------
\1 Established in 1990, this program combines and coordinates the
global change research and policy development interests of all U.S. 
departments and agencies. 


   BACKGROUND
------------------------------------------------------------ Letter :2

The sun provides the energy that determines the climate and weather. 
Solar radiation passes through space and is largely absorbed by
components of the global climate system (the atmosphere, oceans, and
land, as well as the biosphere, which includes all living things);
the remaining radiation is reflected.  The solar radiation absorbed
by the earth's surface is released as infrared radiation.  Some of
this radiation passes back through the atmosphere, and some is
absorbed in the atmosphere by the molecules of gas--principally water
vapor, carbon dioxide, methane, and chlorofluorocarbons--known
collectively as greenhouse gases.  These gas molecules act as a
partial thermal blanket, trapping much of the heat energy and
redirecting it to the earth's surface and lower atmosphere.  This
naturally occurring process, called the greenhouse effect (see fig. 
1), helps to maintain the earth's temperature at an average of
approximately 60 degrees Fahrenheit.\2

   Figure 1:  The Greenhouse
   Effect

   (See figure in printed
   edition.)

   Source:  U.S.  Department of
   State, 1992.

   (See figure in printed
   edition.)

Additional atmospheric warming--called the enhanced greenhouse effect
or global warming--appears to be associated with human activities. 
During the past century, as industry, agriculture, and transportation
have grown, so, too, have atmospheric concentrations of heat-trapping
greenhouse gases (see app.  I).  At the same time, the earth has
gotten warmer, according to historical data.  Recorded temperatures
for the period from 1860 through 1993 show a warming trend that
generally coincides with the increased use of fossil fuels during the
Industrial Revolution--and, hence, with the increased emission of
greenhouse gases. 

During the past 50 to 100 years, volcanic eruptions have combined
with the increased combustion of fossil fuels and emission of
greenhouse gases to increase the concentration of aerosols\3 in the
lower atmosphere.  Scientists believe that because these aerosols
deflect sunlight, they have partially offset the effects of global
warming.  As a result, scientists surmise, temperatures have not
reached the levels projected by GCMs, which do not include the
aerosols' effects. 

To help understand the global climate system's response to emissions
of greenhouse gases, scientists use three types of GCMs: 
atmospheric, oceanic, and coupled.  In general, atmospheric GCMs
predict the physical behavior of the atmosphere.  Oceanic GCMs
represent the physics of the ocean.  Coupled GCMs, which scientists
regard as the most advanced of the models, physically join
atmospheric and oceanic GCMs and treat the evolution of the climate
in both domains.  To improve predictions of the future climate,
modelers are also striving to couple, and to some degree have
coupled, (1) the land surface to the atmosphere and (2) the Antarctic
sea ice to both the ocean and the atmosphere. 

All types of GCMs process vast quantities of data on variables
affecting climate.  Using complex mathematical equations to represent
the actions and interactions of these variables, the GCMs process the
data to project patterns of climatic conditions.  (App.  II shows how
a coupled GCM works.) To test the accuracy of a model's projections,
modelers run the model with their best estimates of historical
climatic data and compare the resulting projections with records of
actual climatic conditions for the period being modeled.  Modelers
assume that if the model can accurately simulate actual climatic
conditions for prior periods, then it can be used to accurately
project future climatic conditions. 

Although the earth's gradual warming since the mid-19th century is
generally consistent with GCMs' estimates of the effects of
greenhouse gases (when adjusted for the effects of aerosols),
scientists have not been able to attribute the warming conclusively
to the enhanced greenhouse effect or to quantify its effects. 
Specifically, they have not been able to uniquely and quantitatively
distinguish the effects of higher concentrations of greenhouse gases
from the effects of other factors that can change the climate.  Such
factors include natural fluctuations in the global climate system,
increases in atmospheric ozone, air pollution, and aerosols emitted
into the atmosphere from volcanic eruptions.  Until more is known
about the relative influence of these various factors on the earth's
climate, GCMs' estimates of global warming will remain uncertain. 


--------------------
\2 All temperatures are reported in degrees Fahrenheit. 

\3 Aerosols are gases that bear other substances. 


   FACTORS LIMITING THE ACCURACY
   OF MODELS' ESTIMATES
------------------------------------------------------------ Letter :3

Over the last decade, GCMs have accurately simulated many elements of
the observed climate, providing useful indications of some future
climatic conditions.  For example, atmospheric models have
demonstrated some skill in portraying aspects of atmospheric
variability, such as the surface temperature of the sea.  Oceanic
models have also simulated the general circulation of the ocean,
including the patterns of the principal currents.  Coupled models,
though still prone to small-scale errors, have simulated the current
climate on a large scale as well as portrayed large-scale atmospheric
and oceanic structures. 

This progress notwithstanding, the models remain limited in their
ability to estimate, with desired accuracy, the magnitude, timing,
and regional distribution of future climatic changes.  These
limitations stem from scientists' imperfect understanding of the
global climate system and computers' insufficient capacity to perform
more detailed simulations.  More specifically, the accuracy of the
models' predictions is limited by (1) incomplete or inadequate
representations of the processes affecting climate and (2)
insufficient computer power.  Research is being conducted to overcome
both the scientific and the technical limitations affecting the
accuracy of GCMs' estimates. 


      INCOMPLETE OR INADEQUATE
      REPRESENTATION OF PROCESSES
      AFFECTING CLIMATE
---------------------------------------------------------- Letter :3.1

According to the U.S.  Global Change Research Program, most GCMs
include the most important processes that affect climate, such as
radiation, convection, and land surface exchanges.  However, some
models do not include or fully incorporate some processes, and even
the most advanced models do not adequately represent the interactions
of some processes.  None of the models fully incorporates certain
components of the global climate system, called feedbacks or feedback
mechanisms, and none adequately represents the interactions of these
mechanisms with greenhouse gases, called feedback processes. 


         SOME PROCESSES NOT
         INCLUDED OR FULLY
         INCORPORATED IN SOME
         MODELS
-------------------------------------------------------- Letter :3.1.1

Atmospheric and oceanic GCMs include fewer processes than coupled
GCMs, and their simulations are, therefore, more limited and, in some
cases, less accurate.  Atmospheric models do not fully portray the
influence of oceanic pressures (currents) and fluctuations in
climate, while oceanic models do not fully account for the effects of
atmospheric surface winds.  The omission or incomplete incorporation
of some processes may introduce errors into these models'
projections.  For example, atmospheric models tested in 1991 produced
systematic errors in their projections of sea level pressure,
temperature, zonal wind, and precipitation.  Compared with
atmospheric and oceanic GCMs, coupled GCMs include more processes and
interactions at the ocean-atmosphere interface, but even they do not
include critical biospheric and chemical interactions with the
atmosphere.  The U.S.  Global Change Research Program is supporting
efforts by modeling groups to include more complete sets of processes
in their models and to identify systematic errors in the models. 


         INADEQUATE
         REPRESENTATIONS OF
         INTERACTIONS AMONG
         VARIABLES
-------------------------------------------------------- Letter :3.1.2

Although coupled GCMs produce more comprehensive simulations of
current climatic conditions than either atmospheric or oceanic GCMs,
their simulations still differ from actual conditions.  Modelers
believe that these models are impaired by a condition known as
climatic drift, which results from imbalances in the models' analyses
of heat and moisture variables.  These imbalances cause the models'
estimates of temperature and precipitation to deviate from actual
conditions.  For example, in an experiment conducted by the National
Center for Atmospheric Research in 1988, the models estimated
wintertime ocean temperatures that were 7 degrees warmer than
observed temperatures for the icebound region of Antarctica and 9
degrees colder than observed temperatures for the tropics. 

Modelers either accept climatic drift or try to correct its effects
by inserting adjustments, called flux adjustments.  Because flux
adjustments artificially improve the models' performance, their use
is controversial.  Scientists believe that an increased understanding
of the interactions between atmospheric and oceanic variables--and,
hence, a more accurate mathematical representation of these
interactions--may eventually remove the need for flux adjustments. 
Reducing the need for flux adjustments is an objective of coupled
model research. 


         INADEQUATE REPRESENTATION
         OF, OR ACCOUNTING FOR,
         FEEDBACK MECHANISMS
-------------------------------------------------------- Letter :3.1.3

GCMs include many of the most important feedback mechanisms, such as
vegetation, water vapor, ice cover, clouds, and the ocean.  However,
the models do not yet adequately represent the interactions of these
mechanisms with greenhouse gases.  Such interactions can amplify,
dampen, or stabilize the warming produced by increased concentrations
of greenhouse gases. 

The influence of feedback mechanisms on climate is likely to increase
as concentrations of greenhouse gases increase; however, modelers do
not fully understand the effects of these mechanisms and have not
learned how to represent them with sufficient accuracy in models. 
Although they have clarified the role of water vapor and improved
their ability to model its effects, they are still seeking to
understand and accurately model the effects of clouds, which have the
greatest potential of all the feedback mechanisms to amplify or
moderate global warming.  Recent studies have shown that different
schemes for modeling cloud formation processes can lead to
substantially different projections of the earth's temperature.  In
1989, for instance, two simulations, which varied only in their
treatment of the cloud feedback process, produced estimates of the
increase in the earth's annual average surface temperature of 4.9 and
9.4 degrees, respectively. 


      INSUFFICIENT COMPUTER POWER
---------------------------------------------------------- Letter :3.2

Insufficient computer power affects the accuracy of GCMs' estimates
because even the most powerful computers are limited in their ability
to store and analyze the vast quantity of data required to accurately
simulate changes in the global climate.  Modelers have tried to
overcome these limitations by introducing assumptions into their
models that deliberately oversimplify some operations in order to
free the GCMs' capacity and time for other, more critical operations. 
For example, modelers have assumed that the ocean was not warmed by
emissions of greenhouse gases before 1985.  Although this assumption
gains capacity for the GCMs, it introduces an error, called the cold
start error, that increases the uncertainty of the GCMs' predictions. 
Another oversimplification, the division of the earth into relatively
large grids for analytical purposes, prevents the GCMs from
accurately predicting regional changes in climate. 


         COLD START ERROR
-------------------------------------------------------- Letter :3.2.1

Simulations by coupled GCMs that are calculated on the assumption
that the ocean was not warmed by increased emissions of greenhouse
gases before 1985 do not adequately account for the ocean's reduced
capacity to absorb these emissions.  In fact, the ocean will reach
its capacity for absorbing these emissions sooner--possibly decades
sooner-- than the coupled GCMs calculate.  It will then deflect more
of the heat-trapping emissions to the atmosphere, thereby enhancing
global warming more rapidly than the models predict. 

While recognizing that the cold start error artificially delays the
onset of global warming in GCMs' predictions, scientists do not know
by how much or by how long it distorts the predictions.  Overall,
they believe that it causes the models to underestimate the change in
temperature that will result from the emissions.  Modelers have shown
that the cold start error can cause projections of the earth's
average annual temperature to differ by as much as 0.7 degrees after
50 years. 

According to scientists, an extraordinary commitment of computer time
would be required to project the timing of future temperature changes
more accurately.  Completing the number of computer runs needed to
arrive at more precise timing projections could take many months even
on a state-of-the-art supercomputer. 


         INABILITY TO PROJECT
         REGIONAL CHANGES IN
         CLIMATE
-------------------------------------------------------- Letter :3.2.2

Still another limitation affecting the accuracy of GCMs' estimates is
the relatively large size of the grids into which the models divide
the earth.  These grids typically cover an area about the size of
South Carolina.  Although their use enables GCMs to depict
larger-scale regional effects in relatively large, homogeneous
regions, it does not allow modelers to incorporate detailed regional
features.  Consequently, the use of large grids prevents the models
from accurately forecasting climatic changes for smaller, less
homogeneous regions.  The use of smaller grids would permit the
incorporation of more detailed features that could be used to project
regional changes more precisely.  However, models using smaller grids
would take longer to run. 

Each grid contains a single value for each variable for the entire
area represented.  Today's grids are smaller than those we described
in our 1990 report on global warming,\4

but they are not yet small enough to produce the information
policymakers and planners need to develop strategies for adapting to
regional changes. 

Researchers believe that the combination of greater computer power,
which would permit the use of smaller grids, and greater
understanding of cloud formation processes, which would permit the
incorporation of this important but often excluded feedback
mechanism, would produce more accurate projections of regional
climatic changes. 


--------------------
\4 Global Warming:  Emission Reductions Possible as Scientific
Uncertainties Are Resolved (GAO/RCED-90-58, Sept.  28, 1990). 


      IMPROVING GCMS' ESTIMATES
---------------------------------------------------------- Letter :3.3

To improve the accuracy of GCMs' estimates, scientists are developing
models that incorporate more of the processes affecting the climate
system (particularly cloud formation processes) and better reflect
interactions among various components of the climate system,
including interactions between or among the ocean and the atmosphere;
the land surface, the biosphere, and the atmosphere; and the
cryosphere (frozen regions), the ocean, and the atmosphere.  They are
also developing larger and faster computers that can manipulate data
for longer periods of time and smaller grids.  In addition, they are
collecting more data and conducting more research on the processes
affecting climate and improving the international exchange of such
data.  Various international programs, such as the World Climate
Research Programme\5 and the Global Climate Observing System,\6
currently have efforts under way to address these actions. 

In commenting on a draft of this report, the Director of the Office
of the U.S.  Global Change Research Program and agency officials
stated that the program has several ongoing efforts to address the
limitations of GCMs discussed in this report.  For example to address
the models' inadequate representation of processes affecting the
climate, the program is devoting approximately 30 percent of its $1.8
billion budget for fiscal year 1995 to conduct research aimed at
improving scientific understanding of these processes.  In addition,
to address the need for increased computer power, the program has,
through NSF, established a dedicated computing facility for modeling
the climate system, known as the Climate Simulation Laboratory, in
cooperation with the National Center for Atmospheric Research.  This
facility will provide state-of-the-art computer resources and data
storage systems for use in major modeling research simulations.  The
goals and funding for the U.S.  Global Change Research Program's
fiscal year 1995 research programs are summarized in appendix III. 
Further information on the program's efforts to reduce the
uncertainties of GCMs' projections appears in a letter from the
Subcommittee on Global Change Research, which is reproduced in
appendix VI. 


--------------------
\5 This program was established as a joint undertaking of the
International Council of Scientific Unions and the World
Meteorological Organization, a United Nations agency, to foster an
improved understanding of the climate's variability and prediction. 

\6 This international organization was established in 1992 to monitor
climatic changes and obtain data for application to national economic
development. 


   FEDERAL EXPENDITURES FOR GCMS
------------------------------------------------------------ Letter :4

Five federal agencies reported spending an estimated $122.6 million
during fiscal years 1992 through 1994 to fund modeling activities to
improve predictions of the future climate.\7 As shown in table 2, the
agencies reported spending approximately $36.9 million, $40.5
million, and $45.3 million for these projects in fiscal years 1992,
1993, and 1994, respectively.  Of the five agencies, DOE had the
largest climate change modeling program, representing about 36
percent of the total cost for all 3 years. 



                           Table 1
           
              Estimated Costs of Modeling Global
             Climate Change, Fiscal Years 1992-94

                    (Dollars in millions)

Agency          1992      1993      1994     Total   Percent
----------  --------  --------  --------  --------  --------
DOE            $14.1     $15.2     $15.2     $44.5       36%
NASA            10.5      11.8      12.6      34.9       28%
NSF              7.4       7.5       9.2      24.1       20%
NOAA             4.1       4.1       6.0      14.3       12%
EPA               .7       1.8       2.3       4.8        4%
Total\a        $36.9     $40.5     $45.3    $122.6      100%
------------------------------------------------------------
Note:  The above costs are approximate because computer resources
were funded separately in some agencies and because some projects had
multiple components.  The costs include expenditures for projects
whose primary focus is modeling the global atmosphere, oceans, and/or
land surface for the purpose of simulating changes in the global
climate.  Projects include (1) predicting changes in the climate that
are likely to occur from the next decade through the next century and
(2) simulating the current climate. 

\a Totals may not add because of rounding. 

Appendix IV presents background and cost information on each agency's
climate modeling program.  Most of the agencies' climate modeling
research was contracted out to universities and research laboratories
throughout the United States.  These modeling activities were
conducted at five major modeling centers in the United States:  1)
the National Center for Atmospheric Research in Boulder, Colorado; 2)
NOAA's Geophysical Fluid Dynamics Laboratory in Princeton, New
Jersey; 3) NASA's Goddard Institute for Space Studies in New York,
New York; 4) NASA's Goddard Space Flight Center in Greenbelt,
Maryland; and 5) DOE's Lawrence Livermore National Laboratory in
Livermore, California. 


--------------------
\7 During fiscal year 1994, the U.S.  Global Change Research Program
was funded at $1.4 billion.  Modeling activities accounted for $45.3
million, or 3 percent of the funding. 


   CONCLUSIONS
------------------------------------------------------------ Letter :5

Although the accuracy of general circulation models' estimates of
climatic change has improved over the past decade, these estimates
are still limited by incomplete and inaccurate representations of the
processes affecting climate and by insufficient computer power. 
These limitations prevent scientists from carrying out analyses that
would yield more precise information about the magnitude, timing, and
regional effects of predicted increases in warming.  Ongoing efforts
to collect and analyze data, improve representations of climatic
processes, and, develop and apply more powerful computers should
improve the accuracy of the models' estimates.  Whether these
estimates will provide policymakers with the information they need to
respond to possible future climatic changes will depend on the degree
of certainty expected from the models, the resources provided to
improve the models, and advances in scientists' fundamental
understanding of the climate system. 


   AGENCY COMMENTS
------------------------------------------------------------ Letter :6

We obtained comments from representatives of DOE, NASA, NSF, NOAA,
and EPA.  According to these comments, which were coordinated by the
Director of the Office of the U.S.  Global Change Research Program,
the agencies found, overall, that this report provided an interesting
and useful perspective on the most important factors that limit the
credibility of general circulation models' projections of future
climatic conditions.  However, the agencies believed that the report
would be more useful if it provided some perspective on what the
modeling community has learned about the models' limitations and what
efforts are under way to address them.  The agencies also believed
that the report focused too heavily on the limitations of the models
while remaining largely silent on their accomplishments.  We have
responded to the agencies' comments by adding information about
ongoing research to overcome the models' scientific and technical
limitations and about recent positive results achieved with the
models. 

Additionally, the agencies believed that the report should include,
in full, the report of the Forum on Global Climate Change Modeling
(Forum), which was developed to inform policymakers about the issues
associated with using general circulation models.  While we believe
that the Forum's document is useful, we did not include it in this
report because its major points are summarized in the agencies'
detailed comments and are included in the body of this report,
insofar as they pertain to the objectives of this assignment. 
Furthermore, since the Forum's report is available to the public from
the Office of the U.S.  Global Change Research Program (USGCRP Report
95-01, May 1995), we believe that persons desiring the additional
detail may request the document.  The agencies' comments and our
response appear in appendix VI. 


---------------------------------------------------------- Letter :6.1

We conducted our work between September 1994 and June 1995 in
accordance with generally accepted government auditing standards.  We
reviewed various scientific documents that discussed the models'
limitations and the implications of these limitations.  Through the
Director of the Office of the U.S.  Global Change Research Program,
we collected data on costs from the five agencies that fund U.S. 
global climate change modeling.  We did not independently verify the
validity of the cost data.  Appendix V more fully discusses our scope
and methodology. 

As arranged with your office, unless you publicly announce its
contents earlier, we plan no further distribution of this report
until 30 days after the date of this letter.  At that time, we will
send copies to the Director of the U.S.  Global Change Research
Program and other interested parties.  We will make copies available
to others upon request. 

Please call me at (202) 512-6111 if you or your staff have any
questions.  Major contributors to this report are listed in appendix
VII. 

Sincerely yours,

Peter F.  Guerrero
Director, Environmental
 Protection Issues


GROWTH IN THE PRODUCTION OF
GREENHOUSE GASES AND THEIR
CONTRIBUTION TO THE ENHANCED
GREENHOUSE EFFECT
=========================================================== Appendix I

Over the past century, human activities have significantly increased
atmospheric concentrations of carbon dioxide, methane, and nitrous
oxide--known, together with water vapor, as greenhouse gases.\8

Emissions of carbon dioxide, the most abundant greenhouse gas after
water vapor, increased by about 25 percent from preindustrial times
until 1993.  Currently, the growth in emissions is primarily
attributable to the increased use of fossil fuels, whereas, in the
19th and early 20th century, it was due to deforestation and the
expansion of agriculture.  Methane emissions increased by about 9
percent between 1978 and 1987 and have more than doubled since
preindustrial times.  Nitrous oxide emissions increased by about 9
percent from preindustrial times until 1993.  Table I.1.  details the
increases in greenhouse gas emissions, the periods when the increases
occurred, and the sources of the emissions. 



                          Table I.1
           
            Growth in the Production of Greenhouse
                            Gases

                    Growth  Applicable
Gas              (percent)  period        Sources
--------------  ----------  ------------  ------------------
Carbon dioxide          25  Preindustria  Fossil fuel
                            l to 1993     combustion
                                          Deforestation


Methane                  9  1978 to 1987  Rice paddies
                                          Cattle and sheep
                                          Natural gas
                                          production
                                          and delivery
                                          Coal production
                                          Landfills

Nitrous oxide            9  Preindustria  Nylon production
                            l to 1993     Nitric acid
                                          production
                                          Use of
                                          nitrogenous
                                          fertilizers
------------------------------------------------------------
Source:  Preparing for an Uncertain Climate, Vol.  I, Office of
Technology Assessment (Oct.  1993). 

The relative contribution of each gas to the enhanced greenhouse
effect is determined by the ability of the gas to absorb infrared
radiation and by its atmospheric abundance.  Atmospheric abundance is
determined by the quantity of gas emitted and by its atmospheric life
span.  For example, although a methane molecule is a more effective
absorber of infrared radiation than a carbon dioxide molecule, it
contributes only about a third as much to the enhanced greenhouse
effect because it is less abundant.  Carbon dioxide is believed to
have contributed 70 percent of the enhanced greenhouse effect from
the beginning of the Industrial Revolution up to 1990.  Figure I.1
depicts the relative contributions to the enhanced greenhouse effect
from the cumulative contributions of carbon dioxide, methane, and
nitrous oxides.  Chlorofluorocarbons are not included in the figure
because, unlike carbon dioxide, methane, and nitrous oxide, their
atmospheric concentrations vary considerably across the globe and are
difficult to quantity. 

   Figure I.1:  Contributions of
   Greenhouse Gases to the
   Enhanced Greenhouse Effect

   (See figure in printed
   edition.)

Source:  John Houghton, Global Warming (Lion Publishing:  Elgin,
Ill., 1994). 


--------------------
\8 Although chlorofluorocarbons may also be referred to as major
greenhouse gases, we did not include them in our categorization of
major gases because their consumption has declined since their
phaseout under the 1987 Montreal Protocol. 


HOW GENERAL CIRCULATION MODELS
WORK
========================================================== Appendix II

General circulation models (GCM) are the most advanced tool that
scientists have to model climate and predict climatic change.  These
models comprise complex mathematical equations that describe various
physical processes and interrelationships, including seasonal changes
in sunlight, global air currents, and other factors that affect the
climate.  Because the equations are so complex, modelers cannot solve
them exactly and consequently must segment the earth into a discrete
number of grids to approximate the solutions.  The coupled model
depicted in figure II.1 calculates solutions for 18 layers above each
grid (extending from the ocean's surface to the top of the
atmosphere) and 20 layers below each grid (extending from the surface
to the floor of the ocean). 

   Figure II.1:  How One General
   Circulation Model Works

   (See figure in printed
   edition.)


U.S.  RESEARCH ON GLOBAL CLIMATE
CHANGE AND ITS ESTIMATED COSTS
========================================================= Appendix III

The U.S.  Global Change Research Program (USGCRP) provides insight
into the causes and effects of changes in the earth's climate system,
especially those related to human activities, and is developing tools
to assess options for responding to global change.  As the depth of
understanding grows, the research results are intended to provide
increasingly valuable support for formulating national and
international policy, as well as for evaluating the impact and
effectiveness of the actions taken.  The research activities are
grouped by major focus in six broad categories: 

  observing the earth's climate system through land, ocean, and
     satellite observation networks;

  managing data and information to ensure that they are preserved and
     available for national and international researchers to use;

  understanding global change processes, ranging from cloud formation
     and hydrologic processes to the accumulation of atmospheric
     ozone;

  predicting the magnitude, timing, and extent of global change;

  evaluating the consequences of global change by analyzing the
     impact of global change on the environment and on society; and

  assessing policies and options for responding to global change. 

The President's fiscal year 1995 budget for the research activities
is summarized in table III.1. 



                         Table III.1
           
           President's Fiscal Year 1995 Budget for
           the U.S. Global Change Research Program,
                   by Category of Research

                    (Dollars in millions)

                                            Fiscal year 1995
Category                                              budget
----------------------------------------  ------------------
Observing the earth's climate system                  $733.7
Managing data and information                          382.0
Understanding processes                                530.7
Predicting change (modeling)                            67.1
Evaluating consequences                                 67.1
Assessing policies and options                          34.7
============================================================
Total                                               $1,814.8
------------------------------------------------------------
Note:  Total may not add because of rounding. 

Source:  Our Changing Planet:  The FY 1995 U.S.  Global Change
Research Program. 


MODELING GLOBAL CLIMATE CHANGE IN
THE UNITED STATES-- ACTIVITIES AND
COSTS
========================================================== Appendix IV

The earth's environmental system encompasses the atmosphere; the
oceans and marine life; the land surface and biosphere (plant and
animal life); and the cryosphere (snow, glaciers, sea ice, and
icecaps).  Because this complex, interconnected system cannot be
reconstructed and experimented with in the traditional laboratory
sense, numerical models are used to simulate the behavior of the
earth's system and its fluctuations, variations, and responses to
disturbances, including the effects of human activities.  Coupled
atmospheric/oceanic GCMs are, within the limits of available
resources and ingenuity, designed to include as much of the important
and relevant physics, chemistry, and biology as is understood and as
is needed to address particular questions posed to the models about
future climatic change. 


   AREAS OF EMPHASIS IN MODELING
   CLIMATIC CHANGE
-------------------------------------------------------- Appendix IV:1

A wide array of modeling activities supports the need to provide
society with the best possible predictions of weather; anomalous
seasonal events, such as floods and droughts; fluctuations in the
frequency of climatic extremes; and long-term changes in climate. 
These activities, which are conducted at research centers,
universities, and government laboratories, are supported by
government agencies that have responsibility for scientific research,
including (1) the Department of Energy (DOE), (2) the National
Aeronautics and Space Administration (NASA), (3) the National Science
Foundation (NSF), (4) the National Oceanic and Atmospheric
Administration (NOAA), and (5) the Environmental Protection Agency
(EPA).  These agencies support research using full models of the
global climate to improve, test, and, in some cases, project the
future climate and its potential changes.  The agencies' roles in
modeling climatic change are discussed below. 


      DEPARTMENT OF ENERGY
------------------------------------------------------ Appendix IV:1.1

DOE's modeling program focuses on changes and variations in the
earth's climate--especially those caused by human activities--that
may occur over periods ranging from decades to centuries.  DOE's
program (1) tests the performance of models from around the world by
comparing their ability to represent the recent climate, (2)
simulates the effects of carbon dioxide emissions on the climate, and
(3) develops global models by taking advantage of the new generations
of highly parallel computers.  These activities are intended to
develop the coupled models of the earth's oceans, atmosphere, and
land surface that are needed to project the climate more accurately
from tens to hundreds of years.  During fiscal year 1994, DOE funded
modeling research at 24 universities and research centers. 


      NATIONAL AERONAUTICS AND
      SPACE ADMINISTRATION
------------------------------------------------------ Appendix IV:1.2

NASA's modeling program focuses on developing and applying a
four-dimensional model that places special emphasis on the role of
data from satellites in providing research-quality information on the
climate system.  NASA's program supports efforts to (1) better
understand the relative roles of the various factors that have
changed or are changing the earth's climate; (2) analyze the global
effects of feedback mechanisms, such as clouds, that can amplify or
moderate climatic change; and (3) develop tools for integrating data
from satellites and other sensors into a coherent record of
atmospheric behavior.  During fiscal year 1994, NASA supported
modeling research on climatic change at its Goddard Institute for
Space Studies and Goddard Space Flight Center and at two
universities. 


      NATIONAL SCIENCE FOUNDATION
------------------------------------------------------ Appendix IV:1.3

NSF's modeling program focuses on climatic change that occurs over
seasons to centuries and provides computer resources to the research
community.  Specifically, NSF's programs emphasize research on
coupling models of the atmosphere, oceans, land surface, and
cryosphere into a single integrated model that can simulate the
global climate system over the long term.  NSF also supports
wide-ranging research activities, including simulations of climates
of the distant past, of natural variations in the present climate,
and of the interactions of the various processes and influences. 
During fiscal year 1994, NSF funded major modeling research projects
at 10 universities and research centers. 


      NATIONAL OCEANIC AND
      ATMOSPHERIC ADMINISTRATION
------------------------------------------------------ Appendix IV:1.4

NOAA's modeling program focuses primarily on seasonal to interannual
(year to year) predictions and on better understanding long-term
climatic variation and change.  NOAA's activities include (1)
developing and improving models of the atmospheric-oceanic system,
(2) comparing models' simulations to observations and analyses of the
processes that most influence climate, (3) simulating the potential
climatic effects of increased concentrations of greenhouse gases, and
(4) separating the effects of natural climatic variations from the
effects of human activities on climate.  In addition, NOAA has tried
to develop models capable of predicting the seasonal to interannual
fluctuations that cause extreme rainfall and other similar
disruptions to regional climates.  During fiscal year 1994, NOAA
supported global general circulation modeling research on 10-year and
longer time scales at its Geophysical Fluid Dynamics Laboratory and
four universities and research centers. 


      ENVIRONMENTAL PROTECTION
      AGENCY
------------------------------------------------------ Appendix IV:1.5

EPA focuses its modeling research on chemical and environmental
interactions within the biosphere.  It supports research to improve
GCMs' representations of ecosystems and of the relationship between
chemicals and plant and animal life in an area so that the effects of
climatic change on the biosphere and of biospheric change on climate
can be projected.  During fiscal year 1994, EPA funded modeling
research at three universities and research centers. 


   COSTS OF MODELING GLOBAL
   CLIMATE CHANGE
-------------------------------------------------------- Appendix IV:2

As previously stated, five agencies support research on modeling
global climate change.  This research--through which models of the
global climate are improved, tested, and, in some cases, used to
project the future climate and the ways it may change--can be grouped
into two broad areas: 

  modeling to predict changes in climate that may occur over decades
     and centuries, and

  modeling to simulate the current climate. 

The cumulative estimated costs of the five agencies' modeling
activities was approximately $123 million during fiscal years 1992
through 1994, as table IV.1 shows. 



                          Table IV.1
           
           Costs of Modeling Global Climate Change,
                     Fiscal Years 1992-94

                    (Dollars in millions)



                                                        1992
Agency              1992  1993  1994  1992  1993  1994   -94
------------------  ----  ----  ----  ----  ----  ----  ----
DOE                 $14.  $15.  $15.  $0.0  $0.1  $0.1  $44.
                       1     1     1                       5
NASA                 1.8   1.6   1.7   8.6  10.2  10.9  34.9
NSF                  7.2   7.3   8.8   0.2   0.3   0.4  24.1
NOAA                 4.1   4.1   6.0   0.0   0.0   0.0  14.3
EPA                  0.7   1.8   2.3   0.0   0.0   0.0   4.8
============================================================
Total\a             $28.  $29.  $34.  $8.8  $10.  $11.  $122
                       1     9     0           6     4    .6
------------------------------------------------------------
\a Totals may not add because of rounding. 


OBJECTIVES, SCOPE, AND METHODOLOGY
=========================================================== Appendix V

The Ranking Minority Member of the House Committee on Commerce asked
us to review the factors that affect the accuracy of GCMs' estimates
of future climatic changes and determine the costs of federally
funded GCMs for fiscal years 1992 through 1994.  We conducted our
work between September 1994 and April 1995 in accordance with
generally accepted government auditing standards. 

To determine the factors that affect the accuracy of GCMs' estimates
of future climatic changes, we reviewed information that we
previously reported in Global Warming:  Emission Reductions Possible
as Scientific Uncertainties Are Resolved (GAO/RCED-90-58, Sept.  28,
1990).  We also met with headquarters officials at DOE, NASA, NSF,
NOAA, and EPA and with the Director of the Office of U.S.  Global
Change Research Program (USGCRP) to discuss these factors.  From
these meetings, we obtained various scientific assessments of GCMs'
strengths and limitations.  In October 1994, the Subcommittee on
Global Change Research held the Forum on Global Change Modeling with
modelers from throughout the United States.  The intent of the forum
was to address requests from the White House Office of Science and
Technology Policy and the General Accounting Office to produce a
consensus document on issues concerning the use of climate models to
inform policy on future climatic changes.  This forum, whose
participants included agency officials, scientists, and academicians
involved in studying the global climate, provided information on the
strengths and weaknesses of GCMs, as well as other relevant topics. 
In addition, we searched four scientific data bases to identify
additional assessments of the models' limitations.  Throughout our
review, we met with the Director of the Office of USGCRP to clarify
technical issues associated with the models' limitations. 

To identify federal funding for GCMs during fiscal years 1992 through
1994, we obtained cost data by agency from USGCRP.  We worked with
the Director of the Office of USGCRP to develop an instrument to
capture all relevant cost components.  We did not independently
verify the validity of the cost data. 

On May 12, 1995, we met with the Director of the Office of USGCRP,
the Manager of the Climate Modeling Program at NASA, the Deputy
Director of the Office of Global Programs at NOAA, and the Manager of
the Global Change Research Program at EPA to obtain their comments on
a draft of this report.  On May 22, 1995, the Chair of the
Subcommittee on Global Change Research provided us with written
comments on the draft.  These comments integrated the responses of
the five agencies included in our review (see app.  VI).  We have
addressed the comments in the text of this report, where appropriate. 




(See figure in printed edition.)Appendix VI
COMMENTS FROM THE SUBCOMMITTEE ON
GLOBAL CHANGE RESEARCH
=========================================================== Appendix V



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



(See figure in printed edition.)



   GAO COMMENTS
--------------------------------------------------------- Appendix V:1

The following are GAO's comments on the Subcommittee on Global Change
Research's letter dated May 22, 1995. 

1.  Under the heading "Factors Limiting the Accuracy of Models'
Estimates," we discussed some of the successes of GCMs to create a
context and provide balance for our discussion of the models'
limitations.  Later, under the heading "Improving GCMs' Estimates,"
we discussed the efforts that are currently under way to address
these limitations and referred to the agencies' discussion of such
activities in this letter.  Under the heading "Agency Comments," we
explained why we did not reproduce the report of the U.S.  Global
Change Model Forum in this report. 

2.  We added a footnote on page 3 of the report to better explain
USGCRP's role in coordinating federal research on global climate
change. 

3.  We revised our discussion of the models' limitations (pp.  8-14
of our draft report) as necessary to address the agencies' specific
comments.  We changed the heading "Exclusion of Critical Processes,"
cited in the agencies' comments, to "Some Processes Not Included or
Fully Incorporated in Some Models" to better describe the supporting
text. 


MAJOR CONTRIBUTORS TO THIS REPORT
========================================================= Appendix VII

RESOURCES, COMMUNITY, AND ECONOMIC
DEVELOPMENT DIVISION, WASHINGTON,
D.C. 

Lawrence J.  Dyckman, Associate Director
William F.  McGee, Assistant Director
Richard A.  Frankel, Science Advisor
Judy K.  Pagano, Senior Operations Research Analyst

NORFOLK REGIONAL OFFICE

Everett O.  Pace, Evaluator-in-Charge
James B.  Hayward, Staff Evaluator
Kathryn D.  Snavely, Staff Evaluator

DENVER REGIONAL OFFICE

Mary A.  Crenshaw, Site Senior

