incompleteness makes it difficult to represent adequately the interannual, interdecadal, and slowly changing nature of the climate. Thus, the models cannot yet be used to investigate changes in climate variability, and in the frequencies of extreme events (e.g., droughts, hurricanes), which is often where there is the tightest coupling between climate and society.

4. Model calculations indicate that past and anticipated changes in atmospheric composition are committing the world to significant climatic change. Regional details, however, remain quite uncertain.

Over the past 17 years, calculations with an increasingly comprehensive set of three-dimensional atmospheric and simple ocean models have indicated that a doubling of the reference CO2 concentration would lead to a global warming of about 1.5 to 5 °C (about 2.5 to 9 °F), once the oceans have had a chance to warm (a time of at least several decades). (The IPCC indicates a range of 1.5 to 4.5 °C and a best guess value of 2.5 °C). Based on paleoclimatic and historical climatic changes, the sensitivity to a CO2 doubling cannot be significantly less than 1.5 °C or we cannot explain what could have caused past climatic changes; nor can the sensitivity be more than about 5 °C or we cannot explain why the global climate is so stable in the face of volcanic eruptions and other changes. Thus, the model estimates are in the proper range if all factors influencing the climate are accounted for. There would also be an intensification of the global evaporation-precipitation cycle by about 5 to 15%.

An atmospheric model intercomparison effort led by scientists at SUNY/Stony Brook and LLNL has identified the representation of cloud-radiation interactions as a primary contributor to the wide range of model estimates of climate sensitivity results. DOE's Atmospheric Radiation Measurement (ARM) program and other agency programs are designed to acquire the data needed to improve representation of the critical cloud-radiation processes.

Despite the uncertainties acknowledged to be present in models, no plausible alternative set of parameterizations and approximations has been able to make significant global warming "go away".

Most models suggest an amplification of the global warming in high latitudes, with the melting back of sea ice disrupting the winter polar temperature inversion; this results in a larger than average, but shallow, warming. Amplified temperature changes in high latitudes are generally consistent with the large polar climatic changes seen in paleoclimatic record. The recent record of temperature changes in the high latitudes, however, does not now show this amplification, which may be pointing to a limitation in the use of equilibrium calculations from models rather than calculations with full, time-dependent models.

Most models suggest up to a few degree warming in low latitudes. Paleoclimatic data, at least for climatic changes controlled primarily by changes in the Sun Earth orbital parameters, suggest that low latitude temperatures have been very stable. Research is intensifying on this apparent discrepancy, with the primary focus being on whether models are inadequately representing cloud and convective processes. For this reason the second ARM field site and other programs are focusing their attention on cloud-radiation processes in the western tropical Pacific. This uncertainty has important policy implications, because without warming in low latitudes (and with a monsoon intensification) the threat of climatic changes may be less of a priority to the large developing nations in those regions. However, other low-lying nations are also particularly concerned about rising sea level.

Although global climate models are not yet constructed to adequately represent regional features, even those as large as the Rocky Mountains and Great Lakes, much less the Sierra Mountains, suggestions have been drawn from the models that continental interiors like the North American corn belt will have more rainfall in winter and less in summer. While plausible, there will remain many significant uncertainties until process representations are improved and grid resolution is refined; model resolution experiments suggest, for example, that large scale atmospheric features are not robustly located until grid resolution is about half of typical values used at present. Thus, estimates of regional scale details should be viewed with considerable skepticism.11

5. Although global warming of 0.3 to 0.6 °C has occurred, quantitative association of the observed warming with the greenhouse gas-induced climatic change is not yet sufficient to significantly reduce uncertainties about the details of future warming.

11 As a consequence, mitigation measures focused on possible local climatic changes should generally be considered only if they are a means of building resiliency to a wide range of possible changes.

With the improving quantification of the moderating effects of aerosols on global warming, however, the models and observations are no longer in troubling conflict. Quantitative association of greenhouse-induced climatic change with observations (often referred to as "detection" of the greenhouse effect) requires both a comprehensive model calculation of the greenhouse effect since the beginning of the Industrial Revolution and a comparable and comprehensive data set. The analysis must then be able to distinguish the greenhouse effect with confidence from the effects induced by other human activities and from noise introduced by natural variations and oscillations.

It has not yet been possible to carry out the needed time-dependent model calculation for several reasons. First, we need to carry out the calculation with a full Earth system model that includes the effects of each greenhouse gas individually, and such models are only now being developed and tested. We need to figure out how to represent the effects of other factors affecting the climate, including volcanic eruptions, solar variations, natural oscillations, and other human activities (e.g., SO2 emissions, biomass burning, land use changes). In lieu of the desired time-dependent calculations with complete models, resort has often been made to interpolation based on the equilibrium sensitivity simulations described above. Preliminary time-dependent simulations have shown however, that the interpolation approach can give quite misleading results in high latitudes, over ocean areas where deep mixing is occurring, and in estimating land-ocean differences. With such limitations, there has been only limited success in quantitatively relating observed and modeled changes.

The available data sets are generally uncertain and limited in spatial extent and quality before the early 1900s. Shorter data sets are usually not as useful as longer data sets due to the inability to average out natural variability and the effects of other short-term processes, regional variations, etc. It is also often difficult to filter out local inconsistencies introduced by station moves, urbanization, desertification, irrigation, changes in measurement technique, etc.

The global temperature record, as best it can be reconstructed, shows a warming of about 0.3 to 0.6 °C since the mid 1800s. Whether some fraction of this warming may be a remaining result of recovery from the Little Ice Age (a cold period lasting roughly from 1450 to 1850 A.D.) to levels typical of the warmer period from about 900 to 1300 A.D. is not certain. The record has not been steady and increasing in intensity in a manner parallel with the changes in greenhouse gas concentrations, as models would suggest. Instead, the record shows relatively short term jumps of about 0.2 to 0.3 °C in the 1910s/20s and 1970s/80s. Whether this intermittent warming pattern is indicative of how the real climate changes (this would indicate that our models are inadequate and that future climatic changes may appear as surprises), is indicative of multiple factors affecting the climate (including natural cooling or warming influences), or is due to limitations in the data sets or other factors is uncertain. There are, however, other less apparent changes (including meltback of mountain glaciers, stratospheric cooling, sea level rise, and a precipitation increase) that are generally consistent with there being a long-term global warming. An important difficulty, however, is that we do not adequately understand natural climatic variations, so some of the warming and of the other changes may be part of the natural fluctuations; of course, natural effects could also be of the other sign and temporarily hiding the greenhouse effect.

A simple best-fit reconciliation of the model results and the observational results suggests an equilibrium climate sensitivity to CO2 doubling of about 1.3 °C. It was thus somewhat surprising that the IPCC 1990 report proposed an uncertainty range for CO2 doubling of 1.5 to 4.5 °C such that the best fit through the data was not within their limits; this was done in recognition that some other factor must be acting to restrain the warming. Since the 1990 report, quantitative estimates of the cooling influence of sulfate aerosols from fossil fuel combustion and of lower stratospheric ozone depletion suggest that they may be playing this role. In fact, when consideration is taken of the potential cooling influences of aerosols from biomass burning and of potential cloud-aerosol interactions, reconciliation with past climatic data suggests that the climate sensitivity to greenhouse gases must be near the high end of the IPCC range to explain the warming that has occurred. An aerosol effect of this magnitude would also explain why most of the warming over Northern Hemisphere land areas has been at night rather than during the day. Thus, while the quantitative comparison of model results and observations is only a little more quantitative than a few years ago, the model results and data are currently no longer in obvious troubling conflict.

6. Although future warming may be somewhat intermittent due to natural climatic variations induced by volcanic eruptions and air-sea interactions, the warming in

fluences of greenhouse gases over the next several decades are likely to increasingly exceed the cooling influences of aerosols from fossil fuel combustion, of aerosols from biomass burning, and of ozone depletion.

Making a prediction of future climatic conditions requires:

(1) a prediction of societal energy and land use,

(2) a comprehensive model tested against many past climatic situations and recent field experiments and observations, and

(3) the ability to project and represent the climatic effects of all possible influ

ences.

All aspects of this process present difficulties that bring to life Neils Bohr's reputed comment that prediction is very difficult, particularly of the future.

The IPCC 1990 report suggested a future rate of warming of 0.2 °C to 0.5 °C per decade if greenhouse gas emissions continued to increase without restriction. The IPCC 1992 report clarifies that this is the expected climate sensitivity to greenhouse gases alone, and that this rate of warming will probably not be realized over the next few decades, and perhaps longer, especially in the Northern Hemisphere, if the emissions of sulfur dioxide continue to increase. Ozone depletion and aerosols from biomass burning may further depress the expected rate of warming from human activities, such that natural climatic variations may even lead to one decade being slightly cooler than the one before. Thus, considering all factors and presuming that natural climatic fluctuations are small, model results would suggest a more modest warming rate, perhaps 0.0 to 0.3 °C per decade averaged over the next few decades, although there is a commitment to the higher warming in the future.

Because the atmospheric lifetime of CO2, is of order a century or more and the lifetime of aerosols is of order a week, the CO2 will accumulate in the atmosphere and increasingly exert its warming influence, especially as aerosol precursor emissions from fossil fuel combustion and biomass burning are reduced for other environmental reasons. Thus, the slow global warming rate of recent decades and the limiting of warming to nighttime in the Northern Hemisphere are likely only transitory moderating influences and are dependent on continuing increases in aerosol loading. Depending on such a trade-off is accepting a philosophy of advertently "geoengineering" the climate. 12

Changes in precipitation and soil moisture will occur, but there is little agreement among model results. There are strong suggestions, however, that continental interiors may dry out and warm during the summer, and that summer monsoons may intensify. However, regional projections are extremely uncertain and range from changes below to well above the global average change, which itself is quite uncertain.

We cannot be confident that the climate will change slowly and steadily rather than in fits and jumps. Such sudden jumps occurred around 1920 and around 1975 and actually make up most of the warming that has occurred. We do not know if these sudden changes were an indication of a set of separate states of the climate system or resulted from coincident occurrence of two or more other factors. Thus, we should not be complacent that the relatively steady warm climate of the 1980s indicates that the long-term warming rate is low.

7. The potential environmental and societal impacts of changes in climate and atmospheric composition are relatively uncertain, due in part to uncertainties in climate model results, in part to poor understanding of impacts and limited development of impact methodologies, and in part to uncertainties about how future society and technology will evolve.

In that fossil fuel energy and agriculture provide essential life support for society, costs associated with changes in their supply and use would seem to require identification of comparably important costs or impacts from climatic change. Some economic estimates suggest impacts will be zero to a few percent of the U.S. GNP, but it is not clear how to extrapolate costs to other nations (i.e., as a % of GNP, on a per capita basis, etc.). Although uncertain, it is possible to identify major classes of potential impacts and to provide a general sense of potential consequences. It is much harder to put these in the context of the costs of altering or changing energy and agricultural systems and of evolving societal activities and tendencies (e.g., urbanization, technological development, population growth, etc.).

Human Health: The effects of changes in atmospheric composition and climate include increased exposure to warmer temperatures and perhaps increased areal extent of tropical diseases. While CFCs tend to deplete stratospheric ozone, the

12 M.C. MacCracken, "Geoengineering the Climate," LLNL report UCRL-JC-108014, June

greenhouse gases together tend to cool the stratosphere, which would slow the gas phase reactions in the upper stratosphere that deplete ozone; at the same time, there may be increased particulate formation that could increase ozone depletion through surface reactions in the lower stratosphere. Overall health impacts are likely negative, but may be modest.

Food and Fiber: With proper soils and fertilizer use, the enhanced CO2 concentration and longer growing seasons may enhance overall agricultural production, especially in technologically advanced countries, if summer moisture levels do not deteriorate significantly and if the crops can grow faster than the weeds and pests. There would, over time, likely be significant changes in cropping patterns, and onecrop countries may suffer severe economic costs as they are forced to try to shift crops. Overall impacts may be positive, but the costs arising from local variations and indirect adaptation effects (e.g., farmers changing crops) may be a significant complication.

Water Resources: Although the global evaporation-precipitation cycle will intensify, it is not certain where and when the increased precipitation will fall and what shifts (involving increases and decreases) in precipitation will occur. Evaporation, however, is likely to increase everywhere. Changes of any kind can be detrimental until adaptation occurs. For example, a rising snowline in California would intensify winter runoff (requiring dams or reduced water retention in reservoirs to protect against flooding) and reduce summer runoff (allowing salt water intrusion and reducing water quality) at the same time warming increases summer demand by urban and agricultural users. The overall effect is likely negative in most areas, at least until management practices and water uses can be adapted.

Coastal Habitat: Sea level will likely rise at a slowly accelerating rate. On ocean islands and atolls, any rise is likely to be detrimental. Around high value property, dikes and levees can likely be built at modest cost, although the potential for damage from storms will rise substantially in low-lying areas. In addition, providing now for some future sea level rise could ameliorate later impacts at modest cost. The Netherlands estimated adaptation costs of about 1% of GNP, but costs might be a much higher fraction in Bangladesh and in island nations. Overall coastal impacts will be negative and generally concentrated in a relatively few hard hit areas.

Ecological Systems: Experience indicates that ecosystems are always changing slowly. Existing systems have adapted roughly to the present climate and an increased rate of climatic change and altered CO2 concentration will start to induce increasing changes. We do not understand the complex linkages involved (EPA is initiating a program to explore these), but it is very unlikely that all species can adapt to rapid change. The overall effect is likely negative, but uncertain and will quite possibly occur in surprising ways.

In total, although the potential for increased agricultural productivity may suggest a beneficial consequence, impacts on health, water resources, coastal habitats, and ecological systems are almost certainly deleterious, especially if climate change is rapid. While there is no way that a world having a population of 5.5 billion (headed for a doubling) can have zero climate or environmental impact, efforts to slow the rate of climatic change and to reduce the eventual total climatic change could moderate negative environmental consequences.

8. Developing approaches for energy supply that take advantage of regional climate characteristics may be an important step in moving toward a more environmentally sustainable global energy system. Research that build's flexibility and a range of energy options would help prepare for the future, as well as being essential for U.S. competitiveness.

With fossil fuels supplying greater than 80% of the world's energy and with the global population growing rapidly, there is no single alternative energy system ready to take over as a global energy source. Fission, fusion and solar power satellites have major global potential, although they would require many decades and significant capital investment to serve for replacement of old systems and for new systems to handle growth in demand. A flexible and regionally adapted approach may prove most practical.

(a) Wet tropics: Replant deforested areas and encourage growth of a hardwood-based forestry system (reportedly made more practical by new forestry techniques); develop hydroelectric sources for electricity; develop biomass plantations for liquid fuels.

(b) Dry subtropics. Accelerate research, development, and deployment of wind energy systems and dispersed photovoltaic, focusing on an all electric approach. (c) Mid-latitude developed countries. Focus on conservation, efficiency improvements, natural gas as a transition fuel, biofuels, windpower, nuclear for

baseload, etc. To go electric, battery and other storage and load shifting technologies are essential.

(d) Large developing countries (e.g., China, and India): These present a significant challenge where all available techniques will he needed. Because it will be very difficult to replace coal as the most economical fuel, promoting efficiency is very critical.

With a breadth of technologies being required, it is essential that the U.S. be investing in research to develop a wide set of alternatives in order to build resiliency and to broaden the set of economically-attractive options that are available. These options will work best if they can be made so cost-effective that they will be the option of choice. Such an effort presents a challenge and opportunity for U.S. industry and such an effort is essential if U.S. industry is to he competitive and hold a major share of the global market in energy technology.

The CHAIRMAN. Thank you very much, Dr. MacCracken.

Next, we will hear from Dr. Stephen Schneider, who is Senior Scientist at the National Center for Atmospheric Research in Boulder, Colorado.

Dr. Schneider.

STATEMENT OF DR. STEPHEN H. SCHNEIDER, SENIOR SCIENTIST, NATIONAL CENTER FOR ATMOSPHERIC RESEARCH, BOULDER, CO

Dr. SCHNEIDER. Thank you very much, Senator Johnston. I appreciate too the much happier weather you conjured up for us today relative to that when I appeared before this committee in the summer of 1988.

Actually something that was a major surprise in the testimony I have heard so far that most of us probably wouldn't have forecast in advance which is Dick Lindzen largely agreeing with Bob Watson, and in that vein, let me also largely agree with a number of statements to maintain that surprise that Dick Lindzen said.

In particular, he said that-I hope this is an accurate quotemuch of the detailed physics is missing. Indeed, I agree with that statement. In fact, I and Mike MacCracken and most of the people who are considered the modelers have made statements like that in their publications from the very beginning of all the publications that we have. In fact, if you want to have a political interpretation, it is in our interests to have uncertainty. It keeps us going, to make people need us to continue to work. So, you will find those statements in the scientific works of virtually everybody responsible, and I think that includes all at this table most of the time.

However, where disagreements can get very acerbic—and I think that Mike MacCracken put his finger on it is that those missing pieces of the puzzle-and there are many-are not clear in advance to responsible people as to whether the sum of those effects are going to make current best guesses more or less. We simply don't know how in advance to be certain in assessing which of the range of certain of uncertainties that all of us could name—we could have a litany on one side with all the so-called negative feedbacks that reduce the magnitude of the problem, and then I could give you a counter-litany on the other side about all the positive feedbacks. The object of research is to try to pin those down. In fact, that is what the world climate research program and many national programs are designed to do.

Well, the time frame in which those research programs should be producing results where most scientists would agree that we have a