Senate Committee on Environment and Public Works
July 10, 1997
Written Testimony of Eric J. Barron
Earth System Science Center
The Pennsylvania State University
University Park, PA 16802

[illustrations omitted]

The prospect of fu~ture human-induced climate change represents perhaps one of the most challenging science and society questions of the century. There is no doubt that humans are altering the environment - both in terms of the land surface and the composition of the atmosphere. In particular, greenhouse gases (carbon dioxide, methane, nitrous oxides) in the atmosphere have increased substantially in concentration over the last several decades.

The best scientific assessments available suggests that the impacts of these changes will be significant, yet the error bars, or uncertainties, are also very large. The real question is how should society respond when the best available science suggests that human activity may substantially alter climate, but at the same time the scientists are seriously debating the magnitude, timing and distribution of the climate changes. Answers to this question depend on two basic sources of information, climate observations and model predictions.

The Observational Record:

Looking first at the observational record, we see continuing debate on its nature. Much of this debate has centered around differences between satellite derived estimates from the Microwave Sounder Unit (MSU) and surface thermometers. The surface observations indicate that recent years have been among the warmest since the late nineteenth century, with 1995 being the warmest on record. The rate of warming from these observations is .13 degrees C per decade. This is in contrast to MSU interpretations of-.05 degrees C per decade. The differences have lead to spirited debate. For example, Hurrell and Trenberth (1997; Nature) have suggested that the negative trend in MSU observations is due to errors when merging records from different MSU satellites.

We are beginning to see distinctive surface signals in precipitation and temperature that separate the later part of this century from the earlier part of the century. Karl et al. (1996; Bulletin of the American Meteorological Society) provide an analysis of U.S. data from precipitation and temperature from 1900 to 1994 - see figure 1 below. These diagrams illustrate substantial trends in key climatic parameters.

[Figure 1: 2 illustrations -- omitted]

Karl's analysis also indicates (figure 2) that there has been an increase in the amount of precipitation from extreme precipitation events (daily events at or above 2 inches of rainfall).

Figure 2:
Percent area of the USA with a much above normal proportion of total annual precipitation from extreme precipitation events [(daily events at or above 2 inches (50.8mm)]
[illustration omitted]

Combined land and ocean surface temperatures (figure 3) provide the basis for examining global trends in temperature, and are the basis for speculation on the importance of anthropogenic greenhouse gas increases as an explanation of the warming. These analyses-indicate that global- mean surface temperatures have increased by .4 to .60 C during the 20~ Century.

Figure 3:
[illustration omitted]

However, our observations of climate change from instrumented records are very short, and they rely on systems designed for weather prediction - not one designed for taking the temperature or pulse of the earth. We lack continuity of satellite observations, surface instruments are subject to change and the level of accuracy is based on weather safety and forecasting needs and not global temperature analysis. Geologic records from ice cores, tree rings, corals and other sources of data suggest that the Earth's climate is naturally highly variable. The record of snowfalls on Greenland (figure 4) illustrate this variation during the last 18,000 years. Changes in snow accumulation rate are often abrupt, suggesting remarkably large climate changes over periods of decades.

Figure 4: Greenland snow accumulation rates
[illustration omitted]

Tree ring data are equally intriguing. For example, Jacoby et al. (1996; Science) report on Mongolian tree rings which indicate much wider tree ring widths for the recent century - a phenomena associated with warmer annual temperatures. The 20~ century warming appears to be unique over the last 450 years

(Figure 5).
[illustration omitted]

Figure 5 caption. Ring widths for the last 450 years from Mongolia illustrating a unique 20~th Century record indicative of warming.

The recent record appears to be unique, but the simple fact is that modern humans haven't experienced the range of variations which occur naturally, nor do we have a real sense of their character or spatial distribution. The record describes change, but without clear attribution as to the causes. Significant natural variability should be expected during the coming decades.

Results from Model Predictions:

The results from model predictions also have limitations. In large measure, scientists agree when the topic is global and the predicted changes are given as a range (e.g. a doubling of CO2 will yield 1 to 4~.5 degrees C globally averaged temperature warming), but we have greater and greater uncertainty when we look at specific regions, specific decades or specific phenomena, such as changes in hurricane intensity or numbers. Yet it is at these scales that human systems intersect and interact with climate.

The reprint that follows is a summary of predictions from climate models with a "ranking" of the uncertainty associated with the predictions. The rankings are based not on some specific criteria, but rather the considered opinions of a large group of climate experts who have sought to place model predictions in an ordered context which would readily be understood by the educated United States citizen. Within the text are two figures which illustrate results from comprehensive climate models. Figure 1 in the reprint illustrates the range in predicted changes in global-mean surface temperature, in degrees Celsius, for the next 80 years based on results from seven different General Circulation Models (the most comprehensive climate models to date) with carbon dioxide increases included at the rate of 1% per year (IPCC 1995 assessment). All seven models suggest an additional 1 degree global-mean increase in temperature by the year 2050. Figure 2 in the attached reprint gives the predicted geographic distribution of an increase in mean- annual surface temperature that would result from a doubling of carbon dioxide based on the GCM simulation of Manabe and Stouffer (1994; Journal of Climate). Increases for the United States range from 3 to more than ~5 degrees C. The predicted changes are substantial given that the 1988 heat wave and drought in the Ohio River Basin was on average less than 1 degree C above normal.

Climate model experiments designed to predict past climates, which are very different from today also yield valuable insights. During the last decade, hundreds of GCM simulations have been completed by a wide variety of models in an attempt to predict climates both substantially warmer and substantially cooler than at present. In no case did a GCM overpredict the warming or the cooling in the geologic record. This suggests that the GCMs may have a sensitivity to factors such as carbon dioxide which is less than that required to explain past climates. Other factors may also be important (e.g. identification of all the factors which may have influenced past climates and difficulty in extracting correct climate information from fossils), but the fact that the models always have underpredicted the changes in the past may be telling. It is also interesting to note that the major warm episodes during the past are also associated with geochemical evidence for higher atmospheric carbon dioxide levels.

The reprint which follows details the strengths and weaknesses of current modeling programs nationally and internationally. It also notes that progress on both observational and modeling fronts over the last decade have been clear, but it is a mistake to promise quick answers. Solution of many of the remaining issues will undoubtably take decades. I suspect that for many years to come, newspapers will continue to explain topics like global warming by quoting scientists who are poles apart on specific points. Yet in the midst of the public confusion that this approach promotes, we can't ignore the fact that even within the range of climate model predictions, the consequences have significance for our economic vitality and national security.

Reprint on Model Predictions: [10 pages omitted]

Evaluating Policy Decisions Based on Climate Model Predictions:

Policy decisions about climate change are particularly challenging given that (1) the results from comprehensive climate models suggest significant changes over the coming decades, but the uncertainties are also large - particularly when examining the aspects of climate model predictions which are most significant for human activities and (2) the increased surface temperatures and changes in precipitation patterns recorded from surface instruments may be a result of human- induced climate change, but may also be a product of natural climate variations. Two types of actions address this conundrum.

(1) We must ensure that we have a healthy observing system and modeling effort in this nation. Obtaining use~ful climate records is a secondary priority of our current observing systems which has been designed for weather safety and prediction. Relatively modest increases in funding could address this issue. Programs designed to provide continuity of satellite observations (e.g. NASA Earth Observing System) are subject to annual review and budget reductions, increasing the risk that continuity of critical measurements will be lost. Interestingly, European countries and Japan are promoting strong space-based observation programs as they recognize the value of these data sets for decision-making and scientific advancement.

The U.S. climate modeling community has expressed strong concerns about the effectiveness of our efforts in climate modeling, with particular emphasis on the fact that IPCC assessments are increasingly being based on long-term simulations completed by other nations. Interestingly, countries like Japan, the United Kingdom and Germany are promoting strong observation and modeling programs with less robust economies than the U.S. The simple fact is that advanced knowledge has economic and societal value.

There is also considerable prospect for advances in knowledge, and at scales which allow us to examine more closely the potential impact of climate change on societies. For example, recent techniques have been applied to produce high resolution climate simulations by embedding or nesting high resolution, limited area climate models within global models. Global models provide the coarse spatial resolution predictions of the large-scale atmospheric circulation, while the high resolution model allows the incorporation of more realistic elevations and model physics. Figure 3 in the reprint illustrates the improvement in the prediction of precipitation for the United States comparing (a) observations for spring 1980, (b)a GCM prediction for spring 1980 showing a relatively poor simulation of this important variable, and (c) the results for the same period from a high resolution model embedded within the same GCM shown in figure 3b. The improvement is dramatic, giving confidence that higher resolution models may provide more useful predictions. Figure 6 illustrates the results from this technique for a doubled concentration of carbon dioxide. The results suggest substantial differences in precipitation (figure 7). Winter precipitation is predicted to increase in the Northwest and Northeast with modest increases across the northern states. California and Arizona show significant decreases in winter precipitation. In summer, the model simulation suggests the largest increases in precipitation occur from Louisiana-Mississippi- Alabama across the across the central U.S. to South Dakota. Again, California has significant decreases. Such results must be viewed with caution- they are a preliminary analysis using a new, and not thoroughly tested technique to achieve high resolution predictions for specific regions.

[three pages of maps omitted]

(2) We need to develop and apply a litmus test to action which is practical and most likely to achieve positive results. Risk and vulnerability to natural variability and climate change must be a key aspect of this test. For example, if a region is already historically and economically vulnerable to droughts or floods, and predictions of fixture climate change also exhibit such tendencies, or even enhanced tendencies, then this should be a call to action. Water and water resources provide a key example of potential vulnerabilities. Two figures follow which describe vulnerability associated with water availability and water quality. Figure 8 illustrates regions with water demand problems in 1980. Each dot or shaded area indicates a problem where water demand approached or exceeded supply during the period of analysis. This suggests a vulnerability to natural variability and to climate change (see figure 1 and 7, for comparisons). Figure 9 illustrates water withdrawals by industry. Note that the industrial withdrawal of water is basically a percentage of the available resource (near 25%). This suggests that water is a critical resource to industry and that industry is co-located with water, using far more in regions where water is abundant. Many regions are susceptible to water quality problems as a result of climate variability or change. Interestingly, decreased river flow, or increased extreme events with decreased median rainfall events, has the potential to dramatically change the dilution power of rivers for pollutants. Water quality may be an unheralded global change issue.

Economic and societal risk should also be a key aspect of decision-making. For example, the emergence or re-emergence of infectious diseases, which are closely related to climate, have become an issue of growing concern in the health community. Human health issues have potential for tremendous costs associated with human life. Human health risks are governed by a large number of factors, ranging from socio-economic status, to the availability of clean water and nutrition, to the quality of the health care infrastructure - factors which generally serve to limit U.S. risks. However, over the last decade, climate and climate change have become recognized as one of the significant factors influencing health risk within the U.S.. Climate change and variability can effect health directly, through extreme thermal events like heat waves and cold episodes, and through severe weather such as hurricanes and tornadoes. Climate change can also influence human health indirectly. The majority of the indirect influences involve (1) changes in the range and activity of vectors and infective agents, (2) changes in water and food-borne infective agents, and (3) altered food (especially crop) productivity. A number of examples of human health vulnerability in the United States serves to illustrate the nature of this problem.

The increases in average temperatures associated with global warming or with extremes in natural climate variability will probably be accompanied by an increase in the number of heat waves. The deaths of 726 people in Chicago during the summer of 1995 heat wave is an example of the potential direct impact of thermal extremes. Mid-latitude cities, already characterized by large urban heat island effects, appear to be the most susceptible to heat waves. The heat-related mortality that has occurred in cities such as Chicago, St. Louis, Washington D.C., and New York City disproportionally affect the young, elderly, the economically disadvantaged, and the ill.

Phenomena, such as El Nino, are associated with changes in rainfall, producing flooding and droughts in different regions. Based on climate model predictions, climatologists have speculated about whether anthropogenic warming will produce increased intensity or an increased number of severe hurricanes along the east coast of the U.S. Severe weather has well-known potential to increase the number of deaths and injuries.

[two pages of maps omitted]

Vector-borne diseases are a major cause of illness and death across the world. These disease vectors (e.g. mosquitoes and rodents) are strongly influenced by climate. For example, Dengue fever is transmitted by the bite of a mosquito (Aedes aegypti and Aedes albopictus). Both mosquitoes are currently present in Florida and Texas (an outbreak of Dengue occurred in south Texas in 1986) but U.S. cases are uncommon, most probably because of high standards of housing, adequate water, sewer and waste management systems. However, the mosquitoes that transmit Dengue are strongly controlled by winter temperatures. Warming, particularly in terms of minimum winter temperatures could substantially increase the range of this Dengue vector, including regions north of the mid-Atlantic states. Figures 10 and 11 show regions of potential outbreak, and the association of the Dengue vector with warm winter temperatures. Malaria, caused by the protozoan parasites of the genus Plasmodium and transmitted by Anopheles mosquitoes, would also substantially extend its range and activity under conditions of global warming.

Wet-Dry cycles also influence human health risks because of its influence on predator-prey relationships. Historically, moving into a wet period following a few years of severe drought, provides advantages to rodent populations which can reproduce faster that their predators (e.g. owls, etc). Population explosions of rodents eventually leads to invasions into human habitats and human food stocks, increasing the risk of disease. This is the primary explanation for the outbreak of the deadly Hanta virus in the Four-Corners region of the U.S. (figure 12).

Lyme disease, which is caused by a bacterium, has a strong climatic association as well. Lyme disease is transmitted by the bite of a tick (Ixodes scapularis) which feeds on the white-footed mouse, the white-tailed deer and other mammals. The number of Lyme disease cases is strongly correlated with the size of the deer population, and in turn, the size of the deer population is correlated with the severity of winter conditions in the northeastern U.S.(figure 13).

The U.S. is less susceptible to problems of malnutrition and crop productivity compared to much of the world because of the breadth of food production and our capability for technological adaptations. None-the-less, climate change and variability may result in the need to change crops and planting practices, and may also influence the activity or emergence of crop diseases.

Health risks associated with climate change and variability have implications for policy. Such policy should involve (1) surveillance efforts, (2) increased research on changes in range and activity of vectors associated with climate change, (3) disease prevention programs, (4) education for medical and public health communities, and (5) public outreach.

[four pages of illustrations omitted]


Two examples are given where action makes sense because of the level of risk and the level of our vulnerability to natural variability as well as the potential for future climate change. In the face of uncertainties associated with the observed record and model predictions, we must adopt practical strategies for dealing with the potential impact of climate variability and change. These strategies should be based on two elements: (1) a strong observation and modeling research program within the U.S. designed to enhance economic vitality and national security, and (2) a litmus test for decision makers based on the level of risk and vulnerability to natural variability as well as future climate change. These two elements provide the most logical basis for policy decisions.