By Eric Barron, The Pennsylvania State University

Climate model predictions provide the estimates that have stimulated much of the national and international debate concerning the importance and the nature of future climate. These model projections provide estimates for many of the climatic variables of importance to society, and hence, the model predictions provide a key framework for assessing the impacts of climate change. Because of limitations in understanding and in computational resources, and because we cannot know precisely what emissions will occur, the climate models employed by scientists around the world have different characteristics and simplifications.

Therefore, these models provide somewhat different simulations of the future climate. Consequently, the model predictions are said to only provide an estimate, or "scenario", of possible future climates. The first U.S. National Assessment is based on a climate information strategy of providing a physically-consistent climate foundation for regional and sector assessments to be utilized by every team, with the opportunity for teams to perform additional independent analyses.

The strategy for providing climate scenarios for regional and national impact assessment is based on several key needs. First, a historical climate record is needed in order to have a basis for assessing the importance of climate and climate change. Second, the range of future climates used in the assessment process must be sufficiently broad to reflect the levels of uncertainty in models and in our projections of how society may evolve.

Third, the assessment must reflect the range and character of natural variability (like El Nino) and a sense of the spatial variability of climate. Fourth, the period of model predictions must overlap with the period of historical observations in order to evaluate the capabilities of the models. Finally, the assessment process should include opportunities to determine thresholds or limits in human and ecosystem adaptability. The selection of the set of climatic information for the assessment must also recognize both time and human constraints. For these reasons, the climate scenarios being provided to each team form a minimum basis for their assessment.

The historical data sets available for the U.S. Assessment include:

1.  The United States Historical Climatology Network (HCN) data set, which is maintained by the National Climatic Data Center and contains monthly averaged maximum, minimum, and mean temperature and total precipitation data for 1200 of the highest quality observing stations in the continental United States for the period 1895 to 1997;
2.  The Daily Historical Climatology Network data set, which contains observations for 187 high-quality stations in the contiguous U.S. for the 1910-1997 time period and observations for 1000 stations in the contiguous USA for the 1948-1997 time period;
3.  The VEMAP (Vegetation/Ecosystem Modeling and Analysis Project) spatially-uniform 1° x 1° record (1895-1993), which is based on 8000 HCN and U.S. Forest Service and Bureau of Land Management Sno-Tel stations for high elevation precipitation.

Each of the climate simulations is based on models that include both the ocean and the atmosphere, and for which the atmospheric greenhouse gas concentrations and sulfate emissions evolve with time. Climate simulations that will be used for the U.S. Assessment include those from the United Kingdom's Hadley Centre for Climate Prediction and Research (HADCM2) and the Canadian Centre for Climate Modeling and Analysis (CGCM1). Model simulations using the National Center for Atmospheric Research Climate System Model (NCAR CSM) and Department of Energy Parallel Climate Model (DOE PCM) runs are also expected to become available to the synthesis team for examination. Variables that will be provided to the assessment teams include surface air temperature, maximum and minimum surface air temperature, total precipitation, soil moisture, solar radiation, wind speeds, humidity, and sea-level. The model output will be provided for each region for the periods 2025-2034 and 2090-2099.

As for all simulations, the model simulations adopted by the assessment process have limitations. Foremost are questions whether these limited models will provide sufficient richness to describe the character of natural variability or the appropriate magnitude of climate response needed to achieve a realistic analysis of climate impacts. Because of this and other limitations, an assessment approach is being encouraged that also allows individual regions or sectors to consider "what if" cases that reflect educated guesses based on the nature and importance of specific regional and sector vulnerabilities. For example, given its importance, the southwestern region might consider intensification or weakening of the monsoonal circulation as a thought experiment in an effort to define or understand regional or sectoral pressure points or high levels of vulnerability.

The historical and model data provide an encompassing basis for assessing climate impacts. The historical record indicates that the greatest U.S. warming over the last century has occurred in the northeast and southwest. The southeast is one of the handful of places in the world indicating some cooling. The historical record also indicates that the U.S. has experienced modest increases in precipitation, with the exception of the precipitation decreases in the upper Great Plains area and parts of Alaska. The climate models predict warmer temperatures throughout the world for increased greenhouse gas concentrations.

The amount of warming more than doubles over the United States in both models between the two periods of study (2025-2034 and 2090-2099). The HADCM2 model predicts a greater than 4° C mean annual near-surface temperature increase over the western U.S. in 2095, whereas CGCM1 indicates a greater than 6°C increase over the central U.S. for the same interval. The total annual precipitation ratios for the decade 2025-2034 and 2090-2099 indicate that both models predict increasing precipitation throughout most of the world as a direct effect of warmer temperatures. On an annual average basis, both models indicate rather modest mean annual precipitation differences for the U.S., with the exception of a large predicted increase in the precipitation off the California coast. On a seasonal basis, however, the changes are larger.