Studies of the regional impacts of climate change must confront the problem of choosing climate change scenarios and global climate model simulations. The task of downscaling global climate model simulations is often so demanding that only a limited selection of models and greenhouse gas emission scenarios may be considered. However, it is preferable to consider a range of scenarios in climate impacts studies (see, for example, IPCC, 2001, page 741). Using multiple models allows a better representation of the uncertainties and range of possible outcomes. Furthermore, each model's performance varies for different regions and it is important to evaluate and choose models that perform well for the particular region and application under consideration. Global climate simulations based on present-day conditions may be used for evaluation.

The models and emissions scenarios included in the recent Intergovernmental Panel on Climate Change (IPCC) (2001) report form a convenient collection of scenarios for many studies, and global model simulations for these scenarios is available via the internet (http://ipcc-ddc.cru.uea.ac.uk/) for six models and up to seven emissions scenarios each. For this talk, selection of three models from the IPCC assessment will be presented. Simulations using historic forcing for 1900-1990 and and using the A2 future emissions scenario for 1990-2100 will be considered.

A simple empirical method (Salathé, 2003; Widmann et al., 2003) is applied to downscale the archived climate change simulations. The method is implemented using readily available data and is fast enough to downscale a large collection of scenarios for long time periods. All data used to fit the downscaling method and the global climate simulations that are downscaled are available via the internet.

The downscaled results are used to simulate streamflow in a mesoscale river basin, the Yakima River in central Washington, USA. The hydrologic response to climate change is an important factor in the impacts of climate on many regions worldwide. Thus the suitability of the global climate model and downscaling procedure for hydrologic modeling is an important test of the climate change scenario. The examples discussed show how downscaled results may be evaluated as part of model selection for an impacts study.

This talk focuses on the quality of the global simulations forced by historic climate conditions as a means to verify the climate change simulations. Since the downscaling method is quite simple and minimally constrains the global simulation, the statistics of the global model are left largely intact. Comparisons among the models presented will help establish how well various global models represent the statistical nature of the meteorological parameters controlling streamflow. Several important considerations in the global climate model simulations are brought to light by this analysis.