Evaluation and comparison of regional and global model simulations of climate over the Susquehanna River Basin

To predict future changes in wetlands and river systems, hydrological models rely on input from meteorological climate models. Regional climate models are often chosen over global models as inputs because they offer a higher spatial and temporal resolution that is closer to the resolution of many hydrological models. To determine if regional models are also more accurate at predicting large-scale climate patterns, simulations of climate over the Susquehanna River Basin by global models from the Coupled Model Intercomparison Project phase 3 (CMIP3) are compared with simulations by nested regional models from the North American Regional Climate Change Assessment Program (NARCCAP). To perform the comparison, data from all of the simulations and observations are interpolated to a one-degree grid, and means, variances, and errors are computed for variables of importance for forcing a hydrological model. The model results are evaluated against observations during the current period (1979-1998) and against other models during the mid 21st century period (2046-2065). The evaluation identifies two regional models that perform significantly better at a one-degree scale than other regional models and most global models. As a whole, however, the group of regional models does not show a significant improvement over the group of global climate models. Caution is therefore needed when selecting a climate model to force another model, as higher resolution is not necessarily correlated with higher large-scale accuracy.