Regional Climate Simulations with Bias Corrected GCM Forcing Data for the Western United States

Performing future regional climate simulations requires the use of global climate model output; however, biases in the global climate model output are passed into and could be magnified in the regional model. For this study, we first force the Weather Research and Forecasting (WRF) model at 32km resolution with National Center for Environmental Prediction Reanalysis I (R1) data for the western United States (U. S.) to find the optimal physics combination. Using the same physics options, WRF was then driven with the output from the Community Climate System Model (CCSM) to assess how biases in such forcing data affect regional climate simulations. Our results show that CCSM forcing produces large positive biases in precipitation throughout the Western U.S., particularly at higher elevations. We find the cause of these precipitation biases to be a result of cold biases in upper tropospheric temperatures and stronger zonal winds when compared to North American Regional Reanalysis (NARR) data. As the physics were unchanged, this indicates that the biases are a direct result of the CCSM forcing. Therefore, we propose a bias correction technique to improve the forcing data before ingestion into WRF. Climatological biases in CCSM data were removed based on R1 data at six hourly intervals for the period of 1949 through 1999 for the forcing variables (e.g, temperature, specific humidity), while the physical consistency among these variables was taken into account. This bias corrected CCSM data was used to drive WRF. The results show improved temperature and precipitation estimates compared to the unmodified CCSM forced simulations. Thus, the methodology used in this study to improve regional climate modeling shows a strong potential to produce reliable future climate forecasts.