Evaluation of Precipitation and Temperature in NARCCAP Regional Climate Models over the North and South Carolina in Southeast US

Climate models are well known tool for predicting future climate condition. However, before application of the climate scenario to predict future water resources availability, comparing current climate model results with measured data are necessary to evaluate modeling performance. Current NARCCAP program provide four regional climate models (RCMs) and atmosphere-ocean general circulation models (AOGCMs) combinations for future climate change. So we evaluate those four results by comparing simulated and measured daily precipitation, daily maximum and minimum temperature in current time period. In this process, the model grid cells which include more than three weather stations are selected for model evaluation. We extract daily precipitation from precipitation file and maximum and minimum temperature from surface air temperature file. 10 year period (1991~2000) measured weather data and climate model results are reorganized by 7 day, 15 day, monthly scale, and seasonal scale aggregated sum (for precipitation) and mean (for daily maximum and minimum temperature) for statistical test. Precipitation simulations have weakness in most of the evaluation time scale; time series and one to one plots do not agree well between simulated and measure data. However, 10 year period cumulative distribution function does not show that much bias. Cold bias is significant in daily maximum temperature though some variations are found in each model result; some simulations, such as GFDL driven RCM3, has more severe cold bias than any other model, whereas WRFG with CCSM shows better fit to the measured temperature. This cold bias seems to be linear offset in cumulative distribution function; for example, CGCM3 driven RCM3 data shows approximately 5°C ~ 7 °C lower than measured data. Daily minimum temperature shows well agreement between simulated and measured data except for the winter season.