Intra-annual and interannual ensemble forcing of a regional climate model

The use of ensemble modeling within the framework of dynamical downscaling of climate change scenarios derived from global climate model scenarios has not been fully explored. This study uses a six member ensemble of RegCM2 regional climate model simulations forced by the CCM3 global climate model to explore the one-way boundary forcing of regional interannual variability of 500 mbar heights, precipitation, and surface temperature. Anomaly pattern correlations (APCs) between the CCM3 and the RegCM2 500 mbar heights, precipitation, and surface temperature show distinct annual cycles. The January ensemble-averaged APCs for 500 mbar heights, precipitation, and surface temperature are 0.95, 0.65, and 0.90, respectively. The July correlations for the same variables are 0.63, 0.14, and 0.52, respectively. This indicates that the RegCM2 winter interannual variability is strongly dependent on the GCM interannual variability. The summer interannual variability of precipitation is found to contain little GCM-supplied signal. The ensemble run variance of the CCM3 and RegCM2, is also explored. The ratio of RegCM2 to CCM3 500 mbar height normalized ensemble run variance (NERV), a measure of climate reproducibility, is near 1.0 for various regions in the simulated domain. The RegCM2 precipitation NERV is greater than CCM3 NERV, suggesting less reproducibility and therefore less predictability. Certain regions show statistically significant reduced RegCM2 surface temperature NERV, suggesting that greater reproducibility may exist in these regions. The effect of increased topographic resolution in the RegCM2 domain was not found to significantly enhance reproducibility.