Tuesday, 24 January 2012
Evaluation of 20th Century Climate Model Simulations of Heavy Precipitation Over North America
Hall E (New Orleans Convention Center )
There is considerable evidence that heavy precipitation events have increased during the period of instrumental records and that this may have been associated with human-induced greenhouse warming. In order to faithfully characterize future changes in intense precipitation due to greenhouse warming and diagnose the physical mechanisms responsible for such changes, the reliability of climate model simulations of precipitation must be better understood. Here, daily precipitation from 20th century simulations from the Coupled Model Intercomparison Project Phase III (CMIP3) is compared with gridded observations from the NOAA Climate Prediction Center over North America for the period 1979-1999. Additionally, the model-simulated large scale environment associated with intense precipitation events is compared with North American Regional Reanalysis data. We found that simulated heavy precipitation is generally smaller than observed in the southeastern United States and does not follow observed topographically forced precipitation patterns along the Pacific coast. The model biases in the southeastern United States are likely due to inadequacies in convective parameterizations in climate models. In the west, inadequate representation of the topography due to the coarse model resolutions likely leads to the observed biases there. In accordance with previous studies, the CMIP3 models are found to overestimate the frequency of lighter precipitation events and underestimate heavier events. Despite quantitative deficiencies in the simulation of heavy precipitation events in some regions, a composite analysis of the simulated and observed large-scale environmental conditions associated with such events suggests that the physical mechanisms that produce heavy precipitation are simulated realistically over most regions. These results suggest that we may have some confidence in using the CMIP3 models to diagnose and understand the physical processes that lead to future changes in heavy precipitation events.