Climate Model Simulation of Point Rainfall Frequency Characteristics

Atmospheric scientists are continuously improving the capability of climate models to predict accurately the frequency of heavy precipitation events in a changing climate. A brief review of recent studies related to the accuracy of precipitation outputs from climate models indicated that most of them concentrate on spatially aggregated precipitation patterns rather than point rainfall frequency characteristics. To fill this gap, this verification study focused on point rainfall frequency characteristics and used the most recent climate model results (that is, results from a regional climate model known as CMM5, by Liang et al.). Both the conventional moving-window analyses and an alternative meteorological event-based approach were used in this study to examine more completely the capability of CMM5 in reproducing rainfall events as observed at individual points. It was found that CMM5 reproduced the major characteristics of point rainfall frequency distributions reasonably well. The meteorological event-based analyses proposed in this study also assisted in the identification of a minor but perhaps systematic bias on the part of CMM5 in overestimating days with small amounts of precipitation. The overall performance of the current generation of climate models as demonstrated in this study encourages more and wider uses of their results in hydrologic engineering.