Event-based comparison between general circulation model precipitation results and point observations

Coupled atmosphere-ocean general circulation models (GCMs) are widely used to study the current climate system as well as its possible changes in the future. The precipitation output from these models has been compared to observations in order to verify the reliability and accuracy of GCMs in simulating the precipitation process. Depending on the specific interest and emphasis, different yearly, seasonal, and extreme condition statistics derived from modeled and observed precipitation time series have been compared. So far, however, there is no report of comparison between modeled and observed rainfall events for any specific locations. In this paper, rainfall events as observed at a individual site are defined first. Event-related statistics are then calculated using modeled and observed precipitation time series. Comparison of these event-based statistics shows that while GCMs cannot model the rainfall process at the two selected locations on a year-by year, or event-by-event basis, they do model fairly well the long-term average number of rainfall events and event volume distributions. Results presented in this paper may render more confidence in the use of certain information derived from GCM precipitation output, and at the same time assist in identifying directions for further improvement in the modeling of precipitation processes.