A comparison of CCM2/BATS simulated precipitation and runoff with observed values over the continental United States

Abstract

In order to evaluate the ability of general circulation models to predict future climate change, it is useful to know how realistically they simulate the current climate. Precipitation and runoff from two simulations of the National Center for Atmospheric Research Community Climate Model 2 coupled with the Biosphere-Atmosphere Transfer Scheme (NCAR CCM2/BATS) are compared to observed values over the United States. Seven different methods are used to compare simulated and observed values over several different temporal and spatial scales. Precipitation and runoff are not realistically simulated over much of the United States. Precipitation is generally too high. Simulated peak runoff occurs one to two months earlier than observed peak runoff in watersheds with a winter snowpack. Some of the errors can be attributed to coarse model resolution, which does not capture the complex sub-grid details of topography which are important factors in determining local climate.