Assessment of NCAR general circulation model precipitation in comparison with observations

Abstract

In this study precipitation fields from six different present-day climate experiments are compared to observed data. The six NCAR general circulation model experiments are: (a) prescribed sea-surface temperature (SST) variable hydrology CCM1, (b) prescribed SST fixed hydrology CCM1, (c) mixed layer CCM1, (d) prescribed SST w/ BATS CCM0B (e) mixed layer CCM0A, and (f) mean annual CCM0A. In general, predicted precipitation compares more favorably with observations for the case of prescribed SSTs, than for the overly zonal energy balance ocean experiments. However, despite shortcomings, the energy balance ocean experiments have utility for climate change experiments due to their lack of bias associated with specifying ocean temperature distributions. The most favorable comparisons are in the subtropics and mid-latitudes, followed by the tropics and poles. The model is best at reproducing large scale features such as the sub-tropical deserts and mid-latitude frontal precipitation. It is poorest at stimulating features that are dependent upon sub-resolution processes such as orographic uplift and convection. The model reproduces the major observed trends between the solstice seasons, such as lower land/ocean precipitation fractions in DJF, although many regional inaccuracies exist. Continental precipitation is generally excessive in the model, although the global combined continental and oceanic (CAO) averages are quite close to the observed value. Observed precipitation typically accounts for between 0.3 and 0.4 of the global spatial variance in the model data.