Analysis of daily variability of precipitation in a nested regional climate model: comparison with observations and doubled CO2 results

We analyze daily mean, variability, and frequency of precipitation in two continuous 31/2 year long climate simulations over the continental U.S., one for present conditions and one for conditions under doubled carbon dioxide concentration, conducted with a regional climate model (RegCM) nested in a general circulation model (GCM). The purpose of the work is to analyze model errors and limitations in greater detail than previously done and to calculate quantities that eventually will be used to form climate change scenarios that account for changes in daily variability of precipitation. The models used are a version of the NCAR Community Climate Model (CCM) and the climate version (RegCM) of the NCAR/Penn State mesoscale model (MM4) at 60 km horizontal grid point spacing. Model output is compared with a 30-year daily observational data set for mainly two regions of the U.S.: the Northwest, and the central Great Plains. Statistics compared include mean daily precipitation, mean daily intensity, frequency, transition probabilities, quantiles of precipitation intensity, and interquartile ranges. We discuss how different measures of daily precipitation lead to different conclusions about the quality of the control run. For example, good agreement between model and observed data regarding mean daily precipitation usually results from compensating errors in the intensity and frequency fields (too high frequency and too low intensity). We analyze how detailed topographic features of the RegCM enhance the simulation of daily precipitation compared to the CCM simulation. In general, errors in all measures are smallest at the Northwest grid points, and the damping of the seasonal cycle of mean daily precipitation from the coast to inland Oregon is basically well reproduced. However, some errors in the frequency and intensity fields can be traced to inadequate representation of topography, even with a horizontal resolution of 60 km. Differences in the control and doubled CO2 runs (for both RegCM and CCM) for these regions are also presented. The most significant changes for the RegCM grid points is increased variability of daily precipitation under doubled CO2 conditions. Areas with significant changes (both increases and decreases) of precipitation frequency and intensity are found. There are some areas where frequency decreases, but precipitation mean daily amounts increase. Such changes, which would be masked by more traditional analyses of precipitation change, are important from a climate impacts point of view. The limitations on the analyses posed by small sample sizes are discussed.