Abstract - PIRCS Precipitation

 

Anderson, C.J., R. W. Arritt, E. S. Takle, Z. Pan, W. J. Gutowski, R. da Silva, and PIRCS modelers, 2003: Hydrologic processes in regional climate model simulations of the central United States flood of June-July 1993. J. Hydrometeor. (in press).

Regional climate model (RCM) simulations of hydroclimate for the central U. S. are sensitive to RCM design, yet comparison of RCM results under common experimental conditions is rare. Thus the degree of and sources for inter-model variability are not well known. We have compared 60-d simulations of 1993 June-July from thirteen RCM simulations to each other and observations. Boundary data and initial conditions were supplied by the Project to Intercompare Regional Climate Simulations (PIRCS) experiment 1b. We have examined water vapor conservation and precipitation characteristics in each RCM for a 10Ox10O sub-region of the Upper Mississippi River Basin (UMRB), containing the region of maximum 60-d accumulated precipitation in all RCMs and station reports.

Results showed that gross features of hydroclimate were well simulated in all RCMs. Specifically, all RCMs produced positive precipitation minus evaporation (PE> 0), and RCM recycling ratios were within the range estimated from observations. The range of P-E in RCMs enveloped the range of estimates of observed P-E, but most RCMs produced P-E below the estimated observed range. We found sensitivity of RCM E to radiation parameterization, including clouds, but inter-model variability of E was spread evenly about estimates of observed E suggesting little, if any, common errors of E among the simulations. In contrast, most RCMs produced P that was below the range of P from observations; thus a common dry bias of the simulations accounted for the low values of simulated P-E compared to observations.

Daily cycles of terms in the water vapor conservation equation revealed that P in most RCMs is driven by the dynamics of atmospheric circulation. In most simulations nocturnal maxima of P and C (convergence) occurred simultaneously, consistent with observations of P and climatological studies of water vapor conservation. Three of the four driest RCMs had maximum P in the afternoon, while the time of maximum C was variable, suggesting that in these RCMs afternoon destabilization by insolation strongly influenced the precipitation process. When 60-d accumulated precipitation was decomposed as the sum of 3-h precipitation totals, a larger fraction of 60-d accumulated precipitation in all RCMs compared to station reports was from low 3-h totals. This tendency was exaggerated in the driest simulations. In station reports, accumulation from high 3-h totals had a nocturnal maximum, whereas accumulation from low 3-h totals had an early morning maximum. Satellite imagery suggests that this time lag between maximum accumulation from high and low 3-h totals occurred, in part, because many mesoscale convective systems had reached peak intensity overnight and had declined in intensity by early morning, while having significant overlap with the UMRB box. None of the RCMs contained this time lag between maximum accumulation from 3-h totals. We therefore recommend additional tests of the ability of RCMs to simulate the effects of mesoscale convective systems.