1. Interannual Variability of Warm-Season Rainfall over the US Great Plains in NCAR/CAM and NASA/NSIPP Simulations: Intercomparisons for NAME. Alfredo Ruiz–Barradas 1 and Sumant Nigam University of Maryland ----o---- 1 st International CLIVAR Science Conference Baltimore, MD 21-25 June 2004 Abstract Interannual variability of summer rainfall and moisture fluxes from two state-of-the-art AMIP model simulations (NCAR/CAM2.0 and NASA/NSIPP) are analyzed over the US Great Plains. The simulations are produced using the observed 1950-1998 lower boundary conditions. The retrospective U.S. and Mexican precipitation data sets, and the NCEP and 40-year ECMWF (obtained from its web site) reanalysis data sets are used as targets for the simulations. The simulations are in some disagreement with each other and with nature in portrayal of rainfall variability over the US Great Plains and its linkages over southern United States and Mexico. Notable Great Plains precipitation anomalies are linked with vertically integrated total (and largely stationary) moisture flux anomalies and associated convergence anomalies from the Gulf of Mexico in nature, but not in the simulations. Models do produce significant rainfall anomalies over central U.S. but these are likely generated from other processes like local evaporation; evaporation, through surface-atmosphere feedbacks, is more prominent in models than in observations. The Great Plains precipitation is found linked to Pacific SSTs in both simulations. Figure 1: Standard deviation of monthly precipitation during June- August. Box marks the Great Plains region used in defining an index. Contour interval is 0.3 mm/day, and values greater than 0.3 are shaded. Figure 2: Standard deviation of monthly summer precipitation in the NCAR/CAM2.0 and NASA/NSIPP AMIP simulations. Contour interval and shading as before. Figure 3: The Great Plains precipitation anomalies in observations; thick line shows the smoothed version (1-2-1 binomial filter applied 4 times). JJA correlations are based on unsmoothed data. Figure 4: The Great Plain precipitation index (smoothed version) is split up into its convective and large-scale condensation components for simulations and ERA-40. Total simulated precipitation is uncorrelated with observations. While models suggest purely convective precipitation, ERA-40 considers equally important large-scale precipitation. Summary Observational data sets are in some agreement in their description of US Great Plains precipitation variability. NCEP and ERA-40 reanalyses however exhibit lesser accord, with ERA-40 having higher correlations with observations. AMIP simulations are in considerable disagreement both with each other and with verifying observations of hydroclimate variability over the Great Plains. Models consider large-scale precipitation of not importance while ERA- 40 considers it equally important as convective precipitation. Great Plains precipitation variability is linked with rather different distributions of large-scale moisture fluxes in models and observations. Evaporation anomalies, through surface-atmosphere interactions in the models, produce precipitation anomalies in simulations in a stronger role than observations indicate. Great Plains precipitation is linked with Pacific SSTs in AMIP simulations, somewhat as in observations. Figure 5: JJA Regressions of precipitation, vertically integrated total moisture flux and its convergence (mfc), and evaporation on unsmoothed Great Plains precipitation index; moisture-flux vector scale is indicated; convergent regions are green. Precipitation anomalies are linked to mfc and southerly moisture flux anomalies. Note the smaller CI for evaporation. Figure 6: Similar to figure 5 but for simulations. Precipitation anomalies are not matched by mfc anomalies but by evaporation. CAM simulation lacks of transient zonal moisture flux data. Regressions of NSIPP ensemble-mean simulation is not too different from those of the ensemble member shown here. Figure 7: SST regressions on the smoothed Great Plains precipitation index during June-August. Top panel shows regressions with observations. SST regression using the index from the NSIPP ensemble-mean looks like that from CAM index over the Pacific. Positive anomalies are red. 1 alfredo@atmos.umd.edu