Potential Predictability of U.S. Summer Climate with “Perfect” Soil Moisture

Abstract

The potential predictability of surface-air temperature and precipitation over the United States was assessed for a GCM forced by observed sea surface temperatures and an estimate of observed soil-moisture content. The latter was obtained by substituting the GCM-simulated precipitation, which is used to drive the GCM's land surface component, with observed pentad-mean precipitation at each time step of the model's integration. With this substitution, the simulated soil moisture correlates well with an independent estimate of observed soil moisture in all seasons over the entire U.S. continent. Significant enhancements for the predictability of surface-air temperature and precipitation were found in boreal late spring and summer over the U.S. continent. Anomalous pattern correlations of precipitation and surface-air temperature over the U.S. continent in the June?August season averaged for the 1979?2000 period increased from 0.01 and 0.06 for the GCM simulations without precipitation substitution to 0.23 and 0.31, respectively, for the simulations with precipitation substitution. The results provide an estimate for the limits of potential predictability if soil-moisture variability is to be perfectly predicted. However, this estimate may be model dependent and needs to be substantiated by other modeling groups.