Subgrid‐scale rainfall variability and its effects on atmospheric and surface variable predictions

A new approach, which combines the Penn State/National Center for Atmospheric Research mesoscale model MM5 with a recently developed statistical downscaling scheme, has been investigated for the prediction of rainfall over scales (grid sizes) ranging from the atmospheric model scale (>10 km) to subgrid scale (around 1 km). The innovation of the proposed dynamical/statistical hybrid approach lies on having unraveled a link between larger‐scale dynamics of the atmosphere and smaller‐scale statistics of the rainfall fields [ Perica and Foufoula‐Georgiou , 1996a], which then permits the coupling of a mesoscale dynamical model with a small‐scale statistical parameterization of rainfall. This coupling is two‐way interactive and offers the capability of investigating the feedback effects of subgrid‐scale rainfall spatial variability on the further development of a rainfall system and on the surface energy balance and water partitioning over the MM5 model grids. The results of simulating rainfall in a strong convection system observed during the Oklahoma‐Kansas Preliminary Regional Experiment for STORM‐Central (PRESTORM) on June 10–11, 1985 show that (1) the dynamical/statistical hybrid approach is a useful and cost‐effective scheme to predict rainfall at subgrid scales (around 1 km) based on larger‐scale atmospheric model predictions, and (2) the inclusion of the subgrid‐scale rainfall spatial variability can significantly affect the surface temperature distribution and the short‐term (<24 hour) prediction of rainfall intensity.