Predictability of mesoscale quantitative precipitation

Controlled numerical model experiments are used to examine the growth of mesoscale precipitation forecast error due to variations in synoptic analyses used to initialize a mesoscale numerical model. A simulated true atmosphere was created by running the Navy's Coupled Ocean Atmosphere Mesoscale Prediction System (COAMPS) on a land-falling precipitating frontal system that occurred during the California Land-falling Jets (CALJET) experiment. MM5 forecasts were generated from 10 slightly different synoptic-scale analyses created by blending fixed first guess fields with a fixed number of observations taken from the simulated true atmosphere at varying locations. Verification was then made against the true COAMPS atmosphere to assess the predictive skill of the MM5 forecasts.

Results for the wind fields show that the mesoscale prediction error depends strongly on the lateral boundary conditions and is dominated by the larger-scale errors. The mesoscale wind error exhibits little growth with time over a 24 h forecast. This study will examine the quantitative precipitation forecasts to document the characteristic error growth, relative prediction skill of larger scale forecasts (coarser nests) versus mesoscale ensembles, and whether numerical precipitation forecasts at 4km grid resolutions have any real predictive skill for this single challenging event.