Predictability of Precipitation in a Cloud-Resolving Model

Abstract

An ensemble methodology is developed and tested to objectively isolate and quantify meso-?-scale predictability limitations in numerical weather prediction (NWP). The methodology involves conducting an ensemble of limited-area simulations with slightly modified initial conditions (representing small-scale observational uncertainties) and identical lateral-boundary conditions (representing perfect synoptic-scale predictability). The methodology is applied using a nonhydrostatic NWP model with a convection-resolving mesh size of 3 km, using a setup covering the entire European Alps. The initial perturbations of the ensemble members have a small-scale structure with predominant scales between 10 and 100 km. Ensembles for four case studies representing different weather conditions are analyzed for 24-h forecasting periods, with particular attention paid to quantitative precipitation forecasting. The simulations show that the predictability of precipitation amounts differs strongly depending upon the weather type and the spatiotemporal scales considered. It is demonstrated that during episodes of convective activity small-scale predictability limitations may be critical even at scales exceeding 100 km. For smaller spatial scales, the uncertainties in precipitation forecasts increase rapidly with decreasing scale as individual convective cells are rendered unpredictable by chaotic aspects of the moist dynamics. However, the results also suggest that the presence of convective activity alone may not necessarily limit predictability. Additional consideration is given to the role of underlying orography, nonlinear processes, and perturbation growth.