Ensemble forecasts of a flood-producing storm: comparison of the influence of model-state perturbations and parameter modifications

摘要: High-resolution ensemble simulations (Δx = 1 km) are performed with the Met Office Unified Model for Boscastle (Cornwall, UK) flash-flooding event of 16 August 2004. Forecast uncertainties arising from imperfections in forecast model analysed by comparing simulation results produced two types perturbation strategy. Motivated meteorology event, one type alters relevant physics choices or parameter settings model's parametrization schemes. The other is designed to account representativity error boundary-layer parametrization. It makes direct changes state and provides a lower bound against which judge spread uncertainties. has genuine skill at scales approximately 60 km an can be estimated within ∼ 10% only eight members. Differences between model-state modification strategies discussed, former being more important triggering latter subsequent cell development, including average internal structure convective cells. Despite such differences, rainfall evaluated skilful shown weakly sensitive This suggests that relatively simple treating uncertainty may sufficient practical, convective-scale forecasting. Copyright © 2012 Royal Meteorological Society British Crown Copyright,