A strategy for high-resolution ensemble prediction. I: Definition of representative members and global-model experiments

摘要: In the last few years, tens of alternative weather forecasts have been made available to forecasters by operational ensemble prediction systems. many forecasting applications, it is useful identify (possibly in an objective way) a representative members, deemed represent most interesting scenarios. this paper, strategy select members (RMs hereafter) from developed, and applied four cases medium-range performed with European Centre for Medium-Range Weather Forecasts (ECMWF) Ensemble Prediction System (EPS). The case-studies correspond events very intense rainfall (leading localized floods) Alpine region, selected as benchmarks numerical simulations Mesoscale Programme. RM selection procedure uses cluster analysis its first step. For each cluster, defined be member smallest ratio between average distance own other clusters. Distances are computed either using L2-norm 700 hPa geopotential height fields or L1-norm precipitation fields. RMs compared centroids extreme rainfall. By definition, RMs characterized synoptic-scale atmospheric flow similar corresponding centroid, but they contain more small-scale features, especially parameters such precipitation. RM initial conditions can used initiate higher-resolution global forecasts; alternatively, may define boundary nested high-resolution limited-area models. Integrations ECMWF model at T1 319 horizontal resolution (compared 159 EPS) were performed. Results indicate that forecast, started conditions, remains closer low-resolution than provides detailed forecast parameters, regions complex topography. Experiments model, same set RMs, described companion paper.