Parametric post-processing of dual-resolution precipitation forecasts

Abstract Recently, all major weather centres issue ensemble forecasts, which differ both in size and spatial resolution – even while covering the same domain. These parameters directly determine forecast skill of prediction computation cost. In last few years, plans upgrading configuration Integrated Forecast System European Centre for Medium-Range Weather Forecasts (ECMWF) from a single with 9 km 51-member 18 induced an extensive study raw post-processed dual-resolution predictions comprising members different horizontal resolutions. We investigate predictive performance censored shifted gamma (CSG) model output statistic (EMOS) approach statistical post-processing help 24h precipitation accumulation forecasts over Europe various horizons. consider operational 50-member ECMWF as high-resolution extend it low-resolution (29-km grid) 200-member experimental forecast. The investigated combinations consist subsets these two ensembles equal computational cost, is equivalent to cost ensemble. Our case verifies that, compared rawensemble combinations, EMOSpost-processing results significant improvement that statistically indistinguishable between any analysed mixtures combinations. Furthermore, semilocally trained CSG EMOS provides efficient alternative state-of-the-art quantile mapping without requiring additional historical data.