Parametric post-processing of dual-resolution precipitation forecasts

TL;DR

This study evaluates a statistical post-processing method for dual-resolution precipitation forecasts, demonstrating significant improvements in forecast skill and efficiency over raw ensemble predictions, with the CSG EMOS approach effectively combining different resolution ensembles.

Contribution

The paper introduces the use of the censored shifted gamma EMOS model for dual-resolution precipitation forecast post-processing, showing it matches state-of-the-art methods without extra historical data.

Findings

EMOS significantly improves forecast skill over raw ensembles.

Dual-resolution combinations yield comparable skill after post-processing.

CSG EMOS performs well without additional historical data.

Abstract

Recently, all major weather centres issue ensemble forecasts which even covering the same domain differ both in the ensemble size and spatial resolution. These two parameters highly determine both the forecast skill of the prediction and the computation cost. In the last few years, the plans of upgrading the configuration of the Integrated Forecast System of the European Centre for Medium-Range Weather Forecasts (ECMWF) from a single forecast with 9 km resolution and a 51-member ensemble with 18 km resolution induced an extensive study of the forecast skill of both raw and post-processed dual-resolution predictions comprising ensemble members of different horizontal resolutions. We investigate the predictive performance of the censored shifted gamma (CSG) ensemble model output statistic (EMOS) approach for statistical post-processing with the help of dual-resolution 24h precipitation accumulation ensemble forecasts over Europe with various forecast horizons. As high-resolution, the operational 50-member ECMWF ensemble is considered, which is extended with a 200-member low-resolution (29-km grid) experimental forecast. The investigated dual-resolution combinations consist of (possibly empty) subsets of these two forecast ensembles with equal computational cost, being equivalent to the cost of the operational 50-member ECMWF ensemble. Our case study verifies that, compared with the raw ensemble combinations, EMOS post-processing results in a significant improvement in forecast skill and the differences between the various dual-resolution combinations are reduced to a non-significant level. Moreover, the semi-locally trained CSG EMOS is fully able to catch up with the state-of-the-art quantile mapping and provides an efficient alternative without requiring additional historical data essential in determining the quantile maps.