Spatially coherent postprocessing of cloud cover ensemble forecasts

Abstract

Statistical postprocessing is commonly applied to reduce location and dispersion errors of probabilistic forecasts provided by numerical weather prediction (NWP) models. If postprocessed forecast scenarios are required, the combination of ensemble model output statistics (EMOS) for univariate postprocessing with ensemble copula coupling (ECC) or the Schaake shuffle (ScS) to retain the dependence structure of the raw ensemble is a state-of-the-art approach. However, modern machine learning methods may lead to both a better univariate skill and more realistic forecast scenarios. In this study, we postprocess multimodel ensemble forecasts of cloud cover over Switzerland provided by COSMO-E and ECMWF-IFS using (i) EMOS 1 ECC, (ii) EMOS 1 ScS, (iii) dense neural networks (dense NN) 1 ECC, (iv) dense NN 1 ScS, and (v) conditional generative adversarial networks (cGAN). The different methods are verified using EUMETSAT satellite data. Dense NN shows the best univariate skill, but cGAN performed only slightly worse. Furthermore, cGAN generates realistic forecast scenario maps, while not relying on a dependence template like ECC or ScS, which is particularly favorable in the case of complex topography. SIGNIFICANCE STATEMENT: Often, statistical postprocessing is applied to the output from weather models to reduce systematic errors. However, traditional postprocessing approaches destroy the spatial patterns of the predictions of the weather model. For instance, cloud cover amounts at two neighboring points on a plain are more likely to be correlated with each other than with the amount on a nearby mountain. In traditional postprocessing, the spatial patterns of the weather model are often imposed on the postprocessed forecasts using the output from the weather model as a template. However, this approach is suboptimal for cloud cover. Therefore, we have applied a modern machine learning method, designed to produce realistic-looking images, to obtain postprocessed and physically realistic maps of cloud cover forecast scenarios.