Postprocessing of NWP Precipitation Forecasts Using Deep Learning

Abstract This study analyzes the potential of deep learning using probabilistic artificial neural networks (ANNs) for postprocessing ensemble precipitation forecasts at four observation locations. We split forecast problem into two tasks: estimating probability and predicting hourly precipitation. then compare performance with classical statistical (logistical regression GLM). ANNs show a higher three stations all Further, more general ANN models are trained merged data from stations. These exhibit an increase in compared to station-specific most However, they significant decay one The seem capable meaningful interactions generalizing these improve other sites, which also causes loss local information station. Thus, this indicates weather forecasting workflows.