ABSTRACT We present, motivate and evaluate Radar Maxima, a calibrated area‐based probabilistic forecast product for heavy precipitation. It is designed to overcome inherent limitations of point‐based forecasts, which often yield low probabilities for extreme events due to spatial displacement errors. The method aggregates radar‐derived precipitation within 40 km neighbourhoods to statistically upscale forecasts from DWD's ensemble system ICON‐D2‐EPS. Evaluation considers both objective verification metrics and feedback from operational weather forecasters based on case studies. Results are compared against both pointwise and spatially aggregated uncalibrated ensemble forecasts in order to disentangle the effects of spatial aggregations and statistical calibration. Spatial aggregation yields improved predictability, reliability and forecast sharpness, while the impact of calibration is mixed. Feedback from forecasters confirms that Radar Maxima provides operational value in specific situations.
Reinhold Hess (Sun,) studied this question.