Fairplay in the verification of operational quantitative precipitation forecasts

摘要: Abstract The accuracy of weather forecasts is not only influenced by the skill forecasting system, but also itself. Here, we propose use odds ratio (benefit), ORB, as a measure, which base-rate event and thus enables fair comparison categorical for different years, regions, events, etc. ORB has simple interpretation it permits split into contributions from non-event. Applying this measure to operational quantitative precipitation reveals that more extreme (rare) events have than ‘normal’ contrary results typically obtained with other measures traditionally used in meteorology subjective perception. Both latter can be interpreted delusive consequence ‘neglect base-rate’ effect. Further consequences are described composition model trials verification forecast warnings. Over recent years there been trends showing small improvement large reduction bias slight precipitation.