A weather-regime-dependent metric for verifying spread-error relationship : Ensemble Performance Index for High-Impact Weather (EPI4HIW)

The spread-error relationship is a crucial measure for an Ensemble Prediction System (EPS). It is found that existing metrics for verifying this relationship are unable to reflect the full picture and can even provide misleading information. This study proposes a new weather-regime-dependent verification metric: Ensemble Performance Index for High-Impact Weather (EPI4HIW). It treats spread and error as a compound variable and correlates them with weather patterns. It quantitatively describes the spatial structure of the spread-error relationship.