A scheme for verifying the spatial structure of extremes in numerical weather prediction: Exemplified for precipitation

Abstract

This paper describes a novel scheme for assessing the ability of a numerical weather prediction (NWP) model to forecast the horizontal spatial structure of local extremes, for example, when comparing accumulated precipitation to a corresponding observed field. Both local maxima and minima are considered. The quality is measured by a score function between 0 and 1, which could potentially be adjusted for specific applications of NWP. The effect of neighbourhood size is included in order to consider the impact of resolvable scales. The scheme complements existing spatial verification schemes with added value and has a potential to be used both in operational verification and as a tool for revealing model deficiencies linked to individual forecast cases. The properties of the scheme are illustrated from various idealized cases. Verification results are presented for a real convective precipitation case and for a frontal precipitation case over several days. The scheme is discussed in the context of operational conditions, using synthetic fields of forecasts and analyses based on statistics of an operational NWP model at the Danish Meteorological Institute (DMI) and related observed data from a 5-year period. Different future options are outlined, including the potential application of the scheme for ensemble prediction systems.