Verification of quantitative precipitation forecasts on short time-scales: A fuzzy approach to handle timing errors with SAL

Abstract

Verification of quantitative precipitation forecasts (QPFs) accumulated over short time periods (hours or less) becomes particularly challenging due to the high temporal variability of surface precipitation. Present-day numerical weather prediction models are, in principle, capable of realistically simulating intense precipitation events, but - also due to the stochastic nature of the triggering of convection - they sometimes fail in predicting the events at the right time. Such temporal prediction errors can have a strong impact on the results of QPF verification and they potentially lead to a misleading interpretation of the verification results. In this study, a fuzzy approach is introduced to handle and quantify timing errors when verifying QPFs accumulated over short time periods, and the approach is illustrated with the feature-based quality measure SAL. Instead of comparing an observed precipitation field only with the simultaneous QPF (the standard approach), it is proposed to compare the observations with a set of QPFs with a time shift of [3. . . + 3] hours and to determine the time shift Δt associated with the smallest QPF location error (the fuzzy approach). Both approaches have been applied to a set of QPFs from the operational weather prediction models COSMO-DE and COSMO-EU in the German part of the Elbe catchment in summer 2007. It is shown that the fuzzy and standard approaches can lead to fairly different verification results, confirming the hypothesis that timing errors significantly impact upon the results from the standard approach. In addition, the interquartile range of the structure and amplitude components of SAL are substantially reduced with the fuzzy approach, indicating that when using the standard approach timing errors in cases are manifested as particularly large errors of the amplitude and/or structure component of SAL. It follows that application of the fuzzy approach is a meaningful way of better identifying key forecast qualities and deficits when considering QPFs with short accumulation times.