Application of object-based time-domain diagnostics for tracking precipitation systems in convection-allowing models

Abstract

Meaningful verification and evaluation of convection-allowing models requires approaches that do not rely on point-to-point matches of forecast and observed fields. In this study, one such approach-a beta version of the Method for Object-Based Diagnostic Evaluation (MODE) that incorporates the time dimension [known asMODEtime-domain (MODE-TD)]-was applied to 30-h precipitation forecasts from four 4-km grid-spacing members of the 2010 Storm-Scale Ensemble Forecast system with different microphysics parameterizations. Including time in MODE-TD provides information on rainfall system evolution like lifetime, timing of initiation and dissipation, and translation. The simulations depicted the spatial distribution of time-domain precipitation objects across the United States quite well. However, all simulations overpredicted the number of objects, with the Thompson microphysics scheme overpredicting the most and theMorrison method the least. For the smallest smoothing radius and rainfall threshold used to define objects [8 km and 0.10 in. (1 in. 5 2.54 cm), respectively], the most common object duration was 3 h in both models and observations. With an increased smoothing radius and rainfall threshold, the most common duration became shorter. The simulations depicted the diurnal cycle of object frequencies well, but overpredicted object frequencies uniformly across all forecast hours. The simulations had spurious maxima in initiating objects at the beginning of the forecast and a corresponding spurious maximum in dissipating objects slightly later. Examining average object velocities, a slow bias was found in the simulations, which was most pronounced in the Thompson member. These findings should aid users and developers of convection-allowing models and motivate future work utilizing time-domain methods for verifying high-resolution forecasts. © 2014 American Meteorological Society.