Spatial Forecast Verification and the Mesoscale Verification Intercomparison over Complex Terrain

Spatial forecast verification is the collection of methods proposed in recent decades for performing more diagnostic and meaningful analyses of forecast performance, as well as to provide summary measures that are not susceptible to problems of double penalites in the face of timing or spatial displacement errors or over accumulations of small-scale (unimportant) errors.  Most of the methods originated in the image analysis and computer vision fields, although some are more unique to weather forecast verification.  The Mesoscale Verification Intercomparison over Complex Terrain (MesoVICT) is the second phase of a meta verification project aimed at investigating the numerous proposed methods and providing guidance about how the methods inform about forecast performance.  The first phase aimed primarily at precipitation forecasts over the central United States, and the second phase is aimed at complex terrain over central Europe, inclusion of ensemble forecasts, ensembles of observations, and also expands the variable sets to include other forecast variables, such as wind. This talk will give an overview of the project and methods, and outline areas that might be ripe for collaborations with statisticians and computer scientists, and perhaps researchers from other fields.