Intercomparison of Spatial Forecast Verification Methods: Identifying Skillful Spatial Scales Using the Fractions Skill Score

• DOI: 10.1175/2009waf2222260.1
• Publication Date: 2009-07-30T14:37:34Z
• Authors: Marion Mittermaier; Nigel Roberts
• Keywords: Quantitative precipitation forecast; Forecast verification

Abstract The fractions skill score (FSS) was one of the measures that formed part Intercomparison Spatial Forecast Verification Methods project. FSS used to assess a common dataset consisted real and perturbed Weather Research Forecasting (WRF) model precipitation forecasts, as well geometric cases. These datasets are all based on NCEP 240 grid, which translates approximately 4-km resolution over contiguous United States. cases showed can provide truthful assessment displacement errors forecast skill. In addition, be determine scale at an acceptable level is reached this usage perhaps more helpful than interpreting actual value. This spatial-scale approach becoming popular for monitoring operational performance. study also shows how responds bias. A biased always gives lower values large scales usually smaller scales. It possible, however, give higher scales, when additional rain overlaps observed rain. However, given sufficiently sample system will lower. use percentile thresholds remove impacts When proportion domain “wet” (the wet-area ratio) small, subtle differences introduced through near-threshold misses lead changes in magnitude individual (primarily because bias changed). Reliable statistics small ratios require larger forecasts. Care needs taken choice verification domain. For high-resolution models, should enough encompass length typical mesoscale forcing (e.g., upper-level troughs or squall lines). If too large, small. fluctuations ratio spatial may missed. good measure accuracy Different methods needed other patterns behavior.