Object-Based Climatology and Verification of HRRR Forecasts

Non-traditional methods of assessing convective-scale forecasts, especially for features-based fields such as bodies of precipitation, radar reflectivity, and updraft helicity and hail swaths, for example, has not yet achieved widespread use, especially for operational NWP forecasting systems such as the High-Resolution Rapid Refresh (HRRR). However, research and development of scripting and metrics for assessing the quality of forecasts of such features-based fields in near-real-time is ongoing, and some preliminary results will be presented here. In particular, the Method of Object-based Diagnostic Evaluation (MODE) and its Time-Domain partner (MODE-TD) from the Developmental Testbed Center’s (DTC) Model Evaluation Tools (MET) software suite was used to compute object-based verification metrics as well as populate climatological distributions of various attributes of components of convective storm objects from the HRRR forecast system. Object-based verification metrics such as the object-based threat score, median of maximum interest, and mean distance between centroids of objects will be discussed. Object-based distributions of attributes such as peak magnitude, length, and orientation angle of UH tracks as well as timing and initiation locations of reflectivity objects from the 2019 warm season, will also be highlighted as a means of determining where within a model’s prior forecast distribution a given event falls, which can allow forecasters to assess the relative severity or generality of a forecast event.