A comparison of grid-point and object-based updraft helicity probabilities for severe weather forecasting

Hourly maximum updraft helicity (UH) objects are identified in convection-allowing ensembles using four thresholds, meant to distinguish intensity, to extract model reports as distinct events. The object-based approach of extracting model reports allows for a more direct comparison, though not identical, tocomparison to observed storm reports than a grid-point approach. The current neighborhood techniques used during the NOAA Hazardous Weather Testbed Spring Forecasting Experiment allow each grid point to contribute to ensemble probabilities, similar to Sobash et al. (2011). In this work, using an object-based approach to extract finer maxima at multiple thresholds reduces the maximum value of the probabilities since the grid points within individual UH tracks are correlated. As a result, the object-based approach reduces the bias and shifts the location of maximum probability on a per-model basis, both resulting in more accuracy over the grid-point approach. Ensemble-mean probabilities from the object-based approach are thus more representative of the uncertainty, rather than conflating areal coverage and certainty, as is common with a grid-point approach. Two detailed case studies will be shown to demonstrate these points.