Using the High-Resolution NSSL-WRF Ensemble to Provide Hazard Guidance

Experimental versions of convection-allowing models (CAMs) have been providing meteorologists with explicit forecasts of convection for the past decade. While these convective forecasts do not explicitly resolve hazards such as tornadoes, hail, and wind, hourly maximum fields of simulated storm diagnostics have been shown to provide valuable information with regard to severe weather potential. While previous works have focused on the ability of CAMs to provide reliable forecasts of any severe weather type, few attempts have been made to produce guidance for specific severe weather types and evaluate the ability of CAMs to discriminate these types. Thus, this study aims to use an experimental CAM-based ensemble initialized daily at the National Severe Storms Laboratory, known as the NSSL-WRF ensemble, to derive separate, calibrated probabilities for each hazard: severe wind, hail, and tornadoes. Four-hourly probabilistic tornado forecasts are developed from the NSSL-WRF ensemble using updraft helicity (UH) as a tornado proxy. The UH fields are used in tandem with simulated environmental fields such as lifted condensation level (LCL) height, the most-unstable and surface-based CAPE, and multi-field severe weather parameters such as the significant tornado parameter (STP). These environmental fields are used to account for the fact that simulated, high-UH storms can occur in many different environments, but only a subset of these environments are strongly supportive of tornadoes. Thus, the simulated environmental information can be used to filter out or dampen the forecast probabilities of tornadoes derived from UH in environments typically not supportive of tornadoes (e.g., high LCLs, a stable near-surface layer, low values of STP).

Probabilities are generated using various thresholds of UH and differing thresholds and combinations of environmental parameters. Calibration of forecast probabilities will be performed based on data from April-June of 2014 and the calibrated probabilities will be tested in real-time during the 2015 NOAA/Hazardous Weather Testbed Spring Forecasting Experiment. The probabilities are evaluated objectively with Receiving Operator Characteristic (ROC) curves, the area under the ROC curve, and reliability diagrams as well as subjectively by participants in the 2015 NOAA/Hazardous Weather Testbed Spring Forecasting Experiment. This presentation will focus on subjective results from the forecasting experiment, as well as results from the objective skill metrics.