Using Machine Learning to Generate Storm-Scale Probabilistic Guidance of Severe Weather Hazards in the Warn-on-Forecast System

Abstract A primary goal of the National Oceanic and Atmospheric Administration Warn-on-Forecast (WoF) project is to provide rapidly updating probabilistic guidance human forecasters for short-term (e.g., 0–3 h) severe weather forecasts. Postprocessing required maximize usefulness from an ensemble convection-allowing model Machine learning (ML) models have become popular methods postprocessing since they can leverage numerous variables discover useful patterns in complex datasets. In this study, we develop evaluate a series ML produce calibrated, WoF System (WoFS) output. Our dataset includes WoFS forecasts available every 5 min out 150 lead time 2017–19 NOAA Hazardous Weather Testbed Spring Forecasting Experiments (81 dates). Using novel storm-track identification method, extracted three sets predictors forecasts: intrastorm state variables, near-storm environment morphological attributes storm tracks. We then trained random forests, gradient-boosted trees, logistic regression algorithms predict which 30-min tracks will overlap tornado, hail, and/or wind report. To rigorous baselines against skill models, probabilities hazard-relevant exceeding tuned thresholds each track. The discriminated well all hazards produced more reliable than baseline predictions. Overall, results suggest that ML-based dynamical output improve short-term, storm-scale guidance.