The Performance of Operational Regional Dynamical Model Forecasts of Tropical Cyclone Rapid Intensification

The prediction of rapid intensification (RI) by operational regional dynamical tropical cyclone (TC) models has improved markedly over the last decade. RI is now predicted by these models at nearly the correct relative frequency, and there are some storms for which the model RI forecasts are consistently accurate (e.g. Hurricane Idalia, 2023). However, there are still many deficiencies in RI forecasts from operational regional dynamical models, with Hurricane Otis (2023) standing out as a stark reminder that not all RI events are accurately captured in the model forecasts.

Here we utilize a very large sample of operational COAMPS-TC and HWRF model forecasts, encompassing TCs worldwide from 2020 – 2023, to study RI forecast performance. This sample enables an approach by which we can robustly assess RI forecast performance as a function of variables such as observed intensity at the forecast initial time or observed peak intensification rate. We also examine RI forecast performance on a storm-by-storm basis, in an attempt to understand why RI forecast performance varies so much from storm-to-storm. Finally, we present results comparing the observed distributions of RI-related quantities (e.g. intensity at the time of RI onset, peak intensification rate) and the corresponding distributions derived from model forecasts. These RI validation diagnostics inform and motivate our recommendations for pathways to improve operational regional dynamical TC model RI forecasts.