Probabilistic Predictions on TC Rapid Intensification

Some of Tropical Cyclones undergo a rapid intensification process, which causes them to become stronger typhoons. Rapid Intensification (RI) is defined as the increase in maximum sustained winds to 30 kt (15 m/s) or more within a 24-hour period (Kaplan and DeMaria, 2003). Typhoons that have undergone RI mainly strengthen into strong LMIs, which can cause significant damage in a relatively short period of time. The recent increase in the number of cases of RI of Tropical Cyclones has highlighted the importance of advanced forecasting. However, achieving accuracy in these forecasts remains a significant challenge. In general, the intensity of typhoons is highly dependent on thermal conditions, such as ocean temperatures. However, the process of rapid intensity development is complex and influenced by dynamic factors, such as upper-level divergence and vertical wind shear. In this study, we developed a guidance for predicting the probability of rapid intensity development in a typhoon using environmental prediction factors at the time of its occurrence. For the purpose of supporting KMA's typhoon forecasting, a statistical based prediction model for the probability of RI was developed and evaluated from 2023 to 2024. It uses logistic regression to provide RI predictions up to the next 24 hours and 48 hours. The predictor variables included upper-level divergence, relative humidity, equivalent potential temperature (EPT), depth-averaged temperature (DAT), tropical cyclone heat potential (TCHP), thermodynamic net energy gain rate (NGR) (Lee et al., 2019), and 30 kt wind radius. To evaluate the accuracy of predictions for typhoons that occurred between June and November 2021-2023, 12 cases of RI and 42 cases of non-RI were analyzed. The POD was 0.78 and 0.76 for the 24-hour and 48-hour prediction accuracy, respectively, with corresponding FAR of 0.32 and 0.38. Predictive models showed good results during the validation period, but predicting favorable conditions for RI is still complex and challenging.