Modeling seasonal typhoon genesis in the North West Pacific using probabilistic approaches

Typhoons are one of the extreme weather phenomena that seriously affect humans and the environment. Modeling typhoon genesis plays a crucial role in evaluating typhoon features, providing valuable information for the early warning systems related to storm hazards. The goal of this work is to establish the probabilistic models for typhoon genesis (TG) across the North West Pacific (NWP) regarding seasonal variations. We examined the congruence between the parametric probability distributions and recorded typhoon using Chi-square and Anderson-Darling tests. In addition, three criteria, consisting of Correlation Coefficient (CC), Mean Absolute Error (MAE), Root Mean Squared Error (RMSE), were used to determine which probability distributions would provide the optimal fit to recorded typhoon dataset. Our findings demonstrated that the majority of typhoons generated during the peak season (PS) spanning between June and November, along the latitudinal belt 5°N − 30°N. Conversely, the low typhoon season (LS), with a lower frequency of typhoons primarily found in the area south of 15°N, was from December to May. The Poisson distribution was the best model to simulate typhoon counts in the two seasons (PS and LS). For modeling TG position in the LS, Loglogistic distribution was chosen to simulate longitude and latitude. For the PS, Gamma distribution was the optimal selection to simulate TG longitude while Nakagami distribution was used to model TG latitude. The geographic patterns of simulated TG in the two seasons exhibited comparability to those of the recorded TG by 1000-year Monte Carlo simulations. This study enhances the understanding of seasonal TG trends and provides a robust framework for future typhoon activity modeling.