Development of an ensemble dynamic-probabilistic prediction model for tropical storm genesis in the Vietnam East Sea using the Logistic Regression approach

Tóm tắt

In this study, a logistic regression model is developed to forecast tropical storm (TS) genesis in the Vietnam East Sea from 2012 to 2019. The model incorporates seven potential predictors including dynamic and thermodynamic parameters at formation time retrieved from the WRF-LETKF outputs. After rigorous testing, six predictors are selected, excluding minimum sea-level pressure. In a broader context, the logistic regression model performs promisingly, generating forecast probabilities that enhance the accuracy of TS genesis predictions, particularly in early forecast cycles. The model’s regression coefficients and forecast outcomes align well with test dataset results, affirming its stability and validity. As a result, the forecast probability from this model can be effectively employed as a probabilistic forecast value for predicting TS genesis status.