Modélisation De La Surface Occupée Par Les Tourbillons Océaniques (Cyclones) Au Sud Et Sud-Est De Madagascar

Abstract

This work aims to build a model based on a random forest to analyze the relationships between several oceanic variables concerning the area occupied by eddies (cyclones) in the South and Southeast of Madagascar. The variables considered include zonal wind (u, m/s), meridional wind (v, m/s), ocean currents (m/s), sea surface height (SSH, m), sea surface temperature (SST, K), sea surface salinity (SSS, PSU), atmospheric pressure (Pa), and the relative area occupied by eddies. Data preprocessing was performed, followed by an evaluation of multicollinearity using the Variance Inflation Factor (VIF) and correlation analysis between variables. Subsequently, causality analysis based on the Granger test allowed the selection of the most significant variables to include in the model. Based on these analyses, the variables retained for modeling are sea surface salinity (SSS), sea surface height (SSH), sea surface temperature (SST), and meridional current. The evaluation of the model's performance shows a MAPE of 1.6008% and a coefficient of determination R² of 0.8523, attesting to its accuracy and reliability.