Author(s): Diah Valentina Lestari; Wei Jian; Edmond Yat-Man Lo

Linked Author(s): WEI JIAN

Keywords: Prediction of compound events CORDEX SEA Support vector machine Metro Manil

Abstract: Interaction between extreme wind and precipitation lead to stronger impact in a coastal urban area compared to the same event on a non-urban coastal area. This study investigates atmospheric conditions linked to compound wind and precipitation extremes (CWP) for Metro Manila, Philippines, a coastal megacity that is prone to extreme weather. We utilize a CORDEX-SEA model to identify CWP days along with eight underlying meteorological variables for linkages to these events. A composite analysis of each meteorological variable over the identified CWP days covering the historical period (June-September, 1975-2005) shows a clear spatial pattern of both outgoing longwave radiation and moisture budget. These eight variables are next used as inputs in a Support Vector Machine (SVM) to predict CWP days under present climate (June-September, 2006-2023) with resulting f1 scores ranging from 0.73-0.78 indicating good predictive capability. This SVM-based approach can further be applied for CWP occurrence predictions under future climate.

DOI:

Year: 2025