LIGHTNING–METEOROLOGY RELATIONSHIPS OVER SRI LANKA AND INDONESIA: A MACHINE LEARNING APPROACH

The relationship between lightning flashes (LF) and various meteorological parameters is analyzed using lightning data from 1995 to 2014. The meteorological parameters considered in this study include aerosol optical depth (AOD), precipitation (P), relative humidity (Rh), convective available potential energy (CAPE), effective cloud droplet size (CER), total precipitable water (TPW), cloud fraction (CF), cloud top temperature (CTT), Richardson number (RN), cloud ice water content (CIWC), and cloud liquid water content (CLWC). This study examines two regions with distinct climates: Sri Lanka (R1) and Indonesia (R2). Results show lower lightning activity in R1 (15.5 flashes/km²/year) than in R2 (21.8 flashes/km²/year), with both peaking in April. Furthermore, the study evaluates the effectiveness of different regression techniques in modeling lightning activity. The Support Vector (SV) regression model performs best for Sri Lanka, while the Random Forest (RF) regression model emerges as the most suitable approach for Indonesia.