Gridded estimation of lightning frequency over the eastern Indian subcontinent using neural networks

Abstract

The present study aims to estimate the severity of lightning occurrences on a real-time basis over the Eastern Indian region during the afternoon hours of the premonsoon season using gridded meteorological and aerosol datasets. A set of 17 surface and mid-tropospheric parameters of thermodynamic, dynamic, and microphysical origin have been found to show a prominent agreement with lightning frequencies. Next, these relationships from both past and present time stamps have been fed to a neural network to provide a reliable estimate of the lightning frequency with a decent hit ratio of 70%.