Harnessing Futuregans For High-Resolution Thunderstorm And Lightning Nowcasting Over Bhopal Region

The accurate prediction and quantification of severe weather events, such as heavy rainfall and lightning; remain imperative for disaster management. The implementation of a realtime predictive model that aims to enhance the precision tracking and development of clouds along with estimation of Vertically Integrated Liquid (VIL) over a region for shortterm forecasting (nowcasting) of thunderstorm events using artificial intelligence (AI) based future prediction models. Integrating AI with remote sensing technologies (Radar and satellite) datasets offers a promising approach to improve the prediction and alerting mechanisms for these extreme weather phenomena. In this work, we leveraged the highresolution Doppler Weather RADAR (DWR) imagery of India Meteorological Department (IMD)-Bhopal, ISRO’s NRSC-LDS Network and other data sources (such as Automatic Weather Station (AWS), electric field (EF) recorder and other sensors) to feed to machine learning (ML) models (Generative Adversarial Networks - FutureGAN) for the prediction of thunderstorm-bearing clouds development, their movement and associated lightning threats. Ultimately, the proposed model produces the high-resolution nowcasting of cloud-to-ground lightning strikes in near-future (30 minutes to 1 hour lead time) in the specified Radar coverage area. The nowcast accuracy for thunderstorm/ lightning predictions obtained from proposed model was found to be $80-85 \%$ in June month of studied years around Bhopal city observatory.