About the Speaker:

Abstract: This talk provides insights from three recent research papers on Indian weather using novel modeling and AI approaches:Palaeoclimate:Utilizes deep learning to reconstruct South Asian monsoon rainfall over the past 500 years.Employs convolutional neural networks (CNNs) and Shapley analysis to interpret spatiotemporal patterns since 1500.Monsoon Low-Pressure Systems (LPSs):Examines the interaction of LPSs with large-scale monsoon circulations and varied surfaces.Uses XGBoost decision-tree ensemble models to predict and analyze LPS behavior, such as intensification rate, propagation, and precipitation rate.Energy Demand:Scrapes, cleans, and models daily energy demand from Indian states.Uses meteorological predictors, population data, and event status (e.g., weekends) to explain over 80% of energy demand variability in many states.Highlights the regional variations in energy responses to temperature, air conditioning, and heating demands.