Deep Learning Insights into the Changing Surface Water of Bhojtal Lake
The research group of Dr. Somil Swarnkar at the Department of Earth and Environmental Sciences, IISER Bhopal in collaboration with Department of Data Sciences IISER Bhopal & MANIT Bhopal, developed an AI-based framework to reconstruct and analyze long-term changes in the Water Surface Area (WSA) of Bhojtal Lake, Bhopal, a Ramsar-designated wetland of international importance. Monitoring lake dynamics over long periods is often difficult because satellite observations frequently contain gaps due to cloud cover, sensor limitations, and missing records. To address this challenge, the team applied a Long Short-Term Memory (LSTM) deep learning model that integrates satellite-derived surface water data with key hydroclimatic variables. The model was trained to learn the temporal behavior of the lake and reconstruct continuous monthly WSA records, even where satellite observations were incomplete. The reconstructed time series, spanning more than three decades, reveals substantial fluctuations in lake surface area, reflecting both climatic variability and hydrological changes in the region. In addition, climate projections under a +2°C warming scenario indicate potential reductions in water extent, suggesting increasing vulnerability of the lake system. These findings highlight the importance of AI-driven monitoring frameworks for supporting wetland conservation and sustainable water resource management. For more details, kindly visit https://www.sciencedirect.com/science/article/pii/S2352938526000881?via=ihub.
