The research group of Dr. Somil Swarnkar, Department of Earth and Environmental Sciences, IISER Bhopal, in collaboration with Dr. Akshay Agarwal from the Department of Data Sciences and Engineering, has developed a Bayesian-optimized recursive machine learning framework to analyze human-induced changes in suspended sediment load (SSL) within the Godavari River Basin. The study uses long-term SSL data, divided into pre- (1969–1990) and post- (1990–2020) dam construction periods, to examine shifts in sediment transport patterns driven by anthropogenic activities such as damming and land-use/land-cover (LULC) changes. Despite a stable monsoonal contribution (~73%), the post-1990 period exhibited a marked decline in peak SSL values and a narrower range of distribution, suggesting reduced sediment availability. These findings were further supported by empirical cumulative distribution function (ECDF) analysis, which revealed altered sediment retention and release dynamics. To capture the non-linear behavior of SSL under these changing conditions, ensemble tree-based models—Extra Trees Regressor (ETR), Random Forest Regressor (RFR), and Gradient Boosting Regressor (GBR)—were trained and evaluated using R², RMSE, and MAE. Hyperparameters were optimized using Bayesian techniques, with ETR outperforming other models (R² = 0.97 training; 0.90 testing). The study offers a robust data-driven framework for sustainable sediment management in regulated river systems. For more details, kindly visit https://link.springer.com/article/10.1007/s10661-025-14039-w.
