Hydrological Implications of LULC Changes in the Netravathi River Basin: Insights from SWAT Modeling and Machine Learning Techniques

This research work explores the impact of land use land cover (LULC) change on basin hydrology using the Soil and water assessment tool (SWAT) model in the Netravathi River basin, south-western part of India. The model inputs, relying on limited data including precipitation (P), LULC, soil, and temperature (T), sets it apart from other data-intensive models. Calibration and validation using 2000 to 2018 data demonstrate strong correlations between observed and simulated streamflows (R
²
of 0.84 and 0.79, NSE of 0.84 and 0.79). Notably, LULC changes from 2000 to 2018, such as dense-forest conversion to cropland and increased built-up area, led to altered runoff patterns, evident 0.9% increase in runoff. In parallel, machine learning (ML) algorithms such as random forest (RF), support vector machine (SVM), and artificial neural network (ANN) were explored for river flow forecasting. The ANN model excelled with a testing accuracy of 93.63%, attributed to its capability to comprehend intricate data relationships, presenting a potent avenue for refined river flow predictions. The work features the capability of the ANN model for water resource management through improved river flow forecasting.