Key points are not available for this paper at this time.
Soil erosion is a major environmental concern in Indian river basins, with the Barak River Basin in Northeast India particularly prone to degradation due to recurrent flash floods and intense monsoonal rainfall. This study integrates the soil and water assessment tool (SWAT) and a multilayer perceptron artificial neural network (MLP-ANN) to assess streamflow, sediment yield, and soil erosion risk in the basin. The SWAT model was calibrated and validated at the Badarpur Ghat outlet (2010–2015) using the sequential uncertainty fitting algorithm version 2 (SUFI-2). The model showed strong performance for discharge (R2=0. 88, NSE=0. 84) and satisfactory accuracy for sediment yield (R2=0. 81, NSE=0. 78). Sensitivity analysis identified CN2 and GWDELAY as key parameters influencing discharge, while USLEC and USLEK were dominant for sediment yield. The MLP model improved discharge prediction (R2=0. 96–0. 98, NSE=0. 91–0. 97) but slightly overestimated peak flows, whereas SWAT better represented sediment transport processes despite underestimating peak loads. Sediment rating curve analysis revealed that fine particles dominated year-round, while coarse fractions were mainly transported during high-flow monsoon periods. Subbasin prioritization indicated that about 62% of the basin faces high to very high erosion risk, especially in downstream areas with steep slopes, croplands, and Hydrologic Soil Group D soils. Overall, the study demonstrates the complementary strengths of SWAT–SUFI-2 and ANN models for hydrological and sediment dynamics assessment, providing a scientific basis for soil conservation and sustainable watershed management in the Barak River Basin.
Pathan et al. (Sat,) studied this question.