Optimising model selection for the morphometric analysis of a drainage basin

ABSTRACT The study investigates the correlation between the morphometric characteristics and hydrological processes within the six drainage basins of the Kakinada district, India. Using Shuttle Radar Topography Mission (SRTM) 30 m DEM data, morphometric parameters, namely, stream frequency, bifurcation ratio, slope, and drainage density, and slope were computed for a 3,019 km2 area. The fuzzy analytical hierarchy process (FAHP) was employed for basin prioritisation, while machine learning models, namely, artificial neural network (ANN), multiple linear regression, and support vector regression, were used to optimise and validate morphometric predictions. The ANN model achieved the highest accuracy (R2 = 0.98), demonstrating a strong agreement with FAHP outputs. The combined FAHP and ANN present a novel framework for morphometric analysis and basin management. The outcomes of the study provide quantitative insights for flood prediction, hydrological modelling, and sustainable water resource planning in the Kakinada district.