Performance Evaluation of Metric and Parametric Models through Rainfall-Runoff Modelling of Sub-Himalayan Catchment

In water resources, management of various hydrological and hydraulic operations requires runoff predictions with the best possible accuracy. Several hydrological models have been widely used to estimate runoff and gained importance in recent decades. The present study aims to evaluate various metric and parametric models of runoff estimation for the Gambhar catchment in the state of Himachal Pradesh, India. The study employed Artificial Neural Network (ANN), and Long Short-Term Memory (LSTM) techniques as metric and Green-Ampt infiltration and statistical models as parametric methods. For this purpose, the five station-based daily rainfall, evapotranspiration, and daily discharge data of Gambhar river for the period from 2010–2018 were procured. The field data of soil infiltration parameters were collected and spatially interpolated over the study area. The metric models perform better as compared to the parametric in runoff estimation, however, the qualitative and quantitative evaluation establish the efficacy of LSTM over other models considered in the study with the coefficient of determination, average relative error and Nash Sutcliffe Efficiency coefficient of 0.91, 0.15 and 0.88 respectively.