Rainfall Runoff Modelling using ANN and SWAT Model for Som Kamla Amba Catchment

Abstract

The modelling of rainfall-runoff for ungauged river basin or catchments is newlinehelpful for the design of hydraulic structures, soil conservation structures, water newlineharvesting structures, flood moderation studies, design of drainage systems etc. The newlinepresent study was conducted for Som Kamla Amba catchment area lies in Dungarpur newlineand Udaipur districts of Rajasthan India. The rainfall forecasted on monthly basis using newlineArtificial Neural Networks (ANN) and stream flow predicted on monthly basis using newlineSoil and Water Assessment Tool (SWAT) and checked their performance using newlinestandard statistical indices such as r, R2, RMSE, MAE, IA, VE and NSE. The input data newlinesuch as daily rainfall, max. temperature, min. temperature, relative humidity, solar newlineradiation, wind speed and stream flow were collected for duration of 24 years (1995 to newline2018) for Som Kamla Amba catchment. newlineThe morphometric parameters for sub watersheds (W1 to W9) of Som Kamla newlineAmba catchment were analyzed in ArcGIS 10.4.1 software using ASTER Dem with newlinespatial resolution 30 m × 30 m. The linear, areal and relief properties of Som Kamla newlineAmba catchment were analyzed. The total drainage area of nine sub watersheds (W1 to newlineW9) were 122.09, 1745.02, 33.88, 21.18, 114.36, 51.88, 786.65, 83.65 and 2382.46 newlineKm2 respectively. Total catchment area of Som Kamla Amba catchment was found newline5342 Km2. The inter-correlation was analyzed among ten dimensionless geomorphic newlineparameters. These parameters shows multicollinearily among them. Ten geomorphic newlineparameters correlation matrix revea1ed strong correlations (correlation coefficient gt newline0.9), as well as good (gt0.75) and moderate (gt0.60) between geomorphic parameters. It newlinewas very difficult at this stage to group characteristics into components and assign any newlinephysical significance since one characteristic (circulatory ratio) do not show any significance correlation with any other characteristics. Contd.... newline