Rainfall-Runoff Modeling of Pindwara Watershed Using SWAT Model
Loading...
Date
2019
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
MPUT, UDAIPUR
Abstract
Land and water are the two most basic yet essential components in agriculture. However, these two natural resources are also the ones that are most subjected to deficiency, deterioration and degradation. Erosion is a common phenomenon which causes deterioration of land and runoff is one of the prime factors responsible for erosion which can be controlled by construction of suitable soil and water conservation structures. Runoff estimation is a key process in the design of these soil and water conservation structures. In this study, the Soil and Water Assessment Tool (SWAT) model was used to estimate the runoff in Pindwara watershed, which spans over an area of 1428 km2, in the Sirohi district of Rajasthan and to assess its feasibility after reasonable calibration and validation of the various parameters, whose individual sensitivities were determined.
The required meteorological data such as rainfall, maximum and minimum air temperature, solar radiation, wind speed and relative humidity for a time period of twelve years from 2001 to 2012 was obtained from four stations in and around the watershed via the Climate System Forecast Reanalysis (CFSR) database operated by the National Centers for Environment Prediction (NCEP). The SRTM 90 m digital elevation model (DEM) of the study area was downloaded from the USGS Earth Explorer. The soil map of the area was extracted from the Digital Soil Map of the World compiled by the Food and Agriculture Organisation (FAO). The land use land cover map of the area was extracted from the land cover maps of the area acquired from the Bhuvan-NRSC database. The daily runoff data for a time period of eight years from 2003 to 2010 was obtained from India-Water Resource Information System website maintained by NRSC.
The simulation of runoff was carried out on a monthly time-step for a period of 12 years from 2001 to 2012 with initial two years provided as warm-up period. The modeling of various hydrologic components in SWAT was based on the water balance equation. The runoff was estimated using SCS-CN model which is one of the techniques of runoff estimation in SWAT. The estimation led to a classification of the entire watershed into 165 sub-basins and 580 HRUs. It was estimated that on an average basis, surface runoff of 174.96 mm, 15.55 mm of lateral flow
100
and 87.06 mm of return flow was generated in the watershed annually. Through feasibility assessment, it was found that the model was not fit enough for direct field application, i.e., without proper calibration, due to poor R2 and NSE values of 0.465 and 0.402 respectively which were unacceptable.
The calibration and validation of the model was carried out in SWAT – CUP using SUFI2 algorithm. Calibration was done on a monthly time-step for five years from 2003 to 2007. Validation was also done on a monthly time-step for three years from 2008 to 2010. Fourteen parameters were calibrated based on the knowledge of literature and conditions of the study area. Through calibration, a good R2 value and an acceptable NSE value of 0.711 and 0.677 respectively, was obtained and the estimated runoff was found to have a more decent fit with the recorded runoff. Through validation, an excellent R2 value and a good NSE value of 0.844 and 0.752 respectively, was obtained and there was an excellent fit between the estimated runoff and the recorded runoff.
Sensitivity and uncertainty analysis of the model parameters was carried out in SWAT – CUP. Sensitivity analysis was done using global sensitivity technique. Twelve of the fourteen parameters considered for calibration and validation were found to be relatively more sensitive. Groundwater parameters such as base flow alpha factor were the most sensitive parameters in the overall sensitive analysis. The result of the uncertainty analysis was obtained at the end of calibration of the model. A P-factor of 0.7 was obtained which was found to be barely acceptable while a relatively high R-factor of 0.61 was obtained indicating a relatively high level of uncertainty in the model estimations.
Description
Rainfall-Runoff Modeling of Pindwara Watershed Using SWAT Model
Keywords
Citation
Upadhyay H. And Mittal H.K.