Shuvra Sangeeta1, Satesh Rahatwal1
CES (I) Pvt Ltd, Gurgaon, India.
For sustainable growth with a rational Engineering and Environmental solutions, the long term records of hydrological observation are of immense value. The current paper is an attempt to apply the physically based, spatially distributed SWAT (Soil Water Assessment Tool) to assess its ability to predict flow at watershed scale ungauged locations in Pranhita sub-basin. The test sub-basin belongs to the Godavari basin located in the central part of India, draining an area of 1,09,079 km2. The SWAT model is tested for streamflow, calibrated at outlet of Wardha, Penganga and Pranhita sub-basins of Pranhita. The spatial heterogeneity in the parameter settings are tested at 16 monitoring gauge locations within the sub-basin, treating them as ungauged sites. The results indicate that SWAT can capture the amount and variability of streamflow acceptably well both at annual and monthly time scale. The model performance at testing sites range between acceptable and good. In general, the results show more than 90% of the stations have annual NSE values greater than 0.5 and about 60% have NSE greater than 0.8. Also, measure for annual R2 of 0.8 were exceeded by 14 out of 16 stations and 0.9 by 10 out of 16 sites. On similar lines, monthly NSE of 0.5 is exceeded by 95% stations and 0.8 by 56%.
Regression is the most widely used technique for transferring information to ungauged catchment. A Regional Analysis has been made for Pranhita sub-basin. Empirical relations have been developed based on the climate and catchment parameter dataset generated by SWAT. The correlation matrix of six variables viz. precipitation, % cropped area, % forest area, mean temperature, relief, and sub basin area with average natural runoff was developed. This was followed by clustering of sub-basins, which involved Principal Component Analysis (PCA) and K-Means Clustering. After dividing the dataset into clusters, empirical equations for the formed clusters were developed for monsoon months from the dataset of SWAT, that related the dependent variable “monthly discharge” with the climate and watershed attributes.
A comparison of SWAT flow simulated at watershed level in ungauged locations is made with flow series developed from Regional equation of Pranhita for monsoon months. Overall, the SWAT model can satisfactorily predict hydrologic budget for the ungauged basins in Pranhita with calibration at basin scale using both the approaches. The Regression based hydrogical response gives a lower NSE as compared to the direct SWAT output. However, the ease of applicability of Empirical equation makes it a viable alternative to adopt for small watersheds, in the absence of other suitable technique.
Keywords: Hydrological Budget, sub-basin, ungauged, Godavari, assessment, watershed, Cluster, simulation, prediction.
Pages: 90 – 106 | Full PDF Paper