MULTI-OBJECTIVE HYDROLOGICAL MODELING USING SWAT-2005 MODEL

Abstract

The problem of water availability in the State of Uttarakhand can be summarized as "Water an abundant, yet scarce resource". Though the state receives abundant rainfall and is the source of many major rivers of the Northern India, yet the complex environmental factors coupled with topography, population pressure, degradation of land and forests, creates tremendous constraint on the sustainability and conservation of water resources. Together with this, unavailability of long term, reliable hydro meteorological data for different watersheds, pose a serious challenge for proper planning and sustainable management of water resources for quantity and quality for the state in particular and the whole of the Himalayan region, in general. This calls for a comprehensive distributed hydrologic modeling to simulate runoff and erosion response. Among the myriad rainfall-runoff models available in literature, the Soil and Water Assessment Tool (SWAT), developed by Agricultural Research Service of USDA, has gained popularity in the recent past, because it is a distributed watershed model and has capabilities to simulate rainfall runoff response, impact of land management practices on water, sediment and agricultural chemical yields in complex watersheds. The present study is undertaken with an aim to test the performance of SWAT model in a predominantly forested hilly mountainous watershed defined at gauging site at Naula in uppermost part of Ramganga River in Kumaon region of Himalaya comprising of an area of 1074 km2. For model application, the catchment area is divided into 8 sub-catchments. The whole catchment area is classified into seven land use 'categories. Available hydrological data (i.e. from 1975 to 1982), split into two groups is used for calibrating and validating parameters of the model. The calibrated model parameters are also used to test the performance of the model at upstream intermediate gauging site at Chaukhutia draining approximately an area of 573 km2. Model performance is adjudged v based - on visual comparison of observed and model computed runoff as well as on statistical measures of mass balance error, coefficient of determination and Nash-Sutcliffe efficiency (NSE). Comparative analysis of results is done on daily and monthly time scale. Analysis reveals that during calibration period, yearly values of Nash-Sutcliffe efficiency varies between 0.58 to 0.85 with an average value of 0.74 when simulation is done on daily time scale and 0.86 to 0.95 with an average value of 0.92 on monthly time scale while during validation period, the Nash-Sutcliffe efficiency varies between 0.52 to 0.68 with an average value of 0.58 on daily time scale and 0.65 to 0.88 with an average value of 0.80 on monthly time scale at Naula gauging site. The performance of the calibrated model at intermediate test site, Chaukhutia (for full data range from 1975 to 1982) has shown good agreement between observed and model computed hydrographs with NSE varying from 0.44 to 0.86 with an average value of 0.69 on daily time scale and 0.63 to 0.95 with an average value of 0.85 on monthly time scale which are comparable to NSE achieved for entire watershed indicating the applicability of the model at intermediate locations also. Visual hydrograph comparison reveals that baring some extreme peak events — which the model underestimated — the agreement between observed and model computed runoff is reasonably good at daily time scale and very good at monthly time scale. Therefore, the model fit can be rated as reasonable to very good.