SWAT model evaluation using generated data and assessing the impact of land use changes

Abstract

Land and water are the primary natural resources which are useful for all the living beings on earth surface. Degradation of the land surface and lack of water availability are the two major important problems mankind is facing in this century. In order to overcome these problems, there is a need of effective management of these resources. Watershed models are the tools which are not only useful for the effective management of these natural resources, but also useful for the proper understanding of the hydrological behaviour of the watershed. These models play a vital role in simulating the hydrology of the watershed. Among the different categories of the model, a physically based, semi distributed hydrologic model SWAT was used for the study. The Kunthipuzha river basin was selected as a study area for the assessment of the calibration and validation of the hydrologic model SWAT adapted to the study area. The data scarcity is one of the major problems in the ungauged watersheds. In order to overcome this problem, CFSR (Climate Forecast System Reanalysis) data which is a global, high resolution, coupled atmosphere-ocean- land surface-sea ice system is available as an alternative option for solving the data deficiency in the watershed. The land use change also plays a vital role in altering the hydrologic system and has a large impact on the stream flow. This is mainly due to the rapid socio economic development. So, based on the above mentioned problems, SWAT output comparison using CFSR & observed meteorological data as inputs was taken up. The impact of land use change on the hydrology of watershed was also studied. The platform used for the study was ArcGIS 10.3 with the Arc SWAT interface. The SWAT model set up was done for the Kunthipuzha river basin and the calibration and validation of the model was also done to make the model suitable for use in the area. This model was later used to understand the hydrologic behaviour of the watershed. The model was simulated for the period 1991 to 2013 for calibration and validation of the model was done using the data for the period 2014 to 2016. Before the model calibration and validation, sensitive parameters were evaluated using SWAT CUP (Calibration andUncertainty Program). CN2 (Initial SCS runoff curve number for moisture condition II) and ALPHA_BF (Base flow alpha factor) were found to be the most sensitive parameters for the study area. The NSE and R 2 before and after calibration were 0.81 & 0.83 and 0.82 & 0.85 respectively. The NSE and R 2 for the validation were 0.70 & 0.87 respectively. Based on the statistical measures and the criteria used, the model performance is “very good” in the calibration period and “good” in validation period. To analyse the possibility of using CFSR data instead on observed meteorological data, the developed model was run with observed meteorological data and predicted meteorological data (CFSR) was done separately without changing any other inputs for the period 1991 to 2013. The NSE, R 2 and RMSE for the observed meteorological data were 0.82, 0.85 and 29.25 respectively, where as for the predicted meteorological data (CFSR) the values were 0.70, 0.72 and 37.18 respectively. Based on the statistical measures, the performance of the observed meteorological data is better than the predicted meteorological data. From the graphical analysis, it was clear that the values of predicted meteorological data were highly correlated with the observed meteorological data except at peaks. Hence, CFSR data can be used as a reliable data source in data scarce areas. The land use change impact play a major role in alternating the stream flow because of the rapid socio-economic development. The land use map for the year 2000 and 2017 were prepared. While comparing the land use for the year 2000 and 2017, it is found that the urban areas drastically increased from 3.01 to 20.01% because of the rapid socio economic development. The forest land reduced from 22.24 to 21.31%. The percentage area under paddy decreased from 17.57 to 6.12%. The model was simulated for the period from 1989 to 2016 with the two years of warm up period. Then the comparison of simulated discharge for the year 2000 and 2016 were evaluated. The results showed that there is no significant change in stream flow when the land use alone is changed keeping all other factors same.