Evaluating the impact of land use uncertainty on the simulated streamfow and sediment yield of the Seyhan River basin using the SWAT model

Abstract

As a result of the increased availability of spatial information in watershed modeling, several easy to use and widely accessible spatial datasets have been developed. Yet, it is not easy to decide which source of data is better and how data from diferent sources afect model outcomes. In this study, the results of simulating the stream fow and sediment yield from the Seyhan River basin in Turkey using 3 diferent types of land cover datasets through the soil and water assessment tool (SWAT) model are discussed and compared to the observed data. Te 3 land cover datasets used include the coordination of information on the environment dataset (CORINE; CLC2006), the global land cover characterization (GLCC) dataset, and the GlobCover dataset. Streamfow and sediment calibration was done at monthly intervals for the period of 2001–2007 at gauge number 1818 (30 km upstream of the Çatalan dam). Te model simulation of monthly streamfow resulted in good Nash–Sutclife efciency (NSE) values of 0.73, 0.71, and 0.68 for the GLCC, GlobCover, and CORINE datasets, respectively, for the calibration period. Furthermore, the model simulated the monthly sediment yield with satisfactory NSE values of 0.48, 0.51, and 0.46 for the GLCC, GlobCover, and CORINE land cover datasets, respectively. Te results suggest that the sensitivity of the SWAT model to the land cover datasets with diferent spatial resolutions and from diferent time periods was very low in the monthly streamfow and sediment simulations from the Seyhan River basin. Te study concluded that these datasets can be used successfully in the prediction of streamfow and sediment yield.

As a result of the increased availability of spatial information in watershed modeling, several easy to use and widely accessible spatial datasets have been developed. Yet, it is not easy to decide which source of data is better and how data from diferent sources afect model outcomes. In this study, the results of simulating the stream fow and sediment yield from the Seyhan River basin in Turkey using 3 diferent types of land cover datasets through the soil and water assessment tool (SWAT) model are discussed and compared to the observed data. Te 3 land cover datasets used include the coordination of information on the environment dataset (CORINE; CLC2006), the global land cover characterization (GLCC) dataset, and the GlobCover dataset. Streamfow and sediment calibration was done at monthly intervals for the period of 2001–2007 at gauge number 1818 (30 km upstream of the Çatalan dam). Te model simulation of monthly streamfow resulted in good Nash–Sutclife efciency (NSE) values of 0.73, 0.71, and 0.68 for the GLCC, GlobCover, and CORINE datasets, respectively, for the calibration period. Furthermore, the model simulated the monthly sediment yield with satisfactory NSE values of 0.48, 0.51, and 0.46 for the GLCC, GlobCover, and CORINE land cover datasets, respectively. Te results suggest that the sensitivity of the SWAT model to the land cover datasets with diferent spatial resolutions and from diferent time periods was very low in the monthly streamfow and sediment simulations from the Seyhan River basin. Te study concluded that these datasets can be used successfully in the prediction of streamfow and sediment yield.