Assessing sediment yield in Kalaya gauged watershed (Northern Morocco) using GIS and SWAT model

Abstract

An efficient design for erosion-control structures of any watershed in the world is entrusted with the delicate forecasting of sediment yields. These outlook yields are usually inferred by extrapolations from past observations. Because runoff, as the transporting vehicle, is more closely correlated with sediment yields than any other variable. So, calibration as well as validation of process-based hydrological models are two major processes while estimating the sediment yield in watershed. The actual survey is fulfilled with the aim of developing a trustworthy hydrologic model simulating stream flow discharge and sediment concentration with least uncertainty among the parameters picked out for calibration so as to verify the effect of the scenarios on the spatial distribution of sediment yield (sediments transported from sub-basins to the main channel during the step of time). Soil and Water Assessment Tool (SWAT, version 2012) model integrated with Geographic Information System (GIS, version 10.1) was used to simulate the stream flow and sediment concentration of Kalaya catchment situated in north of Morocco for the period from 1971 to 1993. Model calibration and validation were performed for monthly time periods using Sequential Uncertainty Fitting 2 (SUFI-2, version 2) within SWAT-CUP using 16 parameters. Our calibration outputs for monthly simulation for the period from 1976 to 1984 showed a good model performance for flow rates with NSE and PBIAS values of 0.76 and −11.80, respectively; also a good model performance for sediment concentration with NSE and PBIAS values of 0.69 and 7.12, respectively. Nonetheless, during validation period (1985–1993) for monthly time step, the NSE and PBIAS values were 0.67 and −14.44, respectively for flow rates and these statistical values were 0.70 and 15.51, respectively for sediment concentration; which also means a good model performance for both. Following calibration, the inclusive effect of each parameter used was ranked using global sensitivity function in SWAT-CUP. From our analysis, the effective hydraulic conductivity in main channel alluvium (CH_K2), USLE support practice factor (USLE_P) and manning's "n" value for the main channel (CH_N2) were found to be the most sensitive parameters during different iterations with different number of simulation but with the same inputs. The least sensitive parameter were found to be different in either cases unlike the most sensitive parameters. As a result, the global evaluated soil erosion rate in the study area varied from 20 to 120t/ha/yr. It was summarized that the entire knowledge of the hydrologic processes happens within the watershed and the consciousness about acceptable meaningful range of the parameters is crucial while developing reliable hydrologic model.