부산대학교 도서관

Erosion is a major issue due to the global availability of land cover change and inadequate land use throughout the world. The topmost fertile soil is predominantly removed via erosion, which lowers both the productivity of the land and the storage capacity of reservoirs. This consequence has a large influence on the global economy; as a result, soil conservation practice is currently being used as a crucial solution to this issue. It is challenging, particularly for developing nations, to carry out soil conservation in a large basin without a significant investment. Therefore, it is essential to determine the most susceptible regions to erosion and implement the necessary remedial measures for those areas in order to lower the investment cost and enhance the efficiency of the work. In this study, erosion-prone areas are identified using the SWAT model. This model has a high level of acceptance globally for managing, allocating, and predicting water resources. However, there is some uncertainty with the model, and this uncertainty may come from the measured data, model structure, or input data. Therefore, extensive sensitivity analysis, uncertainty prediction, and model calibration are required to achieve good simulation results. Model calibration, sensitivity analysis, and prediction of uncertainty are conducted with SUFI-2, ParaSol, GLUE, and PSO, which are integrated with SWAT-CUP software. For this study, SUFI-2 algorithm gives the best simulation results out of the four algorithms, with R2 values of 0.76 and 0.85 under streamflow and R2 values of 0.8 and 0.66 under sediment for calibration and validation, respectively. From the final simulation of 34 subbasins, five subbasins were identified as the most sensitive areas for erosion; these subbasins are 1, 3, 5, 14, and 17, and they have a mean annual sediment yield of 23 tons/ha/year.