5. CONCLUSION
Based on NCRS-CN, this study investigated the spatial distribution of SY and risk of sediment disaster in typhoon events by using big data collected during 1999–2009. The relationship between potential maximum retention and SY was analyzed and applied to simulate the spatial distribution of the risk of sediment disasters. The results showed that different cumulated rainfall levels, land use, soil, and slope could all be essential factors that affect SY. Based on daily river discharge and SY data, the rating curve (Q–Qs rating curve) of runoff–SY indicated that the coefficient of determination of the curve was R² = 0.95; thus, the SY of typhoon events can be estimated using the rating curve.
The analytical results of potential maximum retention and SY showed that the relationship between the two parameters could be expressed by a power function. This function could be divided into three categories according to different cumulative rainfall conditions. If the cumulative rainfall was less than 197.2 mm, Type I was applicable, and the SY was 0. If the cumulative rainfall was between 197.2 and 754.4 mm, Type II was applicable, and the corresponding function was SY = 473.64S–0.866. If the cumulative rainfall was between 754.4 and 1419.7 mm, Type III was applicable, and the corresponding function was SY = 13764S–1.133. The potential maximum erosion in the catchment area could be estimated using the analytical results of potential maximum retention and SY.
The management strategy of this study was to take management subdivisions as a unit and to use the potential maximum erosion and slope to establish the spatial distribution of sediment disaster risks. The risks were ranked, and the landslide rate of each subdivision was estimated. The results indicated that the cumulative number of subdivision was positively correlated with the landslide rate. In areas with limited data, the relational curve between the cumulative number of subdivisions and landslide rate could be used to determine subdivisions with management priorities. In addition, in the 70% landslide rate simulation scenario, the spatial distribution of landslides in selected subdivisions was mostly consistent with actual landslide locations.