Calibrating soil erodibility parameters of landslide dams using back analyses

Landslide dam failures typically result in significant economic losses and human casualties in downstream areas. The soil erodibility of a landslide dam is one of the critical factors controlling the erosion process. However, the complexity of landslide dam formation leads to high spatial variability in soil erodibility parameters, and their varying patterns remain unclear. This paper proposes a framework to analyze and calibrate the varying patterns of soil erodibility parameters of landslide dams. Trend functions were used to characterise soil erodibility parameters at different depths, converting the calibration of soil erodibility parameters into the calibration of the coefficients of trend functions. A multi-criteria decision-making method was employed to determine the optimal trend function types. Two back analysis methods (deterministic and probabilistic) were used to calibrate the trend function coefficients for the two successive Baige landslide dams. The two study cases demonstrated that the trend functions determined in this paper can effectively characterise the varying patterns of soil erodibility parameters of landslide dams. Additionally, factors contributing to the differences in soil erodibility of different landslide dams were briefly discussed, laying a basis for future research on this topic.