Different methods to produce distributed soil thickness maps and their impact in the reliability of shallow landslide modeling at catchment scale

It is well established that soil thickness is one of the most important factors controlling shallow landsliding. Notwithstanding, its spatial organization over large areas is poorly understood and basin scale slope stability modeling is often accomplished using constant soil thickness values or with spatially variable soil thickness maps obtained with simple models based on a single topographic attribute. In this paper we made a comparison between various methods to enter soil thickness as a spatial variable in a deterministic basin scale slope stability assessment. We used a slope stability simulator that couples a simplified solution of Richards infiltration equation and an infinite slope model with soil suction effect. Soil thickness was entered in the stability modeling using spatially variable maps obtained with different methods: a) linear correlation with elevation; b) linear correlation with slope gradient; c) exponential correlation with slope gradient; d) combination of curvature, position along the hillslope profile and slope gradient; e) a more complex model that uses the same topographic attributes of the previous one in conjunction with geological and geomorphological factors. Soil thickness maps were validated using several control points and the derivate Factor of Safety (FS) maps were validated using a landslide inventory. Results confirmed that FS is very sensitive to soil thickness and that infinite slope based models, when fed with more reliable soil thickness maps, provide more consistent spatial distribution of FS values. As a consequence, in a deterministic approach the uncertainty in the FS calculation can be reduced by applying more precise soil thickness input data. At the same time we demonstrated that the same slope stability model can be highly sensitive or highly specific depending on the input data: soil thickness models that systematically overestimate soil thickness underestimate FS values, conversely models that systematically underestimate soil thickness tend to produce FS maps affected by an overestimation of FS. We pointed out that the mean absolute error and the skewness of the frequency distribution of the errors of the various soil thickness models are equally important for a correct definition of the basin scale spatial distribution of FS. Despite the fact that slope-based methods are the most used in literature to derive soil thickness, in our applications they returned poor results. Moreover, we verified that the use of an exponential (rather than a linear) relationship does not improve the results, on the contrary sometimes they can be even worsened. The use of a combination of three morphometric attributes (sGIST model) depicted more consistent soil thickness maps and improved the performance of the stability model. A further substantial improvement was obtained when using soil thickness maps derived from a model that encompasses geomorphological criteria as well (GIST model).