In this study, we evaluated the effect of spatial resolution of topography data on the accuracy of shallow landslide analysis using two different landslide models. First, we used a simple hydrological model combined with an infinite slope stability model to calculate the critical steady]state rainfall intensity required to cause slope failure. In the second model, we used a physical model that linked hydrological processes using a finite element method with an analysis of infinite slope stability. The two models were applied to steep forested hillslope in the Tanakami Mountains of central Japan. We used digital elevation models (DEMs) with different mesh sizes (1, 2, 5, and 10m) and different locations of calculation points of the same mesh size DEMs for this analysis. When the mesh sizes differed, the two models calculated different areas of instability. The differences in the results were attributed to the difference in reappearance of topography and the number and location of calculation points. For the model of critical steady]state rainfall intensity, when the mesh sizes were small, a detailed topography of the slopes was available; however, even for micro]depressions on the hillslopes, the calculated contribution area was relatively large, resulting in an unstable slope area. With the combined model of infiltration and stability analysis, for the larger mesh sizes, the calculated value of the safety factor at each point affected a broad area because of the low density of calculation points in the watershed. In addition, with larger meshes (>5m), the location of calculation points were important for the prediction of unstable areas. These differences led to under]and over]estimates of the unstable areas on the slopes.

Key wordsFhigh resolution, DEM, shallow landslide, forested watershed, safety factor