The effect of DEM resolution on the computation of the factor of safety using an infinite slope model

Abstract

The quality of digital elevation models (DEMs) is essential for reliable landslide susceptibility assessments. In this paper, two DEMs derived from ASTER (ASTER GDEM v.2 with 30m horizontal resolution) and TerraSAR-X (GeoElevation10 with 10m horizontal resolution) data are compared to study the effects of resolution on the derived slope and wetness index parameters in the application of the infinite slope model for the computation of the factor of safety. Several slope stability scenarios representing different wetness conditions with 5, 10 and 100mm d−1 of steady-state recharge were calculated for the eastern flank of Mount Rinjani Volcano on Lombok Island, Indonesia. Each scenario was conducted by computing the static factor of safety with mean values of the bulk density, angle of internal friction, cohesion, and failure depth parameters, as well as for their normally distributed values by Monte Carlo simulation. All scenarios were applied to both DEMs. The scenarios were evaluated by calculating the success/prediction rate using the respective area under the curve (AUC) and an existing landslide inventory. Additionally, uncertainties in the estimated positions of landslides were taken into account.Depending on the particular scenario, the success rate of the GeoElevation10 model shows differences up to 3% compared to the ASTER GDEM model. This apparent improvement is mainly caused by the higher ground resolution in GeoElevation10. However, the success rate increases for the 10mm d−1 and decreases for the 100mm d−1 steady-state recharge conditions. Consequently, the more detailed flow direction in the GeoElevation10 DEM has the highest impact under conditions with lower water saturation. The slight improvement in the total model quality shows that the higher resolution of the DEM has a small impact on poorly parameterized models, in which the material properties are described by roughly estimated parameters. Therefore, the application of a high-resolution DEM to areas with a lack of data on the soil's physical properties is inefficient due to its low cost-benefit ratio. For quick analysis, the ASTER GDEM is of a suitable quality. To fully benefit from the high resolution of the DEMs, other parameters should be available in appropriate detail.