Abstract:
One of the most essential data sets used in landslide susceptibility mapping is the digital elevation model (DEM). As a result,
it is critical to figure out how DEM spatial resolution affects the landslide susceptibility model. In this study, the implications
of DEM spatial resolution of the advanced spaceborne thermal emission and reflection (ASTER) data source on landslide
susceptibility mapping were investigated. The ASTER DEM (30 m) was resampled to 45-, 60-, 75-, and 90-m spatial resolutions. For each DEM, eight landslide-governing factors, and landslide inventory, a set of 10 geodatabases was built using a
Geographic Information System. Maps of landslide susceptibility were produced from the statistical relationship of landslide
and landslide factors using frequency ratio (FR) and certainty factor (CF) statistical methods, which were classified into
very low, low, moderate, high, and very high susceptibility classes. The models’ performance was evaluated using landslide
density and area under the curve (AUC) methods for each DEM resolution. A coarser DEM resolution (90 m) offered the
best performance and prediction accuracy, according to the predictive curve value of AUC. According to this, coarser DEM
resolution produces higher anticipated accuracy in landslide susceptibility mapping than fine resolution, depending on the
size of individual landslides in the area. The FR model performed exceptionally well at coarser DEM resolutions (75 and
90). The AUC analysis prediction rate curve value for the FR model varies from 86 to 92%, while the CF model’s prediction
rate curve value ranges from 81to 89%, revealing that the models are fairly accurate in predicting future landslide occurrence
in the study area. As a result, statistical methodologies (frequency ratio and certainty factor) are acceptable for landslide
susceptibility mapping with regard to DEM resolution
