Аннотация:The sustainable development of mountainous regions in northern Vietnam has been significantly affected by natural disasters, especially landslides. Consequently, studies of landslide susceptibility zonation are essential for planning residential areas and implementing prevention strategies. Digital elevation model (DEM) data is crucial for establishing maps of landslide causative factors. Therefore, the DEM resolution may impact the efficacy of prediction models and the accuracy of landslide susceptibility maps. This study examined the effect of DEM resolutions (12.5m, 30m, 50m, 70m, and 90m) on the accuracy of landslide susceptibility maps in Laocai city, Laocai province. The eight causative factor maps, including elevation, slope, slope aspect, distance to drainage, distance to roads, lithology, distance to faults, and rainfall, were produced. Afterward, the relationship between past landslides and causative factors was analyzed using an integration of the fractal method and certainty method (FCF). As a result, five landslide susceptibility zonation maps were produced, and the study area was divided into five susceptibility zones: very low, low, moderate, high, and very high. ROC analyses revealed that all five models have very good prediction performance (AUC > 80%), with the model using a 70-m DEM resolution providing the best performance. According to the results of this study, there is a relationship between the resolution of the DEM map and the average size of landslides in the study area. In addition, the study results indicated the significance of the landslide inventory map for assessing landslide susceptibility in any region.
