Spatial datasets for benchmarking machine learning-based landslide susceptibility models

The spatial dataset consists of 743 landslide polygons, landslide centroid points, randomly non-landslide points, and 11 landslide-controlling factors. Landslide polygons were delineated through manual interpretation of high-resolution satellite imagery. The landslide-controlling factor data were extracted from topographic maps and Indonesia’s national digital elevation model (DEMNAS). The landslide-event dataset was mapped by comparing pre- and post-event (Tropical cyclone (TC) Cempaka, which occurred on 27–29 November 2017) high-resolution satellite imageries and conducting field surveys. The landslide polygons indicate areas with confirmed landslide occurrences, while the landslide-controlling factors data includes slope aspect, distance to river, distance to road, elevation, lithology, landuse, plan curvature, profile curvature, slope, stream power index, and terrain wetness index. The landslide polygons and points are stored in gpkg format, while the landslide controlling factors are stored in tif format. Files with xml and tfw extensions are text files used to store metadata and georeference of a tif raster file. All data can be opened using GIS software such as QGIS. The datasets can also be accessed and opened using R or Python using specified geospatial libraries such as SF and Terra.