Abstract

Landslide susceptibility mapping is a challenging task if spatial heterogeneity is considered. Considering the difference in topography and neglected landslide events, particularly in transboundary areas of the Himalayas, it is necessary to precisely model the susceptibility in relation to spatial heterogeneity. In this study, we have used multisource remote sensing data, open-source geographical data, and multisource landslide event inventories for the past 20 years (2000−2020). In addition, machine learning methods were implemented to establish a landslide susceptibility model with internal landslide drivers within the transboundary areas. Considering the spatial heterogeneity, the landslide susceptibility of the northern and southern slope in the transboundary areas of the Himalayas was determined using Google Earth Engine (GEE), an online cloud platform. Furthermore, we generated a thematic map of landslide susceptibility at a spatial resolution of 500 m, considering the contribution rate of each landslide internal driver of the landslide. Finally, we analyzed the landslide drivers in areas with high landslide susceptibility and low landslide event point density. The main results are as follows: (i) elevation, aspect, and fractional vegetation cover (FVC) are top three internal drivers of susceptibility on the northern and southern slopes. Meanwhile, the northern slope is dominated by geological, seismic, and soil structure parameters while the southern slope is dominated by terrain, geomorphological, and land cover parameters; (ii) elevation, aspect, and FVC of 3400–4100 m, 260°–335°, and 8 %–30 %, respectively, are related to high landslide susceptibility on the northern slope, while elevation, aspect, and FVC of 980–2100 m, 170°–230°, 70 %–98 %, respectively, are related to high landslide susceptibility on the southern slope, which presents more extreme landslide-susceptible areas compared with the northern slope; and (iii) areas with high landslide susceptibility and low landslide event density are universal. These insights demonstrate that modeling and analyzing the heterogeneity of landslide susceptibility of the northern and southern slopes in transboundary areas of the Himalayas are required to provide a reference for locating and preventing landslide disasters, increase to the focus on landslide-susceptible areas, improve the data gap between landslide inventories and actual landslide events, and reduce the impact of missing data in landslide inventories for this region.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call