Rocky Mountain Section - 73rd Annual Meeting - 2023

Paper No. 23-11
Presentation Time: 11:40 AM

UPDATED LANDSLIDE SUSCEPTIBILITY MAP OF THE UNITED STATES


BELAIR, Gina, U.S. Geological Survey, Geologic Hazards Science Center, Denver Federal Center, P.O. Box 25046, MS 966, Denver, CO 80225

Landslides occur over highly variable terrain in all U.S. states and territories. Landslide susceptibility maps, which spatially characterize the potential for landslide occurrence, can help inform decisions about emergency management, land-use, and infrastructure development. Current national landslide susceptibility maps have coarse resolution and tend to underrepresent potential in moderate-to-low classes. A map with only two classes, either “some” or “no” susceptibility, can be useful at the national scale in lieu of multiple hazard classes, which are more difficult to define. Susceptibility maps based on slope-relief thresholds were previously created for the conterminous U.S. (CONUS) in 2012 and globally in 2021 using 90-m digital elevation models (DEMs), which resulted in 1-km maps. We present an updated national susceptibility map for all landslide types using a similar approach, but with several notable advances. Our new slope-relief threshold was calculated using: (1) higher resolution (10-m) DEMs, (2) ecosystem data, (3) a substantially larger training inventory with over 613,000 landslides, and (4) multiple model iterations using stratified random sampling. It is also higher resolution (90-m) and spans Alaska, Hawaii, and Puerto Rico. Our analysis excluded landslides with the lowest data confidence (defined by national landslide inventory standards) and our threshold was weighted by ecoregion landslide density. For CONUS, 34% of the area has “some” susceptibility and 97% of the landslides fall within this class. This is an improvement to the original USGS map and the CONUS portion of the global map (original USGS map: 31% area has “some” susceptibility, 79% landslides correctly classified; global map: 19% and 63%, respectively). The improvements in predictive capability and resolution can lead to a better, uniform characterization of landslide potential for the entire nation.