CONSTRUCTING A COMPREHENSIVE DATABASE FOR RAINFALL-TRIGGERED LANDSLIDES IN THE UNITED STATES

Landslides are one of the most widespread occurring natural hazards in the continental United States. The documentation of mass movement is challenging because of its temporally variable and spatially discontinuous occurrence. Although, numerous states have constructed landslide inventories, the lack of temporal and geospatial data often provides inadequate information to examine areas susceptible to landslides, particularly across broader spatial scales. Hence, the establishment of a comprehensive national landslide mapping effort is required.

This study establishes a first of its kind semi-homogenous inventory for rainfall-triggered landslides in the United States. This inventory is a product of systematic integration of statewide databases, regional databases, and city-based databases within a Geographic Information Systems (GIS) module. Each landslide event was input into a standardized attribute table derived after consultation with scientists focused on landslide monitoring. A consistent set of attributes reduced data bias and discrepancies. The resulting inventory is a geospatial dataset representing 15,777 landslides from 14 databases spanning 1852-2017. The majority of landslides in the database occur during December and January, particularly in the Pacific Northwest , and in August in the eastern states. Approximately 55% of landslides in the database were triggered by 7 major storms. A spatial bias is evident for southeast and northwest U.S., which experiences frequent landslide activity seasonally. The spatiotemporal characteristics and density of landslides is evaluated with rainfall intensity between 1998 and 2017 in an attempt to highlight the rainfall triggering patterns of landslides within this area. This inventory will serve as a key source for evaluating a nationwide landslide model to evaluate potential landslide activity.