GLOBAL AND REGIONAL LANDSLIDE INVENTORY APPLICATIONS: METHODOLOGY AND HAZARD ANALYSIS

This research explores two landslide inventories at the global and regional scale and examines their potential applicability and validation capabilities within an emerging global landslide prediction framework. The global analysis develops a methodology for compiling rainfall-triggered landslide events, drawing upon news reports, scholarly articles and other hazard databases to develop catalog at the global scale. The events cataloged in the inventory include information on the nominal and geographic location, date, affected population, information source, and a qualitative measure of the landslide event’s size and location accuracy. This global inventory differs from other landslide catalogs by providing a publicly available database of information on rainfall-triggered landslide events globally, which can be compared to other sources. The global catalog is used to evaluate preliminary landslide forecasting work as well as to assess landslide distribution and frequency worldwide.

Due to the empirical and qualitative nature of the global inventory, the catalog only represents a subset of the total rainfall-triggered landslides globally. Regional landslide inventories can provide more detailed information from which to assess landslide susceptibility and hazard. This research draws upon a regional landslide inventory assembled by the USGS and others following Hurricane Mitch’s passage in Central America in 1998 as well as several remote sensing data sources to develop a static regional landslide susceptibility map for four countries in Central America. The regional susceptibility map and inventory are tested within the preliminary landslide forecasting framework. Evaluation of the global and regional forecasting frameworks suggests that the current global landslide inventory can provide a general picture of landslide occurrence within the existing algorithm framework; however, employing a regional inventory and evaluation framework significantly improves the accuracy of this prediction system. Additional regional inventories are greatly needed to better resolve the temporal and spatial distribution of landslide events and allow for more detailed validation within an inter-comparison landslide prediction framework.