Communicating landslide hazard and risk through global catalogs and a forecasting framework

Landslides are one of the most pervasive hazards in the world, resulting in more fatalities and economic damage than is generally recognized. Intense and prolonged rainfall is the most frequent trigger of mass movement events, which have caused over 11,500 fatalities within 70 countries since 2007. Within the United States alone, heavy thunderstorms, landfalling tropical cyclones, and orographically enhanced storm systems have triggered more than 400 destructive landslides over the past four years, which have resulted in significant economic damage from destroyed houses and blocked roads and 6 fatalities. Since 2007, we have systematically cataloged rainfall-triggered landslides (including mudslides, debris flows, etc.) from media reports, online disaster databases, and other credible sources. The catalog, which currently has over 2,400 entries, includes information on the date, location, impacts and rainfall trigger of the event. This database has been valuable for identifying patterns in landslide occurrence and reporting at the global level. Results from the database indicate that the Asian monsoon is the most significant driver of landslide reports, with a consistent peak in the number of reports and fatalities during the northern hemisphere summer in countries such as Nepal, India, China, Philippines, and Indonesia. Tropical cyclones in the Atlantic, Pacific and Indian Ocean basins can also trigger widespread landsliding when they make landfall. The motivation for compiling this global database of events was driven by the development of a prototype satellite-based landslide hazard forecasting algorithm, which uses satellite-based estimates of rainfall and surface cover to assess potential landslide triggering conditions at the global scale in near real-time (http://trmm.gsfc.nasa.gov/publications_dir/potential_landslide.html). Evaluation of the forecasting system using the global landslide catalog indicates that forecasting errors, including both under and over estimation of potential landslide activity areas, are geographically variable and depend on the spatial resolution of the landslide susceptibility information as well as the rainfall information and thresholds used to identify triggering conditions in time. The ability to accurately evaluate the algorithm’s performance is closely linked to the availability and accuracy of landslide event information in the global landslide catalog. Near real-time communication of susceptible landslide areas from the landslide algorithm framework is controlled by the availability and accuracy of satellite-based precipitation information. Current missions such as the Tropical Rainfall Measuring Mission (TRMM) as well as future missions such as the Global Precipitation Measurement (GPM) mission provide critical inputs to near real-time hazard assessment and numerical weather forecasts. Synthesizing landslide report information and the landslide hazard forecasting framework provides the unique opportunity to evaluate and communicate landslide hazard and risk globally to a broad audience, including the scientific community, public, governments, and response organizations.