Prototyping an Early-warning System for Rainfall-triggered Landslides on a Regional Scale Using a Physically-based Model and Remote Sensing Datasets
Abstract
Recent advancements in the availability of remotely sensed datasets provide an opportunity to advance the predictability of rainfall-triggered landslides at larger spatial scales. An early-warning system based on a physical landslide model and remote sensing information is used to simulate the dynamical response of the soil water content to the spatiotemporal variability of rainfall in complex terrain. The system utilizes geomorphologic datasets including a 30-meter ASTER DEM, a 1-km downscaled FAO soil map, and satellite-based Tropical Rainfall Measuring Mission (TRMM) precipitation. The applied physical model SLIDE (SLope-Infiltration-Distributed Equilibrium) defines a direct relationship between a factor of safety and the rainfall depth on an infinite slope. This prototype model is applied to a case study in Honduras during Hurricane Mitch in 1998 and a secondary case of typhoon-induced shallow landslides over Java Island, Indonesia. In Honduras, two study areas were selected which cover approximately 1,200 square kilometers and where a high density of shallow landslides occurred. The results were quantitatively evaluated using landslide inventory data compiled by the United States Geological Survey (USGS) following Hurricane Mitch, and show a good agreement between the modeling results and observations. The success rate for accurately estimating slope failure locations reached as high as 78% and 75%, while the error indices were 35% and 49%, respectively for each of the two selected study areas. Advantages and limitations of this application are discussed with respect to future assessment and challenges of performing a slope-stability estimation using coarse data at 1200 square kilometers. In Indonesia, the system has been applied over the whole Java Island. The prototyped early-warning system has been enhanced by integration of a susceptibility mapping and a precipitation forecasting model (i.e. Weather Research Forecast). The performance has been evaluated using a local landslide inventory, and results show that the system successfully predicted landslides in correspondence to the time of occurrence of the real landslide events in this case.
- Publication:
-
AGU Fall Meeting Abstracts
- Pub Date:
- December 2010
- Bibcode:
- 2010AGUFMNH23A1428L
- Keywords:
-
- 1855 HYDROLOGY / Remote sensing;
- 1922 INFORMATICS / Forecasting;
- 7914 SPACE WEATHER / Engineering for hazard mitigation