INSTITUTO ARGENTINO DE NIVOLOGIA, GLACIOLOGIA Y CIENCIAS AMBIENTALES
Unidad Ejecutora - UE
congresos y reuniones científicas
A PROBABILISTIC APPROACH TO LANDSLIDE SUSCEPTIBILITY MAPPING USING MULTI-TEMPORAL AIRBORNE LIDAR DATA
MORA, OMAR; TOTH CHARLES; GREJNER-BRZEZINSKA, D.; LENZANO, MARÍA GABRIELA
Conferencia; ASPRS-The Imaging & Geospatian Information Society; 2014
A probabilistic approach is proposed to aid landslide susceptibility mapping. The objective of the proposed approach is to identify and predict areas that may develop into landslides and quantify the growth of existing landslides with high probability. Change detection was applied to repeat airborne Light Detection and Ranging (LiDAR) surveys acquired in December of 2008 and April of 2012. The study area was along the transportation corridor of Muskingum State Route 666 in Zanesville, Ohio, an area characterized by high vegetation densities, stream and river channeling, and some residential development. In the investigation, changes between LiDAR-derived Digital Elevation Models (DEM) were computed by analyzing, cell-by-cell, the vertical differences and, consequently, generating a DEM of Difference (DoD) map. Then, a parametric z-test was used to evaluate probabilistically if single-cell differences were real as compared to noise. Next, a non-parametric signed rank test was used to assess local neighborhoods and compute the probability that the median of the samples surpassed a desired threshold. Finally, high-probability neighborhoods (clusters) comprised of a minimum area and desired probabilities were mapped as ?landslide susceptible?. The initial results, obtained by comparison to a reference landslide map, were as expected, indicating that segments of the mapped landslides experienced changes, while others did not. It was also observed that some unmapped areas also experienced changes, indicating that they may be developing landslides. This study demonstrates that the monitoring of existing and identification of newly developing landslides is feasible from multi-temporal airborne LiDAR data.