Validation of prediction patterns for future landslide characterized by spatial data
Speaker: Chih-Cie Huang
Abstract
Validation of prediction results when the predictions were generated from spatial databases for landslide hazard mapping is an important task not only to convince the success degree of prediction, but also to provide the significance information to non-specialists so that decisions can be taken about safer land uses as a consequence of the prediction results. In prediction modeling process, through transform the multi-layered database, the propensity of the land to failure can be obtained. There are two steps in procedure . The first is to divide the study area into several prediction classes according to different hazard level for the likely occurrence of future landslides as the “hazard-mapping step”, and the second step is to empirically estimate the conditional probability in each prediction class given the spatial and geomorphologic data based on cross- validation techniques, as the “probability estimation step”. If the information in the database is sufficiently representative of the typical conditions in which the mass movements originated in space and in time, then the next problem is to confirm the validity of the results of some models. Added values can be used in validation of a prediction, no matter in a fixed time interval, an open-ended time or within the confined space of a study area.
Reference
Chung, C. J. F., Fabbri, A. G. (2003) Validation of spatial prediction models for landslide hazard mapping, Natural Hazards 30, 451–472.
Chung, C. J. F., Fabbri, A. G. (2002) Modelling the conditional probability of the occurrences of future landslides in a study area characterized by spatial data, Proceedings of ISPRS 2000, Ottawa, Canada, July 8-12, 2002.