Cartographie de la susceptibilité aux glissements de terrain en cas de déforestion dans le Northern Negros Natural Park (NNNP) - Philippines

The Philippines is one of the most severely deforested countries in Southeast Asia with around 7 percent remaining forest in 2005. Moreover, due to its geographic circumstances, it is one of the most natural hazard prone countries in the world with frequent occurrence of earthquakes, volcanic eruptions and typhoons, resulting notably in an increasing occurrence of landslides and flash floods. This work focuses on the North Negros Island and especially the recently proclaimed Northern Negros Natural Park (NNNP) that is considered the largest remaining evergreen forest in Negros Island and one of the largest in the Central Philippines. Deforestation continues to be a threat for this forest. The fact that this forest is located in mountainous area and that, due to a very high land pressure, people always creep higher to cultivate the steep slopes of these mountains, increase the landslide susceptibility associated with this deforestation. As an answer to that situation, this research aimed two main objectives. In a first time a land cover map was produced through digital classification of a 2003 SPOT 5 satellite image focusing on attempt for forest types differentiation in NNNP. Given that conventional method using spectral characteristics of the image revealed to be unsuccessful for this differentiation, this was finally achieved with the use of elevation and the distance to river as classifiers. The land cover map produced has an overall accuracy of 89,2 %. In a second time, a Landslide Susceptibility Zonation (LSZ) model was designed and implemented in a Geographical Information System (GIS) environment. The nine selected landslide-controlling factors were built from raw data and then combined through Weighted Linear Combination (WLC).Two scenarios corresponding to a “stable forest scenario” and a “deforestation scenario” were modeled. The integration of the land cover map previously produced served successfully as the key factor that enabled to model the Landslide Susceptibility (LS) change due to potential deforestation in NNNP. A sensitivity analysis of the factors used for this modeling was realized by comparison with a “LSZ in case of deforestation” map based on the four main factors only (land cover, land cover change, slope and proximity to river). It appeared that, the LSZ model was not sensitive to the use of other factors than these four main ones. Finally, in order to give an idea of the reliability of the results, the LSZ maps were compared with a landslide inventory realized during the field survey in North Negros. Some limitations in the landslide inventory did not allow drawing pertinent conclusion from this comparison. In parallel and in order to evaluate the feasibility of integrating the land cover knowledge of local environmentalist into a GIS data base through field survey and 3D-GIS activity in a remote laboratory, a participatory 3D-GIS experience was attempted in Negros. Consistence of results between field and 3D-GIS laboratory experience validate this way of extracting land cover information, which could revealed useful for land cover identification of hardly accessible areas