Assessing and modeling hill fire impact in country parks with SPOT HRV images and GIS

Abstract

Remote sensing and geographic information systems (GIS) techniques provide efficient tools for resource management. In this paper, these techniques are applied for resource management of country-parks in Hong Kong, A spatial database is developed integrating SPOT HRV multispectral data, topographic data, hill fire record and thematic maps. Vegetation resources are mapped based on multidate SPOT and DEM data with an overall accuracy of 74.3%. Integrated with hill fire record, the impact of hill fire on vegetation resources is assessed. Grass is the major vegetation type subject to fire. Bi-variate relationships between hill fire, its frequency of occurrence and various terrain (elevation, slope, aspect and distance to streams) and human activities (accessibility and point sources of fire) variables are examined. A logistic regression model is used to identify hill fire prone sites with a 75.4% of correct prediction of fire vs. non-fire sites. Elevation, vegetation and accessibility are found to be more important factors for hill fire occurrence.