Dynamic behaviors of forest fire in upper Nan wastershed during 1999-2007

Abstract

The applications of geographic information system (GIS) and remote sensing help gathering and analysis the spatial patterns of the forest fire more quickly and more efficiently. This research used the temporal Landsat-5 TM imageries covered the upper Nan watershed, Thailand, acquired between December to April during 1999-2007, to study the changes of the forest fire affected areas. In this study, Landsat-5 TM bands 7-4-2 were chosen to create the false color composite. The visual interpretation combined with Difference Normalized Burn Ratio (dNBR) method was conducted to analyze the forest fire affected areas that were created in GIS database. Field investigations were also used to test for accuracy of the forest fire interpretation. The results were used to produce the burned scar maps and synthesize for the trend of forest fire distribution during those periods. The analysis found that most forest fire patches distributed on high mountainous and steep slope area that are mainly in the northern part of the watershed. The forest fire areas in 1999, 2001, 2004 and 2007 were summarized approximately as 37 km2, 107 km2, 81 km2, and 80 km2, respectively. The relationships of burned areas and relevant parameters, namely slope, aspect, distance from road, distance from settlement, vegetation and land cover were analyzed for forest fire susceptibility assessment. Various maps were constructed from the forest fire relevant parameters derived from the database. The parameters, univariant probability method, and calculation of forest fire susceptibility were applied to modify the forest fire rating risk classes and be presented as the forest fire risk map in the upper Nan watershed. The final results from this research can be used as a spatial data for supporting decision making in handling, monitoring, and controlling the forest fires in Thailand for much better efficient management.