Spatial model of forest fire danger
2001
Kobler, A. (Slovenian Forestry Inst., Ljubljana (Slovenia))
We developed a regional GIS - based model for daily forest fire danger estimation in SW Slovenia depending upon environmental, anthropogenic, and weather factors. The environmental and anthropogenic influence factors take the form of GIS layers, while the weather influences are described indirectly by using estimates of fire danger in natural environment, which are updated daily by the Hydrometeorological Service. The model is based on empirical data about past forest fires. Three modeling methods were comparatively used: linear regression, logistic regression, and top down induction of decision trees. The best results were obtained by logistic regression. We implemented the chosen model in the ArcView GIS environment. Our results indicate that using a sufficiently large and representative sample of fire data it would be possible to develop a useful spatial model for dynamically forecasting fire danger in the natural environment at the country level.
Mostrar más [+] Menos [-]Palabras clave de AGROVOC
Información bibliográfica
Este registro bibliográfico ha sido proporcionado por University of Ljubljana