A RS/GIS–Based Multi-Criteria Approaches to Assess Forest Fire Hazard in Indonesia (Case Study: West Kutai District, East Kalimantan Province)

Abstract

Many forest ecosystems in Indonesia have undergone rapid changes, which certainly influencing in the forest structure and fuel properties. In addition to forest structure and fuel changes, increased number of people living in these environments and frequent extreme droughts as impacts of climate changes has led to extraordinary fire events which leads to several environmental problems including loss of forest as carbon sink, loss of biodiversity, transboundary haze pollution, as well as significant economic losses. Better management of forests and fire problems require spatial-temporal assessment providing important information for identifying and prioritizing fire planning needs. Fire hazard map determining the fire vulnerable areas is important information required to take decisions on fire prevention and suppression activities. To be effective, a comprehensive consideration for forest fire hazard assessment must accomodate a wide range of important determining variables that encompass biophysical and anthropogenic factors leading to occurrence of fire. Combining geomatics technology, including remote sensing and geographic information systems with either statistical or analytical hierarchy process (AHP) approaches offer more efficient and powerful on data acquisition and analysis related to fire hazard assessment. In this research different approaches used for assessing forest fire hazard in West Kutai district at different temporal resolutions: short-term and long-term fire probability assessment. Keetch and Byram Drought Index (KBDI) as a drought index indicating the atmospheric dryness used for short-term fire potential assessment. While long-term assessment encompass two different approaches to spatially model fire probability: binary logistic regression model and analytic hierarchy process (AHP). The first model tried to base weights for fire occurence probability map based on historical fires as recorded by satellites (hotspots), while the AHP approach is based on judgement by asking experts in a systematic fashion. The result of this research showed that by combining between the daily KBDI indexs with hotspots occurences indicated that there is a strong positive association between KBDI index with hotspot occurences. The importance level of variable contributing to forest fire occurences resulted from both binary logistic regression model and the AHP model showed a quite similar pattern.