Forest fire often occurs in mountainous and agricultural areas as a result of a complex processes involving several factors in a particular location. landscape, local weather and human interventions are the most important factors which influence the occurrence of forest fire. Ten Provinces in the upper north region of Thailand are identified as highly susceptible areas to reoccurring hotspots and smoke plumes during January to May. Identification of hotspots will significantly aid in forecasting forest fire in these areas. Using this, forest fire risk assessment may then be undertaken which can be utilize to evaluate its impacts on the society and economy. This research focused on enhancing the application of basic concepts of forest fire risk assessment. A case study was undertaken which entailed forecasting the risk areas of forest fire in 2015 as compared to other years. Two main methodological approaches were applied, spatial modelling for susceptibility assessment and forest fire impact determination. The research generated a risk forecasting model based on major physical-spatial factors in GIS environment. Analysis Hierarchy Process (AHP) was used to define ranges of risk conditions for 7 day forecast. The risk mapping in 2015 was validated by accumulated hotspots in the same year wherein 74 % of hotspots were located in high risk zones for forest fire. Moreover, the forest fire risk forecasting was deployed by local agencies for preparedness and prevention of wildfire incidents in 2016 and 2017. Result shows that burnt areas classified from LANDSAT-8 decreased from 28,803 km2 in 2015 to 16,588 and 13,658 km2 in 2016 and 2017, respectively. The cross matching results between risk maps and burnt areas show better correlation percentage at 52 % in 2016 and 60 % in 2017. Thus, forest fire risk forecasting can efficiently support forest fire management implemented by government agencies and local communities. © 2017 ACRS. All rights reserved.
