Validation of forest fire hotspot analysis in GIS using forest fire contributory factors

Forest fires have been showing an increasing trend in the past few years. Hundreds of hectares of forests are damaged every year. Thus, it is crucial to identify and implement appropriate forest fire management measures. GIS hotspot analysis has become an advantageous technique for the analysis of spatial clustering of forest fires. However, inadequate research has been undertaken on the validation of forest fire hotspot analysis in GIS. The objective of this paper is to validate forest fire hotspots identified by two statistical-based and one non-statistical-based GIS hotspot analysis methods, namely Getis-Ord Gi*, Anselin Local Moran's I and Kernel Density Estimation (KDE), in a study area. The three hotspot analyses were validated by evaluating the spatial interference between the identified forest fire hotspots by each method and existing forest fire contributory factors. The study found that KDE resulted in better spatial matching of forest fire hotspots and forest fire contributory factors, compared to Getis-Ord Gi*. However, Anselin Local Moran's I did not identify any statistically significant forest fire hotspots in the study area.