Cause-specific Spatial Point Pattern Analysis of Forest Fire in Korea

Forest fire occurrence in Korea is highly related to human activities and its spatial distribution shows a strong spatial dependency with cluster pattern. In this study, we analyzed spatial distribution pattern of forest fire with point pattern analysis considering spatial dependency. Distributional pattern was derived from Ripley's K-function according to causes and distances. Spatially clustered intensity was found out using Kernel intensity estimation. As a result, forest fires in Korea show clustered pattern, although the degrees of clustering for each cause are different. Furthermore, spatial clustering pattern can be classified into two groups in terms of degrees of clustering and distance. The first group shows the national-wide cluster pattern related to the human activity near forests, such as human-induced accidental fire in mountain and field incineration. Another group shows localized cluster pattern which is clustered within a short distance. It is associated with the smoker fire, arson, accidental by children. The range of localized clustering was 30 km. Beyond of this range, the patterns of forest fire became random distribution gradually. Kernel intensity analysis showed that the latter group, which have localized cluster pattern, was occurred in near Seoul with high densed population.