Spatio-temporal Pattern Analysis of Wildfires using GIS in Korea: 1991-2014

Abstract

During past 40 years, there have been over 440 wildfires per year in Korea. Wildfire can cause severe damage to ecology, finance, and society. Especially, during the 1996 Goseong Forest Fire and 2000 East Coast Forest Fire, Korea suffered huge economic damage which took a long time to recover from. In addition, Korea is vulnerable to wildfires due to large mountainous areas and coniferous forests. Moreover, the risk and damage of wildfires are increasing because of climate changes. To reduce wildfire damage, more understanding about forest fires is required. The purpose of this study is to analyze spatio-temporal patterns of wildfires in Korea and examine the factors contributing to wildfires. This study also examines spatial pattern analysis of wildfire damage in Korea. This study analyzed data from 9,554 wildfire occurrences during 1991-2014 using GIS. This study analyzed trends in wildfire occurrences, damage caused, and burnt areas in Korea, and spatial distributions for 230 municipalities. Also, this study analyzed temporal characteristics in terms of season, day of the week, and time of occurrence. In order to examine factors contributing to wildfire damage, this study designed four models with different combination of variables. In total six factors, including 23 variables, were examined using OLS regression analysis. In terms of spatial pattern analysis, this study found spatial autocorrelation of wildfire damage in Korea, and conducted cluster analysis for examine wildfire damage hotspot areas. This study found five results. First, in the 2010s, wildfire occurrences and damage in Korea are increasing, but the burnt area is decreasing. The frequency and damage are happening most frequently in Gangwondo and Gyeongsangbukdo regions. Second, the most common wildfire occurrence in Korea is at daytime, on Sunday, in spring season and caused by negligence. Third, the factors contributing wildfires have changed according to different time periods. Commonly, variables regarding conifer forests, population density, spring season, daytime, and negligence showed significant relation to forest fire damage among study periods. Next, hotspot areas of wildfire damage have moved during different time periods. In 2010s, hotspots are located in south-eastern parts of Korea such as Gyeongsangnam-bukdo and Jeollanamdo area. Lastly, municipalities in hotspots showed some significant relationship with wildfires: humidity, days after last rainfall, trails, population density, summer, and morning time. In conclusion, first, reinforcement of wildfire mitigation education to mountain hikers and incineration activity in the spring season is required. Second, local governments who have more coniferous trees and low populations have to intensify their wildfire surveillance system to reduce wildfire damage. Third, local governments in wildfire hotspots should pay attention to wildfire occurrences and make policies regarding a collaborative wildfire management system, such as having joint training, sharing resources, and making comprehensive plans to reduce wildfire damage.