As values and recognitions of forests have been changed, a movement to nurture them also increasingly spreads In Korea. However, forest fires not only extinguish forest resources in a moment but also seriously cause physical and economical damages as well as indirect ones. Since the forest fires occur by artificial reasons in Korea, they are closely related with land uses. Thus, the purpose of this study is to analyze spatial distribution patterns of forest fire occurrence regions in Chungcheongbuk-do and a relationship with land uses through spatial autocorrelation and spatial regression analysis from 1991 to 2012. The forest fire occurrence in Chungcheongbuk-do was highly crowded and forms the dense spacial
density pattern and exhibited the very significant spatial autocorrelation. From examinations by each reason of forest fire, it was suggested that although reason 1 presented a locally strong relation with reason 2, and they were clustered, for reason 3, outliers were often derived, so locally weak relation was shown.
It was found that the land uses which often affected the forest fire occurrences were bare , agriculture and forest ares, and the occurrence of forest fires might have effects on about 40% of average forest fire occurrences. Through examinations of land use effects by each reason of forest fire occurrences, it was suggested that agriculture and forest areas approximately affected 22% of adjacent ones'' forest fires for the reason 1; bare, agriculture and forest areas have effects on as much as 48% of forest fires in neighborhoods for the reason 2; it was analyzed that the explanatory power of the model was very low for the reason 3.
On the basis of findings, it is indicated that a specific forest fire policy suitable for several features of land uses is needed to protect the forest fires. The definite policy depending on the types of land is essential along with the existing forest fire policies, given that the local patterns of forest fires are closely associated with the land uses. This study is meaningful, in that it introduces an effective analysis on spatial data by applying a new approach to a spatial analysis of forest fire research. The more objective findings were derived from application of the spatial statistics to a relation between the existing variables. It was identified as a validate study methodology, especially in researches of forest fires highly affecting through the geographical proximity.