Estimation of Time Series Model on Forest Fire Occurrences and Burned Area from 1970 to 2005

摘要

It is important to understand the patterns of forest fire in terms of effective prevention and suppression activities. In this study, the monthly forest fire occurrences and their burned areas were investigated to enhance the understanding of the patterns of forest fire in Korea. The statistics of forest fires in Korea, 1970 through 2005, built by Korea Forest Service was analyzed by using time series analysis technique to fit ARIMA models proposed by Box-Jenkins. The monthly differences in forest fire characteristics were clearly distinguished, with 59% of total forest fire occurrences and 72% of total burned area being in March and April. ARJMA(1, 0,1) was the best fitted model to both the fire accurrences and the burned area time series. The firetime series have a strong relation to the fire occurrences and the burned area of 1 month and 12 months before.