In this study, spatial patterns and driving factors of fires were identified from 2000 to 2010 using Ripley's K (d) function and logistic regression (LR) model in two different forest ecosystems of China: the boreal forest (Daxing'an Mountains) and sub-tropical forest (Fujian province). Relative effects of each driving factor on fire occurrence were identified based on standardized coefficients in the LR model. Results revealed that fires were spatially clustered and that fire drivers vary amongst differing forest ecosystems in China. Fires in the Daxing'an Mountains respond primarily to human factors, of which infrastructure is recognized as the most influential. In contrast, climate factors played a critical role in fire occurrence in Fujian, of which the temperature of fire season was found to be of greater importance than other climate factors. Selected factors can predict nearly 80% of the total fire occurrence in the Daxing'an Mountains and 66% in Fujian, wherein human and climate factors contributed the greatest impact in the two study areas, respectively. This study suggests that different fire prevention and management strategies are required in the areas of study, as significant variations of the main fire-driving exist. Rapid socio-economic development has produced similar effects in different forest ecosystems within China, implying a strong correlation between socio-economic development and fire regimes. It can be concluded that the influence of human factors will increase in the future as China's economy continues to grow - an issue of concern that should be further addressed in future national fire management.
Keywords: Daxing'an Mountains; Driving factors; Fire occurrence; Logistic regression model; Spatial distribution.
Copyright © 2017 Elsevier B.V. All rights reserved.