The relative importance of driving factors of wildfire occurrence across climatic gradients in the Inner Mongolia, China

Abstract

In recent decades, wildfire regimes have been changing in many parts of the world with significant differences among climatic regions. Thus, understanding the relationship between fires and driving factors in different climatic regions is important. We investigated the relative importance of wildfire drivers across climatic gradients using a Bayesian network model during 2001–2015 in the Inner Mongolia Autonomous Region (IMAR). Our results showed that regional-scale fire relative probability (FRP) on average increased in the IMAR, which was mainly attributed to increases in annual precipitation. However, changes in FRP and driving factors of wildfire occurrence varied among climatic regions. The FRP decreased at rates of −0.02/decade and −0.001/decade in the humid and semi-humid region, respectively, while increasing at an average rate of 0.05/ decade in the arid region. Temperature was recognized as the most influential driver of wildfires in the humid region, in which the relative importance of all driving factors remained relatively stable over time. In the semi-humid region, wildfires responded primarily to human activities, of which distance to railways was of greatest importance. The relative importance of distance to railways significantly increased over time, while the effects of temperature, humidity and elevation decreased considerably. Fractional vegetation cover (FVC) was the dominant factor controlling the wildfire occurrence in the arid region, in which the relative importance of precipitation significantly increased and the effect of settlements significantly decreased. Our study suggests that because of different driving factors of wildfires across climatic gradients, different fire prevention and management strategies are required. The optimal resource allocation for fire prevention should be adjusted according to the relative importance of wildfire drivers and spatiotemporal patterns of the FRP.

Keywords

• Wildfire
• Inner Mongolia
• Bayesian belief network
• Driving factors
• Climatic regions
• Spatiotemporal variations