Global patterns of fire regime analysis based on fire frequency

Background/Question/Methods

Fire management, disturbance ecology, and global biogeochemical cycle modeling all need to map the diversity of fire regimes at global scale. However, most previous studies were at local scales, via tree ring records and fire statistics. We treated 2001-2007 fire counts in the MODIS fire Climate Modeling Grid products to represent global fire frequency. In addition, we compiled nine corresponding climatic and human variables and registrated them together. Then we used a random forest regression tree procedure to explore the relative importance and relationships driving variation in different fire regimes.

Results/Conclusions

Mean annual temperature contributed most in explaining global fire frequency. High frequency fires tended to be in high temperature zones (>19.9°C). Below threshold temperatures, other factors worked together to influence fire frequency. Fire frequency increased with net primary productivity (NPP); however, the highest fire frequency did not coincide with highest NPP, but occurred in zones with intermediate NPP (380 kg C/km^2/month<NPP<956 kg C/km^2/month). Additionally, human activities had strong impacts on fire frequency. To be specific, increasing population density above a certain threshold (>44 people/km^2) decreased fire frequency. Ignition source and topographic roughness were not constraints in driving global fire frequency. According to graphic comparisons, our prediction captured the general pattern of global fire regimes and the most important driving mechanisms, although there are some disagreements in local details. The study provided new insights into global disturbance ecology research and can serve as a tool for fire management and future global biogeochemistry cycle models.