Count Models and Wildfire in British Columbia

Two count models are estimated in this analysis to explain the occurrence of wildfire and area burned by wildfire in the interior of British Columbia, Canada. The main explanatory variable is the 4-month lagged El Niño 1&2 index, which is found to have a strong positive influence on wildfire in the study region. As a result of the lag on the climate index, the count models can be used to predict annual wildfire occurrence and the overall monthly size of the area burned by fire districts. An increase in the mean value of the monthly El Niño 1&2 index is projected to result in a slight increase in the number of fires and an increase in the probability that large areas will be burned. Not unexpectedly, however, the impact in July and August could be quite high (increases of 30%). In conclusion, given the large variance, actual changes caused by climate change are uncertain and could be dramatic.