The El Niño Southern Oscillation index and wildfire prediction in British Columbia

Zhen Xu, G.C. van Kooten

Research output: Contribution to journalArticleAcademicpeer-review


This study investigates the potential to predict monthly wildfires and area burned in British Columbia's interior using El Niño Southern Oscillation (ENSO). The zero-inflated negative binomial (ZINB) and the generalized Pareto (GP) distributions are used, respectively, to account for uncertainty in wildfire frequency and area burned. Results indicate that a four-month lag of the ENSO index has a strong positive influence on monthly wildfire occurrence. Upon fitting the GP distribution with a logit model regressed on the ENSO index, we predict the probabilities that monthly area burned exceeds 1700 ha and find that risks of large fires are significantly higher in northwestern BC. However, the ENSO is likely unable to provide consistent predictions of the total area burned in any month. Sensitivity analysis indicates that increases in the mean value of the monthly ENSO index result in a small increase in the predicted number of fires and an increase in the probability of large burns. This study has several implications for decision-making regarding firefighting budget planning and insurance for firefighting expenditures.
Original languageEnglish
Pages (from-to)592-598
JournalForestry Chronicle
Issue number5
Publication statusPublished - 2014


  • inflated poisson regression
  • wildland fire
  • forest-fires
  • united-states
  • climate
  • canada
  • model
  • weather


Dive into the research topics of 'The El Niño Southern Oscillation index and wildfire prediction in British Columbia'. Together they form a unique fingerprint.

Cite this