An empirical prediction approach for seasonal fire risk in the boreal forests

The ability to predict forest fire risk at monthly, seasonal and above-annual time scales is critical to mitigate its impacts, including fire-driven dynamics of ecosystem and socio-economic services. Fire is the primary driving factor of the ecosystem dynamics in the boreal forest, directly affecting global carbon balance and atmospheric concentrations of the trace gases including carbon dioxide. Resilience of the ocean-atmosphere system provides potential for advanced detection of upcoming fire season intensity. Here, we report on the development of a probabilistic empirical prediction system for forest fire risk on monthly-to-seasonal timescales across the circumboreal region. Quasi-operational ensemble forecasts are generated for monthly drought code (MDC), an established indicator for seasonal fire activity in the Boreal biome based on monthly maximum temperature and precipitation values. Historical MDC forecasts are validated against observations, with good skill found across northern Eurasia and North America. In addition, we show that the MDC forecasts are an excellent indicator for satellite-derived observations of burned area in large parts of the Boreal region. Our discussion considers the relative value of forecast information to a range of stakeholders when disseminated before and during the fire season. We also discuss the wider role of empirical predictions in benchmarking dynamical forecast systems and in conveying forecast information in a simple and digestible manner.