What Fraction of Global Fire Activity Can Be Forecast Using Sea Surface Temperatures?

Thursday, 17 December 2015: 13:55
3014 (Moscone West)
Yang Chen1, James Tremper Randerson2, Douglas C Morton3, Niels Andela3 and Louis Giglio4, (1)University of California Irvine, Irvine, CA, United States, (2)University of California Irvine, Department of Earth System Science, Irvine, CA, United States, (3)NASA Goddard Space Flight Center, Greenbelt, MD, United States, (4)NASA GSFC, Greenbelt, MD, United States
Variations in sea surface temperatures (SSTs) can influence climate dynamics in local and remote land areas, and thus influence fire-climate interactions that govern burned area. SST information has been recently used in statistical models to create seasonal outlooks of fire season severity in South America and as the initial condition for dynamical model predictions of fire activity in Indonesia. However, the degree to which large-scale ocean-atmosphere interactions can influence burned area in other continental regions has not been systematically explored. Here we quantified the amount of global burned area that can be predicted using SSTs in 14 different oceans regions as statistical predictors. We first examined lagged correlations between GFED4s burned area and the 14 ocean climate indices (OCIs) individually. The maximum correlations from different OCIs were used to construct a global map of fire predictability. About half of the global burned area can be forecast by this approach 3 months before the peak burning month (with a Pearson’s r of 0.5 or higher), with the highest levels of predictability in Central America and Equatorial Asia. Several hotspots of predictability were identified using k-means cluster analysis. Within these regions, we tested the improvements of the forecast by using two OCIs from different oceans. Our forecast models were based on near-real-time SST data and may therefore support the development of new seasonal outlooks for fire activity that can aid the sustainable management of these fire-prone ecosystems.