It’s already well known that sea surface temperature patterns play a big role in climate, with El Niño being the classic example. Warmer sea surface temperatures heat the air above, increasing convection and evaporation, all of which can impact the climate over surrounding landmasses for months at a time.

With this in mind, Yang Chen from the University of California, Irvine, US, and colleagues investigated the correlation between sea surface temperature (from 14 different ocean regions), and observed wildfires on land, using data going back over 18 years.

The researchers found that 48% of the wildfires were strongly correlated to a single sea surface temperature variable, and that such fires could have been anticipated up to three months in advance. They believe that this link is due to the longer term climate impacts of sea surface temperature.

"Sea surface temperature modulates the water storage below ground on land, which regulates the fuel moisture and affects fire severity," said Chen. "It also modifies biomass growth, which regulates the fuel amount."

Some regions were much more predictable than others. In equatorial Asia, 92% of the burned area was strongly correlated to sea surface temperature, and in Central America 86% of the burned area was strongly correlated. The Pacific Ocean El Niño–Southern Oscillation was the most significant ocean region when it came to predicting wildfire, with changes here accounting for around one-third of the predictable wildfires.

Taking the study one step further, Chen and his colleagues developed a model that combined sea surface temperature from two ocean regions and found that they were able to provide even more confident hindcasts of wildfire outbreaks. "This suggests that fires in many regions respond to forcing from more than one ocean basin," they write in Environmental Research Letters (ERL).

Using their model the scientists were able to identify 12 "hotspot" regions, where wildfires tended to have a consistent link with sea surface temperature. These regions were: the southern US, Central America, southwest Amazon, southeast Amazon, east Amazon, west Africa, southwest Africa, central Asia, northeast Asia, southeast Asia, southern hemisphere equatorial Asia and Australia.

Within these regions Chen and his colleagues are confident that it should be possible to anticipate wildfire probability up to a few months in advance. "Forecasts like this could be used to strengthen fire management in vulnerable regions and reduce fatalities from reduced air quality," said Chen. "This seasonal forecast can also be combined with instantaneous fire observations and meteorology projections to provide near-real-time fire severity forecasts."

Although Canada is not one of the "hotspot" regions, Chen and his colleagues believe that their model would have identified the heightened risk of the spring 2016 wildfire. "It is clearly associated with El Niño and should be anticipated using sea surface temperatures with moderate confidence," said Chen.

Related links

Related stories