11 November 11
A Nasa-funded research team at the University of California has created a model that can successful predict the severity and distribution of fires in the Amazon rainforest months in advance by analysing satellite data.
Previous research has shown that human settlement patterns are the main factor driving the distribution of fires in the Amazon. However, the new research shows that environmental factors -- including small variations in ocean temperatures -- intensify human influence and help explain the variability in the number of fires in the region from one year to the next.
Researchers analysed nine years of fire activity data collected by Moderate Resolution Imaging Spectroradiometer Instruments (Modis) on Nasa's Terra and Aqua satellites. They compared these to the National Oceanic and Atmospheric Administration's records of sea surface temperatures and changes in precipitation patterns measured by the Tropical Rainfall Measuring Mission -- a satellite co-managed by Nasa and the Japan Aerospace Exploration Agency.
According to the study -- published in Science -- temperature changes as little as 0.25°C in the North Atlantic and 1°C in the Central Pacific can be used to forecast how severe the fires are will be across the region. Higher-than-normal temperatures tend to precede a severe fire season four to six months later.
The team believes that unusually warm sea surface temperatures cause regional precipitation patterns to shift north in the southern Amazon during the wet season. Co-author of the study James Randerson explains: "The result is that soils don't get fully saturated. Months later, humidity and rainfall levels decline, and the vegetation becomes drier and more flammable."
The findings build on a Columbia University-led study published in July 2011 that showed how sea temperatures in the Northern Atlantic could forecast fire severity across a small section of the western Amazon. The new study considers a broader area of South America and takes into account the temperatures of the Pacific as well as the Atlantic.
The team successfully predicted that there would be a prolonged drought and severe fires in the 2010 fire season, but will have to wait to see whether the model's predictions for 2011 (which haven't yet been published) were also accurate, since the activity peaks in September and November.
They also noticed a pattern emerging: fires in the southern and south-western part of the Amazon were most strongly influenced by sea surface temperatures in the North Atlantic, while fires in the eastern part of the Amazon were affected by sea temperatures in the central Pacific.
The team hopes that the findings may be used to build an early warning system for fires to aid South American authorities prepare for fire seasons.