Yang Chen has devised a way to predict the severity and geographical distribution of forest fires in the Amazon months in advance by analyzing historical satellite data.
“Higher than normal sea surface temperatures in the Atlantic and the Pacific proved to be red flags that a severe fire season was on its way in four to six months,” said Yang Chen, a UC Irvine scientist who led a team in analyzing a decade of NASA satellite data to establish the uncanny link between ocean temperature and Amazon fires.
Chen’s team found that temperature changes of as little as .25 degrees C (.45 degrees F) in the North Atlantic and 1 degree C (1.8 degrees F) in the Central Pacific can be used to forecast the severity of the fire season across much of the Amazon.
Previous research had shown that human settlement patterns are the primary factor that drives the distribution of fires in the Amazon, but Chen’s research demonstrates that environmental factors — specifically small variations in ocean temperatures — amplify human impacts and underpin much of the variability in the number of fires the region experiences from one year to the next.
Unusually warm sea surface temperatures cause regional precipitation patterns to shift north in the southern Amazon during the wet season. This shift keeps the soil from becoming fully saturated. As humidity and rainfall levels decline over the ensuing months, vegetation becomes drier and more flammable, according to James Randerson, a UC Irvine scientist who co-authored the study.
To establish the link between fires and sea surface temperatures Chen’s team analyzed nine years of fire activity data collected by Moderate Resolution Imaging Spectroradiometer instruments (MODIS) on NASA’s Terra and Aqua satellites. They compared the number of fires to records of sea surface temperatures maintained by the National Oceanic and Atmospheric Administration.
Years with anomalously cool ocean temperatures had fewer fires, while years that experienced unusually warm ocean temperatures experienced more fires. The team also looked for and found changes in precipitations patterns as measured by the Tropical Rainfall Measuring Mission (TRMM), a satellite managed jointly by NASA and the Japan Aerospace Exploration Agency (JAXA).
A Columbia University-led study published in July showed that sea surface temperatures in the Northern Atlantic could be used to forecast fire severity across a small section of the western Amazon in Peru and Brazil. Chen’s study considers a much broader swath of South America and takes into account how ocean temperatures in both the Pacific and Atlantic Oceans affect the continent’s fires.
The UC Irvine team also developed and validated an innovative computer model that they used to predict 2010 fire activity and could be used to forecast fire season severity in the future. The team’s model predicted that prolonged drought and severe fires would occur during the 2010 fire season. That prediction was proven to be extremely accurate.
“Fire activity can vary dramatically. Satellites detected about twice as many fires during 2010 as they did in 2009,” said Doug Morton, a scientist based at NASA’s Goddard Space Flight Center in Greenbelt, Md. who coauthored the new study.
“For the 2010 season, the model successfully captured not only the severity of wildfire activity, but it also got much of the east-west spatial distribution right,” said Randerson.
A full decade of data revealed a distinctive pattern: fires in the southern and southwestern part of the Amazon were most strongly influenced by sea surface temperatures in the North Atlantic, but fires in the eastern part of the Amazon were strongly affected by sea surface temperatures in the central Pacific.
Chen’s team will have to wait a few more months before they can see whether their model’s predictions for the 2011 fire season were also accurate.
Despite the wide variety of fires in the region — including fires for deforestation, agricultural management, and wildfires in savannas and tropical forests — the model managed to captured the variability in forest and savanna fires from one year to the next when considered separately.
Chen hopes that their findings will help create an early warning system for fires that would help South American authorities prepare for severe fire seasons. An early warning system could play a critical role in helping authorities blunt the negative impacts of heavy fires years such as those the region has experienced in 2005, 2007, and 2010, noted Morton.
The Amazon is a humid region that would experience very few fires in the absence of human activity.
“Deforestation rates in the Amazon have declined significantly in recent years due to government regulations, but fire activity has been holding steady and even going up in some areas due to increases in escaped agricultural fires,” said Ruth DeFries, a scientist at Columbia University, New York, and a co-author on the paper.
South American fires have a particularly important impact on climate. Fires from deforestation contribute about half of the carbon emissions from deforestation in South America. The continent creates about 15 percent of the worldwide carbon emissions from fires, according to a recent study. Climate models predict that the region will receive less rainfall as climate change progresses, which would increase the risk of forest fires and lead to greater carbon emissions.
The UC Irvine study was published in the Nov. 11th edition of Science.
Yang Chen received a bachelors degree in environmental engineering from Tsinghua University in 1997, followed by a masters in 2000. He received a PhD in atmospheric science from the University of Michigan at Ann Arbor in 2006. He spent a year there as a postdoc researcher, then two years at the Jet Propulsion Lab at Caltech. He has been an assistant project scientist at UC Irvine since 2008.
