Sources of the Haze in Southeast Asia, Fires Are Quickly Located with Public GIS Data

June 26, 2013 | 3:33 pm
Doug Boucher
Former Contributor

Over the last two weeks, large numbers of fires have broken out on the Indonesian island of Sumatra, and once again this has led to massive air pollution, carried to neighboring countries such as Singapore and Malaysia. This time, however, people could see not only the choking, dangerous “haze,” but also the locations of the fires, including which ones were on lands used by companies to produce palm, oil, pulpwood, timber, and other commodities. GIS technology and publicly available data, rapidly analyzed by scientists, brought a transparency to the issue that is so sorely lacking in the air over Singapore.

The leader in this analysis has been the World Resources Institute, as part of its Global Forest Watch 2.0 project. With an initial analysis released last Friday, followed up by a second on Monday and a third one just yesterday, they have mapped the fires, calculated how many were on each company’s land, and shown how the distribution of fires among land use types and companies has been changing from day to day. They have also compared this year’s fire season (so far) to previous years, demonstrating that the current situation is not an anomalous one.

As you would expect, there has been criticism of these results from the industries involved, but in fact UCS’s own independent analysis of the data confirms WRI’s findings. Our work on this was completely unplanned; it came about because Stu Sheppard, who does GIS studies as a consultant for UCS, happened to be traveling through Singapore last Friday and witnessed the air pollution firsthand. Here’s part of the message he sent us:

Wow! I’m in Singapore right now and can’t even leave the hotel. The air quality is the worst ever! Far worse than it was in 1997 even. Singapore is, to say the least not happy with Indonesia right now. They have live updates on tv… three hours ago it was 220 and they had announcements to stay indoors. Last live update says it is 368 now!!! (more sat images:

He sent along an initial GIS map of the fires that he had done using NASA data that he downloaded from the National University of Singapore web site:

Map of the fires causing air pollution in southeast Asia, initially prepared by Stu Sheppard using two days worth of data

Map of the fires causing air pollution in southeast Asia, initially prepared by Stu Sheppard using two days worth of data

and also a picture indicating just how thick the pollution was:

Haze picture taken by Stu Sheppard in Singapore, June 2013

Then he and UCS consultant Earl Saxon did a GIS analysis of two days of fire data (June 19-20), overlaying the fire locations with other layers such as the boundaries of concessions, the companies they belonged to, and the areas protected by Indonesia’s moratorium on deforestation and peat clearing.

Although Sheppard and Saxon used only two days’ worth of data (compared to nine in the initial WRI analysis), and got the NASA fire data from a different source, their findings were quite consistent with those of WRI. While few of the fires were  in logging leases, 21 percent were on oil palm concessions and 28 percent were on pulpwood plantations. Both groups also identified the same two commercial groups (Sinar Mas and APRIL) as the largest pulpwood plantation lessees. Here’s the data table showing the distribution of fire pixels in the province of Riau:

Table 1. Ownership, land use, number and proportion of Riau fires (19-20 June) prepared by UCS consultants Stu Sheppard and Earl Saxon.


land use



unknown unknown



Sinar Mas pulpwood



APRIL pulpwood



not known oil palm



Surya Dumai oil palm



Sambu oil palm



Siak Raya oil palm



Wilmar oil palm



Astra oil palm



Hutani Sola Lestrai logging



Bumi Reksa Nusasejati oil palm



Duet Rijja oil palm



Rokan oil palm



Sime Darby oil palm



APRIL oil palm




You can see the full dataset that they used, in Excel format, at this UCS web location. As the table above shows, just over half of the fires are on land that can only be described as not-forest, not-peat, not-protected, and not-concession, because Indonesia has not released land tenure data since 2010. If current data, prepared for the Forest Moratorium maps, were in the public domain, then the relative roles of different business groups and government agencies would be clearer.

The excellent analysis done by WRI, in a matter of days, show just how much the ability of science to show what’s going on has advanced. Using data available on public websites, derived from satellite images and other kinds of remote sensing, analysts can quickly explain what’s happening based on evidence and not just opinion. Furthermore, other scientists can check and confirm their results, as has been so important with deforestation data from Brazil, gathered by the government agency INPE but independently analyzed by the NGO IMAZON. It’s what has long been fundamental to science – conclusions based on shared evidence and reviewed by one’s peers – but now it can be done in a matter of days.