Deforestation and Google Earth

Jun 14 2011
VERTIC Blog >> Environment

Mikael Shirazi, London

A recent article, making use of Google Earth imagery, on the environmental website Mongabay, has called attention to the damaged forests of Sarawak (a state in Malaysian Borneo), comparing their poor health with the apparently pristine forests just across the border in Indonesian and Bruneian territory. These publicly available images—as the authors point out—seem to cast into doubt the local authority’s claim that 70 per cent of the territory’s forest cover is intact. But how much does Google Earth contribute to independent civil society actors’ ability to monitor deforestation? Whilst powerfully illustrative of the effects that different industrial and regulatory policies can have on tropical woodland, these images are of limited use in accurately and systemically tracking changes in the forest over time . A new tool developed by Google, however, may open up the field of more sophisticated and precise deforestation monitoring, which has usually been the preserve of well-resourced official or commercial bodies, to much wider participation.

Sarawak’s damaged rainforest
The spidery formations of brown logging roads permeating the forest in Sarawak, set against the verdant uniformity of the landscape across the border, is striking. The Mongabay article suggests the images ‘seem to lend support to claims from environmentalists that Sarawak's forests have been heavily logged.’ Indeed, some groups have previously claimed that the area retains only 10 per cent of its primary forest cover. Former British Prime Minister Gordon Brown has called it ‘probably the biggest environmental crime of our times.’

The article was posted online a week after Sarawak's Chief Minister, Pehin Sri Abdul Taib Mahmud, made the statement claiming the territory has 70 per cent forest cover. In an interview conducted in response to the ‘exaggerated claims’ reported in the article that only 10 per cent the territory’s forest remained, Taib reaffirmed the 70 per cent figure and claimed that his estimates were the most reliable, since they are based on the calculations of ‘Sarawak’s own foresters, who are charged with replanting and managing Sarawak’s renewable resources in an environmentally sustainable way.’ He continued: ‘Part of the problem is that the activists are making these claims from far away, and are not actually on the ground.’ (This contentious issue last week entered a new phase with the opening of an investigation by the Malaysian Anti-Corruption Commission into Taib for corruption, including his and his family’s alleged ties to the logging industry.)

In such cases, where interested observers call into question official determinations on forest health, it seems there would be a use for objective ‘measurable, reportable and verifiable’ (MRV) analyses to be more widely available and help make sense of these competing claims.

Unfortunately, Google Earth images themselves are unable to shed much light on these conflicting forest cover estimates. Although logging roads are rendered in some detail, Mongabay makes no attempt to glean quantitative information from them – claiming only that they ‘lend support’ to the 10 per cent forest cover claim that some environmentalists have put forward. The relationship between the numbers and spread of logging roads visible from space and the level of deforestation is unclear. For example, it is suggested in the comments page of the article that further logging occurs across the border by light trail rather than road – a driver of deforestation which cannot be traced or measured via the Google Earth images.

The images are also stitched together from several satellite providers, such as GeoEye, TerraMetrics, Tele Atlas and Europa Technologies, making accurate comparisons difficult as they were taken over different time periods, with shifts in resolution quality and colour variations. They may be more useful as an alarm-bell and a tool for the purposes of advocacy rather than systematic monitoring. Historically, satellite monitoring of deforestation has been limited to states and organisations large enough to develop expensive equipment and technical expertise. And capacity-building in this area forms a key part of developing initiatives such as the UNFCCC’s REDD mechanism (Reducing Emissions from Deforestation and Forest Degradation). As the Sarawak example shows, Google Earth does not, by itself, provide actors with smaller resources a tool that allows for measurable conclusions to be drawn, and so does little to improve independent assessments of official figures - but a new programme from the company could change this.

Google Earth Outreach
Google’s charitable arm has recently launched a product that may be able to provide a more accurate and consistent means of ‘MRV-ing’ forest cover in Sarawak: the ‘Google Earth Engine’. Together with the Carnegie Institution for Science, Imazon (Instituto do Homem e Meio Ambiente da Amazônia) and the Gordon and Betty Moore Foundation, the application was developed by Google to enable online users to assess land cover change across the globe. It makes available Google’s extensive computing infrastructure to analyse 25 years of images from the US Landsat archives. Although these images provide as low as a 30-metre resolution (where each pixel represents 30 square metres on the ground), which limits their ability to track logging roads in the detail available from Google Earth’s commercial satellites, they do provide consistent and reliable representations of forest changes over sustained periods of time – essential for MRV purposes. The Earth Engine also makes use of innovative programming, including SAD (Sistema de Alerta de Deforestation), which has been generating deforestation maps and statistics of the entire Brazilian Amazon on a monthly basis since 2008, and CLASlite (Carnegie Landsat Analysis System–Lite), which processes many images and automatically extracts three different land cover classes (live vegetation, non-live vegetation, and bare substrate).

According to Google, ‘unthinkable tasks are now possible for the first time’, and it is dedicating 10 million CPU-hours a year in 2011 and 2012. It has also been working in partnership with the UN-REDD programme to test the application in Tanzania. The accuracy of the system remains to be seen (even established monitoring techniques can vary wildly in their estimates - see Joseph Burke’s contribution to the VERTIC site for more on this), but it may represent a considerable step-up in terms of the tools open to civil society observers that can provide reliable quantitative data on deforestation. In the case of Sarawak, this would be helpful in turning from illustrative but inconclusive images to solid evidence.

Last changed: Jun 15 2011 at 4:55 PM