Landsat imagery and Trimble’s eCognition software prove invaluable to conservation efforts in Laos
The Nam Et-Phou Louey National Protected Area covers almost 600,000 hectares (about 2,300 square miles) of wild, mountainous terrain in the northern reaches of the Lao People’s Democratic Republic (Laos). The area is home to one of the last tiger populations in Indochina and one of the most important remaining populations of northern white-cheeked crested gibbons, a critically endangered primate found only in northern Vietnam, northern Laos and southern China.
One of the principal threats to wildlife in the protected area, along with illegal wildlife hunting, is habitat loss through deforestation and forest degradation. Increasing human population, shifting cultivation patterns and fires have resulted in forests being replaced by large patches of grasslands, bamboo, and other vegetation. These changes threaten local wildlife, including almost 40 other globally threatened mammal and bird species, making the protected area important for conservation.
To help protect this habitat, the Wildlife Conservation Society (WCS) has supported the local Nam Et-Phou Louey management unit since 2003 to implement activities to manage the area. This includes patrolling and law enforcement, community outreach and awareness, and biodiversity monitoring, but securing sufficient funds to conduct the range of activities needed has been a challenge. Laos is classified as a “least developed country,” so government investment in conservation is minimal. This leaves the protected area vulnerable; protecting it is largely dependent on international donors.
The local management unit and WCS joined forces with the Lao-German project called Climate Protection through Avoided Deforestation (CliPAD) to develop the protected area as a Reducing Emissions from Deforestation and Forest Degradation (REDD) project for the voluntary carbon markets. REDD works to establish economic incentives by making the forests worth more standing than if the land was cleared for other uses. Projects that reduce greenhouse gas emissions from deforestation generate carbon credits that can be sold and the revenue used to diversify and supplement the livelihoods of local communities.
Before the project could move forward, a feasibility study was needed to determine if the area was viable for generating carbon credits. WCS enlisted the help of Planet Action, a nonprofit initiative supporting the fight against climate change. The intent was to use Planet Action’s imagery and software to produce an in-depth assessment of the issues facing the Nam Et-Phou Louey region, including:
• A historical analysis of deforestation and degradation.
• The development of a model to predict future deforestation and degradation rates.
• An estimation of the project’s carbon credit-generating potential.
• An assessment of the overall feasibility of the project to access carbon markets.
WCS started by collecting Landsat imagery for the entire Nam Et-Phou Louey National Protected Area for four dates (2000, 2004, 2006, 2010), in order to identify where deforestation was occurring and at what rates. Through a Planet Action grant, WCS obtained Trimble eCognition, an image analysis software for geospatial applications. The software enables geospatial data to be integrated and analyzed in order to quantify features and detect changes over time, allowing for the extraction of georeferenced information. The software allows for automatic mapping, change detection and object recognition, delivering standardized, reproducible, image analysis.
eCognition was used to run segmentation and classification of images into different land classes and ultimately to determine forest and nonforest. WCS started with the year 2000 as a base map with full classification and then performed change analysis from one year to the next to identify areas of deforestation over three time periods.
“The tools provided by Planet Action were absolutely invaluable to quantifying the extent the area was affected by deforestation,” says Colin Moore, regional advisor on REDD and climate change on behalf of WCS in Laos. “Without Trimble eCognition, we would have had to do manual classification on 600,000 hectares, for four time periods, across an area that has very heterogeneous forests. It would have taken forever and may not have been possible within the timeframe of this project.”
The landscape, a region of mixed deciduous, dry evergreen and mountainous forests, is challenging due to its vast and rugged terrain. The Landsat imagery can be difficult to automatically interpret due to shade, heterogeneous forest types and seasonal conditions, which strongly affect image interpretation.
Trimble eCognition’s object-based image analysis technology helped overcome these challenges. Topographic maps coupled with satellite data were integrated into the software to build a customized rule set to assess the Nam Et-Phou Louey region. High-resolution Spot image data was used to conduct accuracy assessments of the final products and ensure it achieved at least an 80 percent mapping accuracy of forested versus nonforested land.
The image data was used to create detailed land cover and land use classification maps to help calculate carbon stocks and predict baseline rates of deforestation. This allowed for accurate predictions of the long-term emission reduction potential, providing a basis to assess the project’s ability to generate carbon finance.
Following the area assessment, it was discovered that deforestation rates within the area were lower than expected; in other words, the project would be financially unsustainable in the long run. So a different approach was needed.
In collaboration with CliPAD and the Lao government, WCS designed a new project encompassing the entire province rather than just the protected area. Trimble eCognition was used to track deforestation across this larger area, an expanse so vast it would not have been possible without software, the experts determined. With these results, WCS developed a fresh REDD approach that targeted drivers of forest loss on a broader level.
This project focuses on developing suitable pilot models for effective forest conservation at the provincial level, which could ultimately extend nationally. Continuous forest change monitoring will be integral, so Trimble eCognition software will remain an essential tool in this endeavor.