Democratic Republic of Congo satellite forest canopy cover, loss and swamp extent mapping,
demonstration pilot in support of Wildlife Works Carbon LLC REDD+ projects
Embedded Files
Project Summary
Project Summary
We will map contemporary forest cover and disturbance using Landsat, Sentinel-2, and airborne laser scanner (ALS) data across nearly 300,000 ha of humid tropical forest in Mai Ndombe Province in the Democratic Republic of the Congo (DRC). The results will be validated in the field and by comparison with high spatial resolution commercial satellite data. The project is funded by the US-based Wildlife Works Carbon (WWC) LLC that pursues the emerging marketplace for carbon offsets as a sustainable and scalable funding mechanism for biodiverse forest protection and community development.
WHY are we funded to do this ? Tropical forests account for about half the world’s terrestrial carbon with deforestation and degradation comprising about one fifth of global annual net carbon emissions from human activities. Tropical forests can grow back as ‘secondary’ forests after natural or human-induced disturbances that cause deforestation. Regrowing secondary forests absorb CO2 from the atmosphere, which partially compensates for the emissions produced by their destruction. Conserving secondary and recovering degraded forests, by allowing them to regrow, provides a pathway to help mitigate climate change, and may play a key role in biodiversity conservation. In last several decades international forest carbon mitigation and monitoring frameworks, including the Reducing Emissions from Deforestation and Degradation and enhancement of carbon stocks (REDD+) program, have been initiated to compensate developing countries for reducing carbon emissions from forested lands. Financial support for REDD+ projects requires measured, reported, and verified (MRV) estimates of forest above ground biomass to quantify avoided emissions achieved by forest conservation. One requirement for the MRV process is to reliably map the forest cover and forest cover change, which can only be undertaken at scale using satellite data.
Study Area
Study Area
This research is being undertaken in the largest REDD+ project in the DRC. It was established by Wildlife Works Carbon (WWC) LLC in 2011 with a 30-year contract with the DRC government in partnership with local communities to generate carbon offset revenue through the voluntary carbon market. Due to centuries of colonialism and exploitation, very poor transport infrastructure, and minimal economic support from the DRC government, the communities are highly marginalized. Approximately 50,000 people live in villages located along the shores of Lake Mai Ndombe and within the forest along tracks and unmetalled roads. The communities agreed, through the Free, Prior and Informed Consent (FPIC) process, to co-create strategies for improved food security, access to healthcare, and education, while maintaining their tradition of living sustainably with the forest. Notably, under the agreement, community members agreed that no timber harvesting will occur but that they can use the forest immediately around their villages, typically for slash-and-burn agriculture.
Please visit https://www.wildlifeworks.com/mai-ndombe for recent media concerning the REDD+ project.
Mai Ndombe REDD+ project area.
Satellite image shows BlackSky 1 m imagery acquired March 2023 over a village (named Kesenge in Lingala) and Lake Mai Ndombe.
Project Implementation
Project Implementation
Satellite and Airborne Remote Sensing
Satellite and Airborne Remote Sensing
Landsat 8 and 9 Operational Land Imager (OLI) 30 m and Sentinel-2 MultiSpectral Instrument (MSI) 20 m surface reflectance imagery acquired in the visible, near-infrared (NIR) and shortwave infrared (SWIR) will be used to map percent tree cover and forest cover loss. A machine learning classification approach will be used. Modern AI machine learning algorithms require large amounts of annotated training data. Rather than manually collect training data by visual assessment of satellite data, we will use canopy height maps defined in ~2 x 2 m grid cells derived from airborne laser scanner (ALS) data.
Landsat 8 and 9 Operational Land Imager (OLI) 30 m and Sentinel-2 MultiSpectral Instrument (MSI) 20 m surface reflectance imagery acquired in the visible, near-infrared (NIR) and shortwave infrared (SWIR) will be used to map percent tree cover and forest cover loss. A machine learning classification approach will be used. Modern AI machine learning algorithms require large amounts of annotated training data. Rather than manually collect training data by visual assessment of satellite data, we will use canopy height maps defined in ~2 x 2 m grid cells derived from airborne laser scanner (ALS) data.
Mai Ndombe REDD+ project area (within the white boundary) superimposed on a 30 m Landsat mosaic. The black rectangles show the locations of seven 2 x 10 km transects where a RIEGL VQ-1560 II-S discrete return LiDAR is flown to provide an average of 5 LiDAR pulses/m2.
Example airborne laser scanner (ALS) canopy height data flown across a single transect.
Fieldwork
Fieldwork
Fieldwork is challenging because of the inaccessibility of the forest interior, a lack of infrastructure, and two rainy seasons. We spent 3 weeks in February 2023 prototyping fieldwork assisted by the REDD+ project managers, forest officers, and local villagers. Future fieldwork is planned to validate the derived percent tree cover and forest cover loss maps, guided by interpretation of PlanetScope 3 m and BlackSky 1 m commercial satellite imagery.
Travelling across the lake from Inongo to Kesenge to undertake the fieldwork (Photo Roy)
Forest officer and Kashongwe on motorbike next to local assistant (Photo Roy)
Backs of Roy, Kashongwe, and forest officer walking along track within secondary forest (Photo Cho)
Walking through marshy area to reach the forest (Photo Cho)
Musanga cecropioides pioneer tree species, and cleared land covered by weeds, shrubs, and cassava in the foreground (Photo Kashongwe)
Millettia laurentii, local name: Wenge, well-known for its valuable hardwood that is "endangered" in the IUCN Red List (Photo Roy)
Taking a pirogue to cross flooded forest to the west of Kesenge (play with sound on) (Video Roy)
Page updated
Google Sites
Report abuse