How Local Restoration Champions in Africa Are Using Data to Drive Change

Originally published by Land and Carbon Lab

From Kenya’s drylands to the tropical forests of the Congo Basin, decades of logging, conversion to agriculture and other forms of ecosystem overexploitation have led to the degradation of about 20% of Africa’s land, a situation that threatens the food security and livelihoods of more than 70% of the continent’s people.

Fortunately, this trend is reversible — and governments, businesses and communities across Africa are ramping up efforts to restore thriving ecosystems and farmlands that can provide for generations to come.

To date, 34 African countries have collectively committed to restoring 100 million hectares of land by 2030 — an area larger than Tanzania. Thousands of local groups across the continent are working right now to turn these goals into reality. But how do we track progress of restoration on the ground, and how will we know if it is successful?

Traditionally, monitoring restoration progress has been labor-intensive and costly. But now, tracking restoration is easier than ever thanks to open-source, high-resolution data sets from Land & Carbon Lab. Across the continent, community groups are using and shaping these data sets to learn from challenges and successes — both of which are crucial to achieving restoration at scale.

Why Tracking Restoration Progress Matters

Tracking restoration projects — part of a process known as monitoring, reporting, and verification (MRV) — is essential to success. It allows those working on restoration projects to assess progress, replicate things that are working well and support local “restoration champions” with training and other resources. It also inspires funders to continue investing in projects where progress has been credibly measured.

In northern Malawi, the NGO Wells for Zoë is providing safe drinking water, educating girls and alleviating poverty. The organization first got involved in restoration when its leaders observed that as forests disappeared, girls were required to walk longer distances in search of firewood. “We started supplying tree planting clubs with a little bit of knowledge and a little bit of training and some seeds,” says Kevin Dalferth, Wells for Zoë’s chief technical officer. “But we were not really strict on follow up due to the costs of logistics, and then you quickly realize: it works, but it could be better, which ultimately motivated us to create a monitoring tool empowering planters to become part of the monitoring process.”

While satellite data has been used to monitor land cover change such as forest loss for a decade, in the past, geospatial data sets have not had the resolution necessary for tracking individual restoration projects.

“It’s not that hard to see forests disappearing from space,” says Emily Averna, associate director for land restoration at the Bezos Earth Fund. “But until recently, it’s been much harder to see tree restoration from space.” In other words, deforestation is relatively easy to capture by satellites. But when trees grow, this change happens at a much slower pace that has been harder for satellites to observe given their limited ability to capture small trees under a certain height.

Land & Carbon Lab combines satellite imagery with the power of AI and machine learning to provide three data sets that are changing that narrative and measure tree cover, tree canopy height and tree count.

The tropical tree cover data set allows anyone in the world to assess the percentage of tree cover in each 10-meter plot, even in cities, croplands and dry forests, where past satellite data sets miss most of the trees. This has important implications for monitoring restoration projects, which by necessity often take place outside of dense forests.

The canopy height data, produced with Meta, shows the height of individual trees at 1-meter resolution. This level of detail is unprecedented, and a breakthrough for tracking progress of tree-planting projects, which previously were monitored by human observation alone, either via drones or field visits. “This data dramatically brings down the cost and time burden of monitoring and verifying restoration efforts,” Averna says.

Building on that underlying foundational AI model, a third experimental data set can count trees within eight months after they are planted if high-resolution satellite imagery is available. This allows the funders and managers of restoration projects to independently verify project progress early in the project lifecycle — and correct mistakes fast.

How Communities are Improving Restoration with Local Knowledge

While comprehensive satellite data can tell us a lot about what’s happening on the ground, what it often can’t tell us is why. That’s where local communities and organizations come in: not only can they provide valuable input on what aspects of restoration initiatives are and aren’t working, they can also recommend which interventions may be needed to make them more effective and feed these lessons back into the creation and improvement of data sets.

Land & Carbon Lab has built, tested and improved these data sets by working directly with more than 200 project sites financed by TerraFund, a funding initiative aligned with WRI’s Restore Local program. It connects restoration projects with grants, low-interest loans and equity investments and with peer organizations undertaking similar work. The majority of those investments are concentrated in Restore Local’s three priority restoration landscapes in Africa: Kenya’s Greater Rift Valley; the Lake Kivu and Rusizi River Basin of Burundi, the Democratic Republic of the Congo and Rwanda; and the Ghana Cocoa Belt.

Wells for Zoë is among the groups that have helped improve how restoration data is collected. In Malawi, tree planting can play a big role in helping fight soil erosion, which is estimated to reduce agricultural productivity in Malawi by 6% every year. Wells for Zoë trains local residents to build catch dams in the deep gullies where erosion has cut through many family farms. These dams catch soil that would otherwise run down the hills during torrential rains and create ideal conditions for new trees to grow. As the trees mature, they hold the soil in place and reduce the chances of further erosion.

Wells for Zoë and other restoration champions supported by TerraFund face a common challenge: many of the local staff leading the work were unfamiliar with geospatial data collection. To improve the organization’s ability to monitor its tree-planting activities, Wells for Zoë developed an app called Flority, designed to help users with limited technical skills use GPS to define geospatial “polygons” that delineate each restoration site, as well as geotagged photos taken at each site. They then worked with WRI’s TerraMatch team to improve the app and reflect the data collection protocol. The app helps champions map out the exact parcels of land where restoration is taking place; Land & Carbon Lab data can then help set baselines for those parcels and eventually track progress against that baseline. TerraFund has trained all its projects to collect and quality assure that data — and champions have successfully enclosed more than 40,000 hectares of land under restoration in polygons.

Using the app has revolutionized how Wells for Zoë tracks its impact: since 2018 the organization has started growing more than 4.2 million indigenous seedlings in Malawi, many of which can now be seen growing on its maps. And because photos of tree-planting sites are often uploaded in near real-time, it’s easy to detect little mistakes — such as planting cuttings upside down or planting a seedling in a non-suitable area — and rectify them before they become bigger ones.

Why Reliable Data is Key to Restoration Funding

Carefully tracking restoration progress is critical to build trust among funders that their money is being well spent.

Averna has experienced this firsthand as the Bezos Earth Fund has used Land & Carbon Lab tools to assess its own impact. This includes TerraMatch, the repository for quality-assured data submitted from TerraFund projects. “Self-reported data is great, but it is not the same thing as verified geospatial data,” she says. “You really do need both, but it’s crazy how often those two things are conflated. By combining self-reported data with geospatial monitoring, Land & Carbon Lab allows us to see and manage what is happening in every single funded project site that we have. This transparency and accountability has given us the confidence to be able to invest multiple times in the TerraFund program. I would argue it’s also a big driver of how TerraFund was able to become so big and drive even more grant funding for these vital restoration efforts.”

Wells for Zoë has been among the recipients of this influx of funding. “By embracing monitoring technology to showcase their impact, they’ve raised a lot more money over the past three years,” says Will Anderson, the senior portfolio manager for TerraFund at WRI. “It’s allowed them to raise millions of dollars from the Priceless Planet Coalition to scale up their work across Malawi and receive interest from reputable investors in the carbon market.”

The Next Leap in Restoration Monitoring

Many of the TerraFund projects are in the early stages, and Land & Carbon Lab data represents a first phase of innovation in geospatial restoration monitoring. In the next phase, Land & Carbon Lab and collaborators will use the foundational model behind the canopy height data produced with Meta to verify tree planting by counting planting holes and detecting growing seedlings as young as eight months after they’re put in the ground. As trees are growing across TerraFund projects, the team will begin to review baseline tree cover and count to develop the first satellite verification protocol for restoration projects.

Challenges remain. TerraFund receives tens of thousands of precise polygons that range from 0.5 to hundreds of hectares in size. The team will need to cross-reference the satellite analysis with self-reported narratives to more accurately interpret data.

It is equally difficult to know precisely which trees are planted when on a site. This is because the majority of restoration projects occur in areas that already have trees, planting takes place twice a year during the rainy season, and dead saplings are replanted. Finding the exact moment to begin verifying 200 projects is not simple — but a standard and adaptable approach is nearing completion.

And verification by satellite is only possible for projects that start as an open canopy system and where highly accurate satellite imagery exists. In closed canopy forests, the team is deploying field collection consultants to gather verification results and compare their accuracy against satellite methodologies. As the team learns and continues to reduce the cost of verification without sacrificing quality, it will update its published methods in 2025.

Another application on the horizon for Land & Carbon Lab data is the ability to measure baseline carbon stocks and removals, which will eventually allow promising TerraFund restoration champions to estimate carbon at the project level for the first time and determine if there’s a potential to create a sustainable revenue stream for the community through carbon credits.

The stakes are high: scaling restoration across Africa will improve the lives and livelihoods of millions of people, help fight the climate crisis and reverse a decades-long decline in health of critical ecosystems. Reliably tracking progress and building trust among funders is crucial to get to the next step: channeling more support to the organizations and communities who are transforming their landscapes, seed by seed.