Tucked on the edges of a biodiversity hotspot, the Tumring project in Cambodia is supposed to prevent a rainforest the size of Chicago from being chopped down.
Its supporters claim it has been doing exceptionally well. The Cambodian government hailed it as the “most successful” community-based forest conservation scheme on the carbon market and a climate solution.
Satellite images tell a different story. Tumring is experiencing dramatic deforestation, losing over 22% of trees in the project area since the scheme began. The Cambodian government does not account for this loss in official monitoring reports.
Nor is this an isolated case. In a joint investigation, Climate Home and Unearthed, Greenpeace UK’s investigative journalism unit, found similar discrepancies in two Brazilian projects, based on data from two different satellite monitoring platforms. Companies like Uber, ArcelorMittal and Marathon are still using credits from these three projects to offset their emissions – and there is nothing to stop them.
It raises serious questions for Verra, the largest standard setter in the voluntary carbon market, which oversees the projects.
Project owners disputed the findings, while Verra said it “is committed to refining and improving its methodologies based on the best available science and data”.
Mind the gap
By protecting trees the Tumring project generates carbon credits – or offsets – which are then used by polluters to compensate for their own emissions elsewhere. Texan oil firm Marathon is a major buyer, while the Cambodian and Korean governments, project partners, are planning to use a portion of the credits as part of their national net zero plans.
But the emissions avoided through the project are likely to be overstated given the deforestation rate appears to be higher than claimed. Project owners recorded just 3,450 hectares (ha) of forest loss in monitoring reports between 2015 and 2019, the most recent data submitted. Our analysis using the online tool Global Forest Watch showed forest loss was four times higher in that period, at 14,000 ha.
Climate Home and Unearthed looked at offsetting projects after a source raised concerns about apparent discrepancies between what project owners were declaring in their monitoring reports, and what could be seen through satellite images.
The team compared project filings with data developed by the University of Maryland and made available on the Global Forest Watch online platform. A second source of satellite data, Forobs, developed by the European Commission’s Joint Research Centre, was used to check the findings. This showed a similar trend.
Redd+ weaknesses
Verra is a major proponent of the UN-backed scheme Redd+, which stands for “reducing emissions from deforestation and forest degradation in developing countries”. It is designed to protect areas at risk of being deforested. Companies can buy carbon credits from these projects to discount their own emissions.
Critics have long raised concerns about weak quality control of this kind of project. An investigation published by The Guardian and Die Zeit earlier this year alleged more than 90% of Verra’s Redd+ projects were not driving emission reductions, largely because developers exaggerated the threat forests were facing. Verra disputed the findings.
Climate Home and Unearthed found that, in addition to inflated baselines, underreporting of forest loss throughout a project’s lifetime and light-touch regulation can lead to far too many credits being generated.
“The findings point out deep flaws in the forest carbon offset mechanism”, said Souparna Lahiri, a climate adviser for the Global Forest Coalition. The fact deforestation is increasing, instead of going down, “is deeply concerning” and “strengthens our conviction that the mechanism of offsetting cannot be fixed”, he added.
Self-reported deforestation
Each carbon credit represents a ton of CO2 kept from being released into the atmosphere by protecting trees. If a larger portion of forest is cleared than project developers claim, the volume of emissions they avoid will be overstated. When used by companies or governments to compensate for their emissions elsewhere, these credits would have a negative climate impact.
Verra says its role is to make sure that, when a company does invest in a carbon project, it has integrity and meaning, verified by the best standards and science. Monitoring reports are a crucial part of how progress is measured, since they disclose setbacks such as rising deforestation.
Monitoring reports are audited by third parties, then submitted publicly on a project’s page, alongside a host of other documents. In practice, they can be difficult for the public to understand and evaluate. There’s no standardised way to monitor projects.
The way the Cambodian government and its partners monitor deforestation in the Tumring area is opaque. They use national land cover data produced by Cambodia’s environment ministry that is not available publicly. It has a low tree cover threshold, meaning an area needs as little as 10% of trees to be counted as forested. To put it another way, you could chop down 90% of tree cover in a previously untouched section and still claim the forest was intact.
Cambodia has one of the highest deforestation rates in the world, according to Global Forest Watch. Photo: Un Yarat / US Embassy Phnom Penh
The Cambodian government has previously tried to discredit independent analysis showing that deforestation is higher in the country than state records.
Wildlife Works, which worked as a technical consultant for project validation and verification, said it “had no connection to the project” since completing the job and directed questions to the Cambodian government.
The Cambodian government did not respond to a request for comment. The Korean government told Climate Home and Unearthed that only credits from 2021 onwards would be used to offset national emissions.
Industry transparency
The Integrity Council for the Voluntary Carbon Market, an independent governance body for the industry, has called for greater transparency, urging offsetting projects to make all their information accessible to a “non-specialised audience” so a project’s climate impact can be better assessed.
Gilles Dufrasne, from the NGO Carbon Market Watch, said: “Current practice on the market simply isn’t up to standard and this lack of transparency needs to be plugged. More credible, and transparent, use of forest monitoring data is part of this.”
Sylvera, a carbon offsets analytics provider, noted in its 2022 State of Carbon report that the majority of the company’s D-rated projects, of which Tumring is one, “grossly under-reported the deforestation in the project area and have exceeded the baseline emissions”.
Samuel Gill, Sylvera co-founder and president, told Unearthed and Climate Home: “The technology to largely resolve issues like underreporting or overcrediting already exist and are being deployed.” He added: “These improvements take time to filter through the system and in the next few years we should see considerable uplift in project quality as a result.”
In theory, Verra already has various mechanisms to prevent worthless credits linked to deforestation from flooding the market and to punish project developers responsible for any irregularities.
Project owners are required to set aside in a “buffer pool”: a portion of credits that cannot be traded on the market. These act like an insurance policy: if trees meant to be protected end up being felled or burned in a fire, credits in the pool should be cancelled to ensure the integrity of the credits previously sold for offsetting purposes.
Additionally, complaints may trigger a project review and, if a developer is found to have issued too many credits, it can be sanctioned or made to pay a compensation.
But carbon market experts have doubts over the effectiveness of the system, saying the size and use case of buffer pools may be too limited. Only one project has ever had credits from the buffer pool cancelled, according to the Verra register.
Recurring problem
Over 17,000 kilometres away from Tumring, the Rio Preto-Jacundá Redd+ project is meant to achieve the same goal and protect an area of the Brazilian Amazon state of Rondonia.
The project has sold more than one million credits, with big name buyers including German utility Entega, Bank of Santander’s Brazilian arm, and Brazilian financial services giant Banco Bradesco.
From when it began in 2012 to 2020, the latest year available in monitoring reports, the project recorded 5,884 ha of loss, with a sharp increase from 2016. Global Forest Watch data shows it lost 8,200 ha of forest – 33% higher than the numbers declared by the project owner, Biofílica Ambipar.
The scheme’s “without project” scenario, to show what would happen under business as usual, predicted 9,922 ha of loss in the same period.
‘On watch’
Sylvera, an offsetting rating agency that independently checks and verifies projects using a combination of satellite imagery and machine learning, has placed the Rio Preto project “on watch”, after noting significant and increasing deforestation within the project area.
Biofílica Ambipar, which runs the Rio Preto scheme, said it “works continuously to monitor, identify and report any illegal activity to the Brazilian public environmental authorities”.
The company says it relies on the Prodes system to monitor forest loss in the area. Created by the National Institute for Space Research in 1988, Prodes is also used by the Brazilian government for its official annual deforestation reports.
“According to the Prodes system, the deforestation rates in the region are lower than those informed by Global Forest Watch, which is not as accurate in classifying deforestation,” Biofílica Ambipar said.
Prodes is used to detect large-scale changes in primary forest, but it can miss smaller changes. The system uses satellite images that only detect clearcut logging of more than 6.25 hectares – an area equivalent to nearly nine football pitches – missing smaller-scale forest loss. The University of Maryland data, made available through Global Forest Watch, captures losses as small as 0.1 hectares, while also picking up forest degradation.
Still selling credits
Another Biofílica project was abruptly cancelled last year after part of it was legally deforested by the landowner. But carbon credits generated by the scheme are still on the market.
The Maísa project covered over 25,000 hectares of forest in the state of Pará controlled by a family-owned agroindustrial company, which runs eucalyptus, Brazil nuts and açaí plantations.
When the project began in 2012, the firm agreed with Biofílica to protect the trees and invest in better forest management practices in exchange for a share of the profits from the sale of carbon credits.
Since then, polluters including steel giant ArcelorMittal have bought hundreds of thousands of its credits.
But starting from last year the landowner began clearing increasingly larger areas of the forest in what Biofílica says was a breach of their agreement.
The project developer decided to stop the project, but it is still listed on the Verra register and its credits continue to be used for offsetting purposes. Over 38,000 credits have been retired since the project was stopped by Biofilica – more than 4,000 of them purchased by Uber to compensate for the emissions spewed by its fleet of cars in Central and South America.
Uber said that it “only invests in projects certified, traceable, and auditable by Verra, the United Nations, Gold Standard, and Climate Action Reserve [other verifying bodies for offsetting schemes] after a thorough investigation”.
Lure of agribusiness
Biofílica told Unearthed and Climate Home that the company had made it a policy to stop selling credits from the Maísa project as soon as it became aware of the legal logging. It added that “the project is currently in the process of being terminated and audited in line with Verra procedures.”
Asked what would happen to old credits in the project that are still available on the market through third-party sellers, Biofílica’s spokesperson said: “It is important to highlight that the credits that are still being sold by traders and brokers refer to credits verified in previous years, when there was still no legal deforestation scenario in the area; that is, they were audited and verified credits.”
However, when trees are cut down, the carbon stored in them is released back into the atmosphere, no matter if they were originally protected, negating any potential climate benefit. Experts say good projects need to ensure the carbon they sequester or avoid will remain out of the atmosphere for at least 100 years.
When asked what happens to credits in projects that are cancelled, a Verra spokesperson said projects are required to deposit a percentage of their credits into buffer pools which can be drawn on if a portion of the forest is lost.
Maísa’s buffer pool contains 131,600 credits which have currently been placed on hold, meaning Verra still needs to decide their fate. That is only 20% of the total credits put on the market for offsetting purposes, most of which have already been used.
Biofílica spokesperson suggested that what happened with the Maísa project was a sign that Redd+ projects can struggle to compete with the economic opportunities offered by agricultural production in the Amazon.
They said: “Maísa shows the reality of the Amazon region and illustrates the difficulties that all actors interested in conservation face in making carbon projects financially viable.”