Lately, there have been a number of articles and discussions about how technology can solve the carbon credit quality problem. In particular, for forest carbon crediting, we often read about how improvements to and use of remote sensing technology “solves” the greenhouse gas (GHG) integrity issues with such projects.

Investors have poured over a billion dollars into companies that offer products that claim to provide assurance on the quality of forest carbon market claims through the use of new technologies. You have heard the buzzwords and acronyms: LiDAR, AI (artificial intelligence), ML (machine learning). A recent article by Reuters suggested that AI and new data, collected using new technologies, will improve the emission reduction claims from forest carbon projects – in particular, REDD projects.  

At Calyx Global, we have assessed over 200 forest carbon projects and, indeed, one key problem we have seen is that projects do not always measure changes in forest carbon well. However, the measurement of changes in forest cover and forest carbon stocks is not the key GHG integrity problem for REDD. 

The most critical issue facing avoided deforestation projects (i.e., REDD) is the assumed amount of forest loss without the project, i.e. the project baseline. The project baseline is a counterfactual, a presumption made by the project about “how much forest would be lost if our project did not implement these activities”. While forest carbon stocks and fluxes are measurable – and indeed something that new technologies will improve the accuracy of over time – the baseline is not measurable because it is a scenario that does not happen. Baseline values are derived from a model that is often opaque and can include assumptions that are dubious or unfounded. 

Risk maps are the best way to predict future deforestation, so how these maps are developed is critical to ensure the integrity of a REDD+ project's emission reduction claims. This requires knowledge of the land uses in the country, the context in which the project is located, the forest dynamics in the project region and how the project applied the selected carbon methodology. 

On one hand, project developers are best placed to develop such risk maps. On the other hand, they are incentivized to overstate the baseline because it allows them to generate more credits. Our analysis suggests this bias has occurred – sometimes to an extreme. We have found REDD baselines that are over 1000% overstated, and we assess that 30% of REDD baselines are over 300% inflated, 40% are at least 200% inflated, and 60% are at least 100% inflated. The “measurement error” that new technologies are tackling is rarely this large – one cannot state there is 200% more biomass in a forest because it’s totally implausible and also verifiable. 

Source: Calyx Global analysis of publicly available project data; See ‘Turning REDD into Green’

What is the answer? We believe that the most important way to build credibility for REDD is to ensure independence in baseline setting. Verra will likely contract with service providers to develop the risk maps that will be critical for setting baselines. Risk maps distill the relationships between factors that drive deforestation, such as access to roads or rivers, population density, terrain and proximity to deforested land. To be clear, this is different than the “activity data” (i.e. forest loss) that Verra has already contracted out to various providers. Project developers should be able to provide input into how risk mapping might be done, but ultimately an independent third party must be setting the baselines. This independent third party alone must decide how to predict where deforestation will occur and at what level. We encourage Verra to develop a clear “conflict of interest” policy that also considers whether a provider of such services is receiving, or has received within a certain timeframe, financial benefits from project developers who will be affected by the risk map.

This has little to do with technology and everything to do with ensuring interests are aligned with high integrity. Most of the problems in the carbon market today are related to this issue around conflicts of interest, not to poor technologies.

On the bright side, it is true that technological innovation can improve the quality of REDD credits. At Calyx Global, we use remote sensing to analyze historical emissions in and around the project area, as well as emissions during the crediting period. We do find inconsistencies between the project estimations and independent datasets, but not on the order of magnitude of baseline overestimations. Furthermore, a simple fix is for REDD methodologies to require uncertainty estimations for key parameters, including for the estimation of forest loss. The new consolidated REDD methodology about to emerge does this – and will go a long way to solving the measurement issue.

We applaud those who are finding ways for technology to improve the carbon market.  But technology alone cannot resolve some of the thorniest issues of GHG integrity. We believe that it is important to recognize the many other issues that will require other means to improve the quality of the market. Calyx Global is working hard to provide a rigorous, independent and comprehensive lens on all facets of integrity - allowing buyers and users of carbon credits to choose better credits. Learn more about our rating approach if you want to find out more.