Reducing tropical deforestation is one of the most immediate and effective nature-based solutions for mitigating greenhouse gas emissions at scale. REDD+ slows deforestation through performance payments to developing countries for emissions reductions. REDD+ projects rely on counterfactual baseline scenarios to determine the amount of greenhouse gas emissions that would have occurred in a without-project scenario. These baselines are what projects are measured against and have recently been the subject of much criticism in the media for being inflated.
Figure 1. Villagers living within the dense forests of the Mai Ndombe region in the Democratic Republic of the Congo. Their lives and livelihoods are deeply intertwined with the surrounding forest, relying on it for shelter, sustenance, and cultural practices. Photo Credit: Filip C Agoo
The REDD+ sector is transitioning to new methods for calculating project baselines. Traditionally, a project would have a reference region with no REDD+ activities to serve as a basis for comparing deforestation. While intuitive, this approach has not necessarily been well connected to national REDD+ strategies or accounting. In the future, most projects will have “nested” baselines; this involves taking a reference level of emissions from an entire jurisdiction (e.g., country or state), and allocating portions to projects. There are different ways to allocated nested baselines, the most prominent of which is based on a project area’s estimated risk of future deforestation.
In our recent paper, we present an approach to map deforestation risk and to allocate baselines for REDD+ projects using risk levels. Named BAAR, the Baseline Allocation for Assessed Risk, we assign future deforestation risk to areas using historic deforestation data. For each jurisdiction, we use a dataset of at least 10 years of land classification data, which allows us to map areas of high deforestation risk. We split this dataset into two: one to develop the models and one to test the models. We develop a series of maps that assign high and low-risk areas using various multiples of the distance from past deforestation. We then test the predictive power of our maps by testing how much actual deforestation was captured in the predicted high-risk areas.
Once an optimal risk map is chosen, the jurisdictional reference level in t CO2e/year is allocated spatially. Governments or risk map developers can allocate their FREL to align with Nationally Determined Contributions (NDCs) to the Paris Agreement by determining which proportion of their FREL should be assigned to high-risk and low-risk areas. BAAR yields a minimum project efficacy, which states how much high-risk area must remain forest for a country to meet its NDC goal. By allocating the FREL using the BAAR approach, REDD+ projects are issued a risk-adjusted reference level, or baseline, decreasing the likelihood of inaccurate baselines that often plague simple area-based models.
We used the Democratic Republic of the Congo (DRC) as an example jurisdiction to demonstrate the BAAR approach. If the national DRC FREL (7.65 tCO2e/ha/year) were allocated simply based on the total accounting area of each REDD+ project, without considering risk, the allocation per year would average 1,533,694 tCO2e/year. BAAR initially results in a risk-based FREL allocation averaging 1,027,552 tCO2e/year (4.93 tCO2e/ha/year) for the projects, which results in an average of 1,050,528 tCO2e/year (5.84 tCO2e/ha/year) when weighted by carbon stock.
The paper is not intended to be prescriptive but serve as a showcase for how results from stress-testing can be publicized prior to implementation, given the massive implications for the Voluntary Carbon Market. This is particularly crucial for Indigenous Peoples and local communities (IPLCs) who rely on this finance. In the authors’ opinions, models need to be fit for purpose, with this fitness being considered as important as scientific accuracy. To our knowledge, this is the first attempt to comprehensively test a baseline setting model of this nature, and we hope to see more of such rigorous testing soon. By doing so, we aim to ensure that the methodologies we implement are both scientifically robust and practically applicable, fostering greater confidence and transparency in nested REDD+ projects.
Figure 2. A woman stands amidst the remnants of a once-dense forest, underscoring the urgent need for financial support directly to Indigenous Peoples and Local Communities (IPLCs) involved in REDD+ projects. This support is crucial to help them pursue sustainable alternatives to the destructive forest practices driven by necessity. Photo Credit: Filip C Agoo
Work has already begun on the next iteration of the BAAR approach. Notably, this entails taking advantage of advancements in remote sensing science and shifting from the use of activity (i.e., deforestation) data to carbon stock and change data. This will allow BAAR users to incorporate biomass change imagery, which will facilitate the inclusion of forest degradation into the approach, among other advantages.
REDD+ is an important mechanism for driving financing to local communities. We believe that using a risk-based mapping approach is imperative for determining accurate baselines that ensure financial flows to communities who need them to halt tropical deforestation.