Forest Carbon Math Gets a Reality Check: Smarter Baselines Cut Overstated Climate Claims

Forests have become one of the most celebrated weapons in the fight against climate change, but a new study suggests that the arithmetic behind forest carbon credits may be inflating the planet’s actual gains. Researchers led by Catherine Chamberlain of The Nature Conservancy, publishing in Frontiers in Forests and Global Change, tested five different carbon accounting approaches across forests in the eastern United States and found that the choice of method alone can swing the estimated climate benefit by more than fivefold. Their conclusion is striking: the most conservative methods—those that update baselines over time and blend multiple plausible futures—produced estimates that were, on average, 63 percent more cautious than the static, single-scenario models still widely used today.

The core problem is something accountants call the counterfactual. When a landowner promises to manage a forest more sustainably—to delay logging, for example, or to harvest less intensively—the climate benefit is defined as the difference between what the forest actually does and what it would have done without the intervention. But that ‘business-as-usual’ scenario can never be directly observed, because the intervention actually happened. Every baseline is therefore an educated guess, and as the new study shows, some guesses are far more generous than others.

The research team focused on improved forest management, or IFM, a suite of practices that boost carbon storage in working forests while keeping timber economies alive. IFM could deliver up to 28 percent of the land-based climate commitments that the United States and Canada have outlined for 2030. But the sector has been dogged by criticism that some carbon credit protocols systematically over-credit projects, particularly when baselines assume unrealistic harvesting that would never have occurred. In California’s forest offset program, earlier work documented over-crediting of this kind, and the Integrity Council for the Voluntary Carbon Market has since begun favoring dynamic baseline approaches, awarding its Core Carbon Principle label to improved forest management methodologies that update their expectations over time.

To untangle how baseline design shapes the numbers, the team homed in on four key decisions that every accounting protocol must make. Should carbon stocks be measured directly from field inventories or inferred purely from computer models? Should the baseline be fixed at the project’s start—a static, ex-ante projection—or updated dynamically as real conditions unfold? Should the counterfactual assume a single deterministic outcome, such as a guaranteed harvest, or blend together a distribution of plausible harvest scenarios? And should comparison forests be statistically matched to the project site, or simply drawn at random from an eligible pool?

To answer these questions, the researchers drew on the U.S. Forest Service’s Forest Inventory and Analysis program, a nationwide network of plots sampled at roughly one 0.07-hectare plot per 2,400 hectares. They assembled simulated project scenarios in four regions—the Northeast, Central Appalachians, Southern Appalachians, and the Northwoods of the Lake States—using Maple/Beech/Birch and Oak/Hickory forest types, the most widespread across the region. The ‘project’ scenario assumed an extended rotation with no harvest over 15 years, while the business-as-usual counterfactuals were built using the Forest Vegetation Simulator, the U.S. Forest Service’s widely used individual-tree growth model, alongside measured inventory data.

The five approaches spanned the full design space, from a modeled, static baseline with a 100 percent certain harvest, all the way to a measured, dynamic baseline that statistically matched each project plot to its ten most similar neighbors using Mahalanobis distances, weighting matches by similarity across site productivity, elevation, stand age, relative density, and distance to roads. Analyzed with Bayesian hierarchical models, the results were unambiguous. The static, single-scenario approach estimated mitigation at 3.32 megatonnes of CO2-equivalent per hectare per year on average—roughly 2.7 times higher than the dynamic, blended approaches, which converged on similar values regardless of whether they were modeled or measured.

Among the four design choices, the blended versus single-scenario decision mattered most, with blending reducing estimates by 1.68 megatonnes of CO2-equivalent per hectare per year on average. That finding has a clear biological logic. Timber harvests are rare events—annual harvest rates in eastern U.S. forests hover around 2 percent, and lower still on lands owned by non-industrial family owners. A baseline that assumes a particular hectare will certainly be logged within 15 years inflates the counterfactual emissions, and therefore inflates the credit. Blended baselines, which incorporate the actual distribution of harvest likelihoods (which ranged from 20 to 90 percent across the study) and removal intensities (5 to 62 percent), better reflect the messy reality of landowner behavior.

Dynamic baselines delivered a second, complementary advantage: they update harvest expectations as timber markets shift, mills close, pests spread, and climate change alters growth. Because project developers cannot know in advance what the baseline will be, they also cannot strategically enroll forests that would have outperformed the baseline anyway—a form of gaming known as adverse selection that has plagued static approaches. The study found no significant difference between modeled and measured baselines when both were dynamic and blended, suggesting the choice between them should depend on context: unpredictable family forest owners favor measured approaches, while industrial owners who respond consistently to market signals fit modeled ones well.

The stakes extend well beyond carbon markets. The researchers calculate that if Inflation Reduction Act investments in climate-smart forestry were measured using the most generous static baseline rather than dynamic, blended ones, mitigation could be overstated by 22.5 to 27.5 million tonnes of CO2-equivalent through 2030. Such a discrepancy could distort capital allocation, policy choices, and public confidence in forest-based natural climate solutions—precisely at a moment when global climate finance needs to grow fivefold, with forest-sector investments needing to rise by roughly $400 billion annually by midcentury to meet Paris-aligned targets of around 6 gigatonnes of CO2-equivalent.

The study also found that stand characteristics shape outcomes: denser stands, measured by relative overstory and understory density, showed the strongest positive effects on estimated mitigation, pointing to where interventions deliver the most additional carbon. The authors are careful to note their limits—the true counterfactual remains unobservable, so no approach can be proven most accurate, and growth models like the Forest Vegetation Simulator may lose accuracy under climate change unless regularly updated. Still, their message is clear: conservative, reproducible, blended and dynamic baselines reduce the risk of overestimation, curb gaming, and direct investment toward projects that genuinely deliver. As timber markets and climate conditions grow more volatile, the integrity of forest climate accounting may depend less on planting more trees than on fixing the math that counts them.

Subject of Research: Carbon accounting baseline methods for improved forest management in eastern US forests

Article Title: Dynamic baselines with blended BAU scenarios ensure conservative mitigation estimates for improved forest management accounting

Article References: Chamberlain, C. J., Riley, L., Busby, S. U., Russell, M., Austin, K. G., Campbell, R., Gunn, J. S., Johnson, T., Martin, D. M., Rifkin, B., Shoch, D., & Belair, E. P. (2026). Dynamic baselines with blended BAU scenarios ensure conservative mitigation estimates for improved forest management accounting. Frontiers in Forests and Global Change, 9, Article 1940074. https://doi.org/10.3389/ffgc.2026.1940074

Image Credits: AI Generated

DOI: 10.3389/ffgc.2026.1940074

Keywords: carbon accounting, improved forest management, dynamic baselines, carbon markets, natural climate solutions, forest carbon, counterfactual, Forest Vegetation Simulator, Forest Inventory and Analysis, voluntary carbon market, additionality, eastern United States forests