Our analysis suggests that 12% of current VCM ARR projects occur in places where albedo entirely negates the mitigation benefit, and a quarter occur in places where albedo halves the mitigation benefit. Yet, the majority are concentrated in places where albedo changes are expected to be minimal, and 9% of projects occur in places where albedo would augment the mitigation benefit. While our analysis has uncertainties, it shows the importance of incorporating albedo in carbon credits from ARR, and we provide a tiered approach for doing so.
One principal source of uncertainty stems from our consideration of only CO2e and albedo, and not other biophysical factors. Changes to albedo are one of several potential biophysical outcomes of ARR activities that may impact a project’s net climate impact30,31, such as changes to cloud cover, evapotranspiration rates, vegetation moisture storage, surface roughness, and snow cover duration6,10,32,33. Not all of these mechanisms influence global climate in a way comparable to the global impacts of CO2e or albedo changes, with several instead altering only the local surface energy budget. These can involve complex interactions and feedback between surface conditions and the atmosphere, involving temperatures, humidity, soil moisture, and clouds, and with variability linked to ecoclimatic and biome settings34. For example, increases in local evapotranspiration-related cooling due to ARR are modest in drier biomes, and the heat can be retained within the regional system when water vapor condenses, negating the cooling effect20,25,34. This suggests that evapotranspiration may play only a small role in the climate forcing of ARR in drier areas, which are also some of the areas with the greatest albedo deductions. In wetter, more humid areas, where there is a relatively large transpiration cooling effect from increased tree cover, the risk of undercounting climate-positive effects from non-albedo biophysical effects is likely largest, though it remains unclear whether ARR would produce clouds that have a net cooling effect on the global climate system33,35,36.
Overall, there remains uncertainty around the global warming or cooling magnitude of these additional, local biophysical effects of ARR activities20,36,37,38 and more work is needed to include them quantitatively within VCM accounting. Furthermore, recent studies suggest that albedo change dominates the biophysical radiative forcing change from ARR activities at a variety of scales39. Although incorporating different factors could change the magnitude of our results (and potentially tip projects into or out of eligibility if they are right at the threshold), we expect the general patterns to hold even as science improves. Moreover, because the bulk of the non-carbon and non-albedo effects would augment the mitigation benefit, it is conservative for VCM accounting to not include these effects until uncertainties are more resolved5. In recognizing the value of these local biophysical benefits to people and ecosystems40, we join others in calling for such work to be prioritized by the scientific community8,25,36. This practice of VCM protocols providing conservative, default datasets (such as the Hasler data) upon which improvements and project-specific contexts can be proposed—such as deforestation risk maps41 or leakage accounting defaults42—can be a path to near-term implementation while creating opportunities for improved data where viable.
Other sources of uncertainties stem from projected carbon accumulation and assumptions around starting and ending land covers. We used publicly reported carbon credit projections; however, uncertainty around ex ante carbon credit projections are only partially addressed within existing ARR protocols. Upon credit issuance, protocols typically require reporting of sampling error but inconsistently require propagation of other error sources such as measurement error (e.g., measurements of stem diameter at different locations over time) or model error (e.g., selection of inappropriate or uncertain allometric models)43,44. Most projects do not report robust uncertainty around projected (ex-ante) estimates of credit production, so we could not directly quantify uncertainty associated with the projected carbon credits. The ex-ante estimates used here are also likely to be refined before verified credits are issued and will vary depending on the successes of the ARR projects.
In many instances, the carbon data may have greater uncertainty than the albedo data14,17. Albedo data benefits from being directly quantifiable from satellite data, whereas carbon sequestration and storage are difficult to estimate accurately7. However, there are several sources of uncertainty around the estimates of albedo change used here. The Hasler data was generated according to the following broad steps: first, generating global maps of “mostly likely” land cover transitions; second, comparing the CO2e of the albedo-driven radiative forcing to the radiative forcing from CO2e associated with maximum carbon storage possible for a given location17; and third, calculating the ratio of albedo- and carbon storage-CO2e changes (see additional details in the “Methods” of Hasler et al.). Our use of the albedo deduction/benefit layer thus assumes that the “most likely” transition per the Hasler data and the geographic extent per the Karnik et al. data45 used for project boundaries reflects the actual transition for each VCM project. However, different pre- and post-ARR land covers can alter the albedo change attributable to a project26, including different ways baseline or historical albedo are defined. Moreover, the 500-m resolution of the Hasler data may fail to capture on-the-ground variation in land cover. Project-specific albedo assessments at finer resolution and with project-specific baseline assumptions would produce more accurate results26,38. For example, a project could select the correct pre- and post-land cover transition from the Hasler data single transition layers and use further-refined estimates of projected carbon sequestration.
For the uses of albedo data proposed here, reliability is most critical if albedo accounting changes project eligibility or credit generation. Thus, we view Tier 1 as the most immediately implementable because albedo information is being used to steer projects towards places that are more likely to be climate positive, but does not alter eligibility or carbon credits. Incorporating Tier 1 considerations into project design and assessment is immediately available to and reproducible for project developers, reviewers, and purchasers pursuing the highest quality of ARR credits, regardless of whether protocols adopt formal considerations of albedo effects.
Tier 2 requires the next level of reliability in albedo information, especially for projects that occur along the threshold of ineligibility. In those locations, higher resolution data or more complete accounting for other biophysical factors could shift projects in or out of eligibility. However, Hasler and colleagues (2024) found that accounting for variation in the albedo data did not substantially shift albedo deductions/benefits14. They found that variation around different radiative kernels shifted albedo deduction/benefits by no more than ±15% in most places, with greater variability in boreal biomes. For only 9% of global land area did this variability shift pixels over or under a 50% albedo deduction threshold.
Notably, there are reasons why lands with substantive albedo deductions may be preferred. For example, they may be lower cost and/or come with important co-benefits beyond climate mitigation. A strict threshold would exclude these locations, whereas a less strict threshold would allow these locations to remain with a more complete estimate of their climate mitigation potential (this paper’s Tier 3). The albedo data available today makes Tier 2 reproducible and operational in the near-term with relatively simple edits to VCM protocols (e.g., adding albedo-based eligibility constraints).
Higher-resolution and/or more project-specific data on albedo effects may be most important for Tiers 3 and 4 because its inclusion would alter the carbon credits for many projects. As we note, reliability can be improved at the project level by using more precise land cover transitions and carbon estimates. Given the uncertainties noted around warming and cooling effects of other biophysical changes of ARR beyond albedo, these tiers would benefit from additional scientific and data development in coordination with policy iteration within the VCM, the result of which could be a more complete net biophysical accounting of ARR activities. Moreover, albedo benefits do not directly reduce ocean acidification in the way that atmospheric CO2 removal does, highlighting a distinction of climate change mitigation approaches that target holistic atmospheric restoration beyond just climate impacts46.
All tiers will require collaborative communication and education with market stakeholders who may not be familiar with biophysical factors, since climate mitigation activities historically have focused specifically on CO2e. Additionally, deeper tiers of ambition will benefit from and in some cases require additional research, datasets, tools, and policy to become feasible. Development of an operational Sentinel-2 and Landsat albedo product47,48,49,50 at 20- to 30-m resolution with global, seasonal coverage has the potential to enhance the Hasler data and provide albedo deduction/benefit information at spatial scales more directly aligned with ARR projects.
To be clear, we view these tiers of albedo accounting as progressively ambitious iterations of applied science within carbon markets. Participants in the carbon markets should be held accountable to the best available science, which requires continuous protocol improvement5. This rapid iteration is happening, for example, with the uptake of projects using VM0047 and other new protocols that incorporate dynamic baselining after academic studies demonstrated gaps in traditional baselining approaches51,52,53. Similarly, tools that calculate non-permanence risk have been updated in light of scientific discourse on the need to incorporate future climate change impacts54. And, while this paper was under review, a new ARR protocol by Isometric was released that follows a similar approach to our Tier 2, with a 100% threshold55, demonstrating potential for market uptake of albedo accounting.
In this vein, we acknowledge that adoption of higher tiers may be limited at this point, but we include them because they show a fruitful direction for future design and implementation, for which we encourage continued research and policy development. Notably, it appears that implementing Tier 2—making ineligible the portion of projects at greatest risk for especially large albedo deductions—may resolve much of the accounting gap, given the smaller credit reductions as the tiers advance (Table 1 and Fig. 4). Additionally, given that many of the locations with high albedo deductions occur in places where tree cover is inappropriate or ineffective (e.g., afforestation of savanna ecosystems56 or other arid sites57), albedo accounting via Tiers 1 and 2 can help direct planned projects to the most effective areas for ARR. Nevertheless, the holistic accounting unlocked by Tiers 3 and 4 will be important to prioritizing climate finance and accurately valuing NCS. Market labels such as ABACUS27 could also be used to differentiate projects that meet these tiers, potentially allowing them to access higher prices in the VCM.
Inclusion of albedo accounting within carbon markets will be most effective where protocols implement consistent, reproducible, standardized processes for its inclusion. The Hasler data was generated by an independent third-party, is publicly available and globally consistent, and was created with reproducible and transparent methods. It also comes in a format familiar to many project developers, and thus presents an opportunity for use by all VCM participants, including use cases beyond ARR projects. As with ARR, albedo and other biophysical factors are likely to be relevant to Improved Forest Management, Reduced Deforestation and Degradation, and biochar projects58,59. Research for all these cases is needed, including reliable, accurate assessments that are extended to practice with maps and tools. It will be important for these datasets to be dynamically updated, as climate change is likely to continue to influence albedo values, for example, through reduced snow cover14,26. When protocols use the best available science to improve their accounting, market confidence grows and climate finance is likely to reach the projects most effectively mitigating climate change5,19,60,61.
