Transparency, robustness, and consistency in aboveground forest carbon quantification methodologies used for tropical forest carbon projects: a review in Southeast Asia

Abstract

Forest carbon projects hold significant potential for mitigating greenhouse gas emissions. However, growing scrutiny has raised concerns about their climate integrity, particularly the gap between scientific knowledge and the practical implementation of carbon quantification methodologies. Southeast Asia, a rainforested tropical region, is a key focus for the development of forest carbon projects. This study critically reviewed the quantification methods and associated reporting of 69 forest carbon projects across Southeast Asia, guided by three essential and interrelated criteria: transparency, robustness, and consistency. The findings reveal limited disclosure in methodological reporting, the adoption of potentially unreliable quantification practices, and substantial variability due to the differing standards adopted by projects. These issues risk undermining the credibility of carbon credits and may hinder their alignment with national and international climate goals. By identifying key methodological gaps and proposing clear evaluation criteria, this study contributes to ongoing debates around forest carbon credit integrity and underscores the urgent need for more transparent, rigorous, and standardised carbon accounting practices within the sector.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13021-025-00352-x.

Introduction

Forests contribute roughly one-third of the greenhouse gas emission removals needed to limit global warming to 1.5 °C [1, 2]. Continued forest degradation and deforestation could, however, convert forest ecosystems from a significant carbon sink to a net source, exacerbating climate change despite their vital mitigating potential [3]. The conservation, restoration, and sustainable management of forests thus play a potentially important role in climate change mitigation strategies [4, 5]. With the aim of reducing net greenhouse gas emissions, forest carbon projects have emerged in various forms that differ according to the type of forest intervention, mode of finance, scale of implementation, and mechanism for carbon credit trading [6]. They have received widespread attention due to their potential to deliver climate and social-ecological benefits and have received substantial investments and funding through various results-based mechanisms, such as donor-funded programmes and compliance and voluntary carbon markets [7–9].

While we fully recognise that the effective implementation of forest carbon projects depends upon a broad array of ecological, social and technological factors [10], the focus of this study is intentionally narrowed to the technical and methodological aspects of carbon quantification, echoing recent discussions about the gap between practice and best science in the accounting of forest carbon projects [11–13]. Both scientific and ethnographic studies [14, 15] have problematised the lack of a rigorous scientific basis for forest carbon credits at the project level to ensure climate integrity [16]. The emergence of rating agencies for carbon projects also signals that buyers may face challenges in understanding the quality and integrity of forest carbon credits [17]. Ensuring the quantification methodology of forest carbon project functions properly in practice forms the basis for carbon credit accounting, and is thus critical for developing carbon market [18].

Early methodological concerns regarding the accuracy and integrity of carbon quantification were temporarily alleviated by technological advancements, particularly through the increased use and improved precision of remote sensing techniques [19, 20]. While methodological challenges such as additionality, leakage, and permanence, along with implementation issues related to monitoring, reporting, and verification (MRV), have been widely discussed, relatively little attention has been given to the credibility and scientific rigour of carbon quantification methodologies adopted by forest carbon projects [21]. However, concerns have re-emerged over the climate integrity of forest carbon projects, such as a lack of methodological transparency, systematic overestimates of avoided emissions, and the application of inconsistent quantification parameters across projects [14, 22, 23].

We contribute to the debate regarding the climate integrity of forest carbon credits by reviewing the implementation of the carbon quantification methodologies and associated reporting in 69 forest carbon projects in Southeast Asia (SEA), with interventions ranging from avoided deforestation, reforestation, and afforestation to improved forest management. Our study is premised on the suggested gap between scientific output and forest carbon accounting practice, which leads to obstacles in the implementation of forest-based, climate change mitigation-oriented projects [24, 25]. We approach our assessment of the state of forest carbon quantification by first establishing a framework comprising three core criteria: transparency, robustness, and consistency (Sect. "Conceptualised criteria") [26]. Subsequently, we identified six key methodological factors to examine the three criteria (Sect. "Six key factors identified for assessment"). Building on these foundations, we developed an analytical framework for assessing carbon quantification methodologies applied in forest carbon projects across SEA (Sect. "Analytical framework"). By focusing on these empirically observable criteria and factors rather than attempting to define “integrity” in its entirety, this study provides a science-based foundation that contributes to the broader integrity discussion.

Our review of forest carbon quantification methodologies is limited to aboveground biomass (AGB) estimation, due to the challenges of scaling-up systematic measurements for other carbon pools, such as belowground biomass, deadwood, litter, and soil organic carbon, in forest carbon projects [27]. To facilitate comparison across projects, we focus on carbon projects in terrestrial forests where AGB represents the largest and most directly measurable carbon pool [28]. Our geographical focus, SEA, constitutes nearly 15% of the global tropical forest inventory [29]. It is also characterised by high rates of forest loss, by a diversity of cultures, and by countries that span a broad range of economic development [30]. Considering that 43% of SEA’s land area is dominated by terrestrial forests that urgently need to be protected from deforestation and degradation [9] and that the operating costs in the region are relatively low [7], SEA has become a key focus of forest carbon projects aimed at climate change mitigation [31]. To date, various types of forest carbon projects of varying scales and involving different stakeholders have been established in the region, under either compliance markets linked to the United Nations Framework Convention on Climate Change (UNFCCC) or voluntary markets largely oriented towards private investors, yet their actual contributions to climate change mitigation remain unclear [14, 32].

Conceptual framing and methodology

Conceptualised criteria

The three criteria that underpin our ideal forest carbon quantification framework (transparency, robustness, and consistency) have been widely recognised as essential to ensuring the credibility of forest carbon projects, including by the international specification for project-level quantification, monitoring and reporting of greenhouse gas emission reductions [11, 12, 14, 16, 22, 26, 33, 34]. Transparency refers to the extent to which the assumptions and methodologies used for a project’s forest carbon quantification are sufficiently disclosed for intended users to make decisions [35]. Robustness refers to the reliability and accuracy of carbon quantification within a project, which is affected by the context-specificity of data and parameters [36–38], the appropriateness of methodologies [39–41], as well as the statistical rigour of uncertainty estimations [42–46]. Finally, consistency refers to the comparability of forest carbon quantification estimates across projects and methodologies [37]. If these three criteria are not met, vagaries and inconsistencies in the carbon accounting process can result in false emission reduction reporting and compromise confidence in forest carbon projects [47].

Rather than viewing each criterion in isolation, we adopt a holistic perspective that emphasises potential interconnections (Fig. 1). That is, transparency enables the assessment of robustness and consistency; robustness provides a clear foundation for transparent reporting; while consistency ensures comparability and standardisation across projects, further strengthening transparency and robustness. Given these interconnections, forest carbon projects need to address all three criteria comprehensively in their quantification.

Fig. 1.

Conceptual framing of interconnections between transparency, robustness, and consistency criteria in forest carbon quantification methodologies

Six key factors identified for assessment

To assess these three conceptual criteria, we identified six factors based on the methodological requirements of the Intergovernmental Panel on Climate Change (IPCC) guidelines [48, 49], major carbon standard methodology guidelines [50–52], and relevant literature [47, 53–55] (Table 1). The projects reviewed by our study incorporated a range of forest interventions, including avoided deforestation, afforestation/reforestation, and improved forest management. Despite this diversity, the six factors remain applicable across interventions, as they directly affect how AGB is quantified and represent fundamental elements of formal carbon accounting [26, 48], thus enabling meaningful methodological comparison.

Table 1.

Key factors for evaluating the quality of carbon quantification methodologies. The table defines each factor and includes how each factor should ideally Meet the conceptual criteria of transparency, robustness, and consistency

Factors examined	Definition	Transparency	Robustness	Consistency
Approach	The process of calculating carbon stocks and carbon stock changes for forest carbon projects [48]	Clearly documents applied approaches	The approach is scientifically rigorous	Follows IPCC guidelines or relevant standard protocols
Activity Data (AD)	Data on the magnitude of activities that result in emissions or removals over a given period [49]	Discloses data sources and classification methods in detail	Employs high-quality, context-specific data	Uses standardised data sources, classification schemes and forest definitions for cross-project comparability
Emission Factor (EF)	The coefficient that quantifies the emissions or removals per unit activity [48]	Provides clear details on measurement and associated model parameters	Derived from local, ground-truth measurements	Follow the standardised practice for EF measurement and estimation
Monitoring Frequency	The interval by which forest carbon is remeasured or recalculated [53]	Reports monitoring intervals and methods explicitly	Frequent monitoring for improved understanding of carbon stock changes	Standardises monitor practices for uniformity
Validation	The third-party audits that confirm compliance with methodologies and standards [47]	Makes validation reports publicly available	Confirms methodological rigour through independent review	Follows standardised validation protocols
Uncertainty	The statistical variability or potential error in estimating carbon stocks due to measurement inaccuracies, methodological limitations, and data constraints [55]	Reports detailed uncertainty metrics, sources of errors, and how errors are propagated	Uses rigorous error propagation to enhance scientific confidence	Applies standardised uncertainty metrics across projects

The first factor is the approach for forest carbon estimation. IPCC guidelines recommend the stock-change approach and gain-loss approach for forest carbon stock estimation [48]. Both approaches rely on Activity Data (AD) and Emission Factor (EF) values to calculate the changes in carbon stock over time. The gain-loss approach breaks down carbon changes into gains from carbon sequestration (e.g., tree growth) and losses from emissions (e.g., deforestation or degradation) for each type of activity. The stock-change approach directly measures the net difference in carbon stocks over a defined period, typically through inventory or remote sensing data, and is often used when detailed AD are limited [56]. Both approaches yield comparable results and are interchangeable if applied correctly [57]. The IPCC guidelines further specify both ex-ante (projected) and ex-post (actual) estimation for carbon quantification. Ex-ante estimations involve forward-looking projections about potential carbon sequestration or emissions reductions that are made before project implementation [58]. In contrast, ex-post estimations rely on actual data collected during or after project execution to measure realised changes in carbon stocks [59]. The two approaches serve complementary roles in the project lifecycle. Ex-ante approaches are essential for planning, financial modelling, and setting expectations, while ex-post approaches provide verification of actual outcomes. However, considering the highly dynamic ecological conditions of tropical forests, relying solely on ex-ante estimates without ex-post verification would likely compromise robustness and introduce bias in carbon accounting [60]. It is also important to note that while the IPCC guidelines provide widely accepted scientific principles for forest carbon quantification, they do not serve as the governing framework for individual projects in the voluntary carbon market. Instead, projects typically follow methodologies prescribed by their respective certification standards. Nevertheless, most of these standards acknowledge and incorporate the principles outlined by the IPCC [50–52].

AD, the second factor, typically refers to information about changes in forest cover, such as deforestation rates or land-use transitions, used to calculate emissions or removals. Biased AD could lead to incorrect allocation of forest cover changes due to human activities, affecting the robustness of the overall methodology. Ideally, there should be consistency in AD across different projects while following national-level carbon accounting frameworks, particularly in terms of land cover types and forest definition. AD differs from a project’s baseline, which represents the expected outcome in the absence of the project and thus the basis for determining additionality, although this is not the focus of the current review [61].

As the third factor, EF values are coefficients that quantify the emissions or removals per unit activity. Each EF is often based on a sample of measurement data, averaged to develop a representative rate of emission for a given activity level under a particular set of operating conditions [48]. For forest-related emissions and removals, EF is generally derived from AGB, usually the most significant carbon pool in terrestrial forest ecosystems and also relatively straightforward to measure [62]. AGB can be estimated by applying allometric models to measurements of the diameter at breast height (DBH) and tree height [63, 64]. EF must be derived from timely ground-truth measurements or locally validated studies to ensure robustness [65]. Projects relying on generalised or outdated EF values risk inaccuracies in their quantification [66]. Consistency and transparency are achieved when protocols for EF determination are comparable across projects, and when the data sources and methodological details of EF are clearly documented in standardised formats [34].

Monitoring frequency refers to the time interval over which forest carbon is remeasured or recalculated [53]. Monitoring ensures that changes in carbon stocks are systematically tracked over time. For robustness, more frequent monitoring can reduce uncertainty by capturing dynamic ecological changes, particularly in tropical forests where rapid but low-intensity disturbances are common [67]. Current major carbon standards typically require monitoring intervals of up to five years, striking a balance between methodological rigour and practical feasibility [51]. However, recent research suggests that longer intervals may limit the detection of subtle but cumulative carbon emissions [68–70]. Enhancing monitoring frequency thus presents an opportunity to improve the robustness of carbon quantification. Furthermore, standardised monitoring protocols that are adapted to project-specific contexts and justified by scientific evidence can facilitate comparability across projects, while the transparent reporting of monitoring schedules enables independent verification.

Validation refers to independent third-party audits that confirm compliance with project methodologies and established standards [47]. Validation enhances transparency by providing verifiable evidence of methodological adherence and contributes to robustness by ensuring that project methodologies are scientifically sound. Consistency is promoted when standardised validation protocols are applied across projects [71].

The sixth factor, uncertainty, ensures the accuracy of carbon quantification [55, 72]. There are many potential sources of uncertainties in forest carbon quantification, including but not limited to methodological errors, human and equipment inaccuracies, and the application of unsuitable allometric models for converting field measurements into AGB [73, 74]. IPCC guidelines suggest that the uncertainty for carbon estimates should be statistically rigorous, complying with accepted statistical theories and practices, and reported as inferences in the form of confidence intervals [48]. A comprehensive assessment and discussion about potential sources of error (e.g. measurement errors or sampling errors) or the statistical properties of errors (e.g. error distribution, bias, precision) in forest carbon quantification are beyond the scope of this study. Instead, we examined how projects addressed uncertainty in their forest carbon quantification process, including the reported uncertainty level (transparency), methods of estimation (robustness), and standardisation of reporting formats (consistency).

Analytical framework

Having conceptualised the three criteria and defined the six key factors, we further developed an analytical framework for assessing carbon quantification methodologies applied in forest carbon projects in SEA (Fig. 2).

Fig. 2.

Analytical framework for assessing carbon quantification methodologies used in forest carbon projects in SEA

We first compiled a project inventory list from public databases, literature, and major voluntary carbon project registries, including all registered and listed projects as of January 2025 (Table 2) for subsequent review and analysis. Notably, no forest carbon projects from Thailand were listed in the databases relied upon. This is because all project documentation in the Thailand Voluntary Emission Reduction Program (T-VER), Thailand’s domestic registry developed exclusively for carbon sequestration initiatives within the jurisdiction, is available exclusively in Thai [75]. Due to the language barrier, we were unable to assess the project documentation registered on T-VER.

Table 2.

Databases, literature, and registries used in project compilation

Database, literature or registry with project information
Asian Forest Cooperation Organization (AFoCO) REDD + Fact Sheet series [76]
REDD + Database - Institute for Global Environmental Strategies (IGES) [77]
An open-access database of nature-based carbon offset project boundaries [78]
Clean Development Mechanism (CDM) registry [79]
Verra registries, including both Verified Carbon Standard (VCS) and Climate, Community & Biodiversity Standards (CCB) [80]
Plan Vivo registry [81]
The International Database on REDD + Projects and Programs: Linking Economics, Carbon and Communities (ID-RECCO) [82]

These projects were then filtered based on three conditions: (a) they focus on terrestrial forest ecosystems (i.e. peatland swamp forests or mangroves were not included for this study, since the methodologies applied for these ecosystems are substantially different from terrestrial forests), (b) they are designed with the aim of reducing emissions or sequestering atmospheric carbon, hence necessitating forest carbon stock assessments (i.e., projects solely focused on social or biodiversity outcomes only were excluded), and (c) their status in relevant registries is neither inactive (denied, decertified, withdrawn), nor under development without complete methodological documentation requirements. After applying these conditions, a total of 69 projects were retained (Fig. 3A). Their distribution highlights inter-country differences in the development of forest carbon projects in the region; Indonesia hosts 32 projects, followed by Viet Nam and Lao PDR with ten and nine projects, respectively. The Philippines and Cambodia each have six projects, while Malaysia hosts four projects. Myanmar and Timor Leste each have only one project.

Fig. 3.

(A) Number of reviewed forest carbon projects by country in SEA; (B) Temporal distribution of project starting years and their operational status.

Although we explicitly excluded projects still under development, some reviewed projects remained in the validation or registration stages, thus not fully implemented. Approximately 13% of projects registered under voluntary carbon standards were at these developed but not implemented stages. Because project reporting can differ according to operational status, we categorised the reviewed projects into three groups (Fig. 3B) comprising “Implemented” (n = 42), “Developed but not implemented” (n = 6) and “Cannot verify” (n = 21). Projects that are officially registered or have demonstrated evidence of operational activity, including both completed and ongoing projects, were categorised as “Implemented”. Donor-funded projects not registered in carbon markets were similarly grouped following verification of their operational status through consulting relevant publicly-available documents. Projects listed in the databases as undergoing validation or in the process of registration were categorised as “Developed but not implemented”. These projects are still required to submit full methodological documents, but validation reports might not yet be available as they are not yet fully implemented [83, 84]. The third and final category, “Cannot verify”, comprised developed projects for which we could not find sufficient evidence to clearly confirm whether or not they have been or are being implemented.

Our review and analysis are based on the following procedures. First, we extracted all available information on quantification methodologies and reporting for each project, based on methodological descriptions corresponding to the six identified factors, from all relevant documents available in English within the public domain (Supplementary materials). To evaluate transparency, we recorded only the presence or absence of each factor, without describing how it was disclosed or the level of disclosure. This is because of an absence of uniform reporting requirements among certification standards. However, we provide an overall measure of transparency for each project using three categories: “No methodological information available,” “Partial factors included,” and “All factors included.”

Due to limitations in transparency and the lack of independent data sources for recalculating carbon stock changes, we did not apply quantitative metrics to assess robustness, consistency, or interrelationships between criteria at the project level. Instead, to assess robustness, we classified methodological choices used in each project across six key factors (as detailed in Results) based on information extracted from project documentation. The classification reflects commonly recognised categories in forest carbon accounting practices. To assess consistency, we examined how the six methodological factors aligned with relevant national or international standards and evaluated their comparability across projects. We focused on identifying methodological discrepancies arising from varied data sources, monitoring intervals, uncertainty quantification methods, and other technical dimensions. Our objective was not to score individual project performance or assess whether the project is compliant with the standards, but to identify methodological patterns and highlight gaps between current practices and best-available scientific approaches.

We also mapped the approximate locations of the projects that we reviewed (Fig. 4), based on the best-available locational information, notwithstanding limitations because not all projects provide explicit locational information.

Fig. 4.

The approximate locations of the 69 forest carbon projects in SEA as of January 2025 that were included in the review. The accuracy of the map and meaningful spatial analysis were limited because some project reports do not provide explicit locational information. Refer to the supplementary materials for details of the 69 projects

Results and discussion

Transparency

35% (n = 24) of the reviewed projects lacked any publicly available information about the methodology used for carbon quantification, despite being officially established and having their funders and proponents documented. Most of these projects are in the “Cannot verify” category. Only around 25% (n = 18) of the reviewed projects disclose information relevant to all six factors examined, while the remaining 40% of the reviewed projects are missing at least one of the factors related to forest carbon quantification (Fig. 5). Notably, for projects classified as “Developed but not implemented”, we assumed validation to be fulfilled regardless of whether validation information was publicly available, as these projects are formally listed in voluntary carbon standards’ registries, where validation is a procedural requirement. The remaining methodological factors for projects categorised in this group were, however, assessed using the same criteria as applied to all other projects.

Fig. 5.

Map illustrating the availability of methodological information for the 69 forest carbon projects in SEA that we reviewed. The color-coded markers represent the level of detail provided in project documentation: purple markers indicate projects with no methodological information available, orange markers denote projects with partially reported methodological factors, and green markers represent projects where comprehensive methodological details have been reported

The number of projects with no methodological information available has declined over time (Fig. 6), with only two such cases identified within the past ten years. However, there is no clear evidence to suggest that recently implemented projects are more likely to report all relevant factors. Even among projects established within the past five years, a substantial portion still omits at least one factor from their methodological reporting.

Fig. 6.

Temporal distribution of methodological information availability in forest carbon projects from 2004 to 2025

Uncertainty reporting is the most frequently omitted factor among the six factors, with 68% of projects failing to report levels of uncertainty (i.e. confidence interval) (Fig. 7A). Independent validations and monitoring frequency were each missing in approximately 57% and 52% of the projects, respectively. AD, EF, and the approach used for carbon quantification were less frequently omitted (missing in ~ 42–45% of projects), with most projects applying methodologies for robust quantification based on peer-reviewed literature or IPCC guidelines.

Fig. 7.

(A) Number of projects missing information for each factor, categorised by project status: “Implemented” (green), “Developed but not implemented” (yellow), and “Cannot verify” (purple). AD = activity data, EF = emission factor; (B) Proportion of absence within each project category for each factor, highlighting the relative frequency of missing information within “Implemented”, “Developed but not implemented”, and “Cannot verify” projects

This pattern of omission varies notably by project status (Fig. 7B). Projects categorised as “Cannot verify” (~ 30% of the reviewed projects) are mostly those that are funded by external donors or government initiatives and are not registered under any voluntary carbon standards. We highlight these cases to draw attention to the challenges associated with extremely limited public documentation in such projects, and also to underscore how voluntary carbon standards have improved transparency through requiring that registered projects disclose project documentation to the fullest extent [83, 84]. Despite this, the “developed but not implemented” projects, which are all registered under a voluntary carbon standard, still exhibit substantial omissions (e.g., 83% did not report their uncertainty). In contrast, “Implemented” projects tend to display higher levels of transparency, yet over 40% of these projects do not disclose validation reports and half of them lack clear documentation of uncertainty. These results suggest that while methodological gaps are most acute among the “Cannot verify” group, significant shortcomings are also present in both the “Developed but not implemented” and “Implemented” groups.

Furthermore, even when projects disclose information that is relevant to the six key factors, it may not be sufficiently detailed to support the criteria of robustness and consistency. For example, projects may publish the EF values they applied, but not all projects clearly identify the method or source from which they derive such values. The same problem occurs with uncertainty reporting, where information on sources of errors and error propagation methods may be missing. In addition, we did not find any project that openly provided the reference data or attachments necessary for verifying and replicating the carbon quantification of the project in a workable format, such as remote sensing classification data in a raster format or plot-based data from field surveys. Similarly, accurate locational information on the boundaries of forest carbon projects is largely unavailable, making further investigation of scaling effects or spatial correlations impossible. Existing research has also shown that the geospatial data files provided by some forest carbon projects may be either damaged or include boundaries that are inconsistent with those of the actual project area [22].

Robustness

Regardless of the limited disclosure in some projects, we reviewed all available information on the six key factors across the full set of projects to evaluate the robustness of their forest carbon stock quantification. We attempted to provide relatively straightforward indicators for each factor that may affect robustness based on a categorical summary (Table 3) with the aim of highlighting where existing practices diverge from best-available scientific principles. We acknowledge that some of these projects or associated methodologies were developed during the early stages of carbon trading mechanisms, when rigorous methodological guidance was still emerging. As such, while these projects should update their methodologies in line with evolving scientific and technical standards, some use of less robust methodological choices is understandable [85].

Table 3.

Classification of methodological approaches used for forest carbon quantification across six key factors in the reviewed projects (n = 69)

Factor	Methodological choices for forest carbon quantification
Approach	Gain-Loss	Stock-Change	Other approach*
	25	15	0
	Ex-post		Ex-ante*
	34		6
AD	Field surveys, potentially supplemented with satellite data	Land use change maps without in-situ validation*	Global level data product*	Pre-defined ex-ante AD*
	18	9	4	6
EF	Ground measurements	National Forest Inventory*	IPCC default factor*	Other sources*
	20	7	3	8
Monitoring	Annual		More than 2 years*
	23		10
Validation	Yes		No*
	30		39
Uncertainty	In standard format		Not in standard format*
	14		8

Methodological choices were extracted from publicly available project documentation and grouped based on commonly adopted practices in forest carbon accounting. Choices marked with an asterisk (*) are considered to have potential limitations in robustness, as identified in IPCC guidelines and related literature

All projects that we examined adopted either the gain-loss approach or the stock-change approach. For those specific methodologies developed by certification bodies, the calculation of emissions/removals still follows the IPCC’s recommended guidelines, and generally uses AD and EF for ex-post estimation. Nevertheless, we found five projects still solely relied on ex-ante estimation for carbon quantification, including two “Implemented” and three “Developed but not implemented” projects established recently.

Our findings show that the AD used for quantification might not always be appropriate. Small-scale projects can rely on field surveys for measuring both AD and EF. When scaling up, projects often use satellite images for land use classification or forest stratification. A matrix of land use changes can then be derived from satellite images of different monitoring periods to determine the AD on a larger scale. However, remote sensing-based maps are not error-free and should be supported by ground validations for quality control purposes. We found approximately 18% (n = 7) of projects failed to incorporate ground observations. Assessing forest areal change directly from remotely sensed data may generate results significantly different from estimates that include additional data from ground-based forest surveys [40]. We also found that around 10% of projects (n = 4) use global-level data products, such as the Global Forest Change maps [86], with no additional evaluation or validation. Such global-level datasets may not be appropriate as AD sources for small to moderate-scale forest carbon projects, as they could underestimate forest loss or have classification errors at the local level [87]. These issues can be particularly prominent in tropical forests [20]. The six projects using the ex-ante approach to quantify forest carbon emissions or removals apply pre-defined ex-ante AD. Such ADs also hinder robustness, because the quantification is made without knowledge of the actual forest changes that have occurred [88].

Ensuring the robustness of EF may also be challenging for some forest carbon projects in SEA because not all forests have the same potential for reducing emissions from deforestation and forest degradation, especially highly complex and diverse tropical forests [89, 90]. An accurate and locally representative EF is essential for quantifying forest carbon. Despite this, we found that 47% (n = 18) of projects among those that provided information on EF sources have no independent EF analysis and use data from other studies or projects, national-level forest inventory, or IPCC default EF data. For those projects that conduct ground surveys to determine their EF, the common approach is to measure field data within forested areas in a set of sampling plots, that is, pre-defined areas selected according to a systematic or stratified random sampling design [91]. The measured forest attributes at the plot level are then used with an allometric model to estimate aboveground biomass [63]. While the field-based measurements and allometric models are widely adopted for EF estimation, many projects lack sufficient methodological details in their public disclosures, such as sampling design, plot size, or model selection. As such, we were unable to further assess the robustness of EF estimates from the ground surveys across projects.

Regarding the frequency of monitoring, 23 of the reviewed projects monitor their forests on an annual basis. Other projects that report their monitoring scheme adopt a monitoring frequency that ranges from bi-annually to every five years. Intervals between resurveys are generally compliant with the flexibility permitted under major carbon standards, which tailor monitoring requirements based on project type, anticipated rate of change, and logistical feasibility. However, recent research shows the length of time between each survey may not be sufficient to detect the subtle changes in forest structure, canopy cover or biomass from less-intensive forest loss or moderate degradation in the tropics [70]. Therefore, relevant methodological guidelines may need to recognise the importance of increasing monitoring frequency in tropical regions like SEA for more robust quantification. Furthermore, we found that repeat surveys may only comprise the monitoring of AD, while the remeasurement of EF is often ignored. For example, while emissions/removals are calculated annually based on the annual deforestation rate, AGB density for unit forested areas is not remeasured in some projects.

Although independent validation is a crucial component of carbon accounting to ensure robustness, only 43% of the projects (n = 30) showed evidence that they were or will be validated by third-party organisations. Such omissions raise concerns over both transparency and robustness.

Our review accords with the existing literature [42, 72, 92–94] in that uncertainty quantification in forest carbon projects remains limited in scope and methodological rigour. Around 36% of the projects failed to follow the standard format of reporting (e.g. without providing confidence intervals (CIs) or the uncertainty levels were not explained in detail). In many cases, even though the projects provide detailed methodologies for uncertainty quantification, the estimates focus primarily on sampling variability. Other key sources of uncertainty in carbon stock estimation, including measurement errors, model-related errors, and residual variability, are often not addressed [73, 94]. In addition, uncertainties related to AD are frequently omitted [36, 95]. This observation likely reflects a broader limitation of the existing methodology guidelines, namely the lack of comprehensive and systematic estimates of uncertainty propagation from all relevant sources. Consequently, the reported carbon stock changes and associated credibility run the risk of being overestimated [96, 97]. Advanced methodologies based, for example, on Bayesian or comprehensive error propagation methods, offer tools to address these gaps [98–100]. However, these methods are rarely applied in practice, suggesting a need for future methodological revisions to incorporate more robust uncertainty accounting into standard project protocols.

Consistency

Although all projects consistently applied IPCC-recommended guidelines, significant variation exists in other aspects of forest carbon quantification. Our comparison of the six key factors in all reviewed projects with national-level carbon accounting practices, as defined by the Forest Reference Emission Levels submitted by each country to the UNFCCC [101], revealed that 63 out of 69 projects exhibit at least one parameter that deviates from their respective national carbon accounting framework. While voluntary carbon projects are not formally required to align with national accounting systems, establishing more consistent and standardised quantification frameworks could help facilitate the integration of project-level mitigation outcomes into national strategies. The observed inconsistencies should not be viewed as methodological weaknesses, however, but a reflection of the diversity of standards and limited harmonisation across project types and contexts. Such methodological variability can be shaped by knowledge evolution, policy, data availability, or differing levels of technical capacity [16, 22].

For AD, given that a standardised framework for quantification is absent, projects adopted different methods and data sources tailored to their specific contexts. Such flexibility is permitted under most major carbon standards, which allow projects to choose context-appropriate methods provided they are properly justified and documented. When implemented rigorously, these approaches can be robust and locally relevant. However, the inconsistencies among ADs are notable, even among those developed from similar data sources. For example, projects relying on satellite imagery exhibit considerable differences in resolution, classification methods, and definitions of land-use schemes. Such inconsistencies can lead to variability in how land-use changes are identified and quantified, thereby affecting the accuracy and comparability of emissions and removal estimates. For instance, differences in the resolution of remotely sensed data may result in varied detection thresholds for deforestation or degradation, while inconsistent classification methods could produce conflicting interpretations of land-use transitions [102]. Furthermore, inconsistencies in the selection and application of land-use categories and definitions can create ambiguities and conflicts in aligning the emission removal efforts across projects or with broader frameworks [103, 104]. For example, the Food and Agriculture Organization (FAO) defines ‘forest’ as “land with tree crown cover (or equivalent stocking level) of more than 10% and area of more than 0.5 ha” [105], while some countries apply a threshold of 20%, reflecting local ecosystem conditions. Without standardised guidelines for adopting appropriate land-use definitions, projects may arbitrarily choose definitions that best suit their objectives, thereby hindering consistency [32, 106, 107]. Although flexibility in methodological choice is important in order to accommodate local conditions, greater standardisation or harmonisation would facilitate more consistent tracking of forest carbon outcomes, enable comparability between projects and potentially reduce estimation errors and ambiguities [61, 108].

The sources of EF are highly varied. Although EF should ideally be derived from in-situ measurements, a considerable number of projects still rely on imported EF values. For example, projects may adopt EF directly from national forest inventories or IPCC default values. While these practices may seemingly be consistent with national or international frameworks, they do not strictly adhere to the overall principles of forest carbon quantification, given that generic or large-scale EF developed for broader assessments can introduce substantial biases when applied at the scale of an individual project. Therefore, unlike AD, achieving consistency in EF determination strongly depends on maximising the use of locally measured data, given the inherent spatial variability of forest biomass.

Even among projects that conduct local measurements for the EF, the allometric models used in biomass estimation also differ between projects. All but one project reviewed here relied on the application of externally-sourced equations to their measured data. Such flexibility in the use of established models is explicitly allowed by certification standards, provided their use at a project site can be justified [50]. However, there is no standardised approach for selecting these models, and many projects did not provide clear documentation on how specific models were selected or validated for local applicability. As shown in Table 4, the allometric models used were developed at different times, for different regions, and at different spatial scales. This lack of consistency in model selection introduces substantial variability in biomass estimates, even among projects operating in similar forest types and conditions [109, 110]. While not inherently problematic, the lack of standardised guidance or transparent justification for model selection can limit comparability [64].

Table 4.

Summary of externally-sourced allometric models used in forest carbon projects reviewed in this study, categorised by their geographic focus and the number of projects adopting each model

Allometric model	Geographic focus	Number of projects using the model by country
Pinard (1995) [111]	Malaysia	1 in Malaysia
Brown (1997) [112]	Pantropical	2 in the Philippines
Ketterings et al. (2001) [113]	Sumatra, Indonesia	4 in Indonesia 1 in VietNam
Magcale-Macandog and Delgado (2002) [114]	Philippines	1 in the Philippines
Chave et al. (2005) [63]	Pantropical	3 in Cambodia 1 in Indonesia 1 in Lao PDR 1 in Timor Leste
Basuki et al. (2009) [115]	East Kalimantan, Indonesia	2 in Indonesia
Kenzo et al. (2009) [116]	Sarawak, Malaysia	1 in Indonesia
Phuong et al. (2012) [117]	Central Highland region, Viet Nam	1 in Viet Nam
Chave et al. (2014) [64]	Pantropical	1 in Malaysia
Manuri et al. (2016) [109]	Kalimantan, Indonesia	1 in Indonesia
Sources not reported	N/A	1 in Cambodia 2 in Indonesia 1 in Myanmar 1 in the Philippines 1 in VietNam

The monitoring frequency of forest carbon stocks in our reviewed projects was on average once every 3.1 years, ranging from once a year to once every 10 years. Such inter-project differences in monitoring frequency may affect the comparability of the quantification results [69, 118, 119].

Our review also identified considerable inconsistencies between projects’ validation practices. While certification requires Validation and Verification Bodies (VVBs) to employ rigorous procedures that involve on-site inspections, independent remeasurement of a subset of sample plots, and recalculation of carbon stock changes and associated accuracy [50, 120, 121], the level of adherence to these requirements varied. For example, we found that the available validation reports may focus more on evaluating the legitimacy of project design, with some lacking independent field verification or recalculation of forest carbon stocks.

Furthermore, significant inconsistencies lie in uncertainty measurement and reporting in forest carbon projects, making comparisons between projects and evaluation of the accuracy of carbon stock estimates challenging. Reported uncertainties range from 7% to 35%, with a majority falling between 10% and 20%. However, these percentages cannot be directly compared due to variations in the reported CIs. Of the projects reporting uncertainty, seven projects use 90% CIs, nine choose 95% CIs, and seven do not specify the CIs at all. Moreover, uncertainty is sometimes reported as a single fixed value and other times as a range, further complicating comparisons. The inconsistency extends to the methods used for calculating uncertainty, such as whether critical error sources other than sampling variability were considered.

Synthesis and prospects

Building on the patterns identified in the previous sections, the technical dimension of quantification for forest carbon projects in SEA faces a wide spectrum of interconnected challenges in relation to transparency, robustness, and consistency. The disclosure in the projects included in our review is highly insufficient, highlighting the transparency issue for carbon accounting. Other reviews also highlight similar differences in methodological and reporting requirements between carbon crediting schemes, and the need to enhance transparency [22, 33]. Confidence in carbon market mechanisms depends on the transparent disclosure of high-quality, accurate information to stakeholders [22]. Although transparency alone may not directly improve climate change mitigation outcomes, it can enable enhanced effectiveness by strengthening the informational systems and infrastructure that underpin MRV within carbon accounting frameworks [122]. Recent concerns and emerging evidence of malpractices in carbon accounting highlight the critical need to maintain transparency and methodological standardisation throughout the carbon quantification process [123]. Without stringent transparency requirements, less-than-fully transparent carbon projects with risky but cheap carbon credits could displace high-quality projects from the market [124] and threaten the viability of forest carbon project-based attempts to mitigate climate change [11].

Whether projects can achieve greater transparency depends on scientifically agreed-upon expectations over what constitutes robustness and consistency [37, 125]. We found that quantification methodologies and relevant reporting practices adopted by some forest carbon projects in SEA may not be able to ensure robustness, while an overall rigorous and specific regulatory mechanism for quantification is also non-existent. Such limitations, in turn, prevent a methodological framework from being reported in a sufficiently transparent but also robust and consistent manner [22]. Similar challenges for forest carbon projects are faced in tropical forest-rich regions around the world [126]. For example, inconsistencies in carbon accounting across different levels of governance are evident in neotropical countries, including Colombia and Peru, where around 70% of projects use at least one parameter that is different from those applied in national carbon accounting [14]. Even in the Global North, significant science-practice gaps persist because of a continued reliance on empirical models developed decades ago [13]. Taken together, this fragmented methodological landscape risks undermining carbon markets and constraining realised emissions reductions, creating an urgent need for quantification frameworks and disclosures that are auditable, defensible, and comparable under a shared scientific consensus [127]. To help bridge these gaps, we underscore the need for science-guided, harmonised methodological frameworks that: (i) are grounded in best available science for robustness, prioritise locally calibrated emission factors, properly propagate uncertainty, and adopt other recognised good practices to maximise accuracy and precision in quantification; (ii) enhance consistency by aligning methodological principles, definitions, and classifications to enable cross-project comparison and aggregation; and (iii) strengthen transparency by disclosing complete methodological details for the six key factors and explicitly reporting any practices that could compromise robustness or consistency.

Furthermore, at the practical level, a core challenge in forest carbon quantification may lie in the varying levels of technical capacity and understanding among practitioners. Such insufficient expertise can result in formulaic application and interpretation of prescribed methodologies without engagement with the underlying scientific and statistical principles, thereby weakening the robustness of quantification [14, 16]. For instance, although the IPCC-sanctioned gain-loss and stock-change approaches provide robust theoretical frameworks, project developers may struggle to interpret the assumptions and technical steps necessary for accurate estimates [13, 92]. A survey of forest carbon reporting experts found that while most acknowledged the importance of including uncertainty in forest carbon accounting, inadequate reporting persists due to a lack of expertise, with only 12% of respondents correctly interpreting the uncertainty calculation methods and 27% demonstrating an understanding of how uncertainty from independent sources should be properly propagated [92]. Such an issue indicates a forward-looking need for targeted capacity building and training, particularly in uncertainty quantification and data interpretation, to strengthen methodological robustness and consistency [13, 92].

Limited availability of high-quality data, particularly for AD and EF, further complicates robust and consistent quantification. Effective data collection often requires advanced remote sensing tools or extensive fieldwork that can be costly and technically demanding [44]. However, many countries in SEA and elsewhere in the tropics lack the resources, infrastructure and domain expertise to enable such comprehensive data collection [38, 128]. Consequently, project developers frequently face trade-offs between methodological robustness and feasibility, and may have to rely on default IPCC factors or global databases lacking local calibration [12]. Although some internationally funded programs provide assistance, this support frequently remains insufficient to meet local data requirements and to plug gaps in expertise and supporting technical infrastructure [129]. This underscores the need to invest in regional and local capacity for data collection and to develop context specific, locally validated estimation models, thereby enhancing the robustness and reliability of carbon quantification.

Certification standards such as the VCS or the CDM mandate certain methodologies and reporting protocols but may not always be sufficiently stringent to ensure comprehensive uncertainty assessments or robust monitoring schedules. This is particularly the case for older methodologies developed by certification standards, the current reliance on which may compromise methodological consistency. In some instances, projects experiencing high measurement uncertainty face discounts from their credited emission reductions [51, 55]. To avoid potential penalties, those running affected projects may be reluctant to transition to updated standards and opt instead for a simpler, less rigorous methodology that under-reports uncertainties [11]. Acknowledging these issues, some certification bodies are gradually improving their protocols. Verra, for example, is updating its quantification methodologies to enable more accurate assessments and more robust uncertainty propagation, as part of its Jurisdictional and Nested REDD+ (JNR) framework [130]. The new approach adopts a top-down, risk-based allocation system that ensures jurisdictionally consistent AD settings. Notably, Verra has announced that all existing projects are required to transition to the updated methodology and released re-quantification procedures [131]. However, our review found that such transitions are yet to be underway in SEA. Many older methodologies are still active, which is usually because projects already validated under previous standards find it cumbersome or costly to transition. Creating clearer pathways to adopt revised protocols and offering incentives for updating project methodologies could greatly improve the consistency and quality of carbon quantification across the region.

Furthermore, the broader JNR framework, along with other emerging standards such as the REDD + Environmental Excellence Standard by the Architecture for REDD + Transactions (ART-TREES) initiative, aims to reconcile project-level activities with larger jurisdictional strategies, reduce double counting, and foster standardised methodologies [107, 132]. These initiatives promote consistency by aligning project-level methods with national and international frameworks and enhance transparency by standardising reporting practices. Still, these frameworks have not yet become widely adopted in SEA and in the wider tropics [80, 133]. Incorporating complete and comparable local data into regional inventories remains a significant challenge [14]. Diverse forest types, data formats, and monitoring intervals complicate efforts to unify approaches, highlighting the need for capacity-building at multiple administrative levels [42, 134]. Supporting widespread adoption of these nested frameworks through policy incentives, technical training, and enhanced local infrastructure represents actionable pathways to improve consistency and transparency across scales [106].

Advancements in remote sensing technologies, computational resources, and analytical algorithms offer opportunities to address some of the data gaps identified in forest carbon quantification. For example, airborne or terrestrial Light Detection and Ranging (LiDAR) data, as well as synthetic aperture radar images, can capture detailed forest structure and enable the derivation of high-resolution, spatially explicit biomass estimates [135–137]. Further, Spaceborne LiDAR missions such as NASA’s Global Ecosystem Dynamics Investigation (GEDI) provide high-resolution three-dimensional views of forest canopies and thus a means of scaling-up biomass estimates [138]. Other initiatives, such as GEO-TREES, aim to establish accurate and consistent biomass baselines and forest inventory networks by comparing and verifying different data products on a global scale using a transparent and standardised approach [139]. When calibrated and validated through local field measurements, these initiatives can help update the generalised EF widely used in carbon project development [140, 141]. The emergence of cloud-based computational tools such as Google Earth Engine (GEE) has also facilitated open-source access to large datasets and expanded carbon quantification capacity in a more cost-effective way [142]. In addition, the growing availability of carbon flux measurements offers a more process-based understanding of the underlying mechanisms governing ecosystem carbon exchange [143]. Integrating such flux data into existing quantification frameworks may present an opportunity to reduce reliance on purely empirical models and improve the accuracy of sequestration potential estimates in a biophysically consistent manner [144]. These tools do not replace the need for context-specific field data; rather, they complement ground measurements to balance the trade-off between spatial scale and accuracy [65]. In many developing countries, including those in SEA, local calibration/validation efforts remain limited and can compromise the efficacy of these products [145]. Promoting open-access data resources, accessible forest monitoring systems and user-friendly EF databases, could lower technical barriers and further facilitate robustness and consistency [146]. Scaling-up the application of these technologies and platforms, ideally with support through international collaboration and regional partnerships, constitutes an actionable strategy to address the aforementioned methodological challenges [147, 148].

Finally, we acknowledge several limitations of this study. First, our review was limited to English-language documents, which may have excluded relevant materials published in other languages, particularly important in the multilingual context of SEA. Second, the analysis relied exclusively on publicly available documentation, limiting our findings to what projects have chosen to disclose or what is accessible through registries. As such, our analysis may not capture additional methodological details that are available internally or through direct engagement with project developers. Third, while we identified gaps between current practices and best-available scientific principles, we did not investigate their underlying drivers, such as institutional constraints, resource limitations, or evolving standard requirements. We recognise that many methodologies applied by the reviewed projects were aligned with the prevailing standards and less stringent practices were adopted by early-phase projects at that time. Our intention is not to critique these efforts but to highlight opportunities for enhancing transparency, robustness, and consistency moving forward. Finally, due to data availability constraints, we did not perform independent recalculations of carbon offset outcomes nor undertake quantitative comparisons across projects. Technological advancements may soon allow for more systematic and quantitative evaluations of forest carbon projects. Future research could build on this study by leveraging such advancements to generate deeper insights into the credibility of carbon accounting practices.

Conclusion

This study compiled and assessed the carbon quantification methodologies of 69 forest carbon projects in SEA using three interrelated criteria: transparency, robustness, and consistency. By identifying and evaluating six key technical factors in accordance with three criteria, we developed both conceptual and analytical frameworks for assessing carbon quantification methodologies in forest carbon projects. Our analysis reveals gaps and inconsistencies in quantification practices and reporting, which may affect the comparability, credibility, and effectiveness of forest carbon offset initiatives. Although these challenges are identified in SEA, they are globally relevant.

To address the challenges highlighted, project developers and market regulators should enhance transparency through clearer, more rigorous, and standardised reporting requirements. Capacity building for stakeholders is equally critical, particularly through methodological training in data collection, local calibration, and uncertainty quantification. Furthermore, facilitating the transition of legacy projects to updated methodologies and emerging jurisdictional frameworks will further align practices with evolving scientific standards. Finally, leveraging advances in remote sensing technologies, computational resources, and open-access data platforms within a broader climate and ecological science framework represents a further key opportunity for overcoming existing challenges in bridging methodological gaps.

If the issues raised through this research are left unaddressed, they could compromise the environmental integrity of carbon credits and diminish trust in carbon markets more broadly and in their ability to contribute meaningfully to climate change mitigation. There is thus an urgent and important need to strengthen the technical foundations of carbon accounting to enhance the reliability of emissions reductions and reinforce the trustworthiness of forest carbon projects as a key component of global climate change mitigation efforts.

Abbreviations

AD

Activity Data

AGB

Aboveground biomass

ART-TREES

The REDD + Environmental Excellence Standard by the Architecture for REDD + Transactions

CCB

Climate, Community & Biodiversity Standards

CDM

Clean Development Mechanism

CEOS

the Committee on Earth Observing Satellites

CI

Confidence Interval

DBH

Diameter at Breast Height

EF

Emission Factor

FAO

Food and Agriculture Organization

GEDI

Global Ecosystem Dynamics Investigation

GEE

Google Earth Engine

JNR

Jurisdictional and Nested REDD+

LiDAR

Light Detection and Ranging

IPCC

Intergovernmental Panel on Climate Change

NASA

National Aeronautics and Space Administration

MRV

Monitoring, Reporting, and Verification

SEA

Southeast Asia

UNFCCC

United Nations Framework Convention on Climate Change

VVB

Validation and Verification Body

Author contributions

Y.Z.: Conceptualisation, methodology, data curation, investigation, visualisation, writing, revision; Y.L.: Methodology, data curation, investigation, writing, revision; Z.D.T.: Methodology, data curation, investigation, writing, revision; H.T.: Conceptualisation, supervision, revision; D.M.T.: Supervision, funding acquisition, writing, revision; All authors read and approved the final manuscript.

Funding

This research was financially supported by the Singapore Social Science Research Council grant entitled ‘‘Climate Governance of Nature-based Carbon Sinks in Southeast Asia” (MOE2021-SSRTG-021) and the Singapore Ministry of Education grant entitled “Validating and Improving Satellite-based Forest Carbon Estimation in Southeast Asia” (MOE-T2EP50122-0006). Any opinions, findings, conclusions or recommendations expressed in this material are those of the authors and do not reflect the views of the Ministry of Education, Singapore.

Data availability

The data that support the findings of this review are publicly available. The compiled project inventory list, with project names, countries, approximate locations, relevant methodological information, and associated documentation sources, can be found in the supplementary materials.

Declarations

Ethics and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interestsEthics declarations

The authors declare no competing interests.