Uncertainty in REDD+ carbon accounting: a survey of experts involved in REDD+ reporting

Abstract

Background: Reducing Emissions from Deforestation and forest Degradation (REDD+) is a program established under the United Nations Framework Convention on Climate Change (UNFCCC) to reduce carbon emissions from forests in developing countries. REDD+ uses an incentive-based approach whereby participating countries are paid to reduce forest carbon loss and increase carbon storage. Country-level carbon accounting is challenging, and estimates of uncertainty in emission reductions are increasingly required in REDD+ reports. This requirement is hard to meet if countries lack the necessary resources, tools, and capabilities. Some REDD+ programs adjust their payments for the uncertainty reported, which presents a perverse incentive because uncertainties are larger if more sources of uncertainty are reported. We surveyed people involved in REDD+ reporting to assess current capacities and barriers to improving estimates of uncertainty.

Results: Representatives from 27 countries (44% of REDD+ countries at the time of survey implementation) responded to the survey. Nearly all respondents thought it important to include uncertainty in REDD+ reports, but most felt that the uncertainty reporting by their countries was inadequate. Our independent assessment of reports by these countries to the UNFCCC supported this opinion: Most countries reported uncertainty in activity data (91%) but not in emission factors (4-14%). Few countries use more advanced approaches to estimate uncertainty, such as Monte Carlo and Bayesian techniques, and many respondents indicated that they lack expertise, knowledge, or technical assistance. Other barriers include lack of financial resources and appropriate data. Despite these limitations, nearly all respondents indicated a strong desire to improve estimates of uncertainty in REDD+ reports.

Conclusions: The survey indicated that people involved in REDD+ reporting think it highly important to improve estimates of uncertainty in forest carbon accounting. To meet this challenge, it is essential to understand the obstacles countries face in quantifying uncertainty so we can identify where best to allocate efforts and funds. Investments in training and resources are clearly needed to better quantify uncertainty and would likely have successful outcomes given the strong desire for improvement. Tracking the efficacy of programs implemented to improve estimates of uncertainty would be useful for making further refinements.

Keywords: Carbon credits; Forest carbon; REDD+; Survey; Tropical deforestation.

Fig. 1

As of 2022, 56 countries, representing most of the forest area in developing countries, reported reference levels of REDD+ emissions for technical assessment under the UNFCCC, with some countries submitting more than once. Numbers show the percentage of submissions that included a combined (not necessarily complete or correct) accounting of uncertainty, as opposed to no reporting of uncertainty, some qualitative discussion but no quantification of uncertainty, or some quantification of uncertainty sources without an estimate of uncertainty in emissions

Fig. 2

Map of countries actively participating in REDD+ as of 2023. Representatives from 27 countries responded to the survey (41% of REDD+ countries at the time it was distributed). Specific countries that responded are not indicated on the map to protect anonymity, and not all countries shown were participating in REDD+ at the time the survey was distributed

Fig. 3

Levels of experience with approaches to error propagation for estimating uncertainty in REDD+ reporting

Fig. 4

This question tested knowledge of Monte Carlo simulation. Some countries have incorrectly reported the smaller interval (uncertainty in the mean of all outputs, which depends on the number of iterations) rather than the distribution of the outputs

Fig. 5

Two questions in the survey tested knowledge of error propagation. Analytical error propagation uses ${\sigma }_{a+b}=\sqrt{{\sigma }_a^2+{\sigma }_b^2+2{\sigma }_a{\sigma }_b}$. Hence, uncertainties of ${\sigma }_a=3$ and ${\sigma }_b=4$ Mt CO₂ yr⁻¹ combine to ${\sigma }_{a+b}$ = 5 Mt CO₂ yr⁻¹ if they are independent (the two errors being equally likely to be in the same or opposite directions) but ${\sigma }_{a+b}$ = 7 Mt CO₂ yr⁻¹ if they are fully correlated in direction and magnitude

Fig. 6

Barriers reported for assessing uncertainty in REDD+ reporting. Percentages include individuals who strongly or somewhat agree that the issue is a barrier using a 5-point Likert scale

Fig. 7

Assistance reported as beneficial for gaining knowledge about improving uncertainty in REDD+ reporting

Fig. 8

Uncertainties in reference levels of emissions reported by 40 countries that have submitted to the UNFCCC, 5 of them twice. 21 additional countries submitted reports without a combined uncertainty estimate. Colors indicate the number of types of uncertainty sources included, ranging from 1 to 5, namely, tree measurement, allometric models, variability in forest carbon (sampling error), land-use change, and other parameters (carbon fraction and root-to-shoot ratio). Not all countries calculated and propagated uncertainty sources correctly