The CHANGING 2024 WEALTH of NATIONS TECHNICAL REPORT Global estimates of carbon stocks in the vegetation and soils of terrestrial ecosystems 20 30 40 60 80 100 120 140 160 © 2024 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202-473-1000 Internet: www.worldbank.org This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does not guarantee the accuracy, completeness, or currency of the data included in this work and does not assume responsibility for any errors, omissions, or discrepancies in the information, or liability with respect to the use of or failure to use the information, methods, processes, or conclusions set forth. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of The World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be construed or considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. RIGHTS AND PERMISSIONS The material in this work is subject to copyright. Because The World Bank encourages dissemination of its knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202-522-2625; e-mail: pubrights@worldbank.org. Design and layout: Clarity Global Strategic Communications www.clarityglobal.net ACKNOWLEDGEMENTS Authors: Alessio Bulckaen, Raul Abad Viñas, Diego Bengochea Paz, Ruben Crespo, and Ferdinando Villa. The authors would like to acknowledge Kenneth Bagstad (USGS) and Stefano Balbi (BC3) for their contributions. The work is supported by the María de Maeztu Unit of Excellence 2023–2027 Ref. CEX2021-001201-M, funded by MCIN/AEI /10.13039/501100011033; and by the Basque Government through the BERC 2022–2025 program. This report received financial support from the Global Program on Sustainability (GPS) trust fund and the PROGREEN trust fund. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS i TABLE OF CONTENTS TABLES AND FIGURES IV ACRONYMS AND ABBREVIATIONS V 1. INTRODUCTION 1 1.1 The Importance of the Ecosystems Providing the Carbon Regulation Service 1 1.2 Latest Studies on Forest Ecosystems 1 1.3 Carbon Stock Valuation in Biophysical and Monetary Terms, Main Drivers of Change, and Interpretation of Results 2 1.4 The ARIES Approach to Interoperability: Ongoing Production of Results Using State-of-the-Art Scientific Knowledge 3 2. OVERVIEW OF THE CARBON STOCK ESTIMATES 5 2.1 Vegetation Carbon Biomass 5 2.1.1 Data Used to Estimate the Vegetation Biomass Carbon Storage of Terrestrial Ecosystems 7 2.1.2 Key Elements for a Correct Interpretation of Model Results 8 2.1.3 Measuring the Accuracy of Results 10 2.2 Estimating Soil Organic Carbon Storage Stocks 11 3. MONETARY VALUATION OF CARBON RETENTION IN ARIES FOR SEEA 12 4. RECOMMENDATION FOR FUTURE DEVELOPMENTS AND IMPROVEMENTS 14 5. DISCUSSION OF RESULTS 16 5.1 Vegetation Carbon Stock Results Aggregated by Country 16 5.2 Vegetation Carbon Stock Results Disaggregated by Land Cover Class 20 5.2.1 Forest Vegetation Carbon Stock 22 5.2.2 Shrub and Herbaceous Vegetation Carbon Stock 26 5.2.3 Agricultural Vegetation Carbon Stock 27 5.2.4 Wetland Vegetation Carbon Stock 32 5.3 Soil Organic Carbon Results Disaggregated by Land Cover Class 34 ii GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 6. COMPARATIVE ASSESSMENT OF VEGETATION CARBON STOCK RESULTS 35 6.1 Introduction 35 6.2 Comparative Assessment of Areas 37 6.3 Issues with Land Classification Systems 39 6.4 Comparative Assessment of Carbon Stocks in Vegetation 41 7. SUMMARY AND NEXT STEPS 45 REFERENCES 46 APPENDIX 1: Country-specific background information used for the comparative assessment included in section 6 49 APPENDIX 2: Aligning the terrestrial results produced in ARIES with the carbon contribution of mangrove ecosystems 55 APPENDIX 3: Large yearly differences in the vegetation carbon stock not explained by land cover changes 57 APPENDIX 4: Report of the causes of large changes in vegetation carbon stock of individual land cover classes 67 APPENDIX 5: Report of the causes of large relative changes in vegetation carbon stock of individual land cover classes 74 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS iii TABLES TABLE 2.1: Extract from the ARIES lookup table 6 TABLE 2.2: Data inputs 8 TABLE 4.1: Carbon stratification 15 TABLE 5.1: Relative contribution to vegetation carbon stock by aggregated class over the whole time series (2001–2020 total vegetation carbon stock) 21 TABLE 5.2: Agricultural carbon stock analysis for Bolivia over the period 2002–2003 30 TABLE 6.1: Comparison of information on total country area submitted by countries with model data 37 TABLE 6.2: Comparison of information on forest area submitted by countries with model data 38 TABLE 6.3: Comparison of forest definition used by countries with model data 40 TABLE 6.4: 2006 IPCC default values for forest aboveground biomass used by the model 42 TABLE A4.1: Analysis of large deviation 67 FIGURES FIGURE 5.1A: Country-level global map (megatonnes) 17 FIGURE 5.1B: Distribution of the vegetation carbon stock among the top 10 countries 17 FIGURE 5.2: Vegetation carbon density map for 2020 (tonnes/hectare) 18 FIGURE 5.3A: Total vegetation carbon stock decline by country (megatonnes) 18 FIGURE 5.3B: Total vegetation carbon stock decline by country (in relative terms) 19 FIGURE 5.3C: Total vegetation carbon stock increase by country (megatonnes) 19 FIGURE 5.3D: Total vegetation carbon stock increase by country (in relative terms) 20 FIGURE 5.4: Distribution of vegetation carbon (%) in 2020 21 FIGURE 5.5A: Forest global vegetation carbon stock (megatonnes) 22 FIGURE 5.5B: Forest global vegetation carbon stock (%) 22 FIGURE 5.5C: Contribution to forest global vegetation carbon stocks for 2001–2020 (megatonnes) 23 FIGURE 5.5D: Forest vegetation carbon stock decline by country (megatonnes) 24 FIGURE 5.5E: Forest vegetation carbon stock decline by country (in relative terms) 24 FIGURE 5.5F: Forest vegetation carbon stock increase by country (megatonnes) 25 FIGURE 5.5G: Forest vegetation carbon stock increase by country (in relative terms) 25 FIGURE 5.6A: Shrub and herbaceous global vegetation carbon stock (megatonnes) 26 iv GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS FIGURE 5.6B: Shrub and herbaceous global vegetation carbon stock (megatonnes) 26 FIGURE 5.6C: Contribution to shrub and herbaceous global vegetation carbon stock (megatonnes) 27 FIGURE 5.7A: Agricultural global vegetation carbon stock (megatonnes) 28 FIGURE 5.7B: Agricultural global vegetation carbon stock (%) 28 FIGURE 5.7C: Contribution to agricultural vegetation carbon stock (megatonnes) 28 FIGURE 5.7D: Agricultural vegetation carbon stock increase by country (megatonnes) 29 FIGURE 5.7E: Agricultural vegetation carbon stock increase by country (in relative terms) 29 FIGURE 5.7F: Agricultural vegetation carbon stock decline by country (megatonnes) 31 FIGURE 5.7G: Agricultural vegetation carbon stock decline by country (in relative terms) 31 FIGURE 5.8A: Wetland global vegetation carbon stock (megatonnes) 32 FIGURE 5.8B: Wetland global vegetation carbon stock (%) 32 FIGURE 5.8C: Contribution to wetland global vegetation for 2001–2020 33 FIGURE A3.1: Land cover map of South Sudan 57 FIGURE A3.2: Presence of burned land in South Sudan 58 FIGURE A3.3: Overlay of layers of presence of burned land and ecofloristic regions in South Sudan 58 FIGURE A3.4: Area covered by the tropical moist deciduous forest ecofloristic region in South Sudan 59 FIGURE A3.5: Carbon stock estimated for a non-burned area in tropical moist deciduous forest ecofloristic region in South Sudan 59 FIGURE A3.6: Carbon stock estimated for a burned area in tropical moist deciduous forest ecofloristic region in South Sudan 60 FIGURE A3.7: Area covered by the tropical rainforest ecofloristic region in South Sudan 60 FIGURE A3.8: Carbon stock estimated for a non-burned area in tropical rainforest ecofloristic region in South Sudan 61 FIGURE A3.9: Carbon stock estimated for a burned area in tropical rainforest ecofloristic region in South Sudan 61 FIGURE A3.10: Area covered by the tropical mountain system ecofloristic region in South Sudan 62 FIGURE A3.11: Carbon stock estimated for a non-burned area in tropical mountain system ecofloristic region in South Sudan 62 FIGURE A3.12: Carbon stock estimated for a burned area in tropical mountain system ecofloristic region in South Sudan 63 FIGURE A3.13: Presence of burned land in Gabon 64 FIGURE A3.14: Presence of burned land in Gabon with legend 64 FIGURE A3.15: Gabon 2002: Presence of fire 65 FIGURE A3.16 Gabon 2003: Presence of fire 66 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS v ACRONYMS AND ABBREVIATIONS AI Artificial intelligence ARIES Artificial Intelligence for Environment and Sustainability BC3 Basque Centre for Climate Change CWON Changing Wealth of Nations ESA European Space Agency ESA-CCI European Space Agency Climate Change Initiative FAO Food and Agriculture Organization of the United Nations FIRMS Fire Information for Resource Management System GHG Greenhouse gas IPCC Intergovernmental Panel on Climate Change IUCN International Union for Conservation of Nature REDD+ Reducing Emissions from Deforestation and forest Degradation program SEEA System of Environmental-Economic Accounting SEEA EA System of Environmental-Economic Accounting Ecosystem Accounting UNFCCC United Nations Framework Convention on Climate Change UNSD United Nations Statistics Division vi GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS INTRODUCTION l 1 Introduction This section provides a general introduction to efforts have been undertaken at different scales, present the work done to produce global maps from local to global, and by different stakeholders, representing modeled vegetation carbon stocks ranging from academic and research institutions for the period 2001–2020, available at https://data. to governments and private institutions. Some of integratedmodelling.org/dataset/global-vegetation- the most important are the European Space Agency carbon-soil-2001-2020, as well as summary statistics (ESA), particularly the ESA Climate Change Initiative and results derived from these outputs. (CCI); the United Nations Statistics Division (UNSD); the Intergovernmental Panel on Climate Change 1.1 THE IMPORTANCE OF THE (IPCC), which focuses on studying human-induced climate change; as well as the institutions advancing ECOSYSTEMS PROVIDING the Reducing Emissions from Deforestation and THE CARBON REGULATION forest Degradation program (REDD+). SERVICE The global climate crisis is one of the most pressing 1.2 LATEST STUDIES ON FOREST challenges of our time, exacerbating existing social ECOSYSTEMS and environmental issues. Regulating ecosystem services, and climate regulation in particular, plays a Forest ecosystems play a critical role among fundamental role in mitigating the effects of climate terrestrial ecosystems, especially in the context change. In this regard, terrestrial ecosystems of climate regulation, and for this reason they are play a crucial role as stocks for carbon that would the focus of many important studies. Mengist and otherwise be released into the atmosphere, causing Soromessa (2019) provide a comprehensive review further warming, as well as being potential carbon of the research approaches and trends in studying sinks. Terrestrial ecosystems, particularly forests, forest ecosystem services, including carbon are thus essential to addressing the global climate regulation. Their overview shows that studies on crisis. For these reasons, it is important to include forest ecosystem services mainly focus on three carbon in the World Bank’s Changing Wealth of types of services: (i) provision, such as timber Nations (CWON) wealth estimates. production and water supply; (ii) climate regulation, Terrestrial ecosystems, which are under threat, are such as carbon sequestration; and (iii) cultural the main focus of this analysis, given their important ecosystem services, such as recreation. They also contribution to global carbon regulating services. To found that the methodologies applied are diverse try to understand the dynamics behind the human- and that results are dispersed and inconclusive induced changes in the climate balance, as well as at the global level. In our analysis, we provide a to identify ways to mitigate such effects and adapt to globally consistent methodology to quantify climate their impacts, significant work has been done. Such regulation services from terrestrial ecosystems. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 1 l INTRODUCTION Other relevant studies that are worth highlighting to aims to produce a comprehensive collection of give an idea of the ongoing efforts in this field include measurements, in which all assets are reported Taye et al. (2021) and Costanza et al. (2014). The 1 2 individually and aggregated. first study (Taye et al.) uses meta-analysis to estimate This analysis was carried out by the Basque the economic values of global forest ecosystem Centre for Climate Change (BC3), which develops services. A regression is run, considering studies multidisciplinary knowledge to support decision- published between 1990 and 2018, to derive the making on this global challenge. Within the research economic value of a hectare of forest in a particular center, the Artificial Intelligence for Environment year. By controlling for specific characteristics that and Sustainability (ARIES) team developed an make individual applications differ from others, integrated modeling system with a platform the authors try to capture the intrinsic value of the capable of combining data and models to study the service, which is then used to derive values across the interaction of human and natural systems, with a globe. The second study (Costanza et al.) highlights strong focus on the environment and sustainability the contribution of land use to human well-being dimensions. More information is provided in and emphasizes how ecosystem services are best the next section (1.4) to briefly describe how the considered public goods, requiring new institutions. modeling integration based on artificial intelligence There are numerous other studies assessing and (AI) and semantics is used to obtain results aligned valuing ecosystem services, but in most cases, they with the World Bank’s CWON approach. The global focus on specific locations or services, and few use model to estimate vegetation carbon stock used in methods that are replicable at the global scale. this analysis quantifies how the global landscape is changing as a result of ecological transformations 1.3 CARBON STOCK VALUATION that occur at biophysical (environmental), social, IN BIOPHYSICAL AND and economic levels. In the context of the CWON MONETARY TERMS, MAIN program, the World Bank collaborates with the DRIVERS OF CHANGE, AND BC3 team, who developed the ARIES for System INTERPRETATION OF RESULTS of Environmental-Economic Accounting (SEEA)3 platform, to create global baseline biophysical and A growing number of scientific publications quantify monetary accounts of the carbon storage ecosystem the contribution of the carbon regulation service service. in the context of climate change. While the works described in the previous section offer an important It is important to emphasize that changes in these perspective, their purposes are different from the results are mainly driven by land cover changes, and approach taken in CWON: rather than analyzing potentially by land cover re-classification, which previous studies on forests around the globe, the do not always match with the real, but rather the World Bank focuses on biophysical observations modeled, ecological information. The complexity to estimate both physical and monetary values of of an observed area is greatly simplified to allow a the services reported in their work. This approach global representation of the reality. The landscape 1 Table 3. 2 Appendix 2 (spreadsheet). 3 https://seea.un.org/. 2 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS INTRODUCTION l is classified into discrete classes, which are of This report starts with a general overview of the course artificial constructs. This trade-off is well model used to estimate global climate regulation known by experts in spatially explicit analyses, services, then covers specific methods, data, and with no completely accurate solution, though models used in ARIES. The report then discusses some approximations are better than others. Other the limitations and caveats to be considered when dynamic variables used in the model that can interpreting and using the results. It concludes contribute to change include the occurrence of fire with a discussion of the improvements planned and the presence of pristine forests. Fire, as well to enhance future outputs and align results from as data gaps and information on fire intensity and different analyses. vegetation regrowth, can also be the main reason The monetary valuation based on the biophysical for changes in the forest carbon stock of a country results is also discussed in section 4. The main (more details are given in Appendix 3). The model ecosystem service priced in monetary terms is is built to allow annual carbon stocks all over the world to be globally comparable, and the reader carbon retention, which consists of carbon storage should interpret the outputs of this analysis as expressed as an annual flow. The total physical such. Such an approach is fundamental to be able carbon stock from all the different pools (above- to observe trends and produce comparable results and belowground biomass and soil organic carbon) over a long time period, and the main reason for is first priced to estimate the avoided damages and choosing a global model as opposed to, for instance, subsequently multiplied by a rate of return to obtain global biomass datasets that exist for a single year or an annuity. only several recent years. 1.4 THE ARIES APPROACH TO Considering this primary purpose, caveats to the use of the data are clearly stated in section 2.1.2. INTEROPERABILITY: ONGOING Users should be aware that the main goal of the PRODUCTION OF RESULTS results is to observe trends in global changes of USING STATE-OF-THE-ART carbon stock over time. Indeed, the results are SCIENTIFIC KNOWLEDGE intended to support climate policy makers to verify The results produced in this work are an important trends observed in related reports, contribute to output, but as scientific knowledge evolves and filling gaps in time series, or raise awareness of better models and data become available, the results potential hotspots of carbon fluxes—all with a view will be adjusted accordingly. Some improvements to informing actions to combat the adverse effects are already planned and future results will include, of climate change. Annual changes in vegetation at the minimum, the incorporation of IPCC carbon stocks throughout the time series reflect the 2019 guidelines, and further information on the effect of updated information on land cover classes uncertainty associated with these outputs. These or other dynamic model inputs. They should not be planned revisions are further described in section 4. interpreted as a proxy for complete and accurate figures of net carbon fluxes and other non-CO2 The process to produce results based on improved emissions from lands. The results include a dataset information is made easier by the inherent and tabular time series that can be downloaded and interoperable nature of ARIES, a modeling platform used for the above-noted purposes. that enables the integration of science-based GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 3 l INTRODUCTION knowledge. Interoperability  4 is the ability of humans that experts can model the particular phenomena and computer systems or software to exchange and that they know best, while the AI combines that make use of information, and is at the core of the particular piece of knowledge with the rest of the strategy that the ARIES team is working to promote, information contributed by others in order to answer and that other institutions, like the World Bank and larger questions. Since all the information used in UNSD, also want to bring forward. ARIES has been added by experts specializing in specific domains and with different expertise, its Whenever new and reliable data and methodologies AI can answer a particular question by distilling (or specific model components) are added in the the complexity of the query into a workflow that is system, ARIES can automatically prioritize the use executed in order to provide the final user with a of these new inputs if they help to improve the clear and fully documented answer, regardless of results. Thus, new data and models are prioritized the complexity of the needed analysis. when they better suit a specific context or more precisely answer a question posed by a user. ARIES works at different scales and automatically adjusts its results to the spatio-temporal context ARIES is powered by its cutting-edge integrated under analysis. While using local data and models modeling technology, in development since 2007. is not the purpose of this analysis, as global Its artificial intelligence is based on machine comparability is emphasized here in order to support reasoning, 5 which guides the AI system to the a global analysis, ARIES can digest different data, most appropriate data and models to answer each at different spatio-temporal scales. Its underlying question, providing clear and fully documented AI automatically accesses, assembles, and runs answers to critical questions in the context of the data and models needed to adjust the output to a ecological crisis. These answers inherently contain particular spatio-temporal context, or automatically a wide range of information, spread across multiple uses a better model for that particular context, if it areas of knowledge (such as ecology, hydrology, exists in the system. For this work, a unique global and sociology), which are compiled by integrating model and global data are used, but context-specific the information through the power of AI. This (regional, national, or subnational) modifications allows ARIES to provide the answer based on the are possible and encouraged in ARIES. most up-to-date knowledge across different areas of expertise. In this context, interoperability means 4 https://seea.un.org/sites/seea.un.org/files/seea_interoperability_strategy.pdf. 5 Machine reasoning is a type of artificial intelligence that draws conclusions from ideas and concepts using logic and rules introduced to the system by humans. 4 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S l 2 Overview of the Carbon Stock Estimates The terrestrial carbon storage model, as currently integration of the mangrove carbon storage data, coded in ARIES, computes terrestrial carbon stock with some recommendations on how to produce as the sum of vegetation stocks, composed of a harmonized but comprehensive and consistent aboveground biomass and belowground biomass, result. Specifically, attention is given to the best and soil organic carbon stocks. None of these way to reconcile the vegetation carbon stock results components are limited to forests, but the model with the maps and outputs produced for mangroves does not take into account woody debris. The first 6 (outside ARIES). Preliminary discussions were held two stocks (aboveground biomass and belowground among the groups to explore how these outputs biomass) are modeled simultaneously using the could be included in ARIES in the future, which vegetation carbon model for the years 2001 to 2020. would then automatically harmonize these results. To estimate vegetation stocks, it is assumed that each terrestrial land cover class, with a few exceptions,7 2.1 VEGETATION CARBON BIOMASS contributes to storing carbon to varying extents. The classes with the highest contribution to carbon stock For vegetation, the carbon accounted for is both are forests and wetlands, particularly mangroves, the carbon stock in aboveground biomass, with but each class contributes proportionally to a particular focus on forest ecosystems, and in the total carbon stock of an area. The estimates belowground biomass. The model quantifies the provide annual snapshots of the stock for every total carbon stored in vegetation in each year. The year considered in the time series. In contrast, the information is produced using IPCC 2006 default information provided on soil organic carbon stocks factors8 and by incorporating yearly updated land is based on a static dataset produced in 2020 that cover data and other inputs described below. maps global soil organic carbon. Using a multi-layer lookup table, the model merges Separately, the World Bank is estimating mangrove spatially explicit data on land cover classes and other carbon storage using higher resolution data from features with carbon stocks derived from the IPCC Global Mangrove Watch for a subset of years (1996, guidelines to map the carbon in vegetation globally. 2007, 2010, 2015, 2018, and 2020). The results of this This approach replicates the method developed by work have been produced by Silvestrum Climate Gibbs and Ruesch (2008).9 The current inputs used Associates. Appendix 2 of this report includes in the model are as follows (further details on the technical aspects considered for the potential data sources are included in the next section): 6 Fallen dead trees and the remains of large branches on the ground. 7 For example, the ESA-CCI includes classes for water bodies, and glaciers, which do not store carbon, and bare soil/rocks, which only store limited amounts of carbon: Legend of the global CCI-LC maps, based on LCCS. 8 Guidance that officially governs the current submission of GHG inventories. A refinement of this document was published in 2019 (IPCC 2019 refinement to the 2006 IPCC Guidance), although this has not yet been adopted by the UNFCCC reporting framework. 9 New IPCC Tier-1 Global Biomass Carbon Map for the Year 2000 (dataset): https://www.osti.gov/biblio/1463800. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 5 l OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S 1. Data about land cover classes to estimate changes which have no vegetation that stores carbon. in the extent of the vegetation. The first items in the lookup table (Table 2.1) show 2. Ecological region (Food and Agriculture how inputs are combined to obtain the carbon stock Organization (FAO) ecofloristic zone) to in each grid cell. For each land cover class, located distinguish different regional ecological in a specific ecofloristic zone, within a particular characteristics that influence carbon storage. continent, the model assigns a specific value for the vegetation carbon stock in that area. In the case of 3. Presence of primary forest to identify areas of a forest, the carbon content also depends on how greatest carbon storage. pristine, old, and/or intact the forest is. A so-called primary forest stores a greater quantity of carbon. 4. Occurrence of fire to take into account loss of The model also identifies burned areas and assigns a carbon stored in vegetation in burned areas. significantly lower to zero vegetation carbon stock for The code used in ARIES to estimate these values that year in the burned area. To provide an example, is publicly available in this repository. An example a broad-leaved forest, located in the African tropical of the first lines of the lookup table is shown in rainforest and in an area where there was no record Table 2.1. The complete table, which summarizes of fire in that year, is estimated to store, for each the inputs to implement the original methodology hectare, 200 tonnes of vegetation carbon, contained by Gibbs and Ruesch (2008), can be explored at this in its aboveground and belowground biomass. link. As mentioned above and shown in the code, Table 6.4 provides more information on the default all the terrestrial land cover classes are included, values set in the IPCC guidelines, which is discussed except for bare soil/rock, water bodies, and glaciers, in more detail in section 6. TABLE 2.1: EXTRACT FROM THE ARIES LOOKUP TABLE FAO LAND COVER CONTINENTAL PRIMARY BURNED CARBON ECOFLORISTIC CLASS REGION FOREST LAND STOCK (t/ha) ZONE Broadleaf Tropical Africa Either False 200 forest rainforest Broadleaf Tropical North and Either False 193 forest rainforest South America Broadleaf Tropical Continental Either False 180 forest rainforest Asia Broadleaf Tropical Asia Either False 180 forest rainforest Broadleaf Tropical Insular Asia Either False 225 forest rainforest Broadleaf Tropical Australia Either False 199.5 forest rainforest 6 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S l 2.1.1 Data Used to Estimate the 4. Occurrence of fires (annually available from Vegetation Biomass Carbon 2000 to 2020). The information is needed to Storage of Terrestrial Ecosystems incorporate the effect of wildfires on the carbon budget of the ecosystem in the model. An The inputs used in the model and briefly listed in approach is also suggested to smooth out post- the previous sections include: burned effects (see section 2.1.4 for further detail). 1. ESA-CCI land cover data used to retrieve a The data initially considered for this work are the complete time series of land cover classes from ESA-CCI land cover maps10 from 1995 to 2020; MODIS burned land maps11 for 2000 to 2020; Intact Forest 1992 to 2020. Land cover classes are used to Landscapes maps12 for 2000, 2013, 2016, and 2020; assign IPCC values of carbon stock. It should be and FAO ecofloristic and continental regions.13 The noted that some classes, such as water bodies and original goal was to estimate a time series starting in glaciers, are excluded from the computation in 1995 to align results with the comprehensive wealth this first set of results, while others, like lichen estimates in other CWON products, but limited data and moss-covered areas or bare areas, only availability restricted the analysis to a shorter time contribute modestly to carbon stock. series from 2001 to 2020. 2. Ecofloristic zones and continental regions. The integration of several alternative datasets was The zones are assumed to be static and are used to considered but eventually discarded because of select default IPCC values and to produce results limited geographical coverage or shorter temporal at different spatial scales. coverage. For example, the Joint Research Centre’s tropical moist forest dataset was considered as an 3. Spatially explicit layers on intact/primary alternative for the Intact Forest Landscapes data to forests. The layers provide information on the capture deforestation and degradation, but it was presence of intact forests, which are used as ultimately dismissed due to its limited geographic a proxy for primary forests. The information coverage. Similarly, several alternatives to the is available for the years 2000, 2013, 2016, and MODIS burned land datasets were considered, as 2020. The location of intact and non-intact they might provide over- or under-estimates in forests allows for a proxy of the degree of forest some instances, due to the coarse resolution (1km2) degradation and other anthropogenic effects that of the input data from the Fire Information for affect the carbon stock in the vegetation in non- Resource Management System (FIRMS).14 However, intact areas. the alternatives were either not accurate enough15 10 https://doi.org/10.24381/cds.006f2c9a. 11 https://modis-fire.umd.edu/ba.html, product MCD64A1. 12 Intact forest landscapes data are available at https://intactforests.org/data.ifl.html. 13 New IPCC Tier-1 Global Biomass Carbon Map for the Year 2000: https://cdiac.ess-dive.lbl.gov/epubs/ndp/global_carbon/carbon_documentation.html. FAO’s ecofloristic region dataset: https://databasin.org/datasets/dc4f6efd1fa84ea99df61ae9c5b3b763/. 14 The FIRMS data detect thermal anomalies in either MODIS AQUA or TERRA data, at https://www.earthdata.nasa.gov/learn/find-data/near-real-time/firms. 15 The MODIS/Terra Vegetation Continuous Fields Yearly L3 Global 250m SIN Grid, LAADS DAAC. (https://ladsweb.modaps.eosdis.nasa.gov/missions-and-measurements/products/MOD44B#overview), is a possible alternative, but has several limitations. First, inter-annual comparisons of vegetation continuous fields must be interpreted cautiously, considering the high potential of inter-annual fluctuations in the percentage of tree cover due to various forms of forest disturbance and regrowth. Second, the data are only validated to stage 1, so the accuracy was estimated through an assessment of the training data’s accuracy and from limited in situ field validation datasets (Data maturity levels, Science Mission Directorate, https://science.nasa.gov/earth-science/earth-science-data/data-maturity-levels). GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 7 l OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S or had a too limited time series16 for the purpose 2.1.2 Key Elements for a Correct of building a consistent, global, and relatively long Interpretation of Model Results time series for fire occurrence. 2.1.2.1 General considerations Previous literature reviews, namely Humber et al. (2019) and Pessôa (2020), have also indicated that At least three considerations should guide correct compatible data products do not exist. Should interpretation and use of the results produced using new data become available, the possibility of this modeling approach. incorporating them into ARIES and extending the First, when viewing model results, any change current estimate to cover the period 1995–2000 can identified across the time series to vegetation carbon be considered.17 The full list of data inputs is shown stocks in a given area is mainly driven by a change in Table 2.2. in the land class information. If the underlying land cover class changes, so does the carbon stock value associated with that area, in accordance with the IPCC methodology. Net carbon fluxes from lands, and other emissions of non-CO2 gases, are recognized as highly complex to accurately model, and are excluded from the approach. TABLE 2.2: DATA INPUTS TEMPORAL TEMPORAL SPATIAL SPATIAL NAME DESCRIPTION COVERAGE RESOLUTION COVERAGE RESOLUTION ESA-CCI Land cover 1992–2020 Annual Global 300m MODIS Burned land 2000–2020* Monthly* Global 500m Ecofloristic Single FAO Static Global Shapefile regions observation Intact Forest 2000, 2013, Primary forest 2000–2020 Global 300m Landscapes 2016, 2020 * The dataset only covers the last two months of 2000. More information is provided in the next sections to explain the consequences of the results and the decision to exclude this year from the final results. 16 For example, the timeseries of the ALOS PALSAR FNF (forest/non-forest) products (https://asf.alaska.edu/data-sets/sar-data-sets/alos-palsar/alos-palsar-about/) is limited to the years 2007 to 2018 (Global PALSAR-2/PALSAR Forest/ Non-Forest Map, Earth Engine Data Catalog, Google Developers, at https://developers.google.com/earth-engine/datasets/catalog/JAXA_ALOS_PALSAR_YEARLY_FNF). 17 Benchmarking is a standard practice in generating economic statistics (Hillmer and Trabelsi 1987) but can involve complex and formal analyses, which are beyond the scope of this project. 8 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S l Second, working at a global scale hampers the because the necessary data are simply not available accuracy of the results, mainly because of the lack and consistent at a global level. Such information of available information covering the full time frame is not available for top-down approaches that aim and spatial resolution necessary to successfully to produce global results. This limitation should represent natural and anthropogenic effects in the be considered if comparing global results to those carbon cycle. of local models that can account for more detailed ecological and management processes. When Third, some factors need to be considered when analyzing results produced using different data using the global outcomes of the model at lower sources, users should be aware that the use of spatial or shorter time scales. Regional or forest- different land classification systems or land cover stand models can account for important drivers datasets introduces well-known divergences in needed to quantify vegetation carbon stocks and results of two such modeling efforts. their changes—for instance, management practices, climate conditions, species composition, and Ultimately, model results represent snapshots of market demand. For example, direct oxidation to the worldwide terrestrial carbon stocks in vegetation atmosphere can be assumed when a deforestation that are assigned on the basis of yearly spatially event occurs. However, it is not feasible to accurately explicit information on land cover classes, FAO capture the effects of vegetation growth, natural ecofloristic zones and location in the continental disturbances, aging of trees, climate conditions, regions, and presence of primary forests and fire species compositions, and management practices activity. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 9 l OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S 2.1.2.2 Spatial and temporal coverage In terms of temporal coverage, the results are produced for the period 2001–2020. This decision is In terms of spatial coverage, there are some areas mainly based on the forward-looking applicability in which the model could not be run due to lack of of the results and because the lack of pre-2001 data inputs. These areas are mainly very small islands for some inputs hampers the consistency of the and administrative areas, particularly archipelagos results produced before 2001. In particular, the year and overseas territories, which are not covered by 2000 was excluded after a careful review, because the FAO ecofloristic layer. They include: burned area data do not cover some areas of the world and are only available for the last two months • Clipperton Island (France) of the year, meaning that fire activity would be • Palmyra Atoll (United States) strongly underestimated. For the same reason, years • Midway Island (United States) prior to 2000 are excluded, due to the lack of the input data on the occurrence of fire and presence • Kingman Reef (United States) of primary forest. Conducting the analysis without • Jarvis Island (United States) complete and consistent input data would lead to incomparable outputs between the areas around the • Johnston Atoll (United States) globe where data gaps influence the results. • Howland Island (United States) 2.1.2.3 Moving from spatial to country • Baker Island (United States) aggregate values • Heard Island and McDonald Islands (Australia) The results are produced yearly at the global level. • Bouvet Island (Norway) The raw model outputs from ARIES are tiled raster • Ashmore and Cartier Islands (Australia) data, which are merged together into a single map to give annual global results. Country-level results • Wake Island (United States) are obtained by aggregating grid cells within the • Tokelau (New Zealand). country boundaries and summarizing results in the tables by country, by land cover class,19 and by year. Other regions, such as the Antarctic, were intentionally excluded due to the insignificant or In future, as new iterations of the model are complete absence of vegetation. However, the model endorsed by expert groups (for example, should the is capable of producing consistent results for these forest ecosystems modeling be vetted by the SEEA excluded areas and incorporating them in the global Ecosystem Accounting (EA) Forest Working Group), outcome, provided that necessary information the results could be summarized by ecosystem type. becomes available.18 This is a complex endeavor; for example, it is an open 18 An advantage of the model is that new data made interoperable with ARIES can be readily incorporated in future analyses, offering the opportunity to generate results focusing on different temporal or spatial scales using ecosystem or region-specific data. However, the implications and interpretations of future analyses go beyond the scope of this work and are not discussed here. 19 According to the ESA-CCI land cover classification. 10 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS OV E R V I E W O F T H E C A R B O N S TO C K E S T I M AT E S l question in the environmental and natural capital 2.2 ESTIMATING SOIL ORGANIC accounting communities how forests are defined by CARBON STORAGE STOCKS different countries. To represent the final results, the ESA-CCI land cover classes have been grouped In this analysis, the latest available spatially explicit into six aggregated classes, each representing a information on soil organic carbon was used. generalized group of ecosystems. These classes are Earlier global soil carbon datasets were generated described in the results section. in the past, covering previous years,22 but they were produced using different methodologies, 2.1.3 Measuring the Accuracy of Results and they are not comparable or cannot be used as a time series. The incomparability of these As mentioned above, a number of limitations hinder products is due to the increasingly sophisticated the ability to develop uncertainty estimates for these machine learning methods used to estimate global results. One approach could propagate uncertainties soil sample databases, as well as the larger and associated with all the input data, but such more accurate number of Earth observation and uncertainty information is not available. Moreover, other environmental datasets used as covariates, this approach could provide misleading information combined to map the spatial distribution of soil if user-supplied land classification systems are properties around the globe. For this reason, this inconsistent with the model inputs. The same caveats analysis uses the most recent dataset, covering apply to other data used to feed the model. the year 2020, which is incorporated into the final For transparency and whenever possible, results as a separate output from vegetation carbon information on the uncertainty of the input data stock, to increase the completeness of the outputs. was incorporated. For example, various studies The dataset used for this purpose is a product from note accuracy dimensions for the ESA-CCI data at ISRIC—World Soil Information,23 which includes soil 70 percent20 and 82 percent.21 carbon information at different depths, up to 200cm. The generated results account for the soil organic carbon at 30cm of depth. Since these outputs are the best global estimates of soil organic carbon stock, they are added to the final results without any manipulation and their values are shown in addition to the vegetation carbon stock. Results are disaggregated by country, and by land cover class, but the results should be carefully interpreted, considering that the ESA-CCI layer was not an input to the model. 20 https://www.esa-landcover-cci.org/?q=webfm_send/84: pp. 31–33, chapter 4.2.2. Validation results. 21 Annual dynamics of global land cover and its long-term changes from 1982 to 2015: Classification accuracy of the ESA-CCI European Space Agency Climate Change Initiative: https://www.researchgate.net/figure/Classification-accuracy-of-the-ESA-CCI-European-Space-Agency-Climate-Change-Initiative_tbl3_341894401. 22 ISRIC, SOTER, and SOTWIS among others. 23 https://www.isric.org/explore/soilgrids . GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 11 l M O N E TA R Y VA L U AT I O N O F C A R B O N R E T E N T I O N I N A R I E S F O R S E E A 3 Monetary Valuation of Carbon Retention in ARIES for SEEA In the ecosystem services literature, the dominant sequestration. Eventually, a compromise was approach to the monetary valuation of carbon is found for SEEA EA, which considers global to apply monetary values to carbon sequestration. climate regulation as a single service consisting This is, however, problematic from an accounting of a sequestration and a retention component: perspective when ecosystems are net emitters of “In principle, carbon retention and carbon carbon, as this would lead to a negative service flow sequestration components should be measured for and hence negative production. Moreover, a strict all ecosystem assets. In practice, it is likely that focus on sequestration may lead to counterintuitive different ecosystem assets will provide different outcomes: for instance, old-growth ecosystems contexts for measurement. In stable ecosystems, that may be near equilibrium (and hence provide carbon retention will be the primary component relatively low or even zero sequestration) would while in those ecosystems where there is clear receive low asset values, undermining the case for expansion in the stock of carbon, then carbon their conservation. sequestration may be the focus of measurement” (NCAVES and MAIA 2022). To address these issues, an alternative framing was developed for the SEEA EA revision process, focusing Arguably, for the majority of developing countries instead on carbon storage as the main ecosystem carbon retention will be the primary component, service. As the ecosystem service (in monetary which is why this was implemented as the default units) would need to be expressed as an annual flow, in ARIES for SEEA. For the monetary valuation, one this proposal was described as “carbon retention” to needs two key inputs: an appropriate price for each set it apart from carbon storage. Essentially, carbon tonne of carbon retained and a discount rate. retention in physical units consists of a stock of carbon in the various carbon pools (for example, As described in Edens et al. (2019), there are various aboveground biomass + belowground biomass + soil ways to determine the price of carbon: organic carbon). In monetary units, this physical • An estimate based on the value stock is first multiplied by a suitable carbon price of damages avoided to estimate the avoided damages and subsequently • The marginal costs of abatement of carbon turned into an annual service flow by multiplying this number by a rate of return to obtain an annuity. • Observed market prices. While the proposal received broad support, it was As the carbon retention framing is based on the idea also criticized for failing to account for the fact that of avoided damages, the valuation report (NCAVES different ecosystems vary in their risk of release, and MAIA 2022) recommends applying a social cost and that a focus on carbon storage would be out of of carbon: “Different Tiers may be distinguished touch with policy contexts that focus on (increasing) depending on the sophistication of the model used 12 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS M O N E TA R Y VA L U AT I O N O F C A R B O N R E T E N T I O N I N A R I E S F O R S E E A l for deriving the [social cost of carbon]. A Tier 3 The default ARIES for SEEA model (inspired by approach may consist of using a national model Turpie, Letley, et al. 2021) applies a 3.66 percent (such as Brucknall et al. 2021). A Tier 1 approach discount rate based on Kotchen et al. (2019) and a may consist in choosing a value from the scientific 100-year time horizon. The price is adjusted 3 percent literature.” The ARIES for SEEA model, as a Tier 1 upward/downward for years later/earlier than 2015 approach, has opted for the latter. (as suggested by Nordhaus 2017). For this project, the monetary value of carbon storage is expressed While multiple studies on the social cost of carbon in constant 2020 US dollars, and discounted at a exist, ARIES for SEEA has opted to use a value of 4 percent rate over time to be consistent with CWON $33.70 per tCO2 based on Nordhaus (2017). The practices for other resource valuations. social cost of carbon measures the net present value of future damage costs calculated for a long time This approach has been implemented to produce horizon. As ecosystem services are (annual) flows, homogenous results on a global scale. This allows we need to turn the avoided damage costs into an for direct comparison of monetary valuations across annual service flow. The ARIES for SEEA model does the globe, reflecting the differences in biophysical this based on the approach from Turpie, Forsythe, terms across years and countries. If the valuation et al. (2021) by annualizing the social cost of carbon moves to a country-specific economic valuation, as follows: considering regional markets, local economic conditions, and other social aspects, results can A S C C   =   ( δ × S C C ) × [ 1 - ( 1 + δ ) -t] -1 diverge greatly. in which: ASCC stands for annualized social carbon cost, δ is the discount rate, SCC stands for social carbon cost, and t is the number of years. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 13 l R E C O M M E N DAT I O N F O R F U T U R E D E V E LO P M E N T S A N D I M P ROV E M E N T S 4 Recommendation for Future Developments and Improvements Several developments could be considered for land-based conditions. In addition, using the producing the next generation of carbon storage 2019 IPCC refinement introduces several cross- results for CWON. ARIES by design supports cutting improvements. For instance, it allows a re-running of updated models that reflect measurement of the uncertainty associated with improvements, incorporating updated knowledge certain IPCC parameters. into the platform as it becomes available. It should be noted that the integration of the 2019 The model currently uses default factors and IPCC refinement into the model will not replace approaches based on the 2006 IPCC guidelines (IPCC the 2006 IPCC guidelines. Instead, users will have 2006), which is the reference officially adopted for the option to choose between these two approaches reporting and reviewing under the international depending on which is the most suitable for their reporting framework of greenhouse gas (GHG) purposes. inventories and related information that is framed by the United Nations Framework Convention on In addition, work is ongoing to produce in-house Climate Change (UNFCCC). results for vegetation carbon stock estimates using the 2006 IPCC guidelines. The model intrinsically However, during the 49 session of the IPCC, held in th uses these guidelines, because the data produced May 2019, an update of these guidelines was adopted by Gibbs and Ruesch (2008) are based on the 2006 (IPCC 2019) to incorporate more recent science- IPCC guidelines. However, it was decided that using based knowledge. These guidelines, hereafter available expertise on land use, land-use change, named the IPCC 2019 refinement, are expected to and forestry to produce results by implementing soon replace the 2006 IPCC guidelines as the official the 2006 IPCC guidelines from scratch would lead source of data that will govern the international to improved results. Specifically, it is expected that reporting framework of GHG information. these refinements will: To align the current model with this more recent • Avoid the uncertainty of incorporating inputs development and to meet future users’ demands, the pre-processed by third parties that are prone to ARIES team is collaborating with land use, land-use misinterpretation due to difficulties in tracking change, and forestry experts to update the model to the entire data production process. reflect the 2019 IPCC refinement. • Allow for verification by comparing final outcomes and the model results, and, when Incorporating data from the 2019 IPCC refinement relevant, assist with identifying artifacts and areas into the model offers the possibility of selecting for further refinement. more disaggregated input data. Without losing sight of the unavoidable uncertainty associated • Offer the possibility of providing users with with default factors, the new data are expected to transparent information on each step of the produce more accurate results for representing data production process. 14 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS R E C O M M E N DAT I O N F O R F U T U R E D E V E LO P M E N T S A N D I M P ROV E M E N T S l TABLE 4.1: CARBON STRATIFICATION CARBON IPCC CARBON ROOT TO CARBON EMISSION STOCK (GIBBS & ABOVEGROUND FRACTION SHOOT RATIO COMPUTED (+) RUESCH 2008) BIOMASS 200 0.47 0.37 310 199.61 -731.9 193 0.47 0.37 300 193.17 -708.29 180 0.47 0.37 280 180.29 -661.071 225 0.47 0.37 350 225.37 -826.338 200 Ave Ave Ave 199.50 -731.5 Finally, for the sake of transparency, and without the most reliable approach to model the effects of modifying the results based on using Gibbs and natural disturbances including fires, which play Ruesch (2008) inputs, disaggregated information an important role in the carbon cycle, discussions about the calculation steps followed by the authors to are ongoing with experts from the SEEA EA Forest derive carbon stock values will be included. Table 4.1 Working Group, as well as experts who developed shows a practical example of the implementation of the International Union for Conservation of Nature this development for carbon stratification. (IUCN) Global Ecosystem Typology. As mentioned above, efforts are being made to As mentioned previously, UNSD and the World include, to the extent possible, complete information Bank also plan to reassess the monetary valuation of the uncertainty associated with the model inputs. approach. This represents an important improvement that is recognized by the UNFCCC as a good practice A separate analysis of the carbon stock contribution for reporting data. However, due to the lack of of mangroves is also available, but it is currently data for some sources of uncertainty, and partial not comparable with the results produced due to information for others, any attempt to propagate the different spatial resolution (level of detail of the this uncertainty in order to measure the uncertainty spatial information used) and temporal scales (time of model results is deemed meaningless. period and the frequency of the observations) of Lastly, approaches are being discussed to improve this study (for more information, refer to Appendix how the model treats the post-fire carbon dynamics. 2). Depending on the demand for the inclusion of Although for stands that have reached maturity, it specific models to estimate these results, and on the can be assumed that the carbon stock is in a steady availability of data, mangroves are also expected to balance over time, all the changes, either naturally or be computed as a separate model in the future, while human-induced, produce dynamic effects that need preserving consistency with the results obtained in to be considered. With the intention of selecting this analysis. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 15 l D I S C U S S I O N O F R E S U LT S 5 Discussion of Results The produced maps of vegetation carbon stocks are 5.1 VEGETATION CARBON broken down by the 36 ESA-CCI land cover classes and STOCK RESULTS AGGREGATED aggregated at the country level to provide information BY COUNTRY on carbon stocks for given years and their changes over time. This section presents a summary of results Several graphs show country-level results: for the country-level data. The figures below serve as • Figure 5.1A depicts the status of vegetation examples of the results that can be obtained with this carbon stock, by year, at the country level dataset and as inspiration for further, more in-depth • Figure 5.1B shows the relative change analysis. All the code run to obtain the results from of vegetation carbon stocks. the vegetation carbon stock maps, the main outputs of this analysis, is also made available in this public Note that the carbon stocks depicted in the map are repository. Using this code, users can replicate the in megatonnes, and not in tonnes per hectare, thus outcome or obtain other statistics of interest. the absolute value of the whole country is ranked, without taking into account how large or small its area Since not all changes are caused by a change in the is. Intuitively, larger countries have more land area, land cover data, Appendix 3 provides examples and which results in a greater capacity for carbon storage. the interpretation of changes resulting from different In Figure 5.1, small countries with large carbon stock causes. Appendix 4 provides an overview of all the densities can thus show small carbon stocks, as seen large changes observed over the time series, and for for Ecuador and some Central African and East Asian each case, describes the main drivers for that result. countries. Figure 5.1B shows that most of the world’s Whenever the change of vegetation carbon stock for vegetation carbon stock is concentrated in large a country was mainly driven by the lack of input data, countries or tropical countries, with Brazil being both this was excluded from the graph, and the next country vast and tropical, accounting for almost 20 percent of ranking for the largest change (increase or decrease the world’s stock. of vegetation carbon stock) was instead shown in the figure. The last section presents the results for the soil organic carbon in 2020. 16 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l FIGURE 5.1A: COUNTRY-LEVEL GLOBAL MAP (MEGATONNES) Total global vegetation carbon stock (Mt), 2020 FIGURE 5.1B: DISTRIBUTION OF THE VEGETATION CARBON STOCK AMONG THE TOP 10 COUNTRIES 2020 Brazil Democratic Republic of Congo Indonesia Russian Federation United States Australia China Canada Peru Colombia 0.0 0.2 0.4 0.6 0.8 1.0 Vegetation carbon stock (Mt) 1e5 It is also useful to visualize the density of the carbon regulating service provided, expressed in tonnes per hectare (as in Figure 5.2). This map can be used to identify hotspots with high carbon storage density. The map below shows the vegetation carbon density for the year 2020. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 17 l D I S C U S S I O N O F R E S U LT S FIGURE 5.2: VEGETATION CARBON DENSITY MAP FOR 2020 (TONNES/HECTARE) The following graphs illustrate the dynamics of initial year is depicted. Four out of the five countries vegetation carbon stock for the countries that with the largest decrease in vegetation carbon stock experienced the greatest vegetation carbon stock are among the 10 countries with the largest stocks: declines and increases, in absolute terms, between Brazil, Democratic Republic of Congo, the United 2001 and 2020. The curves of absolute change look States, and Russia. Peru is one of the 10 countries flat because of the disparate scales of carbon stock with the largest stocks, and the only country in across countries. The yearly variability is better this group with increasing stocks. This is aligned observed in the relative change curves, as all the with the global decline of vegetation carbon stock stocks are rescaled and their evolution from the observed in the final results. FIGURE 5.3A: TOTAL VEGETATION CARBON STOCK DECLINE BY COUNTRY (MEGATONNES) Top 5 countries decreasing their vegetation carbon stock (Mt), 2001–2020 80,000 Vegetation carbon stock (Mt) 60,000 40,000 20,000 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Democratic Republic of Congo United States Paraguay Brazil Russian Federation 18 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l FIGURE 5.3B: TOTAL VEGETATION CARBON STOCK DECLINE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries decreasing their vegetation carbon stock (% change from 2001), 2001–2020 0 –2 Relative change (%) –4 –6 –8 –10 –12 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Democratic Republic of Congo United States Paraguay Brazil Russian Federation FIGURE 5.3C: TOTAL VEGETATION CARBON STOCK INCREASE BY COUNTRY (MEGATONNES) Top 5 countries increasing their vegetation carbon stock (Mt), 2001–2020 14,000 Vegetation carbon stock (Mt) 12,000 10,000 8,000 6,000 4,000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Ethiopia Nigeria Peru Philippines GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 19 l D I S C U S S I O N O F R E S U LT S FIGURE 5.3D: TOTAL VEGETATION CARBON STOCK INCREASE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries increasing their vegetation carbon stock (% change from 2001), 2001–2020 3 2 1 Relative change (%) 0 –1 –2 –3 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Ethiopia Nigeria Peru Philippines 5.2 VEGETATION CARBON STOCK While artificial surfaces, waterbodies, and RESULTS DISAGGREGATED BY permanent snow and ice do not contribute to vegetation carbon storage, the other aggregated LAND COVER CLASS classes’ contribution over the time series is The results were also stratified by land cover class to summarized below. As expected, forest ecosystems observe the contribution of each individual class to are the greatest contributor to global vegetation the vegetation carbon stock. carbon stocks, accounting for more than 70 percent. Shrub and herbaceous vegetation, which includes Since the classification of land cover contains 36 grasslands, shrublands, and savannas, accounts classes, which would be cumbersome to visualize, for about 14 percent of vegetation carbon stock. we grouped those into six aggregated classes: Agricultural vegetation covers an extensive area, • Agricultural vegetation but its carbon density is much lower, and accounts for about 9 percent. Wetlands are ranked fourth • Forest for contribution, accounting for about 5 percent • Shrub and herbaceous vegetation of vegetation carbon stock. Wetland carbon stocks • Wetland are substantially greater when also accounting for soil organic carbon, which is an important part of • Sparse vegetation wetland carbon stocks. Sparse vegetation and bare • Bare area. area contribute less than 0.5 percent each. 20 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l TABLE 5.1: RELATIVE CONTRIBUTION TO VEGETATION CARBON STOCK BY AGGREGATED CLASS OVER THE WHOLE TIME SERIES (2001–2020 TOTAL VEGETATION CARBON STOCK) CONTRIBUTION TO GLOBAL VEGETATION AGGREGATED CLASSES CARBON STOCK (%) Forest 70.50% Shrub and herbaceous vegetation 14.18% Agricultural vegetation 9.29% Wetland 5.31% Bare area 0.40% Sparse vegetation 0.33% Artificial surface 0.00% Waterbody 0.00% Permanent snow and ice 0.00% FIGURE 5.4: DISTRIBUTION OF VEGETATION CARBON (%) IN 2020 Forests 70.2% Shrub and herbaceous 14.29% vegetation Agricultural 9.37% vegetation Wetland 5.4% Bare area 0.39% Sparse vegetation 0.34% 0 1 2 3 4 5 6 7 Vegetation carbon stock (%) 1e1 The relative contribution of the vegetation carbon stock of the aggregated classes is described in the following sections, which analyze class-level changes in greater detail. Overall, from 2001 to 2020, forest vegetation carbon stocks decrease, while shrub and herbaceous vegetation and agricultural vegetation stocks increase, but not by enough to compensate for forest carbon stock declines. The distribution across categories, though, stays relatively stable over the time series, and only small changes, in relative terms, are observed. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 21 l D I S C U S S I O N O F R E S U LT S 5.2.1 Forest Vegetation Carbon Stock In the following figure, the vertical scale, representing the vegetation carbon stock, has been compressed to focus on the interval between the maximum and minimum values observed in the time series, so that the bars are not completely flat. FIGURE 5.5A: FOREST GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 371,467 Mt 371,000 Vegetation carbon stock (Mt) 370,000 369,000 367,922 Mt 367,965 Mt 367,934 Mt 368,000 367,000 366,770 Mt 366,000 2001 2005 2010 2015 2020 Year FIGURE 5.5B: FOREST GLOBAL VEGETATION CARBON STOCK (%) Vegetation carbon stock change from 2001 (in %) 0.5 –0.95% –0.94% –0.95% –1.26% 0.0 –0.5 –1.0 –1.5 2005 2010 2015 2020 Year 22 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l The global forest vegetation carbon stock decreased by almost 1.3 percent over the period analyzed. The greatest decline occurred from 2001 to 2005. Global forest carbon stocks stayed relatively constant from 2005 to 2015 before declining further from 2015 to 2020. FIGURE 5.5C: CONTRIBUTION TO FOREST GLOBAL VEGETATION CARBON STOCKS FOR 2001–2020 (MEGATONNES) Forest global vegetation carbon stock (Mt), 2001 Forest global vegetation carbon stock (Mt), 2020 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 23 l D I S C U S S I O N O F R E S U LT S FIGURE 5.5D: FOREST VEGETATION CARBON STOCK DECLINE BY COUNTRY (MEGATONNES) Top 5 countries decreasing their forest vegetation carbon stock (Mt), 2001–2020 70,000 60,000 Vegetation carbon stock (Mt) 50,000 40,000 30,000 20,000 10,000 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Argentina United States Paraguay Brazil Russian Federation FIGURE 5.5E: FOREST VEGETATION CARBON STOCK DECLINE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries decreasing their forest vegetation carbon stock (% change from 2001), 2001–2020 0 –5 Relative change (%) –10 –15 –20 –25 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year United States Paraguay Brazil Argentina Russian Federation Aside from Argentina, all countries recording the largest declines of forest vegetation carbon stocks also experienced the greatest declines in total vegetation carbon stocks. This shows the importance of forests for vegetation carbon storage. 24 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l FIGURE 5.5F: FOREST VEGETATION CARBON STOCK INCREASE BY COUNTRY (MEGATONNES) Top 5 countries increasing their forest vegetation carbon stock (Mt), 2001–2020 12,000 Vegetation carbon stock (Mt) 10,000 8,000 6,000 4,000 2,000 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Ethiopia Nigeria Peru Philippines FIGURE 5.5G: FOREST VEGETATION CARBON STOCK INCREASE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries increasing their forest vegetation carbon stock (% change from 2001), 2001–2020 12.5 10.0 7.5 Relative change (%) 5.0 2.5 0.0 –2.5 –5.0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Ethiopia Nigeria Peru Philippines Collectively, Figures 5.3 and 5.5 show that deforestation is the main driver of global vegetation carbon stock loss. As shown in the next section, agricultural and shrub and herbaceous vegetation areas and carbon storage increased from 2001 to 2020, suggesting that these expansions come at the expense of forests. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 25 l D I S C U S S I O N O F R E S U LT S 5.2.2 Shrub and Herbaceous time series, so that the bars are not completely flat. Vegetation Carbon Stock Figure 5.6A shows a clear trend of constant growth In the following figure, the vertical scale, for the carbon stored in shrub and herbaceous representing the vegetation carbon stock, has been vegetation, which increased by slightly less than compressed to focus on the interval between the 1,000 megatonnes, or about 1.35 percent of its initial maximum and minimum values observed in the value in 2001. FIGURE 5.6A: SHRUB AND HERBACEOUS GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 75,000 74,678 Mt Vegetation carbon stock (Mt) 74,500 74,236 Mt 74,057 Mt 74,000 73,783 Mt 73,684 Mt 73,500 73,000 2001 2005 2010 2015 2020 Year FIGURE 5.6B: SHRUB AND HERBACEOUS GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 2001 26 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l A land cover transitional analysis could correctly does not replace forests. Because shrub and determine which shrub and herbaceous vegetation herbaceous vegetation is largely replacing forests, expansion has net positive and negative effects on the above results contribute to an overall decrease total global vegetation carbon stocks. One must of vegetation carbon stock. determine which land cover type replaces another in order to interpret the results of changes in the Different countries contribute to the vegetation overall carbon stock. As mentioned at the end of carbon stock for shrub and herbaceous vegetation, the previous section, a positive overall increase notably Australia, Mexico, Argentina, as well as in vegetation carbon stock only occurs when the countries in Sub-Saharan Africa and in the Eastern expansion of shrub and herbaceous vegetation Asia region. FIGURE 5.6C: CONTRIBUTION TO SHRUB AND HERBACEOUS GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 2020 5.2.3 Agricultural Vegetation Agricultural vegetation carbon stocks increased glob- Carbon Stock ally by about 2 percent or almost 1,000 megatonnes, with nearly all this change occurring between 2001 In the following figure, the vertical scale, and 2005. Agricultural vegetation carbon stocks by representing the vegetation carbon stock, has been country are shown in Figures 5.7C–7E. compressed to focus on the interval between the maximum and minimum values observed in the time series, so that the bars are not completely flat. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 27 l D I S C U S S I O N O F R E S U LT S FIGURE 5.7A: AGRICULTURAL GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 49,000 48,944 Mt 48,896 Mt 48,920 Mt 48,867 Mt Vegetation carbon stock (Mt) 48,500 48,000 47,955 Mt 47,500 47,000 2001 2005 2010 2015 2020 Year FIGURE 5.7B: AGRICULTURAL GLOBAL VEGETATION CARBON STOCK (%) 3.50 3.25 Vegetation carbon stock change 3.00 from 2001 (in %) 2.75 2.50 2.25 2.01% 2.06% 1.96% 1.90% 2.00 1.75 1.50 2005 2010 2015 2020 Year FIGURE 5.7C: CONTRIBUTION TO AGRICULTURAL VEGETATION CARBON STOCK (MEGATONNES) 2020 28 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l FIGURE 5.7D: AGRICULTURAL VEGETATION CARBON STOCK INCREASE BY COUNTRY (MEGATONNES) Top 5 countries increasing their agricultural vegetation carbon stock (Mt), 2001–2020 8,000 7,000 Vegetation carbon stock (Mt) 6,000 5,000 4,000 3,000 2,000 1,000 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Brazil Bolivia Argentina Malaysia Indonesia FIGURE 5.7E: AGRICULTURAL VEGETATION CARBON STOCK INCREASE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries increasing their agricultural vegetation carbon stock (% change from 2001), 2001–2020 14 12 10 8 Relative change (%) 6 4 2 0 –2 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Brazil Bolivia Argentina Malaysia Indonesia GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 29 l D I S C U S S I O N O F R E S U LT S Intuitively, agricultural vegetation carbon stock carbon stock declines, it had the same pattern as increases are usually associated with decreases Brazil: the extent of agricultural land increased by in the total vegetation carbon stock, because 10 percent, while agricultural carbon increased agricultural expansion comes at the expense of land by 16 percent. However, agricultural carbon only cover types with greater per-area carbon storage. accounts for about 3 percent of the total carbon This is clear for the case of Brazil when analyzing stock in the country, and this increase, estimated at graphs of the total decrease of carbon vegetation 0.5 megatonnes, did not offset the total vegetation and the increase of agricultural vegetation. The carbon stock loss in that year, which was estimated drop in the first graph, representing forest carbon at 1.3 megatonnes. storage, takes place over the period 2002–2003, the same period in which the latter, illustrating The opposite effect is observed when natural agricultural carbon storage, increases significantly. vegetation replaces agricultural vegetation, which This is also the case for Bolivia, where a similar trend usually stores greater carbon content. So a reduction occurs from 2002 to 2003. Although Bolivia does not in this aggregated class can also contribute to an rank among the top countries for total vegetation increase in the overall level of vegetation carbon TABLE 5.2: AGRICULTURAL CARBON STOCK ANALYSIS FOR BOLIVIA OVER THE PERIOD 2002–2003 CARBON MOSAIC TREE OR TOTAL TOTAL MOSAIC HERBACEOUS STORAGE NATURAL CROPLAND SHRUB AGRICULTURAL CARBON CROPLAND COVER (t) VEGETATION COVER CARBON STOCK STOCK Year 2003 1,195,260 1,802,835 44,915 127,260 10 3,170,279 106,963,909 Year 2004 1,310,846 2,179,271 52,445 128,545 15 3,671,122 105,633,346 Net change 115,587 376,437 7,530 1,285 5 500,843 -1,330,563 % change 10% 21% 17% 1% 50% 16% -1% MOSAIC AREA MOSAIC HERBACEOUS TREE OR NATURAL CROPLAND TOTAL (km²) CROPLAND COVER SHRUB COVER VEGETATION Year 2003 15,771.94 21,846.96 8,645.92 24,484.76 1.93 70,751.51 Year 2004 17,088.51 26,019.14 10,103.22 24,732.96 2.9 77,946.73 Net change 1,316.57 4,172.18 1,457.31 248.2 0.97 7,195.23 % change 8% 19% 17% 1% 50% 10% Vegetation 88 90 5 5 5 carbon storage density (t/ha) 30 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l stock. Clearly, a reduction of agricultural carbon effect on the overall vegetation carbon stock, since stock substituted by the other aggregate classes, the already relatively low carbon stored in crops such as artificial surface (urbanization) or bare disappears completely without being replaced by area (desertification), has an even more severe any type of vegetation. FIGURE 5.7F: AGRICULTURAL VEGETATION CARBON STOCK DECLINE BY COUNTRY (MEGATONNES) Top 5 countries decreasing their agricultural vegetation carbon stock (Mt), 2001–2020 4,500 4,000 3,500 Vegetation carbon stock (Mt) 3,000 2,500 2,000 1,500 1,000 500 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Democratic Republic of Congo China India Türkiye FIGURE 5.7G: AGRICULTURAL VEGETATION CARBON STOCK DECLINE BY COUNTRY (IN RELATIVE TERMS) Top 5 countries decreasing their agricultural vegetation carbon stock (% change from 2001), 2001–2020 0 –2 –4 Relative change (%) –6 –8 –10 –12 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 2019 2020 Year Tanzania Democratic Republic of Congo China India Türkiye GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 31 l D I S C U S S I O N O F R E S U LT S 5.2.4 Wetland Vegetation Carbon Stock clear temporal trends globally. Carbon stocks were stable from 2000 to 2005 and 2010 to 2015, declined In the following figure, the vertical scale, from 2005 to 2010, and increased from 2015 to 2020. representing the vegetation carbon stock, has been This could indicate ecosystem recovery in recent compressed to focus on the interval between the years or accuracy issues in detecting wetlands at maximum and minimum values observed in the global scales. Future work can shed further light on time series, so that the bars are not completely flat. the evolution of wetland vegetation carbon stocks at Wetland vegetation carbon stock does not show the global scale. FIGURE 5.8A: WETLAND GLOBAL VEGETATION CARBON STOCK (MEGATONNES) 28,750 28,500 28,392 Mt 28,389 Mt Vegetation carbon stock (Mt) 28,250 28,228 Mt 28,000 27,985 Mt 27,967 Mt 27,750 27,500 27,250 27,000 2001 2005 2010 2015 2020 Year FIGURE 5.8B: WETLAND GLOBAL VEGETATION CARBON STOCK (%) 1.0 Vegetation carbon stock change 0.5 from 2001 (in %) –1.50% –0.58% –0.01% –1.44% 0.0 –0.5 –1.0 –1.5 2005 2010 2015 2020 Year 32 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS D I S C U S S I O N O F R E S U LT S l Over the entire period, there is a loss of about 0.6 by wetland vegetation roots and rhizomes. Such percent from the 2001 vegetation carbon stock level matter decomposes slowly due to anaerobic and to the 2020 level. An increase or decrease of this size poorly drained wetland soils. Thus, this category should have a straightforward positive or negative contributes more significantly to the total carbon impact on the overall vegetation carbon stock. stock, but the vegetation carbon model does not capture this effect. The next generation of carbon Notably, the above results exclude the contribution stock models could account specifically for this to the carbon stock from soil organic carbon, which phenomenon, completing the information on soil accumulates greatly in the wetland subsoil and organic carbon, and increasing figures for wetlands, plays an important role in global carbon stocks. which are underestimated when only considering The accumulation is caused by sediment trapping vegetation carbon stock. FIGURE 5.8C: CONTRIBUTION TO WETLAND GLOBAL VEGETATION FOR 2001–2020 Wetland global vegetation carbon stock (Mt), 2001 Wetland global vegetation carbon stock (Mt), 2020 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 33 l D I S C U S S I O N O F R E S U LT S 5.3 SOIL ORGANIC CARBON are inherently different and unrelated, these results RESULTS DISAGGREGATED must be carefully interpreted. Inconsistencies are BY LAND COVER CLASS expected due to the very different geographical scales and the methodologies used to build them. The results for soil organic carbon in 2020, While soil organic carbon has been estimated by disaggregated by land cover class, are produced as interpolating soil samples collected on the ground, additional information to this work. Since there the ESA-CCI dataset is being obtained from the are no data available to produce a long time series, interpretation of satellite images. Moreover, the ESA- the most recent results, obtained using the best CCI land cover map is not a source of information techniques, are summarized in the tables https:// used in computing soil organic carbon, whereas data.integratedmodelling.org/dataset/soil-organic- it was a primary input in the computation of the carbon-stock-2020. vegetation carbon stock. The disaggregation by land cover class has been Nevertheless, the disaggregation of this map obtained by superimposing the soil organic carbon and the 2020 ESA-CCI land cover maps. The latter can provide useful information when correctly map (ESA-CCI dataset) was used as a geographical interpreted, and the results of the aggregation of reference to identify which land cover class would all land cover classes’ contributions to soil organic correspond to carbon stocks in the soil organic carbon for a country can more confidently be carbon, but since the two products superimposed considered the correct estimates for that region. 34 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S l 6 Comparative Assessment of Vegetation Carbon Stock Results 6.1 INTRODUCTION their information, national submissions undergo expert reviews or technical assessments before In line with the good practices from the 2006 IPCC the international reporting frameworks make the guidelines (volume 1, chapter 6) on implementing country submissions publicly available. quality assurance, quality control, and verification, this section provides a comparative assessment of For developing countries, the comparative model results with country-specific information. assessment used information included in the last The IPCC recognizes: “Discrepancies between submission of forest reference emissions levels alternative methods do not necessarily imply that under the UN REDD+ 24 program. For developed the inventory data are in error. When analyzing countries, the main data source was the latest discrepancies, it is important to consider that there GHG inventory submission under the UNFCCC.25 may be large uncertainties associated with the Moreover, information from FAO’s Global Forest alternative calculations themselves.” Resources Assessments 26 was used in the comparison for certain parameters or used as an auxiliary data This section, which presents the results of the source when the other sources lacked information comparative assessment, aims to: needed for the comparison. • Enhance the understanding of the model’s outputs The assessment focuses on four countries: Gabon, • Identify potential areas for future improvement Mexico, Finland, and Papua New Guinea. These countries were selected based on the availability • Contribute to future developments of country-specific data under the international in the monitoring, reporting, and reporting frameworks (with a view to covering verification of emissions and removals different ecoregions) and reporting capacities. from the land-based sector. Attention was also given to two criteria: country Model results were compared with publicly available area and degree of forest heterogeneity. 27 The country-specific forest-related information included use of these criteria is intended to facilitate the in submissions by countries under international understanding of the effects, and potential artifacts, reporting frameworks. In addition to the quality that the country area and different forest types assurance and quality control procedures considered in national submissions might have on implemented by the countries before submitting the data comparison. 24 https://redd.unfccc.int . 25 https://unfccc.int/ghg-inventories-annex-i-parties/2022 . 26 https://www.fao.org/forest-resources-assessment/fra-2020/country-reports/en/ . 27 Based on expert judgement. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 35 l C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S Despite the attempt to minimize artifacts, the impact • Land use versus land cover: Whether national of some elements cannot be filtered out. And, as with forest monitoring systems are land cover- or any other comparison, the meaningfulness of results land use-oriented impacts the real meaning of is a function of the comparability of the elements the forest land information. A good example that are being compared. In this comparative is when the extent of deforested areas is assessment, the degree of comparability of areas overestimated by including unstocked areas and carbon stocks is also closely linked to the (resulting from management practices) where, internal consistency across time and space of the although the forest coverage is reduced, sources used to obtain country data for this section. the land use has not changed and the forest on the land is expected to regenerate. The scientific community has recently given attention to the challenges of comparing forest-related • Consistency of forest monitoring systems: information and the need to bridge differences In line with the comment above, how when setting science-based strategies to combat the countries identify and classify lands across adverse effects of climate change, using comparable time and space also has an impact on the background information (Grassi et al. 2018). meaning of the forest information reported. Implementing different methods across a A case-by-case analysis of the results, which goes time series of data hampers the internal beyond the purpose of this section, may point out consistency of the information. An example specific drivers of differences. Common reasons for of this effect is when a country uses remote- the differences in forest information across data sensing products with different resolutions sources, which could help to correctly interpret the to produce a time series of forest areas. results, include the following: Although this list is not comprehensive, it shows real • Definition of forest: In principle, most countries examples from country submissions to international count on an official forest definition, which should frameworks, highlighting the challenges of determine the areas classified as “forest land” by comparing forest information. In some cases, the the forest monitoring systems. In practice, these lack of transparency in the submissions also impedes official definitions are often modified to fulfill the a full understanding of the information needed to reporting requirements of international reporting better grasp differences among data sources. frameworks. In some cases, these definitions are also modified because they are not technically Therefore, the results of this comparative feasible to implement due to the current methods assessment should not be used as an indicator of the used by countries to obtain forest information transparency, accuracy, consistency, comparability, (for example, minimum area adopted for and completeness of any of the datasets involved, defining forest versus minimum mapping including our model results. units used in the land monitoring system). The use of different forest definitions in a single country is often the main driver of differences when the information for the same parameter is compared across official submissions. 36 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S l 6.2 COMPARATIVE ASSESSMENT information or across different submissions, when OF AREAS the land representation system used for the datasets is different, or when the scope of the information is Table 6.1 includes information on the total area of not the same. For example, a country may include countries selected (Gabon, Mexico, Finland, and or exclude information related to small islands, Papua New Guinea) for this assessment. Further overseas territories, or territories occupied in a information on areas and other parameters used particular submission. in this section for each country can be found in Appendix 1 of this report. Information on country The differences found when comparing total areas refers to the year 2020 when available, or to the country area among datasets are acceptable as their most updated reference found in the data sources. potential impact on the model results is minimal. It is expected that the total country area remains In all cases, the total area used in the model differs constant over time. In practice, comparing country from national submissions by less than 5 percent. areas across national submissions can reveal In addition, while official information on total different values for the extension of the national area submitted by countries is inconsistent among territories due to boundary disputes. data sources, the model runs with a constant and In some cases, natural effects may change total consistent total country area for the entire time country area (for example, coastal erosion), but series. Nevertheless, information on country area the differences would be captured only when represents the sum of different land classes. As such, comparing references over a long time span. the value from each dataset intrinsically balances Commonly, technical issues are found that explain out the effect of using inconsistent definitions of the differences in national areas. For example, the land classes among data sources. By contrast, a a single country may report inconsistent land consistent classification of lands plays a key role in area across the years for the same submission of comparing areas when only one category is involved. TABLE 6.1: COMPARISON OF INFORMATION ON TOTAL COUNTRY AREA SUBMITTED BY COUNTRIES WITH MODEL DATA TOTAL AREA (KILOHECTARES) COMPARISON DIFFERENCE AVERAGE COUNTRY (%) MODEL VS. VALUE OF FAO-FRA REDD+ / UNFCCC MODEL RESULTS AVERAGE OF OFFICIAL OFFICIAL DATA SUBMISSIONS SOURCES Gabon 25,767 26,767 26,511 26,267 1% Mexico 194,395 194,078 196,854 194,236 1% Finland 30,391 33,843 33,375 32,117 4% Papua New Guinea 45,286 46,100 46,600 45,693 2% GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 37 l C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S Table 6.2 compares the forest areas included in official submissions by countries with model results. The table also includes parameters to facilitate the interpretation of the comparison in terms of the level and trend of results. TABLE 6.2: COMPARISON OF INFORMATION ON FOREST AREA SUBMITTED BY COUNTRIES WITH MODEL DATA FOREST AREA (KILOHECTARES) GABON 2000 2010 DELTA 2015 DELTA 2020 DELTA FAO-FRA 23,700 23,649 -0.2% 23,590 -0.3% 23,531 -0.3% Forest land 23,700 23,649 -0.2% 23,590 -0.3% 23,531 -0.3% Other woody lands 0 0 NA 0 NA 0 NA REDD+ (*) 23,663 23,600 -0.3% 23,546 -0.2% 23,523 -0.1% Model 23,896 23,823 -0.3% 23,817 0.0% 23,830 0.1% (*) Values included in table 6 of the Forest Reference Level submission. The value in the last column refers to the year 2018. MEXICO 2000 2010 DELTA 2015 DELTA 2020 DELTA FAO-FRA 88,906 86,932 -2.2% 86,216 -0.8% 85,525 -0.8% Forest land 68,381 66,943 -2.1% 66,331 -0.9% 65,692 -1.0% Other woody lands 20,525 19,989 NA 19,885 NA 19,833 -0.3% REDD+ (*) 101,269 99,561 -1.7% 98,061 -1.5% 97,711 -0.4% Model 76,796 75,973 -1.1% 75,617 -0.5% 75,464 -0.2% (*) Values included in table 26 of the Forest Reference Level submission. The value in the last column refers to the year 2016. FINLAND 2000 2010 DELTA 2015 DELTA 2020 DELTA FAO-FRA 23,269 23,031 -1.0% 23,155 0.5% 23,155 0.0% Forest land 22,446 22,242 -0.9% 22,409 0.8% 22,409 0.0% Other woody lands 823 789 NA 746 NA 746 0.0% UNFCCC-GHGI (*) 22,106 21,943 -0.7% 21,885 -0.3% 21,849 -0.2% Model 22,317 22,517 0.9% 22,491 -0.1% 21,816 -3.0% (*) Values included in the 2022 GHGI submission (CRF table 4.1 for specific years). 38 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S l TABLE 6.2: CONT. PAPUA NEW 2000 2010 DELTA 2015 DELTA 2020 DELTA GUINEA FAO-FRA 36,278 36,179 -0.3% 36,024 -0.4% 35,856 -0.5% Forest land 36,278 36,179 -0.3% 36,024 -0.4% 35,856 -0.5% Other woody lands 0 0 NA 0 NA 0 NA REDD+ (*) 36,051 36,051 0.0% 36,051 0.0% 36,051 0.0% Model 40,659 41,257 1.5% 41,282 0.1% 41,114 -0.4% (*) Values included in figure 7.1 of the Forest Reference Level submission. The values refer to the year 2013. For most of the case studies, the comparison shows 6.3 ISSUES WITH LAND that the model matches the country data. However, CLASSIFICATION SYSTEMS it also raises some differences that would need to be individually analyzed to explain their origin, The model uses information from the ESA-CCI for instance, the forest area reported by Papua to classify land cover. This data source allows the New Guinea to FAO and under REDD+ and the area construction of worldwide complete and consistent classified as forest in this country by the model. time-series information on land cover classes across Nevertheless, a preliminary analysis of these results time and space. suggests that, as acknowledged by the countries, The land classification of the ESA-CCI does not different forest definitions drive differences in forest explicitly assign classes under a “forest land” areas across their submissions. For this reason, category. For this comparative assessment, it was the semantic differences between the model’s land necessary to merge the area of certain ESA-CCI land cover classes encoded as “forest land” and national cover classes to derive the area of a theoretical forest forest definitions seem to have a significant impact land category. To ensure a consistent comparison of on the result of this comparison and could, at least the model information on forest land area with the partly, explain the differences in forest areas. country data, it was decided to always use the sum of areas for the same ESA-CCI classes considered This problem could be greatly reduced if each forest lands without considering the country country could provide specific information on how being compared. This approach aims to avoid the the land cover classes in the global dataset differ uncertainties of a subjective selection of ESA-CCI from the national definition of forest or their land classes, based on which country is compared. cover classification. In practical terms, this problem cannot be solved due to the intrinsic nature of the Therefore, the only criteria used to select the global data, which provide a global perspective ESA-CCI land cover classes that conform with the that is insufficiently detailed to distinguish those area of the forest land used in this comparative characteristics. assessment is the likelihood of them satisfying GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 39 l C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S the minimum threshold values that are commonly Table 6.3 also indicates whether the forest definition used by countries for classifying forest lands. When used by countries in their submissions to REDD+ selecting the land cover classes, special attention or the UNFCCC is consistent with the one used for was given to the canopy cover value included in the their reporting under the FAO-FRA. The inclusion of legend of the ESA-CCI classification system. this information aims to improve the understanding of differences among data sources included in other Table 6.3 provides the forest definitions used by tables of this section, and increase awareness of countries in their reporting under the REDD+ the challenges of comparing data sources on forest- program or in their GHG inventory submission to related information (Grassi, Conchedda, et al. 2022; the UNFCCC. It also includes the ESA-CCI land cover Grassi, Schwingshackl, et al. 2022). classes that were classified as forest land for the purpose of comparing model data on forest areas with country data. TABLE 6.3: COMPARISON OF FOREST DEFINITION USED BY COUNTRIES WITH MODEL DATA DATA CONSISTENCY COUNTRY FOREST DEFINITION SOURCE WITH FAO-FRA Tree formation covering at least 30 percent of the soil over more than 1 hectare and more than 20 meters wide, with trees at least Gabon 5 meters high at maturity, but not subject to any agricultural UN REDD+ Not consistent practice. It does not include land that is predominantly under agricultural or urban land use. Forest lands with canopy cover greater than 10 percent, with Mexico woody species over 4 meters high or capable of reaching that UN REDD+ Not consistent height, and with a minimum surface of 1 hectare. The definition of forest that Finland employed for FAO’s FRA is applied in the GHG inventory to define forest land. The FAO definition for forest in FRA 2005 was: “Land spanning more than 0.5 hectares with trees higher than five meters and a canopy Finland cover of more than 10 percent, or trees able to reach these UNFCCC Mostly consistent thresholds in situ. It does not include land that is predominantly under agricultural or urban land use.” This definition was used for national data in the 2005 FRA report except for the requirement of the minimum area of 0.5 hectares. Land spanning more than 1 hectare, with trees higher than 3 Papua New meters and a canopy cover of more than 10 percent. This excludes REDD+ Not consistent Guinea land that is predominantly under agricultural or urban land use. 40 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S l DATA CONSISTENCY COUNTRY FOREST DEFINITION SOURCE WITH FAO-FRA The following ESA-CCI land cover classes are used to represent the most likely alignment with areas considered forest by the countries: Mosaic natural vegetation (tree, shrub, herbaceous cover) (>50%) / cropland (<50%) Tree cover, broadleaved, evergreen, closed to open (>15%) Tree cover, broadleaved, deciduous, closed to open (>15%) Tree cover, broadleaved, deciduous, closed (>40%) Tree cover, broadleaved, deciduous, open (15–40%) Tree cover, needleleaved, evergreen, closed to open (>15%) MODEL ESA-CCI28 NA Tree cover, needleleaved, evergreen, closed (>40%) Tree cover, needleleaved, evergreen, open (15–40%) Tree cover, needleleaved, deciduous, closed to open (>15%) Tree cover, needleleaved, deciduous, closed (>40%) Tree cover, needleleaved, deciduous, open (15–40%) Tree cover, mixed leaf type (broadleaved and needleleaved) Mosaic tree and shrub (>50%) / herbaceous cover (<50%) Tree cover, flooded, fresh or brackish water Tree cover, flooded, saline water 6.4 COMPARATIVE ASSESSMENT guidelines (Table 4.7). Further details on this methodology are provided in section 2, which also OF CARBON STOCKS IN includes information on the methods for estimating VEGETATION belowground biomass and carbon stocks in The model provides global estimates of vegetation vegetation. carbon stocks for terrestrial ecosystems using a This section includes 2006 IPCC default values for multi-layer lookup table for merging geographically aboveground biomass in forests used in the model explicit information that is needed to correctly assign in an attempt to compare the model outcomes with aboveground biomass values to lands. Specifically, country data. the table allows one to merge information on land cover classes, ecofloristic regions, and geographical It should be noted that most countries have in zones, which is then used to assign aboveground place national forest inventories that provide field biomass stocks. measurements of tree parameters, which are used to retrieve information on aboveground biomass, Information on aboveground biomass is obtained belowground biomass, and carbon stocks. This from the default values included in the 2006 IPCC information, based on field measurements, is, 28 https://cds.climate.copernicus.eu/cdsapp#!/dataset/satellite-land-cover?tab=overview. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 41 l C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S in principle, more representative of the natural prevents a one-to-one comparison of information conditions of the countries than the IPCC default on aboveground biomass among data sources. values because it intrinsically captures the effects Bearing all this in mind, it is recommended that the of national circumstances, such as management comparison of model data with country data should practices, natural disturbances, and ecosystem focus solely on aboveground biomass information, features, on the vegetation carbon stocks. which is the main contributor to forest carbon stock, However, using default aboveground biomass values disregarding the comparison of information on from the 2006 IPCC guidelines allows a globally other parameters involved in estimating land-based comparable and consistent treatment of the data emissions and removals, for which a dedicated case- that would be difficult to achieve using country- by-case comparative analysis would be needed to specific information. obtain meaningful results. Moreover, the interpretation of the results of The information in Table 6.4 below includes (i) comparing this information should consider the aboveground biomass values used in the model, as different stratifications of forest land used by included in the 2006 IPCC guidelines, and (ii) country- countries as compared with the model approach. specific information on aboveground biomass, or in- Countries often implement random sampling or house estimates for aboveground biomass retrieved systematic approaches in their national forest from information included in country submissions, inventories that allow them to retrieve information in order to enhance the comparability of model data on biomass stocks disaggregated by forest types, with country-specific information. The result of this or for strata based on biological characteristics comparison must consider the discussion provided of the vegetation or ecozones. This difference in above on the challenges and artifacts of comparing stratification by countries and the model approach forest-related information. TABLE 6.4: 2006 IPCC DEFAULT VALUES FOR FOREST ABOVEGROUND BIOMASS USED BY THE MODEL ABOVEGROUND BIOMASS IN FORESTS ABOVEGROUND ECOLOGICAL DOMAIN CONTINENT BIOMASS REFERENCES ZONE (tonnes d.m ha-1) Tropical Tropical rain forest Africa 310 (130–510) IPCC 2003 North and South America 300 (120–400) Baker et al. 2004a; Hughes et al. 1999 Asia (continental) 280 (120–680) IPCC 2003 Asia (insular) 350 (280–520) IPCC 2003 Tropical moist deciduous Africa 260 (160–430) IPCC 2003 forest North and South America 220 (210–280) IPCC 2003 Asia (continental) 180 (10–560) IPCC 2003 Asia (insular) 290 IPCC 2003 42 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S l ABOVEGROUND BIOMASS IN FORESTS ABOVEGROUND ECOLOGICAL DOMAIN CONTINENT BIOMASS REFERENCES ZONE (tonnes d.m ha-1) Tropical Tropical dry forest Africa 120 (120–130) IPCC 2003 North and South America 210 (200–410) IPCC 2003 Asia (continental) 130 (100–160) IPCC 2003 Asia (insular) 160 IPCC 2003 Tropical shrubland Africa 70 (20–200) IPCC 2003 North and South America 80 (40–90) IPCC 2003 Asia (continental) 60 IPCC 2003 Asia (insular) 70 IPCC 2003 Tropical mountain Africa 40–190 IPCC 2003 systems North and South America 60–230 IPCC 2003 Asia (continental) 50–220 IPCC 2003 Asia (insular) 50–360 IPCC 2003 Subtropical Subtropical humid forest North and South America 220 (210–280) IPCC 2003 Asia (continental) 180 (10–560) IPCC 2003 Asia (insular) 290 IPCC 2003 Subtropical dry forest Africa 140 Subei et al. 2001 North and South America 210 (200–410) IPCC 2003 Asia (continental) 130 (100–160) IPCC 2003 Asia (insular) 160 IPCC 2003 Subtropical steppe Africa 70 (20–200) IPCC 2003 North and South America 80 (40–90) IPCC 2003 Asia (continental) 60 IPCC 2003 Asia (insular) 70 IPCC 2003 Subtropical mountain Africa 50 Montès et al. 2002 systems North and South America 60–230 IPCC 2003 Asia (continental) 50–220 IPCC 2003 Asia (insular) 50–360 IPCC 2003 Temperate Temperate oceanic Europe 120 – forest North America 660 (80–1,200) Hessl et al. 2004; Smithwick et al. 2002 New Zealand 360 (210–430) Hall et al. 2001 South America 180 (90–310) Gayoso and Schlegel 2003; Battles et al. 2002 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 43 l C O M PA R AT I V E A S S E S S M E N T O F V E G E TAT I O N C A R B O N S T O C K R E S U LT S ABOVEGROUND BIOMASS IN FORESTS ABOVEGROUND ECOLOGICAL DOMAIN CONTINENT BIOMASS REFERENCES ZONE (tonnes d.m ha-1) Temperate Temperate continental Asia, Europe (≤20y) 20 IPCC 2003 cont. forest Asia, Europe (>20y) 120 (20–320) IPCC 2003 North and South America (≤20y) 60 (10–130) IPCC 2003 North and South America (>20y) 130 (50–200) IPCC 2003 Temperate mountain Asia, Europe (≤20y) 100 (20–180) IPCC 2003 systems Asia, Europe (>20y) 130 (20–600) IPCC 2003 North and South America (≤20y) 50 (20–110) IPCC 2003 North and South America (>20y) 130 (40–280) IPCC 2003 Boreal Boreal coniferous forest Asia, Europe, North America 10–90 Gower et al. 2001 Boreal tundra woodland Asia, Europe, North America (≤20y) 3–4 IPCC 2003 Asia, Europe, North America (>20y) 15–20 IPCC 2003 Boreal mountain systems Asia, Europe, North America (≤20y) 12–15 IPCC 2003 Asia, Europe, North America (>20y) 40–50 IPCC 2003 44 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS SUMMARY AND NEXT STEPS l 7 Summary and Next Steps This analysis provides results at the global level for meant to deduce net carbon fluxes from lands or aboveground and belowground vegetation carbon other emissions of non-CO2 gases. stock covering the period 2001–2020. It prioritizes Future results will integrate more details without the comparability of the outcome, providing jeopardizing the comparability of the countries’ consistent global results by using relatively long results. New items will include: time series of global data and adopting an approach aligned with the 2006 IPCC guidelines. • The 2019 IPCC refinement. • Further stratification of the ecological variables These results offer a global overview of the considered to estimate carbon vegetation stock. vegetation carbon stock trends by country and by land cover class, but are not meant to be used • The uncertainty range as per the IPCC guidelines. to derive information or to be directly compared • More advanced forest aging and post-fire against results for a specific country, which are dynamics. obtained using different data, on different spatial and temporal scales, and for different purposes. The current results also include information on The results obtained are in line with country-level soil organic carbon, which are not comparable with estimates and do not show unexpected trends. each other due to the different methodologies used Nevertheless, the changes in the outputs for each in their calculation. These results will be made country are driven by the landscape changes comparable in case the necessary information observed in the global datasets used, and are not becomes available. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 45 l REFERENCES References Costanza, R., de Groot, R., Sutton, P., van der Government of Gabon. 2021. “Gabon’s Proposed Ploeg, S., Anderson, S.J., Kubiszewski, I., Modified National REDD+ Forest Reference Farber, S., and Turner, R.K. 2014. “Changes in Level.” https://redd.unfccc.int/files/gabon_ the Global Value of Ecosystem Services.” Global frl_modified_oct2021_clean_final.pdf. Environmental Change 26: 152–158. Government of Mexico. 2021. “Nivel de https://doi.org/10.1016/j.gloenvcha.2014.04.002. referencia de emisiones forestales de Dagum, E.B., and Cholette, P.A. 2006. México (2007–2016): Modificado.” https:// Benchmarking, Temporal Distribution, and redd.unfccc.int/files/nref_2020_modificado_ Reconciliation Methods for Time Series. mexico_23072021_nt_20220103_clean.pdf. https://doi.org/10.1007/0-387-35439-5. Government of Papua New Guinea. 2017. Edens, B. 2021. “The Treatment of Global “Papua New Guinea’s National REDD+ Forest Climate Regulation Service in SEEA EA and Reference Level: Modified Submission for Its Potential for Informing Climate Policy.” UNFCCC Technical Assessment in 2017.” Paper presented at the 9th IMF Statistical https://redd.unfccc.int/files/png_frl_ Forum, “Measuring Climate Change.” resubmission_modified_201700710_final.pdf. Edens, B., Elsasser, P., and Ivanov, E. 2019. Grassi, G., Conchedda, G., Federici, S., Abad “Discussion Paper 6: Defining and Valuing Viñas, R., Korosuo, A., Melo, J., Rossi, S., Carbon Related Services in the SEEA EEA.” Paper Sandker, M., Somogyi, Z., Vizzarri, M., and submitted to the Expert Meeting on Advancing Tubiello, F.N. 2022. “Carbon Fluxes from Land the Measurement of Ecosystem Services for 2000–2020: Bringing Clarity to Countries’ Ecosystem Accounting, New York, January 22–24, Reporting.” Earth Syst. Sci. Data 14: 4643–4666. 2019, and subsequently revised. Version of March https://doi.org/10.5194/essd-14-4643-2022. 15, 2019. https://seea.un.org/sites/seea.un.org/files/ discussion_paper_6_-_valuing_carbon_final.pdf. Grassi, G., House, J., Kurz, W.A., et al. 2018. “Reconciling Global-Model Estimates and FAO (Food and Agriculture Organization Country Reporting of Anthropogenic Forest of the United Nations). “Global Forest CO2 Sinks.” Nature Clim Change 8: 914–920. Resources Assessments: Country Reports https://doi.org/10.1038/s41558-018-0283-x. 2020.” https://www.fao.org/forest-resources- assessment/fra-2020/country-reports/en/. Grassi, G., Schwingshackl, C., Gasser, T., Houghton, R.A., Sitch, S., Canadell, J.G., Cescatti, Gibbs, H.K., and Ruesch, A. 2008. New IPCC Tier- A., Ciais, P., Federici, S., Friedlingstein, P., 1 Global Biomass Carbon Map for the Year 2000. Kurz, W.A., Sanz Sanchez, M.J., Abad Viñas, R., United States. https://doi.org/10.15485/1463800. Alkama, R., Ceccherini, G., Kato, E., Kennedy, 46 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS REFERENCES l D., Knauer, J., Korosuo, A., McGrath, M.J., Nabel, Mengist, W., and Soromessa, T. 2019. “Assessment J., Poulter, B., Rossi, S., Walker, A.P., Yuan, W., of Forest Ecosystem Service Research Trends and Yue, X., and Pongratz, J. 2022. “Mapping Land- Methodological Approaches at Global Level: A Use Fluxes for 2001–2020 from Global Models Meta-Analysis.” Environmental Systems Research to National Inventories.” Earth Syst. Sci. Data 8: 1–18. https://doi.org/10.1186/s40068-019-0150-4. (preprint). https://doi.org/10.5194/essd-2022-245. NCAVES (Natural Capital Accounting and Hillmer, S.C., and Trabelsi, A. 1987. Valuation of Ecosystem Services) and MAIA “Benchmarking of Economic Time Series.” Journal (Mapping and Assessment for Integrated of the American Statistical Association 82 (400): Ecosystem Accounting). 2022. Monetary Valuation 1064–1071. DOI: 10.1080/01621459.1987.10478541. of Ecosystem Services and Ecosystem Assets for Ecosystem Accounting. Interim version, 1st Humber, M.L., Boschetti, L., Giglio, L., and edition. New York: United Nations Department of Justice, C.O. 2019. “Spatial and Temporal Economic and Social Affairs, Statistics Division. Intercomparison of Four Global Burned Area Products.” Int J Digit Earth 12 (4): 460–484. https:// Nordhaus, W.D. 2017. “Revisiting the Social Cost www.ncbi.nlm.nih.gov/pmc/articles/PMC6178238/. of Carbon.” PNAS 114 (7): 1518–1523. https://www. pnas.org/doi/full/10.1073/pnas.1609244114. IPCC (Intergovernmental Panel on Climate Change). 2006. 2006 IPCC Guidelines for National Pessôa, A.C.M., Anderson, L.O., Carvalho, N.S., Greenhouse Gas Inventories. Prepared by the Campanharo, W.A., Silva Junior, C.H.L., Rosan, National Greenhouse Gas Inventories Programme, T.M., Reis, J.B.C., Pereira, F.R.S., Assis, M., Jacon, Eggleston, H.S., Buendia, L., Miwa, K., Ngara, A.D., Ometto, J.P., Shimabukuro, Y.E., Silva, T., and Tanabe, K. (eds). Japan: IGES. C.V.J., Pontes-Lopes, A., Morello, T.F., and Aragão, L.E.O.C. 2020. “Intercomparison of Burned Area IPCC. 2019. 2019 Refinement to the 2006 Products and Its Implication for Carbon Emission IPCC Guidelines for National Greenhouse Gas Estimations in the Amazon.” Remote Sensing 12 Inventories. Calvo Buendia, E., Tanabe, K., (23): 3864. https://doi.org/10.3390/rs12233864. Kranjc, A., Baasansuren, J., Fukuda, M., Ngarize, S., Osako, A., Pyrozhenko, Y., Shermanau, P., Fitalew Agimass, T., Folkersen, M.V., Fleming, and Federici, S. (eds). Switzerland: IPCC. C.M., Buckwell, A., Mackey, B., Diwakar, K.C., Le, D., Hasan, S., and Saint Ange, Kotchen, M., Fenichel, E., and Addicott, E. C. 2021. “The Economic Values of Global 2019. “Even the Representative Agent Must Forest Ecosystem Services: A Meta-Analysis.” Die: Using Demographics to Inform Long- Ecological Economics 189: 107145. https:// Term Social Discount Rates.” Journal of the doi.org/10.1016/j.ecolecon.2021.107145. Association of Environmental and Resource Economists 7 (2). https://doi.org/10.1086/706885. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 47 l REFERENCES Turpie, J.K., Forsythe, K.J., Knowles, A., Blignaut, J., and Letley, G. 2021. “Mapping and Valuation of South Africa’s Ecosystem Services: A Local Perspective.” Ecosystem Services 27B: 179–192. https://doi.org/10.1016/j.ecoser.2017.07.008. Turpie, J.K., Letley, G., Schmidt, K., Weiss, J., O’Farrell, P., and Jewitt, D. 2021. Towards a Method for Accounting for Ecosystem Services and Asset Value: Pilot Accounts for KwaZulu-Natal, South Africa, 2005–2011. NCAVES project report. https://seea.un.org/content/knowledge-base. United Nations. 2021. “System of Environmental Economic Accounting—Ecosystem Accounting.” Final draft, background document for the UN Statistical Commission, February 2021. https://unstats.un.org/unsd/statcom/52nd-session/ documents/BG-3f-SEEA-EA_Final_draft-E.pdf. United Nations. “National GHG Inventory Submission 2022: Finland.” https://unfccc. int/ghg-inventories-annex-i-parties/2022. The World Bank. Vardon, M. 2024. CWON Technical Report. “Estimating Global Carbon Storage in Mangrove Ecosystems.” 48 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 1 l Appendix 1: Country-specific background information used for the comparative assessment included in section 6 This appendix contains further information on each of the countries analyzed to compare the results produced in the study against other global data assessments. 1. Gabon: Modified submission of the REDD+ Forest Reference Level, 2021 Gabon’s submission of the Forest Reference Level under the UN REDD+ program does not provide information on aboveground biomass. Instead, information on carbon stocks by hectares is provided for different forest types and carbon pools. To facilitate the comparison of the information submitted by Gabon with the aboveground biomass stocks used in the model, the default IPCC carbon fraction (0.47) for converting the dry matter of biomass into carbon stocks was used. The tables below show the information on carbon stock as provided by Gabon and in-house estimates of aboveground biomass stocks retrieved from the country’s information. DEAD ORGANIC FOREST DISTURBANCE ABOVE BELOW MATTER SOIL ABOVE TOTAL SOURCE TYPE HISTORY GROUND GROUND (DEADWOOD CARBON AND ECOSYSTEMS & LITTER) BELOW- GROUND TREES LIVING TOTAL 0–100CM >10CM DBH ROOTS DOM MEAN U MEAN MEAN U MEAN U MEAN U MEAN U Old- Undisturbed 151.6 12% 35.6 23.5 16% 161.3 41% 187.2 12% 372.0 19% NRI (Poulsen et al. growth 2020) Logged Undetermined 172.8 11% 40.6 42.8 14% 127.1 10% 213.4 11% 383.3 7% NRI (Poulsen et al. 2020) Secondary Undetermined 95.6 22% 22.5 26.8 31% 98.3 7% 118.0 22% 243.2 11% NRI (Poulsen et al. 2020) Forest Undetermined 141.7 9% 33.3 32.2 11% 125.1 14% 175.0 9% 332.3 7% NRI (Poulsen et al. (average 2020) old growth, logged, secondary) Mangrove Undisturbed 111.8 20% 37.0 18.9 18% 254.6 11% 148.8 19% 422.3 9% (Kauffman and Bhomia 2017) https://data.cifor. org/dataverse/ swamp Colonizing Undisturbed 47.2 21% 11.1 Data not available 72.3 2% 58.3 21% 130.6 104% (Chiti et al. 2018; Cuni-Sanchez et al. 2016) U = Uncertainty GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 49 l APPENDIX 1 ABOVEGROUND ABOVEGROUND FOREST TYPE (t dry matter/ha) (t C/ha) (IN-HOUSE ESTIMATES) Old-growth forest 151.6 322.55 Logged 172.8 367.66 Secondary 95.6 203.40 Forest (average old growth, 141.7 301.49 logged, secondary) Mangrove 111.8 237.87 Colonizing 47.2 100.43 2. Mexico: Modified submission of the REDD+ Forest Reference Level, 2021 Mexico’s submission of the Forest Reference Level under the UN REDD+ program does not provide information on aboveground biomass. Instead, information on carbon stocks for aboveground biomass, by hectare, is provided for different forest types. To facilitate the comparison of the information submitted by Mexico with the aboveground biomass stock values used in the model, the default IPCC carbon fraction (0.47) for converting the dry matter of biomass into carbon stocks was used. The tables below show the information on carbon stock of aboveground biomass as provided by Mexico and in-house estimates of dry matter in aboveground biomass retrieved from the country’s information. 50 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS Sub-estrato 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 CalMed – 10x20 – TF 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.79 0.78 0.78 CalMed – 20x20 – TF 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 0.08 CalMed – 5x5 – TF 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.86 8.36 8.36 8.36 DesAN – 10x10 – TF 1.71 1.71 1.71 1.71 1.71 1.71 1.71 1.70 1.70 1.70 1.70 1.70 1.70 1.70 1.69 1.69 1.69 DesAN – 20x20 – TF 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 0.04 DesAN – 5x5 – TF 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.79 4.78 4.78 ElevSM – 10x10 – TF 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.67 4.63 4.59 4.59 ElevSM – 20x20 – TF 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.48 1.47 1.47 1.47 1.47 1.47 ElevSM – 5x5 – TF 9.71 9.71 9.71 9.71 9.71 9.71 9.70 9.70 9.70 9.70 9.70 9.70 9.70 9.70 9.67 9.67 9.67 GraPla – 10x10 – TF 1.78 1.78 1.78 1.78 1.78 1.78 1.78 1.78 1.77 1.77 1.77 1.77 1.77 1.77 1.77 1.77 1.77 GraPla – 20x20 – TF 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.50 0.49 0.49 0.49 0.49 0.49 0.49 GraPla – 5x5 – TF 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 9.77 SCH – 10x10 – TF 12.52 12.52 12.52 12.52 12.52 12.52 12.52 12.52 12.52 12.52 12.48 12.48 12.48 12.48 12.47 12.34 12.26 SCH – 20x20 – TF 3.74 3.74 3.74 3.74 3.74 3.74 3.74 3.74 3.70 3.67 3.67 3.67 3.67 3.62 3.59 3.56 3.35 SCH – 5x5 – TF 24.87 24.87 24.87 24.87 24.87 24.86 24.82 24.71 24.61 24.50 24.42 24.36 24.28 24.20 24.12 24.05 23.97 SCS – 10x10 – TF 9.37 9.37 9.37 9.37 9.37 9.37 9.37 9.34 9.33 9.33 9.29 9.27 9.25 9.25 9.23 9.18 9.16 SCS – 20x20 – TF 1.96 1.96 1.96 1.96 1.96 1.96 1.96 1.96 1.96 1.96 1.96 1.92 1.92 1.90 1.90 1.89 1.88 SCS – 5x5 – TF 15.24 15.24 15.24 15.24 15.24 15.22 15.21 15.20 15.18 15.18 15.16 15.11 15.10 15.05 15.04 14.99 14.94 SieTem – 10x10 – TF 10.43 10.43 10.43 10.43 10.43 10.43 10.41 10.41 10.41 10.41 10.41 10.36 10.36 10.33 10.33 10.33 10.33 SieTem – 20x20 – TF 7.59 7.59 7.59 7.59 7.59 7.59 7.59 7.59 7.59 7.59 7.59 7.57 7.56 7.56 7.56 7.56 7.56 SieTem – 5x5 – TF 26.46 26.46 26.46 26.46 26.46 26.45 26.43 26.43 26.41 26.38 26.36 26.30 26.24 26.21 26.15 26.11 26.05 CalMed – California Mediterrànea GraPla – Grandes Planicies SieTem – Sierras Templadas DesAN – Desiertos de América del Norte SCH – Selvas Cálido-Húmedas TF – Tierras Forestales ElevSM – Elevaciones Semiáridas Meridionales SCS – Selvas Cálido-Secas GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 1 l 51 52 l ABOVEGROUND BIOMASS (TON DRY MATTER) (IN-HOUSE ESTIMATES RETRIEVED FROM CARBON STOCK INFORMATION INCLUDED IN THE REDD+ MODIFIED FOREST REFERENCE EMISSIONS LEVELS SUBMISSION, 2021) APPENDIX 1 Category 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 California Mediterranea – 10 × 10 – TF 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.68 1.66 1.66 California Mediterranea – 20 × 20 – TF 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 0.17 California Mediterranea – 5 × 5 – TF 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 18.85 17.79 17.79 17.79 Desiertos de America del Norte – 10 × 10 – TF 3.64 3.64 3.64 3.64 3.64 3.64 3.64 3.62 3.62 3.62 3.62 3.62 3.62 3.62 3.62 3.62 3.62 Desiertos de America del Norte – 20 × 20 – TF 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 0.09 Desiertos de America del Norte – 5 × 5 – TF 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 10.19 Elevaciones Semiaridas Meridionales – 10 × 10 – TF 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.94 9.85 9.77 9.77 Elevaciones Semiaridas Meridionales – 20 × 20 – TF 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.15 3.13 3.13 3.13 3.13 3.13 Elevaciones Semiaridas Meridionales – 5 × 5 – TF 20.66 20.66 20.66 20.66 20.66 20.66 20.64 20.64 20.64 20.64 20.64 20.64 20.64 20.64 20.57 20.57 20.57 Grandes Planicies – 10 × 10 – TF 3.79 3.79 3.79 3.79 3.79 3.79 3.79 3.79 3.77 3.77 3.77 3.77 3.77 3.77 3.77 3.77 3.77 Grandes Planicies – 20 × 20 – TF 1.06 1.06 1.06 1.06 1.06 1.06 1.06 1.06 1.06 1.06 1.06 1.04 1.04 1.04 1.04 1.04 1.04 Grandes Planicies – 5 × 5 – TF 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 20.79 Selvas Calido-Humedas – 10 × 10 – TF 26.64 26.64 26.64 26.64 26.64 26.64 26.64 26.64 26.64 26.64 26.55 26.55 26.55 26.55 26.53 26.26 26.09 Selvas Calido-Humedas – 20 × 20 – TF 8.23 8.23 8.23 8.23 8.23 8.23 8.23 8.23 8.09 7.79 7.79 7.79 7.79 7.68 7.62 7.55 7.11 Selvas Calido-Humedas – 5 × 5 – TF 53.17 53.17 53.17 53.17 53.17 53.13 53.04 52.81 52.55 52.30 52.11 51.98 51.77 51.60 51.38 51.26 51.06 Selvas Calido-Secas – 10 × 10 – TF 19.94 19.94 19.94 19.94 19.94 19.94 19.94 19.87 19.87 19.87 19.79 19.74 19.70 19.64 19.62 19.51 19.47 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS Selvas Calido-Secas – 20 × 20 – TF 4.28 4.28 4.28 4.28 4.28 4.28 4.28 4.28 4.28 4.28 4.28 4.17 4.17 4.15 4.13 4.13 4.11 Selvas Calido-Secas – 5 × 5 – TF 32.36 32.36 32.36 32.36 32.36 32.32 32.30 32.30 32.26 32.23 32.19 32.06 32.04 31.94 31.85 31.74 31.64 Sierras Templadas – 10 × 10 – TF 22.30 22.30 22.30 22.30 22.30 22.30 22.23 22.23 22.23 22.23 22.23 22.15 22.11 22.04 22.04 22.04 22.04 Sierras Templadas – 20 × 20 –TF 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.11 16.09 16.09 16.09 16.09 16.09 Sierras Templadas –5 × 5 –TF 56.30 56.30 56.30 56.30 56.30 56.28 56.23 56.23 56.19 56.15 56.11 56.02 55.91 55.85 55.74 55.68 55.62 APPENDIX 1 l 3. Finland: 2022 GHG inventory submission under the UNFCCC To ensure consistency of the data source used to obtain country-specific information for Finland for other sections of this comparative assessment, the 2022 GHG inventory submission was first consulted for aboveground biomass stocks of the country’s forests. However, due to Finland’s method for estimating forest-related GHG emissions and removals, the GHG inventory submission does not include information on aboveground biomass stocks, which is publicly available in FAO’s Global Forest Resources Assessments for the country. The table below provides information on forest aboveground biomass stocks in Finland, as included in the 2020 FAO-FRA country report. FOREST BIOMASS (tonnes/ha) FRA CATEGORIES 1990 2000 2010 2015 2016 2017 2018 2019 2020 Aboveground biomass 43.86 49.14 53.64 59.77 59.77 59.77 59.77 59.77 59.77 Belowground biomass 13.05 15.20 16.51 17.31 17.31 17.31 17.31 17.31 17.31 Deadwood 1.37 1.37 1.49 1.49 1.49 1.49 1.49 1.49 1.49 4. Papua New Guinea: Modified submission of the national REDD+ Forest Reference Level, 2017 In its submission of the Forest Reference Level under the UN REDD+ program, Papua New Guinea includes explicit information on aboveground biomass by forest type, which is further divided into classes according to different anthropogenic impacts on those forest types. This information is included in the table below to facilitate the comparison of the country data with the aboveground biomass stocks used in the model. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 53 l APPENDIX 1 ABOVEGROUND BIOMASS BELOWGROUND BIOMASS FOREST HUMAN SOURCE ECOLOGICAL DRY MATTER DRY ROOT TO TYPE IMPACT ZONE AS (tonnes/ha) MATTER SHOOT PER IPCC (tonnes/ha) RATIO GUIDELINES Low altitude forest on Primary 223 83 0.37 plains and fans Logged 146 54 0.37 Other disturbances 146 54 0.37 Low altitude forest Primary 223 83 0.37 on uplands Logged 146 54 0.37 Other disturbances 146 54 0.37 Littoral forest Primary 223 83 0.37 Fox et al. Tropical Logged 146 54 0.37 (2010) rainforest Other disturbances 146 54 0.37 Seral forest Primary 223 83 0.37 Logged 146 54 0.37 Other disturbances 146 54 0.37 Swamp forest Primary 223 83 0.37 Logged 146 54 0.37 Other disturbances 146 54 0.37 Lower montane forest Primary 140 38 0.27 Logged 92 25 0.27 Other disturbances 92 25 0.27 Montane forest Primary 140 38 0.27 Tropical Logged mountain 92 25 0.27 system Other disturbances 92 25 0.27 Mountain coniferous Primary 140 38 0.27 forest Logged 92 25 0.27 Other disturbances 92 25 0.27 Dry seasonal forest Primary 130 36 0.28 Logged IPCC 85 24 0.28 Guidelines Other disturbances (2006) 85 24 0.28 Tropical Woodland Primary 130 36 0.28 dry forest Other disturbances 85 24 0.28 Savanna Primary 130 36 0.28 Other disturbances 85 24 0.28 Scrub Primary Tropical 70 28 0.4 shrubland Other disturbances 46 18 0.4 Mangrove Primary Tropical 192 94 0.49 wet mangrove Other disturbances 126 62 0.49 Forest plantation Primary Tropical 150 56 0.37 rainforest (plantation) Other disturbances 98 36 0.37 54 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 2 l Appendix 2: Aligning the terrestrial results produced in ARIES with the carbon contribution of mangrove ecosystems THE DIFFICULTY OF INTEGRATING The integration of current approaches is technically MANGROVE AND VEGETATION feasible but beyond the scope of this work. Doing so would primarily require reconciling issues related CARBON STORAGE RESULTS to spatio-temporal scale, since the mangrove carbon The current results include the carbon contribution storage results are provided at a much higher spatial of mangrove ecosystems estimated using the model resolution than those for terrestrial carbon stock. currently available in ARIES and described in the On the other hand, the input data30 used to identify second segment of the report. the mangrove areas cover only the years 1996, 2010, 2015, and 2020, while the results of the terrestrial This section shows how the vegetation and mangrove carbon stock cover a complete time series of annual carbon stock results could be aligned to allow future information for the period 2001–2020. comparison or facilitate a potential integration of these outputs. However, given the important If the two datasets were to be combined, priority contribution of mangroves to the global carbon should be given to the layers estimating mangroves budget, the World Bank commissioned Silvestrum carbon stock due to the higher spatial resolution Climate Associates to estimate the carbon stock of the original information used as input in the of mangroves separately.29 Given the focus of this model. To fill the gaps in the time series, a linear analysis, datasets specifically developed to observe interpolation approach could be applied. To avoid changes in these ecosystems were used, which use double-counting, the interpolated vegetation carbon inconsistent assumptions to those in this global stock that overlaps with the observed mangroves assessment approach. This section discusses the would be excluded. The overlapping values would implication of merging these two different data be neglected for all years in which overlapping sources to ensure that integrating these data does areas are observed until a more recent observation not affect the consistency of the final result. confirms that the overlap no longer exists. This approach would minimize error, assuming that To integrate the different data sources, the data must the spatial extent of mangroves stays constant in be consistent across spatial and temporal scales. between available spatially explicit information. 29 CWON Technical Report 2024: Estimating Global Carbon Storage in Mangrove Ecosystems. 30 https://geowetlands.org/knowledge-base/datasets/global-mangrove-watch-gmw-v3-0/. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 55 l APPENDIX 2 While this approach is achievable from a purely RECONCILING SOIL ORGANIC mathematical perspective, the final result would CARBON AMONG DATA SOURCES be difficult to defend. Most of the numeric value resulting from this approach would result from As indicated above, ISRIC—World Soil Information31 aggregation error rather than carbon stock. It is a provides information on soil organic carbon at well-known phenomenon in geography that results different depths up to 200cm. Information on soil will change, often substantially, at different spatial organic carbon in mangrove areas, as provided by resolutions. Sensitivity analyses can be used to Silvestrum, is given to a depth of 100cm. Techniques understand the divergence in results obtained to infer the carbon contents at different depths for using coarser and finer-scale models, but they a given value are available. If setting the carbon fall outside the scope of this work. Therefore, content at a unique depth is seen as a way to align accounting separately for the overall vegetation data sources, such processing of the data would in carbon stock, and for mangrove carbon storage, principle be feasible. However, using the data as results in the best approach, at least until there is a originally provided in each dataset is seen as a more clear and recognized way to compute carbon stock realistic option because (i) it covers a wider range of at comparable resolutions. options for satisfying user demand, and (ii) it avoids introducing new uncertainties associated with processing two outcomes from different data sources. 31 https://www.isric.org/explore/soilgrids. 56 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 3 l Appendix 3: Large yearly differences in the vegetation carbon stock not explained by land cover changes Over the time series, there are cases where the The impact of fire in regions where fires are vegetation carbon stock of a unique land cover class abundant in the year observed leads to a loss of contributes significantly to the overall change in the vegetation carbon stock. But, given the variables vegetation carbon stock for that country. Intuitively, involved in the model, the final results analysis is this might be due to large yearly fluctuation of the not always straightforward. An example is provided extent of that land cover in the ESA-CCI dataset, to help the reader interpret the results correctly. which drives the reduction in carbon stock. Some Between 2001 and 2002, in South Sudan, the large deviations can be caused by: vegetation carbon stock of open deciduous broadleaf 1. Large areas burnt in a given year, with fire as an forests 32 decreased by 5.5 megatonnes of CO2. This important influence on vegetation carbon storage. loss corresponds to more than 98.5 percent of the total vegetation carbon stock decrease in that year 2. Data gaps in the presence of fire for large, in that country. In the same year though, the change fire-prone areas, which make it challenging to in the extent of the open deciduous broadleaf forest accurately estimate the vegetation carbon stock. cover is less than 0.15 percent. FIGURE A3.1: LAND COVER MAP OF SOUTH SUDAN However, large areas of deciduous forests burned in 2001. 32 These forests are classified as tree cover, needleleaved, deciduous, open (15-40 percent) in the ESA-CCI dataset (code 82). GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 57 l APPENDIX 3 FIGURE A3.2: PRESENCE OF BURNED LAND IN SOUTH SUDAN In this model, the combination of other inputs, including land cover class and the ecofloristic region, determines the carbon storage lost due to fire (more sophisticated models account for burn severity in modeling carbon emissions from fires, but global burn severity products do not exist in time series adequate to support CWON). FIGURE A3.3: OVERLAY OF LAYERS OF PRESENCE OF BURNED LAND AND ECOFLORISTIC REGIONS IN SOUTH SUDAN 58 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 3 l For example, when subjected to fire, open deciduous forests in the African tropical moist deciduous region are estimated to lose half of their carbon content, dropping from 152 tonnes per hectare to 76 tonnes per hectare. FIGURE A3.4: AREA COVERED BY THE TROPICAL MOIST DECIDUOUS FOREST ECOFLORISTIC REGION IN SOUTH SUDAN FIGURE A3.5: CARBON STOCK ESTIMATED FOR A NON-BURNED AREA IN TROPICAL MOIST DECIDUOUS FOREST ECOFLORISTIC REGION IN SOUTH SUDAN The light blue regions located in the south-central part of South Sudan are those burned by fire. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 59 l APPENDIX 3 FIGURE A3.6: CARBON STOCK ESTIMATED FOR A BURNED AREA IN TROPICAL MOIST DECIDUOUS FOREST ECOFLORISTIC REGION IN SOUTH SUDAN The same proportional loss of carbon stock is observed for deciduous forests in the African tropical rainforest region that burned; stocks in burned areas decreased from 200 tonnes per hectare to 100 tonnes per hectare. FIGURE A3.7: AREA COVERED BY THE TROPICAL RAINFOREST ECOFLORISTIC REGION IN SOUTH SUDAN 60 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 3 l FIGURE A3.8: CARBON STOCK ESTIMATED FOR A NON-BURNED AREA IN TROPICAL RAINFOREST ECOFLORISTIC REGION IN SOUTH SUDAN FIGURE A3.9: CARBON STOCK ESTIMATED FOR A BURNED AREA IN TROPICAL RAINFOREST ECOFLORISTIC REGION IN SOUTH SUDAN The same type of forests located in the African tropical mountain system ecofloristic region are affected more severely by fire, and the model estimates a much greater loss, dropping from 152 tonnes per hectare to 15 tonnes per hectare—about 90 percent of initial carbon stock in relative terms. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 61 l APPENDIX 3 FIGURE A3.10: AREA COVERED BY THE TROPICAL MOUNTAIN SYSTEM ECOFLORISTIC REGION IN SOUTH SUDAN FIGURE A3.11: CARBON STOCK ESTIMATED FOR A NON-BURNED AREA IN TROPICAL MOUNTAIN SYSTEM ECOFLORISTIC REGION IN SOUTH SUDAN 62 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 3 l FIGURE A3.12: CARBON STOCK ESTIMATED FOR A BURNED AREA IN TROPICAL MOUNTAIN SYSTEM ECOFLORISTIC REGION IN SOUTH SUDAN The above examples illustrate how reductions in may result from missing data. A significant example carbon stock can be caused by burned vegetation was the increase shown in vegetation carbon stock as well as changes in the extent of land cover, and in Gabon between 2001 and 2002. Almost 99 percent the magnitude of the change also depends on the of this total increase could be attributed to data ecofloristic region in which the change takes place. that showed higher stocks of Gabon’s broad-leaved evergreen forests,33 which had actually declined Another concern relates to how the model treats a in extent during that time by 10,000 hectares due lack of information on the presence of fire. In most to fires, corresponding to a 0.04 percent reduction cases, only a few limited areas are not covered by from the previous year. This counterintuitive output the data, but when data gaps are extensive in areas is due to the missing information on the presence where fires play an essential role, modeled changes of fires. 33 Equivalent to the tree cover, broadleaved, evergreen, closed to open (>15 percent) class in the ESA-CCI classification, identified with the code 50. GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 63 l APPENDIX 3 FIGURE A3.13: PRESENCE OF BURNED LAND IN GABON In the maps above, red corresponds to burned areas, blue to unburned areas, and pale green areas (highlighted with yellow circles in the image below) to data gaps in the MODIS burned area product, most likely due to persistent cloud cover.34 FIGURE A3.14: PRESENCE OF BURNED LAND IN GABON WITH LEGEND 34 https://lpdaac.usgs.gov/documents/875/MCD64_User_Guide_V6.pdf/section 3.1.3 Metadata. 64 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 3 l A simple fix for this situation would be to assume Data gaps for burned areas can also lead to an that the areas with no data experienced no fires, but increase in vegetation carbon stock between two domain experts strongly discouraged this approach, observed years. For example, comparing vegetation because this assumption would be difficult to defend carbon stocks in Gabon in 2002 and 2003, the extent and would deviate from the original methodology of data gaps is much smaller, and so the unburned proposed by Gibbs and Reusch. Indeed, data gaps area increases significantly, which in turn leads to a for fire presence only affect model results in regions significant increase in 2003 vegetation carbon stock, prone to fire. Thus, forest ecosystems are the most as there is enough evidence to assume that a large affected by this lack of information, whereas the extent of the region was not burned in that year. effect on other ecosystems is relatively insignificant. FIGURE A3.15: GABON 2002: PRESENCE OF FIRE GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 65 l APPENDIX 3 FIGURE A3.16: GABON 2003: PRESENCE OF FIRE For the above reasons, such cases have been flagged in the results, so that the user can accurately interpret these changes, which are based on the availability of information, rather than changes in the ecological characteristics of the forests over a single year. Planned model improvements related to forest aging and post-fire dynamics will improve results for such cases and smooth out these issues whenever sufficient information exists to provide more accurate outputs. As such, large changes, such as the case shown above, should be observed less frequently in the future. 66 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 4 l Appendix 4: Report of the causes of large changes in vegetation carbon stock of individual land cover classes For full transparency of the results, the table below shows the main driver of all the large changes observed for a land cover class, over each period observed in the time series. The cases report such information for any change in the vegetation carbon stock of a land cover class that accounted for a significant increase or decrease in the total vegetation carbon stock in a country, as shown in the equation below: | ( L C 1 t 2 - L C 1 t 1 ) / To t .V C S t 1 | > 3 % where: • LC1t2 is the contribution to the vegetation carbon stock of the land cover class #1 at time 2 • LC1t1 is the contribution to the vegetation carbon stock of the land cover class #1 at time 1 • Tot.VCSt1 is the total vegetation carbon stock from all LC classes at time 1. For each case in which a single class is directly responsible for a change in the country’s total carbon stock (loss or increase) of at least 3 percent, the reason for that change is explicitly reported in the tables below. TABLE A4.1: ANALYSIS OF LARGE DEVIATION LOSS OF VEGETATION CARBON LEGEND FIRE OCCURRENCE STOCK DUE TO FIRES No data (fire) Lack of data (fire) LC change Loss of vegetation carbon stock due to land cover change Fire* Fire occurrence + lack of data (fire) LC change* Land cover change + lack of data (fire) GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 67 l APPENDIX 4 TIME PERIOD 2001–2002 TREE COVER, TREE COVER, TREE COVER, BROADLEAVED, BROADLEAVED, TREE COVER, LAND COVER/ HERBACEOUS BROADLEAVED, EVERGREEN, DECIDUOUS, FLOODED, COUNTRY COVER DECIDUOUS, CLOSED TO CLOSED TO SALINE WATER OPEN (15–40%) OPEN (>15%) OPEN (>15%) South Sudan -11% Gabon 24% Equatorial Guinea 10% São Tomé and -9% Príncipe Congo 4% Central African -3% Republic Togo -4% Guinea -3% Ghana -5% Côte d’Ivoire -6% Benin -8% Dominica 11% Grenada 13% Saint Kitts and Nevis 9% Montserrat (UK) 5% Panama 11% Colombia 5% Singapore -5% Monaco -5% Papua New Guinea 6% Vanuatu 3% Wallis and Futuna (Fr.) 4% American Samoa (US) 6% Saint Vincent and the 10% Grenadines Curaçao (Neth.) 3% Sint Eustatius (Neth.) 8% Saint-Martin (Fr.) 5% 68 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 4 l TIME PERIOD 2002–2003 MOSAIC SHRUB OR CROPLAND TREE COVER, TREE COVER, TREE COVER, HERBACEOUS LAND COVER (>50%)/NATURAL BROADLEAVED, BROADLEAVED, NEEDLELEAVED, SHRUB- COVER, FLOODED, CLASS/ VEGETATION EVERGREEN, DECIDUOUS, EVERGREEN, LAND FRESH/SALINE/ COUNTRY (TREE, SHRUB, CLOSED TO OPEN CLOSED TO BRACKISH HERBACEOUS OPEN (>15%) (15–40%) OPEN (>15%) WATER COVER) (<50%) Uganda -4% São Tomé and Príncipe 53% Central African -6% Republic Senegal 8% Benin 3% Grenada 3% Montserrat (UK) 4% Panama 6% Ecuador 7% Monaco -6% -3% Wallis and Futuna (Fr.) 5% Saint Vincent and the 8% Grenadines Sint Eustatius (Neth.) 6% -6% TIME PERIOD 2003–2004 TREE COVER, TREE COVER, SHRUB OR TREE COVER, LAND COVER BROADLEAVED, BROADLEAVED, HERBACEOUS BROADLEAVED, TREE COVER, CLASS/ EVERGREEN, DECIDUOUS, COVER, FLOODED, DECIDUOUS, FLOODED, COUNTRY CLOSED TO CLOSED TO FRESH/SALINE/ OPEN SALINE WATER OPEN OPEN BRACKISH WATER (15–40%) (>15%) (>15%) Uganda 4% São Tomé and Príncipe -13% Central African Republic 5% Togo 4% Senegal -5% Nigeria 4% Ghana 9% Côte d’Ivoire 5% Benin 3% Bolivia -4% Vanuatu -9% 17% Sint Eustatius (Neth.) 5%* -6% GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 69 l APPENDIX 4 TIME PERIOD 2004–2005 2005–2006 TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, LAND COVER BROADLEAVED, BROADLEAVED, BROADLEAVED, BROADLEAVED, NEEDLELEAVED, BROADLEAVED, BROADLEAVED, CLASS/ EVERGREEN, EVERGREEN, EVERGREEN, DECIDUOUS, EVERGREEN, DECIDUOUS, DECIDUOUS, COUNTRY CLOSED TO CLOSED TO CLOSED TO CLOSED TO CLOSED TO OPEN (15–40%) OPEN (15–40%) OPEN (>15%) OPEN (>15%) OPEN (>15%) OPEN (>15%) OPEN (>15%) South Sudan 8% Uganda -6% 14% São Tomé and 18% 5% Príncipe Togo -6% 4% Nigeria -4% 3% Ghana -7% Côte d’Ivoire -5% Benin -7% 5% Cambodia 4%* -3%* Guernsey (UK) -3% American Samoa -3% (US) TIME PERIOD 2006–2007 2007–2008 TREE COVER, TREE COVER, TREE COVER, MOSAIC TREE TREE COVER, TREE COVER, BROADLEAVED, NEEDLELEAVED, BROADLEAVED, AND SHRUB LAND COVER BROADLEAVED, BROADLEAVED, EVERGREEN, EVERGREEN, EVERGREEN, (>50%)/ CLASS/ DECIDUOUS, DECIDUOUS, CLOSED TO CLOSED TO CLOSED TO HERBACEOUS COUNTRY OPEN OPEN OPEN OPEN OPEN COVER (15–40%) (15–40%) (>15%) (>15%) (>15%) (<50%) South Sudan -3% Uganda -8% São Tomé and -13% 10% Príncipe Guinea-Bissau 4%* Benin 3% Ecuador -3% 3% Macau; SAR -3% UN Buffer Zone -6% 4% in Cyprus 70 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 4 l TIME PERIOD 2008–2009 2009–2010 2010–2011 TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, BROADLEAVED, BROADLEAVED, BROADLEAVED, NEEDLELEAVED, LAND COVER BROADLEAVED, BROADLEAVED, BROADLEAVED, EVERGREEN, EVERGREEN, EVERGREEN, EVERGREEN, CLASS/ DECIDUOUS, DECIDUOUS, DECIDUOUS, CLOSED TO CLOSED CLOSED TO CLOSED TO COUNTRY OPEN OPEN OPEN OPEN TO OPEN OPEN OPEN (15–40%) (15–40%) (15–40%) (>15%) (>15%) (>15%) (>15%) South Sudan -3% Uganda 4% -3% 5% São Tomé and -7% 7% -13% Príncipe Central African 5% Republic Togo -3% Ghana -4% Montserrat (UK) 3%* Panama -4% Ecuador 4%* TIME PERIOD 2011–2012 2012–2013 TREE COVER, SHRUB OR TREE COVER, SHRUB OR TREE COVER, TREE COVER, BROADLEAVED, HERBACEOUS BROADLEAVED, HERBACEOUS LAND COVER BROADLEAVED, BROADLEAVED, EVERGREEN, COVER, FLOODED, EVERGREEN, COVER, FLOODED, CLASS/ DECIDUOUS, DECIDUOUS, CLOSED TO FRESH/SALINE/ CLOSED TO FRESH/SALINE/ COUNTRY OPEN OPEN OPEN BRACKISH OPEN BRACKISH (15–40%) (15–40%) (>15%) WATER (>15%) WATER Uganda -3% São Tomé and -17% 27% Príncipe Central African 4% Republic Togo 5% Guinea-Bissau -4%* 4%* Ghana 4% Ecuador -6%* Macau; SAR 20% -6% Sint Maarten (Neth.) -3% Saint-Barthélemy (Fr.) -5% 4% Saint-Martin (Fr.) -3% GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 71 l APPENDIX 4 TIME PERIOD 2013–2014 TREE COVER, SHRUB OR TREE COVER, BROADLEAVED, HERBACEOUS LAND COVER TREE OR BROADLEAVED, HERBACEOUS EVERGREEN, COVER, FLOODED, CLASS/ SHRUB DECIDUOUS, SHRUBLAND COVER CLOSED TO FRESH/SALINE/ COUNTRY COVER OPEN OPEN BRACKISH (15–40%) (>15%) WATER Mauritius 8% São Tomé and Príncipe -10% Côte d’Ivoire 3% Antigua and Barbuda -7% Aruba (Neth.) -4% Palau 7% 6% Sint Maarten (Neth.) -3% Saint-Barthélemy (Fr.) -4% TIME PERIOD 2014–2015 2015–2016 TREE COVER, TREE COVER, TREE COVER, TREE COVER, TREE COVER, LAND COVER BROADLEAVED, BROADLEAVED, BROADLEAVED, BROADLEAVED, BROADLEAVED, CLASS/ EVERGREEN, EVERGREEN, DECIDUOUS, DECIDUOUS, DECIDUOUS, COUNTRY CLOSED TO CLOSED TO CLOSED TO OPEN OPEN OPEN OPEN OPEN (15–40%) (15–40%) (>15%) (>15%) (>15%) South Sudan 4% -4% São Tomé and Príncipe 12% -8% Senegal 4% -4% Guinea -6% Ecuador 6%* Kuril islands -5%* TIME PERIOD 2016–2017 2017–2018 TREE COVER, TREE COVER, LAND COVER BROADLEAVED, BROADLEAVED, CLASS/ DECIDUOUS, SHRUBLAND EVERGREEN, COUNTRY OPEN CLOSED TO OPEN (15–40%) (>15%) Guinea 4% Aruba (Neth.) -4% Ecuador -5% Saint-Barthélemy (Fr.) -4% 72 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 4 l TIME PERIOD 2018–2019 MOSAIC NATURAL TREE COVER, LAND COVER VEGETATION TREE COVER, TREE COVER, MOSAIC TREE AND BROADLEAVED, CLASS/ (TREE, SHRUB, BROADLEAVED, MIXED LEAF TYPE SHRUB (>50%)/ EVERGREEN, COUNTRY HERBACEOUS DECIDUOUS, OPEN (BROADLEAVED AND HERBACEOUS CLOSED TO OPEN COVER) (>50%)/ (15–40%) NEEDLELEAVED) COVER (<50%) (>15%) CROPLAND (<50%) South Sudan 8% São Tomé and Príncipe 18% Monaco -8% 4% No Man’s Land Area -3% 3% TIME PERIOD 2019–2020 MOSAIC TREE SHRUB OR TREE COVER, TREE COVER, LAND COVER AND SHRUB HERBACEOUS TREE OR BROADLEAVED, BROADLEAVED, CLASS/ (>50%)/ COVER, FLOODED, SHRUB EVERGREEN, DECIDUOUS, COUNTRY HERBACEOUS FRESH/SALINE/ COVER CLOSED TO OPEN OPEN COVER BRACKISH (>15%) (15–40%) (<50%) WATER South Sudan -8% Ghana -4% Macau; SAR -6% Monaco -4% Palau 11% Wallis and Futuna (Fr.) 7% GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 73 l APPENDIX 5 Appendix 5: Report of the causes of large relative changes in vegetation carbon stock of individual land cover classes Over the time series generated, there are large relative changes within each land cover class. The contribution of a land cover class to the vegetation carbon stock of a country can disappear or experience large changes in relative terms over just a single year. Those changes, however, are always associated with land cover classes whose extent was underrepresented in that context, thus small absolute changes in the underlying data (such as ESA-CCI land cover change or the availability of the information on burned land) could lead to significant relative changes. This is unsurprising, even though time-series products should be built to avoid such situations. However, it is challenging to maintain perfect consistency when producing global maps covering such a long period of time. This appendix lists the cases in which large relative changes (for an individual land cover class contribution to vegetation carbon stock) take place, and the reason for each case: • Djibouti: Mosaic cropland (<50 percent) increase of 200 percent in 2018: ESA CCI LC change • UAE: Mosaic cropland (<50 percent) drops to zero in 2006: ESA CCI LC change • Brazil: Tree cover mixed leaf type peaks and valleys between 2001 and 2005, then steady: Burned land (NO DATA) • Iceland: Tree cover mixed leaf type switching between hundreds and zeros: ESA-CCI LC change and Burned land (NO DATA) • Central African Republic: Tree cover mixed leaf type drops to zero after 2003: ESA-CCI LC change • Niger: Tree cover, broadleaved, deciduous, closed to open (>15 percent) goes from zero to thousands after 2014: ESA-CCI LC change • Eritrea: Tree cover, broadleaved, deciduous, closed to open (>15 percent) goes from zero to thousands after 2017: ESA-CCI LC change • Ecuador: Tree cover, broadleaved, deciduous, closed to open (>15 percent) rapidly increases after 2016: ESA-CCI LC change 74 GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS APPENDIX 5 l • Rwanda: Tree cover, needleleaved, evergreen, closed to open (>15 percent) constant but with several valleys: ESA-CCI LC change • Argentina: Tree cover, needleleaved, evergreen, closed to open (>15 percent) constant with a valley in 2018: Burned land (NO DATA) • Mozambique: Tree cover, needleleaved, evergreen, closed to open (>15 percent) large value in 2001 then all zeros: ESA-CCI LC change • Angola: Tree cover, needleleaved, evergreen, closed to open (>15 percent) large value in 2001 and 2002 then all zeros: ESA-CCI LC change • Kenya: Tree cover, needleleaved, evergreen, closed to open (>15 percent) rapid drop to zero: ESA-CCI LC change • Nicaragua: Tree cover, needleleaved, deciduous, closed to open (>15 percent) high peaks in 2006 and 2010: ESA-CCI LC change • Honduras: Tree cover, needleleaved, deciduous, closed to open (>15 percent) rapid increase in 2007: ESA-CCI LC change • Malaysia: Tree cover, needleleaved, deciduous, closed to open (>15 percent) all zeros after 2006: ESA-CCI LC change • Belize: Tree cover, needleleaved, deciduous, closed to open (>15 percent) all zeros after 2012: Burned land (NO DATA) GLOBAL ESTIMATES OF CARBON STOCKS IN THE VEGETATION AND SOILS OF TERRESTRIAL ECOSYSTEMS 75