Revised and updated geospatial monitoring of twenty-first century forest carbon fluxes
Abstract. Forests are a key component of climate change mitigation strategies because they both emit and remove atmospheric carbon dioxide. Earth observation data are increasingly used to estimate the magnitude and geographic distribution of greenhouse gas (GHG) fluxes and reduce overall uncertainty in the global carbon budget, including for forests. Here we report on a revised and updated geospatial, Earth observation-based forest carbon flux modelling framework that maps GHG emissions (Gibbs et al. 2024a), carbon removals (Gibbs et al. 2024b), and the net balance between them (Gibbs et al. 2024c) globally from 2001 onwards at roughly 30-meter resolution (Harris et al. 2021, hereafter referred to as the Global Forest Watch (GFW) model). Beyond updating the model to include the most recent years, revisions address some of the original model’s limitations, improve model inputs, and refine the uncertainty analysis. We found that between 2001 and 2023, global forest ecosystems were, on average, a net carbon sink of -5.5 ± 8.1 (one standard deviation) gigatonnes CO2 equivalent yr-1 (Gt CO2e yr-1), which reflects the balance of 9.0 ± 2.7 Gt CO2e yr-1 of GHG emissions and -14.5 ± 7.7 of carbon removals, with an additional -0.20 Gt CO2e yr-1 transferred into harvested wood products. Uncertainty in gross removals was greatly reduced compared to the original model due to refinement of temperate secondary forest carbon removal factor uncertainties.
To increase the conceptual similarity between fluxes from the GFW model and national greenhouse gas inventories (NGHGIs) and further policy relevance, we translated (re-allocated) GFW’s estimates of gross emissions and removals into fluxes from forest land and deforestation, i.e. the land use categories that countries use to report anthropogenic forest-related fluxes from managed land in their NGHGIs. We estimated a global net anthropogenic forest-related sink of -3.5 Gt CO2e yr-1 (-3.7 Gt CO2e yr-1 including transfers into harvested wood products). Emissions from deforestation ranged between 3.3 and 5.0 Gt CO2yr-1 and the net anthropogenic sink in managed forest land ranged between -6.8 and -8.5 Gt CO2e yr-1, reflecting ambiguity about the reporting category to which countries assign emissions from loss of secondary forests within shifting agriculture systems. We categorized the remaining net flux of -2.2 Gt CO2e yr-1 reported by the GFW model as non-anthropogenic (0.37 Gt CO2e yr-1 emissions and -2.5 Gt CO2e yr-1 removals). The magnitude of the GFW model’s annual average deforestation emissions and the global anthropogenic forest sink aligned well with aggregated NGHGIs, although their temporal trends differed; NGHGIs reported a slightly increasing forest land sink and fluctuating deforestation emissions, while the GFW model reported a declining sink and increasing deforestation emissions.
Updates to the model and the revised uncertainty analysis demonstrate a spatially explicit forest carbon flux monitoring framework that is increasingly transparent, operational, timely, and flexible enough to answer research and policy questions. Moreover, the translation of Earth observation-based flux estimates into the same reporting framework as countries use for NGHGIs can help build consensus on land use carbon fluxes, support the independent evaluation of progress towards Paris Agreement goals, and assist national policymakers in locating sources and sinks of forest carbon and their drivers.