Exploring the application of Earth Observation datasets for SEEA carbon accounting and its comparison with national GHG reporting to the UNFCCC

Global biomass and carbon datasets derived from Earth Observation (EO) are rapidly increasing. Here we assess how these datasets can be used to compile carbon accounts aligned with the Systems of Environmental Economic Accounting (SEEA). Five carbon pools are considered: above-ground, below-ground, deadwood, litter and soil organic carbon (SOC) using EO datasets that include CCI Biomass and CCI Land Cover - both subjected to independent map validation. We compiled carbon accounts across multiple accounting periods, from annual to decadal intervals, spanning 2010-2021 for six countries: Brazil, Mozambique, the Netherlands, the Philippines, Sweden and USA. Next, we compared the SEEA-aligned EO-derived carbon accounts with the national greenhouse gas (GHG) inventories submitted to United Nations Framework Convention on Climate Change (UNFCCC) after considering the areas covered by managed forests and emissions from forest degradation for better comparability of the two frameworks. The compiled SEEA carbon accounts indicate that the above-ground component is the dominant carbon pool in Brazil while SOC outweighs other carbon pools in Netherlands, Sweden, Philippines and Mozambique. Results show substantial inter-annual variation in carbon fluxes, exhibiting a strong inverse relationship with accounting period (|r|=0.57-0.85). Such variability is notably higher than the values reported by countries to the UNFCCC. While UNFCCC and SEEA fluxes show moderate overall agreement (r=0.47, 58% agreement whether fluxes are sequestration or emission), there were differences across countries and flux categories. Our compiled accounts showed lower SOC emissions than UNFCCC reports potentially underestimating peatland emission; and minimal carbon emissions from forest degradation. Other sources of disagreement could be influenced by country definitions of managed forests which can be inconsistent with EO-based forest management maps, and the dependency of countries on national forest inventories which are rarely updated annually. The findings and outputs from this study echo the potential and flexibility of EO datasets for carbon accounting and inter-comparison exercises.