Constraining African Wildfire Carbon Emissions Using Satellite XCO₂ Retrievals

Abstract Wildfire carbon emissions are critical in the global carbon cycle, but their estimates remain highly uncertain. Here, we developed an inversion framework to jointly constrain wildfire carbon emissions and net ecosystem exchange using OCO-2 XCO2 retrievals. The observing system simulation experiment shows that this approach significantly improves wildfire carbon emission estimates, especially for Africa where wildfires are concentrated in non-growing seasons. We estimate annual wildfire carbon emissions of 1.18 ± 0.22 PgC/yr in Africa during 2015-2016, approximately 20% higher than the mean of GFED4s and GFAS estimates. Specifically, wildfire carbon emissions in northern Africa were underestimated by ~0.25 PgC/yr, mainly due to the underestimation of burned area, while those in southern Africa were overestimated by ~0.05 PgC/yr, largely due to inflated fuel consumption associated with high tree cover fraction. This study presents a novel approach for constraining wildfire carbon emissions, which can help us to improve wildfire carbon emission models and better understand the carbon cycle dynamics.