Current estimates of the gross tropical deforestation flux from recent carbon cycle synthesis reports are about 2.0 Pg C/y during 2000-2020. This flux is primarily associated with large-scale, commodity-driven agriculture and small-scale shifting cultivation. These two deforestation drivers are known to be closely coupled to fire emissions because cutting, aggregating, drying, and burning residual biomass, often in several cycles, is the fastest way to prepare the land for cultivation. In contrast with these global-scale carbon cycle assessments, the sum of fire emissions from tropical forest deforestation, forest degradation, and peat fires from the 5^th version of the Global Fire Emissions Database (GFED5) is about 0.4 Pg C/y during 2003-2020. Estimates of fire emissions from GFED5 may be low due to extensive cloud cover and low active fire detection efficiencies in tropical regions with high forest cover.

Here we explore the impact of fire emissions from GFED5 on atmospheric composition across different tropical continents during 1997-2022, focusing on carbon monoxide, ozone, and aerosol optical depth using the Energy Exascale Earth System Model (E3SM). New observations and syntheses of emission factors from peat, deforestation, and savanna fires reduce uncertainties with respect to fire impacts in these simulations. We conduct two sensitivity experiments with E3SM. In one sensitivity simulation, we exclude all fire emissions but retain all other emission sources (a no-fire simulation). In a second simulation, we increase the tropical deforestation and forest degradation flux by a factor of 4. We then compare these simulations to observations of column CO from MODIS, tropospheric column O₃ from OMI/MLS, and aerosol optical depth from MODIS and VIIRS. In a final step, we describe the range of gross deforestation fire emissions compatible with the satellite-derived constraints and the implications of these adjustments for the land component of the global carbon budget.