Dynamic savanna burning emission factors based on satellite data using a machine learning approach

Abstract. Landscape fires, predominantly in the frequently burning global savannas, are a substantial source of greenhouse gases and aerosols. The impact of these fires on atmospheric composition is partially determined by the chemical breakup of the elements in the fuel into individual emitted chemical species, which is described by emission factors (EFs). These EFs are known to be dependent on, amongst others, the type of fuel consumed, the moisture content of that fuel and the meteorological conditions during the fire, indicating that savanna EFs are temporally and spatially dynamic. Global emission inventories, however, rely on static biome-averaged EFs which makes them ill-suited for the estimation of regional biomass burning (BB) emissions and for capturing the effects of shifts in fire regimes. In this study we explore the main drivers of EF-variability within the savanna biome and assess which geospatial proxies can be used to estimate dynamic EFs for global models. We collected over 4500 EF bag measurements of CO₂, CO, CH₄ and N₂O using an unmanned aerial system (UAS), and measured fuel parameters and fire severity proxies during 129 individual fires. The measurements cover a variety of savanna ecosystems under different seasonal conditions, sampled over the course of six fire seasons between 2017 and 2022. We complemented our own data with EFs from 85 fires with known locations and dates listed in the literature. Based on the locations, dates and time of the fires we retrieved a variety of fuel-, weather- and fire severity proxies (i.e. possible predictors) using globally available satellite and reanalysis data. We then trained random forest (RF) regressors to estimate dynamic EFs for CO₂, CO, CH₄ and N₂O and calculated the spatiotemporal impact on BB emissions over the 2002–2016 period using the Global Fire Emissions Database version 4 with small fires (GFED4s). We found that the most important field indicators for the EFs of CO₂, CO and CH₄ were tree cover density, fuel moisture content and the grass to litter ratio. The grass to litter ratio and the nitrogen to carbon ratio were important indicators for N₂O EFs. RF models using satellite observations performed well for the prediction of EF variability in the measured fires with out-of-sample correlation coefficients between 0.80 and 0.99, reducing the error in EF estimates by 60–85 % compared to static biome averages. Using dynamic EFs, global savanna emission estimates for 2002–2016 were 1.8 % higher for CO while CH₄ and N₂O emissions were respectively 5 % and 18 % lower compared to GFED4s. On a regional scale we found a spatial redistribution compared to GFED4s with higher CO, CH₄ and N₂O EFs in mesic regions and lower ones in xeric regions. Seasonal drying resulted in a decrease of the EFs of these species with the fire season progressing, with a stronger trend in open savannas than woodlands. Contrary to the minor impact on annual savanna average emissions, the model predicts localized reductions in EFs of CO, CH₄ and N₂O exceeding 60 % under seasonal conditions.