A global model for estimating fuel consumption and fire carbon emissions at 500-m spatial resolution

Fires constitute a key source of emissions of greenhouse gasses and aerosols. Fire emissions can be quantified using models, and these estimates are influenced by the spatial resolution of the model and its input data. Here we present a novel global model based on the Global Fire Emissions Database (GFED) modelling framework for the estimation of fuel consumption and fire carbon emissions at a spatial resolution of 500 m. The model was primarily based on observation-derived data products from MODIS, reanalysis data for meteorology, and an updated field measurement synthesis database for constraining fuel load and fuel consumption. Compared to coarser models, typically with a resolution of 0.25°, the 500-m spatial resolution allowed for increased spatially resolved emissions and a better representation of local-scale variability in fire types. The model includes a separate module for the calculation of emissions from fire-related forest loss, using 30-m Landsat-based forest loss data. We estimated annual carbon emissions of 2.1 Pg C yr^-1, of which around 24% was from fire-related forest loss. Fuel consumption was on average a factor 10 higher in case of fire-related forest loss compared to fires without forest loss. Up to now, emission estimates from our new model are based on MODIS burned area with a 500-m resolution, leading to global emissions similar to GFED4s. However, novel high-resolution burned area datasets based on the Landsat and Sentinel-2 missions reveal substantially more global burned area. Our 500-m global fire model provides a suitable framework for converting these burned area products to emissions, with the prospect of substantially higher global emissions.