The role of climate and humans in shaping interannual variability in biomass burning emissions: The Global Fire Emissions Dataset (GFED v2)

The amount of carbon that is lost from the biosphere to the atmosphere as a result of biomass burning is poorly quantified, but these emissions are thought to contribute substantially to interannual variability in the growth rate of many atmospheric trace gases and are a main source of aerosols. Traditionally, biomass burning emissions are quantified using field work or satellite derived information on burned area and fuel loads ("bottom up"), but recent studies using inversion techniques and atmospheric measurements ("top down") have provided new means to constrain regional and global levels of fire emissions. Here we have combined lessons learned from several recent inversion studies, with new MODIS data, to improve and extend an existing "bottom up" biomass burning dataset (the Global Fire Emissions Dataset). The dataset now covers the 1997-2003 period but will be updated whenever more recent input data is available and uses a new cluster-based algorithm with MODIS data for estimating burned area. We show how interannual variability in precipitation (tropics) and temperature (boreal region) drives the yearly variations in emissions. Interannual variability of emissions is mostly due to variations in deforestation rates, mainly in tropical America and Southeast Asia, while interannual variability in savanna regions is relatively low. Year-to-year variability is also large in the boreal region. Although most of the area burned occurs in the savannas of Africa, we find that absolute emissions in tropical America exceed those from Africa.