Sensitivity of Estimated Wildfire Emissions to Variations in Vegetation Carbon Pools

Fires are a global phenomenon with diverse influences on vegetation distribution, biogeochemical cycles and the atmospheric composition. Global gridded carbon emissions are a useful dataset for various applications including climate and atmospheric sciences. The estimation of global carbon emissions is based on the combination of estimates of burned area, available fuel and combustion completeness. Different approaches for the estimation of biomass are being used: from static approaches where the amount of biomass only depends on the land cover type to dynamic approaches where vegetation models provide the variations in available fuel due to, e.g. vegetation phenology, meteorology or disturbances. In this study we assess the influence of seasonal variations, spatial variations driven by climate and long term changes due to CO2 fertilization in vegetation carbon pools on burned area and fire emissions on global scale. We use a land surface model (JSBACH) including an implementation of the mechanistic fire model SPITFIRE. We perform a reference simulation with daily updated carbon pools and simulations with the fuel load fixed to the mean fuel load per grid cell, to maximum monthly fuel load and the minimum monthly fuel load to address the influence of seasonal variations. We address the influence of climate driven spatial variations by using constant fuel loads per plant functional type (PFT) in an additional simulation. The influence of CO2 fertilization on fire occurrence and emissions is covered by comparing a simulation with historical CO2 increase, with a simulation where we use constant preindustrial CO2 with the same meteorological forcing. The simulation using the mean fuel load per grid cell shows similar results to the reference simulation with respect to global values, spatial distribution and variability in time. The simulations using minimum and maximum fuel show a large difference in global values, but similar spatial distributions. In the simulation based on constant fuel loads per PFT the peak in emissions along a precipitation gradient is shifted towards lower precipitation values. The simulation with preindustrial CO2 fertilization shows a strong reduction in carbon emissions and in burned area compared to the historical simulation.