Item

Abstract

Atmospheric composition is strongly influenced by wildfire emissions, which have a strong variability over time and space. Estimates of fire emissions on large scale are based on a combination of burned area, combustion completeness, and fuel load. Approaches differ in the derivation of this information, which involves models and observations to different degrees. Due to the lack of highly spatially and temporally resolved observations the variability of fuel load is often not fully taken into account.The fuel load can differ between seasons due to variations in the vegetation productivity, decomposition rates and fire occurrence. On the longer time scale the effect of CO2 fertilization is expected to influence the vegetation productivity and therefore overall fuel load abundance. All these processes are accounted for in land carbon cycle models. We use the land surface and vegetation model JSBACH as a tool to understand the influence of fuel load seasonality, fuel load variability within land cover types and CO2 fertilization on fire occurrence and wildfire emissions.We find that using the mean fuel load over time for each grid cell instead of seasonally varying fuels leads to comparable burned area and emissions (only 3% deviations from the reference). Using minimum or maximum values, however, leads to strong under (0.54 times the reference) and overestimation (1.85 times the reference) of the emissions. When using constant fuel load for each vegetation type strong regionally varying, over and underestimations of emissions are found. Over the 20th century CO2 fertilization strongly impacts fuel availability. As a consequence, burned area and carbon emissions are almost 20 and 40% higher at present day.In general, our results confirm the applicability of time constant fuel loads in emission estimation methods for present day, as the seasonality is of minor importance. However, we suggest that considering the variability of fuel driven by climate variability in space can improve the estimates. This result is in line with a number of studies highlighting the importance of fuel limitation for the occurrence of fire. On the longer time scale the influence of CO2 fertilization is not negligible according to our results, but high uncertainties in the understanding of the process increases the difficulty to account for it in fire carbon emission approaches. This assessment of potential errors in fire emission datasets should help to further improve approaches to estimate fire emissions and to interpret available datasets and differences between them. © 2015 The Authors.