Abstract. Fires are a major source of trace gases and aerosols to the atmosphere. The amount of biomass burned is becoming better known, most importantly due to improved burned area datasets and a better representation of fuel consumption. The spatial and temporal variability in the partitioning of biomass burned into emitted trace gases and aerosols, however, has received relatively little attention. To convert estimates of biomass burned to trace gas and aerosol emissions, most studies have used emission ratios (or emission factors (EFs)) based on the arithmetic mean of field measurement outcomes, stratified by biome. However, EFs vary substantially in time and space, even within a single biome. In addition, it is unknown whether the available field measurement locations provide a representative sample for the various biomes. Here we used the available body of EF literature in combination with satellite-derived information on vegetation characteristics and climatic conditions to better understand the spatio-temporal variability in EFs. While focusing on CO, CH4, and CO2, our findings are also applicable to other trace gases and aerosols. We explored relations between EFs and different measurements of environmental variables that may correlate with part of the variability in EFs (tree cover density, vegetation greenness, temperature, precipitation, and the length of the dry season). Although reasonable correlations were found for specific case studies, correlations based on the full suite of available measurements were lower and explained about 33%, 38%, 19%, and 34% of the variability for respectively CO, CH4, CO2, and the Modified Combustion Efficiency (MCE). This may be partly due to uncertainties in the environmental variables, differences in measurement techniques for EFs, assumptions on the ratio between flaming and smoldering combustion, and incomplete information on the location and timing of EF measurements. We derived new mean EFs, using the relative importance of each measurement location with regard to fire emissions. These weighted averages were relatively similar to the arithmetic mean. When using relations between the environmental variables and EFs to extrapolate to regional and global scales, we found substantial differences, with for savannas 13% and 22% higher CO and CH4 EFs than the arithmetic mean of the field studies, possibly linked to an underrepresentation of woodland fires in EF measurement locations. We argue that from a global modeling perspective, future measurement campaigns could be more beneficial if measurements are made over the full fire season, and if relations between ambient conditions and EFs receive more attention.
Estimation of forest canopy cover: a comparison of field measurement techniques
