Wildfires are a significant source of nitrogen oxides (NOx = NO + NO2) to the global atmosphere, representing approximately 15% of the total NOx budget. Fire conditions that govern NOx emission vary significantly from fire to fire, resulting in highly variable emissions. Emissions from fires burned in a laboratory setting fail to reproduce the conditions in which large wildfires occur, such as fire size and meteorology; however, in situ measurements of fire emissions are challenging to make, in part due to the destructive nature of large wildfires. As a result, systematic variability of NOx emissions--even when normalized for biomass burned--across or within biomes is poorly understood and documented.
In this dissertation, I demonstrate that the high spatial and temporal coverage of space-based observations can be used to greatly increase the number and scope of available observations of actively burning wildfires. I derive a method to estimate NOx emission coefficients (ECs in g NOx MJ-1) using NO2 column densities from the Ozone Monitoring Instrument (OMI) and fire radiative power from the Moderate Resolution Imaging Spectroradiometer (MODIS), two Earth-observing satellite instruments. I show that this method, when applied in California and Nevada, reproduces differences in fire emission factors (EFs in g NOx kg-1) between fuel types that have been previously observed using in situ measurements. I then identify and explore sources of variability in NOx ECs in fires across the globe. I compare mean ECs for fires in different locations but similar biomes (e.g. grasses) and find that while most ECs cover a relatively narrow range, ECs for several locations are significantly different from the mean biome EC. I examine seasonal variability in ECs, finding that ECs in African woody savannas have a strong seasonal dependence that is not observed in open savannas; this behavior may be related to reallocation of nitrogen to below ground during the fire season by plants and/or the seasonal variation of fire fuel composition in woody savannas. I also find that this behavior extends to other woody savanna regions in South America and Australia, and that ECs in several biomes exhibit a dependence on wind speed.