Application of a combined standard deviation and mean based approach to MOPITT CO column data, and resulting improved representation of biomass burning and urban air pollution sources

Abstract This work presents a new methodology to simultaneously account for both the mean and variance of 17 years of Carbon Monoxide [CO] measurements from the MOPITT satellite (from 2000 to 2016) over Southeast and East Asia. We demonstrate that the new technique is stable and produces a set of results which are both consistent with the understood geographical and temporal distribution of CO sources, as well as those of some other co-emitted species. These regions were chosen because they have high levels of CO loadings and complex factors driving their underlying emissions profiles. We first successfully categorize the region, based on the total CO column measurements, into those locations impacted by intense urbanization (high climatological mean, low climatological variance, high loading throughout most of the year, and variation mostly due to global-scale chemistry and climatology), large-scale biomass burning (low climatological mean, high climatological variance, mostly clean but with short and intense peak events occurring around a similar time year-to-year), those regions undergoing a significant change from one type to another, and those regions which are clean. We further reproduce the temporal and spatial distributions of other co-emitted species measured by other measurement platforms, including aerosols (AOD) and gasses (NO2), and demonstrate consistency across all three platforms. Third of all, we produce an important scientific finding using these new results, in terms of a significant geographic expansion of the known biomass burning regions from Myanmar through Northern Vietnam, as compared with previous research. Fourth, we also find that urbanization dominates emissions of CO over both known urban agglomerations in East and Southeast Asia, including around Beijing, Shanghai, the Pearl River Delta, Seoul, Shandong, Chengdu, Chongqing, Hanoi and Bangkok, but also extend into otherwise unclassified or recently emerging mega-cities including Shanxi Province, Henan Province, and Liaoning Province. Additionally, we have found that there is a 37% overlap with emissions from FINN, providing both validation of the technique over known biomass burning regions, and an ability to quantify regions that are burning but not presently identified. Finally, we have a consistent and value-adding comparison and contrast with an EOF/PCA analysis in terms of both our regions classified as biomass burning and mixed/urban, with each approach offering its unique advantages and drawbacks. Although our approach has made some very simple assumptions in terms of cutoff that can and should be improved by the community, its robust results provide new insights into the rapid changes impacting CO throughout Asia, and will allow models to have an improved chance at representing peak events or areas undergoing rapid change.