Revising the Global Budget of Glyoxal (OCHCHO) Based on OMI Vertical Columns

Abstract:

Glyoxal is, like formaldehyde, a short-lived intermediate in the oxidation of non-methane VOCs emitted by plants, vegetation fires and anthropogenic activities. It is also a precursor of secondary organic aerosols. Both compounds absorb in the UV-visible spectral region and have been measured by the SCIAMACHY satellite sensor since 2003, and more recently, by OMI and GOME-2. Previous modelling studies using SCIAMACHY data have pointed to the existence of large additional sources, in particular over forests (Stavrakou et al. 2009), and more recently over Eastern China, most likely due to aromatic hydrocarbons (Liu et al. 2012), suggesting that glyoxal can serve as an indirect estimator of urban VOC sources. The current study is motivated by (i) recent advances in our understanding of chemical pathways leading to glyoxal formation, in particular from the oxidation of isoprene, the most largely emitted NMVOC, (ii) the existence of numerous in situ concentration measurements for the key anthropogenic glyoxal precursors (e.g. acetylene, aromatics) over industrialized areas, which can be used to narrow down the anthropogenic emission estimates in these regions, and (iii) substantial improvements in retrieval algorithms for glyoxal columns from UV-visible satellite instruments, which has led to an significant reductions of the number of unphysical negative columns over the oceans as well as to generally lower glyoxal columns over continents. In this study, the chemical mechanism and NMVOC emission inventories of the global CTM IMAGESv2 are revised based on recent investigations. The relative importance and possible uncertainties of different chemical pathways leading to glyoxal formation in the oxidation of isoprene are determined by box model simulations. Next, OMI (also possibly GOME-2) glyoxal and formaldehyde data are used to constrain the emissions of biogenic, pyrogenic and anthropogenic VOCs. To that effect, the inverse modelling technique using the adjoint model of IMAGESv2 is used. The role of model uncertainties is explored through a number of sensitivity studies. The model results are evaluated against ground-based measurements (MAX-DOAS, SOAS, CABINEX) and aircraft observations (SENEX) for HCHO, OCHCHO and their ratio (R_GF).