14 years of OMI HCHO observations reveal VOC emission trends over large cities worldwide

Formaldehyde (HCHO) observations from satellites have been widely used to constrain volatile organic compound (VOC) emission estimates. The oxidation of anthropogenic organic compounds accounts for only a small fraction(~7%) of the total HCHO column on global average (Stavrakou et al., 2009). Therefore, the use of satellite observations to infer information about anthropogenic VOC emissions is generally very challenging . However, the relative contribution of anthropogenic VOCs in and around metropolitan centers is expected to be significant. In this study, we use HCHO column data retrieved from the OMI sensor between 2005 and 2018, and calculate monthly averages for every city of more than 500,000 inhabitants based on data within 20 km of the city centers. Because of the dependence of the background and especially of the biogenic VOC source on temperature and solar radiation, and because these contributions might be significant even around large cities, it is not possible to directly infer the anthropogenic contribution to the long-term observed HCHO trends based on HCHO data. To remove these non-anthropogenic contributions, we first regress the monthly averaged columns either onto the monthly maximum surface temperature, obtained by ECMWF reanalysis data, or onto the monthly isoprene flux, calculated with the MEGAN-MOHYCAN model (Guenther et al., 2012, Stavrakou et al. 2018). Only cities for which anthropogenic emissions are estimated to exceed biogenic emission by more than a factor of 3 are considered. In this way, positive trends of up to 3% yr^-1 are found over many Asian cities, especially in China and in the Indo-gangetic Plain, whereas over European cities, South Africa and South America negative trends up to -2% yr^-1 are derived. The deduced trends are compared to the corresponding trends of global bottom-up anthropogenic VOC emission inventories and are found to be in good overall agreement. Model simulations are further needed to quantify the relationship between anthropogenic emission trends and HCHO columns, accounting for the effect of non-anthropogenic emissions and potential changes in the oxidizing capacity.