Composition and variability of gaseous organic pollution in the port megacity of Istanbul: source attribution, emission ratios, and inventory evaluation

Abstract

Abstract. In the framework of the TRANSport Emissions and Mitigation in the East Mediterranean (TRANSEMED/ChArMEx) program, volatile organic compound (VOC) measurements were performed for the first time in Istanbul (Turkey) at an urban site in September 2014. One commercial gas chromatograph coupled to a flame ionization detector (GC–FID) and one proton transfer mass spectrometer (PTR-MS) were deployed. In addition, sorbent tubes and canisters were implemented within the megacity close to major emission sources. More than 70 species including non-methane hydrocarbons (NMHCs), oxygenated VOCs (OVOCs), and organic compounds of intermediate volatility (IVOCs) have been quantified. Among these compounds, 23 anthropogenic and biogenic species were continuously collected at the urban site. VOC concentrations show a great variability with maxima exceeding 10 ppb (i.e., n-butane, toluene, methanol, and acetaldehyde) and mean values between 0.1 (methacrolein + methyl vinyl ketone) and 4.9 ppb (methanol). OVOCs represent 43.9 % of the total VOC concentrations followed by alkanes (26.3 %), aromatic compounds (20.7 %), alkenes (4.8 %), terpenes (3.4 %), and acetonitrile (0.8 %). Five factors have been extracted from the Positive Matrix Factorization model (EPA PMF 5.0) and have been compared to source profiles established by near-field measurements and other external variables (meteorological parameters, NOx, CO, SO2, etc.). Surprisingly, road transport is not the dominant source, only explaining 15.8 % of measured VOC concentrations contrary to the local emission inventory. Other factors are toluene from solvent use (14.2 %), biogenic terpenes (7.8 %), natural gas evaporation (25.9 %) composed of butanes, and a last factor characterized by mixed regional emissions and composed of most of the species (36.3 %). The PMF model results point out the influence of industrial emissions while there is no clear evidence of the impact of ship emissions on the measured VOC distribution. For the latter additional measurements of organic compounds of lower volatility like IVOC would be helpful. The sensitivity of PMF results to input data (time resolution, meteorological period, peak episode, interpolation method) was tested. While some PMF runs do not perform as well statistically as the reference run, sensitivity tests show that the same factors (number and type) are found with slightly different factor contributions (up to 16 % of change). Finally, the emission ratios (ERs) of VOCs relative to carbon monoxide (CO) were established. These ratios are usually higher than the ones of other cities worldwide but in the same range of magnitude. These ERs and the road transport factor from PMF were used to estimate VOC emissions and to evaluate three downscaled global emissions inventories (EDGAR, ACCMIP, and MACCity). It was found that the total annual VOC anthropogenic emissions by global inventories were either within the same range by a factor of 2 to 3 for alkanes and aromatics or underestimated by an order of magnitude, especially for oxygenated VOCs.