Optimizing the emission inventory of volatile organic compounds (VOCs) based on network observations

Hourly observations of 56 non-methane hydrocarbons (NMHCs) performed by a network of photochemical assessment monitoring stations (PAMS) at 11 locations across Taiwan were used to evaluate 56 speciated emissions and the resulting simulations of an air quality model. Based on the PAMS observations at two urban sites, emission modification was made for the 56 PAMS species in the model. To further test the applicability of this emission correction approach, the same modified emissions were subject to seven different meteorological conditions and comparison with observations of all the 11 PAMS sites. Originally there was a minimum of only 8 of 56 species showed agreement with observations for the worst of the 11 PAMS sites and 28 of 56 species for the best site. With modified emissions, the number increased to 13-52 species across the 11 PAMS sites, demonstrating that the simple urban based correction procedure has broad applicability. When applying this modification of PAMS emissions to the simulations of other air quality gases, SO2 and NOx showed small changes compared with observations (−0.27% and −2.51%, respectively), while total VOC concentrations showed significant changes (+15.28%) as a result of the adjustment in VOC emissions (+26.7%). Although VOCs are the precursor of ozone, the relatively large changes in VOC did not seem to affect ozone formation to the similar extent, only resulting in the changes of average O3 by 2.9 ppb (+9.41%). It shows that although the emission modification improves individual VOC simulations, the performance in oxidant simulation is still largely unaltered. Although the original U.S. VOC emission profiles can capture the general features of ambient VOCs, further optimization of emissions may still be needed by referencing extensive observations, so that emissions can better fit domestic conditions and accuracy in model simulations can be improved.