Impact of mobile source emission inventory adjustment on air pollution photochemical model performance

Coupled weather forecasting and chemical transport models are useful tools to evaluate air pollution episodes in big cities for the purpose of forecasting and air pollution abatement measures' evaluation. However, large set of accurate data of various sources and modeling calibrations are needed for such complex modeling system to be reliable. The problem becomes more obvious when the model is operated over a domain in which there is a general lack of accurate input data such as emission inventory data. The current study investigates the possibility of model tuning for more accurate prediction of air pollutant concentrations in the city of Tehran in an air pollution episode as a case study. In the last several years, Tehran has frequently experienced long episodes of air pollution, often resulted in forced holidays in the city. WRF/CAMx modeling system were used to simulate the spatial distribution of pollutants concentrations over the city of Tehran in a highly polluted episode during December 2017. In order to improve model performance, a methodology was developed and used to adjust the emission inventory of the system. Due to lack of national emission inventory and background data, effects of introducing a top-down national emission inventory to the modeling system was investigated. This emission inventory can also be used to prepare more accurate boundary concentrations for countries located near Iran. In addition, emission rates of the most dominant emission sources of the city of Tehran – mobile sources – were adjusted using the results from chassis dynamometer tests. Also, the hourly variations of emission were improved using measured ambient NO2 concentrations. Comparison of modeled and measured concentrations at air quality monitoring stations for various pollutant showed that emission adjustment was the most effective method which resulted in considerable model performance improvement over Tehran. Results showed that due to better and more realistic emission factors used in the model, bias error for CO, NO2, O3 and PM2.5 reduced by 0.36 ppm, 3.4 ppb, 0.19 ppb and 2.14 μg/m3, respectively and correlation coefficient for CO, O3 and PM2.5 increased by 0.13, 0.02 and 0.6. This proves the possibility of using such method to improve model performance when there is a general lack of accurate input data. © 2020