Loading...

Adjoint Inverse Modeling of PM2.5 Emissions in Order to Improve Performance of Air Quality Models

Shahbazi, Hossein | 2022

236 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 54981 (08)
  4. University: Sharif University of Technology
  5. Department: Mechanical Engineering
  6. Advisor(s): Hosseini, Vahid; Mozafari, Ali Asghar
  7. Abstract:
  8. In atmospheric studies, chemical transport models are formulated to simulate the spatial and temporal distribution of pollutant concentrations. However, the performance of these models is strongly dependent on the input parameters such as emissions. Inverse modeling is a widely used mathematical approach for estimating model parameters by minimizing the discrepancy between model output and observations. For air quality studies, inverse modeling is often used for emission inversion as emissions are associated with significant amount of uncertainties.This research aims to estimate optimal values for anthropogenic PM2.5 emission through a four-dimensional variational (4D-Var) inverse modeling approach using ground-network observations. The modeling period was chosen January 8th -14th, 2021 over north of Iran. The Community Multiscale Air Quality (CMAQ) model version 5 and its adjoint is used along with Weather Research and Forecasting model (WRF) for meteorological modeling. Adjoint model provides gradients that can be used to provide directions for gradient-based optimization. In this study, background emissions were obtained from Tehran bottom-up emission inventory and national top-down emission inventory.In most of the previous studies, the focus of the research was on the gaseous pollutants or dust and in low spatial resolution. In this work, not only the inverse modeling of emissions for a higher spatial resolution was addressed, but also ability to constrain anthropogenic PM2.5 emission were studied. BComparing background and optimized emissions for Tehran showed that the level of uncertainty in background emissions is higher in northern and western part of the city. Daily scaling factors in different regions of the city varied between 0.44 to 9.44. In the central and southern areas of Tehran, with higher emissions rates, the uncertainty in the background emissions were lower compared to other regions. Resulted daily scaling factors showed that in the central, eastern and southern areas of Tehran and Rey city, the scaling factors on Friday (weekend) were larger than weekdays. In these areas, on weekdays, the daily scaling factor varied between 0.44 and 2.19, and the range of values for Friday ranged from 1.67 to 3.06. The values of scaling factors for the western regions of the city were between 4.28 to 7.06 and for the northern regions were between 5.85 to 8.73.In urban areas around Tehran, the higher level of uncertainty in the background emissions were obtained for the cities of Karaj, Islamshahr, Pakdasht and Qarchak/Varamin, respectively. In these areas, the minimum daily scaling factor was 2.7 in Varamin and the maximum scaling factor was 9.44, 9.31 and 9.2 in Karaj, Pakdasht and Islamshahr, respectively. In Qom and Saveh cities, the quality of background emissions were better than other cities compared to the optimized values, and the daily scaling factor for these two cities varied between 0.82 to 1.12 and 0.76 to 2.14, respectively. The results show that in future updates of emission inventory more details on activity data and emission factors should be considered for the cities of Karaj, Islamshahr, Pakdasht and Qarchak/Varamin.Discrepancy between modeled and observed concentrations before and after emission inversion showed significant improvement in CMAQ performance in calculating PM2.5 concentrations. Results showed that four-dimensional variational inverse modeling approach is an efficient method in reducing the uncertainty of input parameters of air quality models.Finally, in order to investigate the comprehensiveness of the optimized emissions obtained from inverse modeling, the CMAQ model was performed in another simulation episode in two cases. CMAQ outputs when utilizing optimized emissions were compared with the baseline CMAQ run using background emissions. During this period, only the forward model was implemented and the results showed an improvement in the performance of the model when using optimized emission rates
  9. Keywords:
  10. Air Quality Modeling ; Inverse Modelling ; Data Assimilation ; Community Multiscale Air Quality Modeling System (CMAQ) ; Weather Research and Forecasting (WRF)Modeling System ; Diffusion Information ; Uncertainty

 Digital Object List

 Bookmark

No TOC