Development, Implementation, and Application of an Improved Protocol for the Performance Evaluation of Regulatory Photochemical Air Quality Modeling Public Deposited

Downloadable Content

Download PDF
Last Modified
  • March 20, 2019
  • Kim, Byeong-Uk
    • Affiliation: Gillings School of Global Public Health, Department of Environmental Sciences and Engineering
  • Ozone is a secondary pollutant resulting from complex reactions of two precursors: nitrogen oxides (NOX), and volatile organic compounds (VOCs) under ozone-conducive meteorological conditions. Thus, the ozone modeling becomes complex and needs rigorous model performance evaluations (MPE) before the modeling results are used for air quality decisions. In the past regulatory ozone modeling, however, virtually all MPE practices were over-simplified by following the EPA's current MPE method. That is, modelers cannot answer the most important question in applying air quality models for ozone decision-making processes with the EPA's MPE method: "why should I believe this modeling?" In this study I investigated a solution by integrating the theoretical advances of MPE for environmental modeling with my practical knowledge in regulatory ozone modeling. As a result, I developed an MPE method with which modelers must (1) gather and examine graphical/statistical measures in a systematic manner, (2) conduct in-depth analyses with respect to potential ozone control options, and (3) report their performance assessments explicitly in light of policy questions. Because the existing analysis tools showed significant shortcomings in implementing the new MPE method, a new tool was developed to exercise the new MPE method efficiently. With the new tool, modelers can accomplish MPE tasks necessitated by the new MPE method in a timely manner. iii The Houston-Galveston Mid-Course Review (HGMCR) modeling was re-evaluated as the case study to demonstrate the advantages of new MPE method. I could reveal that the HGMCR modeling showed significantly low reliability even though the model could pass the majority of EPA's simple statistical tests. That is, the model showed significantly high biases in winds, NOX, and VOCs. Two major roots of high biases were identified: (1) the highly reactive VOCs (HRVOC) adjustment that was not scientifically defensible and (2) the insufficient modeling grid resolution with respect to the nature of ozone problems in Houston. Ultimately, the application of new MPE method led me to develop an alternative modeling case with which I showed that the alternative case could be used in a limited way to test a certain type of HRVOC control strategies by reducing VOCs biases.
Date of publication
Resource type
Rights statement
  • In Copyright
  • Jeffries, H. E.
  • Open access

This work has no parents.