PAMS-Constrained Top-Down Calibration of VOC-Speciated CMAQ Simulations
Abstract. Accurate simulation of volatile organic compound (VOC) emissions and their role in ozone (O3) formation remains a persistent challenge in chemical transport models (CTMs). Most models rely on lumped surrogate species, limiting their ability to represent speciated VOCs and directly compare with observations. In this study, we develop an enhanced version of the Community Multiscale Air Quality model, termed CMAQ-PAMS, which explicitly incorporates 54 VOC species targeted by the Photochemical Assessment Monitoring Stations (PAMS) network in Taiwan.
We evaluate model performance during a representative high-ozone event in fall 2021 and apply a top-down calibration approach using hourly VOC data from 12 PAMS sites. The original simulation (OrigSIM) significantly misrepresents key species, largely due to reliance on U.S.-based speciation profiles. After adjustment, the modified simulation (ModSIM) shows substantial improvements in both individual species concentrations and group-level composition (e.g., alkanes, aromatics). Notably, acetylene, a key tracer of incomplete combustion, was underestimated in OrigSIM but successfully recovered in ModSIM.
Despite accounting for only ~32 % of total VOC emissions, PAMS species contribute up to 52 % of modeled domestic O3 formation, highlighting their disproportionate impact in VOC-limited regimes. Additionally, the CMAQ-PAMS framework enables the use of diagnostic ratios (e.g., propylene/acetylene) to identify emission sources and assess air mass aging. These findings underscore the importance of localized VOC profiling and demonstrate that the PAMS-constrained CMAQ-PAMS model provides a more chemically detailed and observationally anchored platform for ozone modeling and regulatory applications.
This manuscript by Chen et al. modified the chemical mechanism of the CMAQ model to incorporate 54 VOC species included in PAMS. Furthermore, the anthropogenic VOC emissions in the emission inventory were adjusted under the constraint of PAMS observation data from Taiwan, resulting in modeled VOC concentration results that are close to the observed levels. This study conducts model evaluation for individual PAMS VOC species.
Main comments:
1. The authors appear to frame the inclusion of PAMS species as a key feature of this study. Based on current research in this field, lumped mechanisms are widely recognized as a reliable method for improving model operational efficiency—specifically by grouping species with similar photochemical reaction behaviors. Different gas-phase chemical mechanisms may treat certain species (deemed important) as individual components, whether in the context of emissions or chemical reaction mechanisms. More notably, the explicit Master Chemical Mechanism (MCM) includes no fewer than 6,000 explicit species. Therefore, the novelty of developing the CMAQ-PAMS modeling system needs to be discussed in Introduction and Discussion section.
2. Although Figure 1 is provided to introduce the CMAQ-PAMS modeling system, it only shows where modifications were made to the model. The process of inventory adjustment and model modification are needed. For instance, how to generate emissions of PAMS species into the CB mechanism via SPECIATE. Mabe it is necessary to include at least one table in the appendices that presents the proportion of each PAMS species within lumped species, preferably disaggregated by emission source.
3. There is no doubt that using VOC observation data to constrain emissions is an effective method for improving VOC simulation performance. However, details are needed for the approach used to adjust emissions:
4. Lines 90–91: Adjustments to PAMS emissions would likely have a significant impact on O3 formation—insights that could help us better understand the role of VOC emissions in pollutant simulation. What are the results of O3?
Minor comments:
1. There are multiple instances of inconsistent subscript formatting for O3 and NOx throughout the manuscript, such as in Line 9, Line 29, Line 35, and Line 37. Please standardize subscript formats; as inconsistencies may be interpreted as evidence of AI-generated content.
2. Lines 34–39: This section summarizes the research status of VOCs in air quality models, but every sentence lacks supporting literature citations.
3. There are numerous issues with the reference citation format in the manuscript. For example: Line 50: Cardelino and Chameides (Cardelino and Chameides, 1990)→Cardelino and Chameides (1990); Line 87: (Knote et al.) → (Knote et al., 2015). Please standardize all reference citations to comply with the journal’s formatting requirements.
4. Sections 2–4 can be merged into a single chapter titled "Materials and Methods".
5. Lines 142–144: For the 2021 simulation, the 2010 emission inventory used in the manuscript appears outdated. Although the simulation performance of NOₓ and O₃ shown in Figure S3 seems acceptable, statistical parameters (e.g., R2, NMB, NME) should be added to verify the simulation accuracy.
6. Improve the standardization of figures and tables. This encompasses, but is not limited to, Figure 3, Table 3, Figure 5, Figure 6, Figure 8, and most figures in the supplementary information. Specifically: Use a consistent Arial font across all figures; Standardize axis scales and labels; Ensure no single figure spans multiple pages. These adjustments are necessary to meet the journal’s formatting standards.
7. Lines 350–356: Does Table 2 present the average values of adjustment coefficients? If so, this should be clearly indicated in the table title to avoid ambiguity for readers.
8. Line 375: The manuscript mentions that the improved simulation includes explicit ISOP compared to previous mechanisms. However, many current research already considered and analyzed ISOP as an individual species. The authors should clarify: What distinguishes your results from these prior studies?