Referee’s comments on Development and application of a street-level meteorology and pollutant tracking system (S-TRACK)

General comments

The authors present a new multi-modeling system S-TRACK which combines the Weather Research and Forecasting (WRF) model for the regional meteorological modeling and the STAR-CCM+ model for the street-scale air flow modeling and the FLEXPART model for a backward trajectory modeling.

The multi-scale modeling is an important issue to improve modeling performance of street-scale meteorology and air quality. In this manuscript, downscaling is conducted from the linking between the regional-scale model to the street-scale model. However, more detailed description for the linking is necessary (see my specific comments 1 and 5).

The most interesting part is a new research method to track the sources of pollutants in the street level using a backward trajectory modeling. The authors should give more analysis on the results as suggested in my specific comments 2 and 3.

A major revision is needed to improve the manuscript before the publication.

• Scientific significance: Good
• Scientific quality: Good
• Presentation quality: Fair

Specific comments

1. The authors need to explain in more details how three components of S-TRACK are coupled. What is the major development by the authors to obtain the final simulation results? For example, in Figure 1, a regular and fine grid having the spatial resolution of 10 m is constructed by combining WRF and CFD results. Why is this refined grid necessary for FLEXPART? How is it done? An another example is in l. 164. According to the authors, WRF results fill some missing data from CFD results. It is not clear. What data are missing in CFD model? How have the authors filled this gap?
2. I think that one of the major findings in this manuscript is the potential impact of the traffic source is the largest at 15 m (l. 309). However, the authors’ interpretation is not clear. They linked the result to the distance of the building from the source. This result should be explained in details taking into account the background wind field. And this result should be added in Conclusion.
3. Another major finding is in l. 329. The authors compare the contribution ratio during westerly wind and easterly wind. The higher ratio during westerly wind is explained by the building distribution. Why does the dense building distribution of upwind of the receptor site lead to a higher contribution ratio? If the authors can explain it, this result should be added in Conclusion and Abstract.  
4. In l. 21, “which is more obvious for high buildings and influencing air pollution transport at the street-level.” is not discussed in the text. Is it related to any results of this study?
5. One of important issues on the atmospheric modeling using CFD model is the computation time. How have you conducted a simulation on such a large domain in Figure 3 for about one month? Could you briefly explain the technical part in Appendix?

Minor comments

1. l.12, l.86, l.93, and many other lines: the CFD model name should be given.
2. l.26-27: please rewrite the sentence.
3. l.37: why is the term “diffusion” used? Many times, “transport and diffusion” appear in the text (l.203, l.224). “diffusion” is one phenomena of transport.
4. l.69: remove “In 2006”
5. l.69-70: the sentence is not clear.
6. l.74-75: the sentence is not clear. What do you mean with “the spatial resolution is affected by the numerical dispersion in the Eulerian model?”
7. l.96-98: as Major comment 1, explain which data of WRF are used to fill the gap.
8. l.116: remove “powerful”. Please add the references that the STAR-CCM+ has been used for street-level simulation to show its modeling performance.
9. l.122-123: correct the section title to “3D street-level grid generation”. What do you mean “geometric model”?
10. l.135 and l.138: correct “grids” to “grid cells”. Please don’t mix “grid” and “grid cells”
11. l.139: I think the equations are not useful to explain the STAR-CCM+ model. Please rewrite this section focusing on the coupling of WRF model to STAR-CCM+.
12. l.168: why are the turbulence options needed in FLEXPART? Presenting the main idea of the FLEXPART may be helpful. “CBL” can be removed.
13. l.179: add Wind direction. It is presented in Figure 4.
14. l.181: correct mean deviation to mean bias
15. l.189-190: T and RH are underestimated. P and WP (correct W to WP) are overestimated.
16. l.192: explain what is the significance test at least in Appendix.
17. l.197-199: This sentence does not have proof. Please remove it.
18. l.206: the averaged wind speed of 0.92 m/s is very now speed. Models cannot easily reproduce the low speed. Why is the observed wind speed very low? Is it a typical value on the simulation domain in winter?
19. l.218: This sentence does not have proof. Please remove it.
20. l.235: How is the influence of buildings on the wind field estimated? How do you know it is diminished?
21. l.299-l.302: the authors may remind the results in Section 3.2 to link the differences in the wind field to potential contribution ratio.
22. Figure 4: Why have you not compared WRF wind direction?
23. Figure 5: Please make figures bigger and explain what is the divergence.
24. Figure 7: Explain what is the density distribution in the text.
25. Table 3: Are the potential contribution ratio the averaged value of all wind directions?

The authors presented an interesting work that integrates the Weather Forecasting Model (WRF),  Computational Fluid Dynamics (CFD) and Flexible Particle (FLEXPART) to study local circulations at a neighbourhood scale and the potential contribution of one street as a source of air pollution.  The authors analyzed local wind circulation under different regional wind regimes. The methods used by the authors are robust and the meteorology simulated shows good agreement as shown in figure 4. However, conclusions, as stated in lines 338-339, authors say: "In general, the S-TRACK system is effective in simulating meteorological and air pollution problems" has no strong basis. This is because the is no real emissions and only one street was evaluated. Specifically:

1) in the abstract, the authors state: "The results of the study are helpful to understand the characteristics of wind environment and effect of traffic emissions in the area...", however, there are no traffic emissions in the study.

2) The authors presented the potential impacts of only one street. however, we do not know anything about that street, type of traffic, the diurnal cycle of traffic by type of vehicle and as shown in Figures 3a, there are many streets surrounding the area.  For instance, Figure 7b shows that the street might be important for site S2, but this site is closer to a perpendicular street (east), as shown in figure 3a. In a summary, despite that the results are interesting, more work is needed to provide the basis for the conclusions.

3) Figure 1 shows that the results are interpolated but the authors do not show the methods and reasons for this interpolation.

minor issues:

The authors simulated between December 30 2018 and January 31, 2021.  Why this period?

Line 69: add space before Fast and remove space in "( 2006"

Lines 84-88. Only one phrase for a 5 lines paragraph. Each paragraph must be formed by at least three phrases, introduction, body and conclusion.

Line 124: Why do you need drone aerial photography? I think it is very good, but I'm surprised that you are not trying other sources.

Lines 162-164: "This... FLEXPART", this phrase is redundant.

Lines 172: What was the number of particles released?

Lines 197-198: Simulating well the PBL is important to the dispersion of pollutants. How are the results? The authors can include more results in the supplementary material.

Lines 219: This line seems incomplete.

Section 3.2.1. There are many parts where English needs to be improved with a more technical language. For instance line 232 authors say: "The wind is larger", maybe more intense is more appropriate. Lines 234-237 needs edit/improvement.

Line 240: change climb or fall subsidence or other.

Line 244: Change pile up for other.

Line 261: Line seems repeated.

Figure 5 showing streamlines and divergence. Consider merging streamlines with shaded values for divergence. In that way, it would be easier to see diffluence.