The manuscript investigates the triggering mechanisms of new particle formation (NPF) under various O₃ pollution scenarios, focusing specifically on the dynamic balance between precursor sources and the background condensation sink. The study concludes that NPF is highly dependent on the synchronicity between source enhancement and sink attenuation—a process regulated by the evolution of the planetary boundary layer. Overall, this is a highly interesting and well-structured study that provides valuable quantitative insights through the application of interpretable machine learning methods. The methodology is scientifically sound, and the data robustly support the final conclusions. I recommend publication following minor revisions. The authors are requested to consider and respond to the following specific comments:

Major comments

1. Section 3.2 emphasizes the breakup of the morning inversion layer and the subsequent rapid vertical mixing as the key process leading to the synchronization of source enhancement and sink weakening. Considering that the temporal resolution of the UAV vertical observations is 3 hours, whereas this process may occur on a much shorter timescale, is this resolution sufficient to capture such rapid evolution?
2. Section 3.3: In the Non-O₃ scenario, RH shows a strong positive SHAP contribution, whereas T, SO₂, and O₃ show negative contributions. Could the authors provide more scientific interpretation for this?

Minor comments

1. Line 29-33: The first paragraph of the Introduction is rather difficult to read and it is recommended to be reorganized.
2. Lines 107-109: Please delete the space after the hyphen in “Low- O₃” and “High- O₃”.
3. Line 120: “coagulation sink (Coags, S^-1)” is recommended to be revised to s^-1.
4. Line 122: In the equation, should “Cogas” be corrected to Coags? It seems to be a spelling error.
5. Section 2.4: Why was GAM chosen instead of random forest or XGBoost?
6. Line 178: “As show in Table 1” should be corrected to “As shown”.
7. Line 181 & Table1: “Cogas” is also a spelling error here.
8. Lines 219-220: “In contrast…” is repeated from Lines 165–166, please delete it.
9. Lines 226-227: The statement “According to the classification criteria in Sect. 2.2, both the NPF and Low-O₃ scenarios had similarly low O₃ pollution levels (160-215 μg/m³)” does not seem to be mentioned in Section 2.2. Please clarify.
10. Line 254: The citation “Ail et al., 2025” is inconsistent with Ali in the References (Line 392); please revise it.
11. Lines 229-300: The phrase “turning negative beyond approximately 0.085 m·s^-1” is discussing CS, so the authors are advised to check the unit. Should it be s^-1 instead?
12. Lines 316–317: Please revise the caption of Figure 6, as “(a) Mean SHAP values across for scenario” is not grammatically smooth.

This manuscript demonstrates that atmospheric new particle formation (NPF) is governed not by ozone (O₃) levels alone, but by the temporal synchronization between precursor production and background aerosol removal. The authors show that such synchronization is primarily controlled by the evolution of the planetary boundary layer, with NPF occurring when post‑sunrise boundary‑layer development simultaneously enhances near‑surface oxidation capacity through mixing of O₃ from above and reduces the condensation sink. On the other hand, stable PBL conditions can either lead to low O₃ levels or high background aerosol concentations leading to low low source rate or high sink of newly formed particles.

In my opinion, methodological strength of this work is the combination of NPF classification with interpretable machine‑learning tools. By integrating generalized additive models with SHAP analysis and vertical observational data, the authors quantitatively attribute nucleation‑mode particle variability to individual meteorological, chemical, and aerosol parameters under different pollution regimes. This approach represents an advance over traditional correlation analyses and offers a clear framework for diagnosing nonlinear source–sink competition in NPF studies.

In addition to the comments provided by Referee #2, I suggest adding a short discussion on how the authors think the identified source–sink mechanisms would be expected to generalize to other regions, seasons, or pollution regimes. Since this study is based on an intensive observation period of roughly two weeks at a single site, it would help setting this study into broader context and to prevent overinterpretation of the results obtained here.

I recommend the manuscript to be accepted for publication.