In this manuscript, the authors gave a very details analysis mostly based on 3 year ozone and other gas pollutants observations on a 488m high Canton Tower with 4 levels. The data collected were precious, and the topic is of great interesting to recognize ozone vertical exchanges within boundary layer (BL) and related to ozone diurnal variation. The analysis is mostly sound, but some details need clarify. I recommend a minor revision and my specific comments listed below.

Specific comments:

1. Refer to the discussion in 341-344, the study address the cases that rule out wind speeds above 2 m/s at four vertical layers, and say “exclude the possible influence of horizontal transport”. My concern is that in a consecutive event (NOE), some periods could be in smaller wind speed, sometimes in bigger. While, I do not think that means horizontal transport is not important in a NOE. My interesting is what about the influence of horizontal transport, because horizontal transport from rural sites may also induce high surface ozone and the emission from high stacks could also contribute to high CO and NO2 in RL. I suggest a detailed analysis of some typical cases, also at least compared the model results of ΔVDIF and ΔADV in different levels.
2. This study identify a strong ozone residual capacity, defined as the ratio of the ozone concentration averaged over nighttime to that in the afternoon (14:00-17:00 LT). As the ratio in mathematics, I suggest to replace the definition of “ozone residual capacity” to “ozone residual ratio”.
3. It’s not exact to say “weakens the titration effect by diluting NOx concentrations.” during NOE. It could be better to say “offset the surface ozone decrease by NO diluting”.
4. 6 and the explanation in line 270-279 were not sufficient. Except the “ozone-rich air at higher altitudes to mix with air in the lower boundary layer.”, the higher ozone produced by photochemistry near surface should also convectively transport to upper BL or low free troposphere.
5. In fig. 12, what’s the result if you statics the relation of surface ozone before noon (for example 9:00-11:00) and before the sunrise (for example 5:00-6:00)?
6. In introduction, in line 58-61, I suggest move and combined these sentences that introduce your study to line around 95. Also, please polish the context and concisely present the analysis.

General comments

The paper by He et al., investigated vertical distributions and process contributions of the nighttime boundary-layer ozone in Southern China using 3-year tower-based measurements. As indicated by the authors, the continuous gradient measurements of ozone in the lower boundary layer, particularly in urban regions, are very important for clarifying vertical exchange characteristics of ozone and thus elucidating the reasons that regulate surface ozone air quality. The paper is well written and organized. The analysis regarding vertical distributions and key drivers of ozone was reasonable and has been well supported in the literature. Therefore, I only have some small concerns that may be further explained by the authors before its publication.

Specific comments:

Line 115: How the wind information was measured on the tower? was it measured inside or outside the outer shell of tower? The structure of the tower may cause complicated and different turbulences affecting the measured winds on different altitudes.

Line 205: The vertical gradients of ozone in the nighttime boundary were much stronger than in the daytime due to the inhibition of the vertical mixing. The authors also stated that the vertical gradients of Ox mixing ratios are much weaker than those of ozone. Therefore, the positive gradients of vertical ozone profiles were mainly determined by the gradually reduced NO titration effect. Were any vertical gradient measurements of NO on the tower that can be used to support this conclusion?

Line 235-240: As highlighted by the authors, the discrepancies between modeled and measured ozone in urban regions aloft may be caused by many factors. In my opinion, a grid with a small spatial scale of 3×3 km² in the urban region may be not the dominant factor causing these significant differences. I suggest that the authors can provide a comparison between the modeled and measured vertical profiles of NOx in the boundary layer to check whether exist significant discrepancies. In addition to the state of vertical mixing, the vertical distribution of NOx is also an important factor to shape the vertical profile of ozone in the boundary layer. Furthermore, as reported in the literature, ambient NOx concentrations declined rapidly in recent years in China and thus I am not sure whether the emission inventory of NOx used in the model can well reflect these changes.

Line 245-250 and Figure 4: These results and conclusions are quite confusing. In nighttime, process contributions of the change in ozone at different altitudes are plausible. However, process contributions of the change in ozone in daytime are confusing. As shown in Figure 4, the increase in surface ozone in daytime were mainly contributed by vertical diffusion and horizontal advection. The authors also highlight that chemistry is not a major source/sink of ozone at 488 m. According to these results, the boundary-layer ozone budget in urban Guangzhou was mainly contributed by transport from adjacent regions or from even higher altitudes? Local formation of ozone from photochemistry has negligible contributions to the increase in the boundary layer ozone in daytime?

Line 260-265: In daytime, the enhancement of air turbulence could drive the well mixing of ozone in the boundary layer. Therefore, the defined “nighttime ozone residual capacity” may be not suitable to assess the influence of the ozone at a certain height in the nighttime on the ozone budget at the same height in the daytime boundary layer.

Line 340-344: How Ox changes at different altitudes during the NOE events?

Line 370-375: I agree with the authors’ opinion that the improvement of the model capability in simulating nighttime ozone in the RL is a key to decreasing errors of the modeling results. The timely update of the NOx emission inventory may be another important approach to improve the accuracy of the ozone modeling results.

Nighttime ozone in the lower boundary layer and its influences on surface ozone: insights from 3-year tower-based measurements in South China and regional air quality modeling

This is an interesting manuscript on vertical structure of ozone in the lower atmosphere based on 3-year tower-based measurements in South China and air quality model analysis. The manuscript is clearly written and well organized. Here are some specific comments:

• The title is confusing a little bit. “Nighttime ozone in the lower boundary layer” includes surface ozone, right? How to understand “its influences”?
• How many nighttime ozone enhancement (NOE) events occurred in the 3-year tower-based measurements? Does NOE happen at the same day in other air quality stations in Guangzhou? Are there seasonal differences in NOE? Authors stated that “During the surface nighttime…a typical feature of enhanced vertical mixing” (lines 32-34). Is there another way leading to NOE in Guangzhou, for example, horizontal advection. A period of 3 years is not short.
• “oxidation capacity” is not clearly defined in the abstract. How to understand “This indicates a persistent high ozone level and oxidation capacity aloft the surface” (line 23)?
• What is the weather condition favoring “significant influences of nocturnal RL ozone on both nighttime and the following day’s daytime surface ozone air quality”?
• Model performances are evaluated only on monthly bases, while NOE does not happen every day. How does the model capture typical NOE? I suggest two or three typical cases are analyzed in oer to see its major impact factors.