Wang et al., present a field measurement report about dinitrogen pentoxide (N₂O₅) and relevant parameters in an island site in the South China Sea, where the nighttime chemistry is less studied compared with those in urban regions in China. They showed that this site is strongly affected by the outflow of urban polluted plumes from the Pearl River Delta, China, although the local anthropogenic emission is weak and has ~50 km from the coastline. High nitrate radicals (NO₃) production rate, moderate N₂O₅ concentrations, and short N₂O₅ lifetime are well characterized in the outflow plumes. The budget analysis is also convincing especially in the aspect of volatile organic compounds oxidation during the nighttime.

The data presented in this report has high quality although only valid for about half a month, this data set is valuable with respect to the nighttime chemistry in the marine regions that are frequently affected by anthropogenic activities, which can be helpful to the understanding of the interactions of anthropogenic and marine air masses. The results inspire that human emissions in the coastal cities may have a significant impact on the air quality in the marine regions over a large spatial scale. Overall, the paper is well written and certainly within the scope of the measurement report type in ACP. I would like to recommend minor revisions before the publication.

General comments:

1. As shown in Figure 2, the N₂O₅ data is also available in one night (11-14), but it is not presented in Figure 3 as well as Table 2, I can understand the limited data did not have representativeness of average condition for CAM, but it should be clarified in the legend of Figure, Table, and the main text.
2. I strongly encourage the author to conduct more analysis about the nocturnal oxidation capacity of the different types of VOC by considering the nighttime ozone oxidation as well as the nitrate radicals.
3. Line 402, the concentration of phenol and cresol is below 10 ppt on average, considering the high contribution to the NO₃ reactivities, I suggest the author add some discussion about the instrumental detection limit of the species.
4. Line 515, the two cases presented in Figure 9 are 0.04 ppb and 0.12 ppb, which is not consistent with the statement of 40-400 ppt, the author should clarify it. By the way, I don't know why the two concentrations of NO are chosen in the two cases, respectively, since the budget is still not closed in the second half of the night.
5. The reaction rate constants of NO₃ with VOCs can be added as a table in the support information.

Technical comments:

1. Figure 6a, the standard deviation should be added to the bar plot.
2. Figure 5c, the font size of the percentages is too small, it can be further improved.
3. Line 138, 3000 change to 3, 000.

Wang and coauthors describe a recent study of nighttime N₂O₅ and NO₃ chemistry at a marine site in the South China Sea that was alternately influenced by inland and coastal air. They measured N₂O₅, NO_x, O₃, VOCs, and meteorological parameters with a suite of instruments, and calculated the expected NO₃ based the N₂O₅-to-NO₃ equilibrium constant. They calculated the NO₃ production rate and attribute its loss processes to a combination of the reaction with NO, heterogeneous uptake of N₂O₅ on aerosol, and reactions with of NO₃ with VOCs, and find that the loss is dominated by reaction with NO, even at very low NO concentrations. They compare the concentrations and lifetimes of NO₃ and N₂O₅ with other recent urban, costal, and marine studies.

The study adds to our understanding of nighttime pollution processes, and may be suitable for publication, but there are some key details missing in the description of the instrumentation and the data analysis that need to be addressed first. I recommend major revisions.

General comments:

• The authors essentially have three categories for NO₃ loss: N₂O₅ uptake, NO₃ + VOCs, and NO₃ destruction by NO. But the NO₃ + NO reaction makes two NO₂ With an excess of O₃ available, the NO₂ molecules with react to regenerate two NO₃ molecules, unless it reacts instead with another oxidant such as RO₂. So in that sense, NO destruction could be considered a null cycle. I am concerned that the authors’ statement on line 466 that “nighttime NO₃ chemistry may be almost negligible” isn’t the right conclusion to make once you consider that NO + NO₃ is part of the nighttime chemical cycling, which NO₃ + VOCs and N₂O₅ uptake represent termination steps. Have the authors considered instead of looking at production of NO₃, looking at some kind of “net” production that accounts for the reformation of NO₃ back into the budget?
• The authors separate the model study into two types of air masses: inland air masses (IAM) and coastal air masses (CAM). The HYSPLIT trajectories show that the CAM air follows the southeastern coast of China, though only one example trajectory was shown. How much variability was observed in the trajectories? Did any trajectories imply an influence from the megacities on the coast? Unless the authors observed distinct tracers of purely marine air, it seems like the CAM air is also continental air, just more aged than the IAM air. But in several places such as line 332, the CAM-influence measurements are described as coming from “clean areas”. The site is also described as an “island” site (i.e. Table 2), but based on wind speeds, the site is only about 2-3 hours downwind from the major cities. So I would suggested reframing these descriptions as IAM = fresh urban emissions, and CAM = aged urban emissions.
• More details are required for the methods section. For example, the CEAS technique measures N₂O₅ by thermally converting N₂O₅ to NO₃, but there is also some ambient NO₃ in the measured sample. How do you solve for [NO₃] in on line 216 if [N₂O₅] is actually [N₂O₅ + NO₃]? This needs to be described in more detail. Additionally, the PTR-MS measurement is not described in enough detail. How were sensitivities for each compound assessed? How were backgrounds and calibrations done? Which of the VOCs were measured by PTR-MS and which by canister and was there overlap in the species that allowed a comparison to assess instrument accuracy? Please see the specific comments below for additional questions?
• The parameterization of the N₂O₅ uptake coefficient is not yet well understood, with many different parameterizations out there. Numerous papers have shown that there are other dependences besides temperature and relative humidity, including aerosol composition, aerosol pH, and aerosol liquid water content. This study didn’t measure aerosol composition, so it would be difficult to parameterize here, but the authors should discuss why this parameterization was selected, given that the results in Figure 8 show the results are highly sensitive to N₂O₅

Specific comments:

Line 124 – Change “downward of the city” to “downwind of the city”

Lines 140-142 – Where on the island was the measurement site? The listed latitude and longitude correspond to a location in the ocean, likely because there are not enough digits. This is relevant because the authors state several times that there may have been some local NO emissions, perhaps from soils. But they also mention that there is some local fishing boat activity. Figure S1 shows that the highest levels of NO come from the north, but must be fairly local. Could there be a local source such as a fishing boat or generator just north of the site?

Line 143 – “local air flow was consistent from the northwest to southeast”. Consistent seems like the wrong word here. The wind wasn’t evenly distributed between those directions. Perhaps “local air flow was consistently from either the northwest or southeast”.

Line 177 – More details are needed about how mirror reflectivity was measured. What was the effective pathlength of the optical resonator?

Line 180 – “The loss of N₂O₅ in the sampling line and filter was considered in the data correction”. More detail, or a reference to a previous paper, is needed here.

Line 200 – “VOCs were also sampled every 2 h using 2 L canisters on the days when the hourly O₃ mixing ratio exceeded 70 ppbv”. Why was it only sampled only those days? And if the peak O₃ was only reached in the afternoon, how could the full day before the peak be sampled?

Line 191 – More details, or a reference, should be included to describe how the SMPS inversion was calculated to generate the particle size distributions. Additionally, information about the peak diameter, whether it changed between CAM and IAM, would be useful information.

Figure 2 – The y-axis for NO₃ should clearly state that it was calculated, not measured, to prevent misunderstandings

Line 259 – 261 – It is not clear what is meant when the authors say the O₃ “hourly maximum level” was exceeded “for 6 days out of 37 days of measurements”. Was O₃ above the standard for all 24 hours of those days? Or at least 1 hour?

Line 273 – What is the LOD for NO? That wasn’t stated in the description of the instrument

Line 276 – “NO is likely to come from a local source such as soil emission”. Could boats, generators, or cooking emissions be responsible for this emission?

Line 281 – “only lasting three days”. But there was lots of data missing. How many days was N₂O₅ actually measured? Those three days might be a significant fraction of them.

Line 294 and Table 2 – Does “All day” and “daily average” mean the 24-hour average or just the daytime hours? And why do these two numbers not match?

Line 298 – Where does 1.4 ppbv / hour number come from?

Line 300 – It should be easy to calculate how much the rate constant will change as the temperature increases. The authors should state what percentage increase they would expect given their higher temperatures.

Line 306 – If there is no data at a given time, it should be given as a blank or a null, not a 0. A zero implies that the data was measured at 0.

Line 314 – Can the authors estimate how much nocturnal NO₂ occurred? After all, they say that NO + NO₃ à 2NO₂ is the dominant pathway for NO₃.

Figure 3 – These data should have error bars to show the variability in the average diurnal cycle. Also see the note about line 306 regarding non-data being labeled as “zero”

Line 324 – VOC concentrations are described here as being “higher”, but no VOC data are shown in either the main text or the supplement.

Line 327 – The way this sentence is phrased makes it sounds like 155 pptv was the peak for both N₂O₅ and NO₃. Rephrase this for clarity.

Figure 4. Please include the CAM/IAM bar at the top of this graph, like in Figure 2, to help the reader see which data came from which period.

Line 359 – Why is 600 um²/cm³ a considered a threshold for cleanliness? Is that a value from literature? If so, it should be referenced.

Line 372 – How was distinction between anthropogenic and biogenic VOCs made? Is there overlap between these two categories? A table would be helpful here, that also contains the rate constants for each VOC that is part of the k(NO₃) calculation, as well as references for each rate constant.

Line 377 – This is the first time that outflow from Hong Kong and Shenzen in CAM-influenced air is mentioned, and should be mentioned earlier.

Figure 5 – See comment about Figure 4 regarding putting an IAM/CAM bar on top to help guide the eye

Line 403 – “significantly higher” requires some quantification, especially because just before that you said “despite their lower concentrations”. Are VOCs here lower or higher than expected given the location?

Line 420 – Error bars on the calculated k(NO₃) are required here to quantify whether this is really “significantly” different between the IAM and CAM periods.

Line 453 – Please define the “loss ratio”. Is it (loss by process X) / (loss by all processes)?

Line 453 – This phrasing is unclear. Are the authors saying that we can assume that total loss = total production because there is very little NO₃? Is this a valid assumption given the fact that NO₃ is measured above 0?

Figure 9 – There are two blues in this figure. And it isn’t clear what the second blue and the red traces represent from the figure caption. Please include a legend.

Figure S1 – Why does this wind rose look different than the one in Figure 1?

Figure S4 – Was this loss ratio calculated assuming NO was between 40 and 400 ppt? Or with the measured values? If the latter, it would be helpful to include the calculation with both values as well, to show the range of expected loss ratios.