This paper examines the sources of formic acid (HCOOH) in the somewhat polluted marine boundary layer.  The authors have an excellent site on the coast near Hong Kong that samples a range of polluted conditions from coastal to marine.  The mix of air masses was a bit unusual as the marine is was highly polluted in NOx with low O3, and the land(coastal) air had high O3 and biogenics.  They identify aerosols as a key co-existing species that is related to high formic acid levels.  They track the observed air parcels with HYSPLIT back trajectories.  They then pursue laboratory-chamber studies to quantify the net production of HCOOH from different aerosols and follow up with a box model study of the kinetics involved.  Their conclusions that a major a major source of HCOOH comes from photochemical aging of organic compounds in aerosols (particularly nitrate-containing aerosols) is indisputable.  They note that inclusion of this additional source would reduce some (all) of the model-measurement discrepancy in global models.  The paper is very clearly written; and from my fast read-through, I did not find any typos.  Altogether, impressive.

I have few minor gripes:

"191.1 ± 167.2 ppt in marine air masses" – the 4^th decimal place is unnecessary and only clutters up the major numbers:  191 ±167.

The lack of atmospheric HCOOH sources in models is duly noted, but do the aerosols in the dominant remote marine atmosphere have the organics and nitrates to generate the missing source?  Can the authors assess/speculate on this based on published data, e.g., from ATom. I do not expect them to analyze other observations, but they can comment on whether the aerosols observed over the Pacific would likely produce ~0.1 ppb/hr as at their site.

In their study, Jiang et al. estimate the production of formic acid from photochemical aging of ambient particles by combining observations, laboratory experiments, and HYSPLIT and box modelling. Their proposed HCOOH production pathway can explain some of the major underestimation in their performed box model studies. Overall, the manuscript is well written but further sensitivity studies are necessary before the manuscript can be published in ACP.

Major comments:

My major concern is related to the box modelling performed. Overall, the box model setup used is in part too simplified and some assumptions by the authors influence the predicted HCOOH concentration. For all box model simulations performed, the model fails to properly predict the diurnal cycle of HCOOH. From midnight to about 2pm (Fig. 4), a steady increase in HCOOH is observed but the model predicts a decrease until sunrise. After sunset, a sharp decrease in HCOOH is observed, which the model fails to reproduce. The authors only discuss uncertainties in the deposition velocities, which do not resolve issues in reproducing the diurnal cycle. I suggest performing further modelling sensitivity simulations to challenge some of the assumptions made by the authors. These simulations should at least be concerned with:

1. The authors acknowledge that there are direct biogenic and anthropogenic emissions of HCOOH in the vicinity of the station and in line 280, they acknowledge a marine and anthropogenic influence. However, these influences are ignored in the box model. It should be checked if influx from other regions might explain some of the discrepancies in the early morning.
2. I checked out METEOSAT images and there were clouds reported for that day. Even though there was no precipitation, by vertical mixing, the production pathway proposed by Franco et al. could still be important and its contribution should be tested in the box model.
3. After sunset, there is a sharp decrease in the observed formic acid. I suspect that changes in the transport pattern might play a role, due to the station being so close to the ocean. I checked the raw data provided by the authors and noticed that there was a substantial change in wind direction and speed in the afternoon. I suggest performing a HYSPLIT analysis for this day and investigating if the change in transport pattern might explain the sharp decrease. In the model, this in-/outflow could be added as an additional production-/loss term. A similar approach could be used for point 1.

The authors limit their modelling to two separate days but limit their main analysis to only one day.  Why not performing longer simulations? From Figure 1, we clearly see a different behavior on the next day (29 September 2021). On this day, a sharp increase in formic acid is observed early in the morning. Is the production pathway proposed in this study able to reproduce this behavior? I strongly suggest performing long term simulations for which a high variability in formic acid is observed. From Fig. 1 and the raw data provided there should be sufficient data to perform this analysis from 9 September to 6 October.

Performing these additional simulations and their discussion will greatly improve confidence in the author’s findings and allow publication in ACP.

Detailed comments:

Line 54-56: The recently updated formic acid budget presented in Franco et al. 2021 needs to be discussed in this context. Their proposed production pathway resolves the global model bias to some degree.

Line 163: Is there any particular reason why a higher endpoint height of 100m was selected for your HYSPLIT simulations?

Line 176-179: Could you please justify why you set all these parameters to 3? Why are you ignoring gas-phase species in this context? What uncertainty in aerosol properties do you expect from this assumption?

Line 179-180: This statement might be confusing. Please rephrase!

Line 191: What characterizes a typical day? Which typical day criteria did you use? In Figure S3, you provide backward trajectories for many days. Could you please add a separate HYSPLIT analysis to the supplement for the 28 September 2021. It would be useful to color code the backward trajectories by time to understand transport patterns and the influence of different air masses during the day.

Line 235-236: Please justify this assumption and discuss related uncertainties. In the result section, you discuss this but the reader is unaware of these details up to this point.

Line 301: I am confused by the registered trademark sign listed after the Pearson correlation coefficient. What is the meaning of this?

Line 408: I disagree. The model is not capable of reproducing the steady increase in HCOOH early in the day (from midnight to 9am) nor the rapid decrease in HCOOH after sunset. In addition, I would not label the observed increase from midnight to 2pm as rapid but rather a steady increase.

Technical corrections:

Caption of Table 2: Again, confused by the registered trademark sign.

Reference list: The reference order is not according to Copernicus guidelines.

References:

Franco, B., Blumenstock, T., Cho, C. et al. Ubiquitous atmospheric production of organic acids mediated by cloud droplets. Nature 593, 233–237 (2021). https://doi.org/10.1038/s41586-021-03462-x