|The Authors have made a thorough revision of their work and addressed most of the concerns raised. Subject to some further major revisions and a thorough job of revising the manuscript for improved clarity, this work will be acceptable for publication in Atmospheric Chemistry and Physics.|
1. Clarity of the manuscript is lacking. In some instances, this is due to formatting issues (e.g. reactions listed with some assigned arbitrary letters, instead of consecutive numbers), while in other instances ideas need to be separated in order to be clearly understood. In most cases, this is possible to achieve by breaking the current statements into smaller sentences. These are noted in problematic areas through the technical comments but should be revisited throughout. It is likely that this very interesting work will not receive the attention, or citations, it deserves if the clarity is not improved. The Authors should also ensure that their manuscript and reference formatting meets the journal guidelines.
2. Instrument details to validate quality of the intercomparison are still lacking and need to be improved. Specifically, the following need to be addressed:
The MARGA accuracy and precision statistics for quantifying nitrite must be provided in this manuscript. Referral to when the instrument operated similarly in another study, using different solvent sources, and potentially being disassembled and reconstructed prior to use here, does not qualify the performance it operated at in this work. Nor should the reader need to look for critical relevant details for this work in another manuscript. This is a critical revision.
Pages 4-5, Lines 142-150: The MARGA blanks description is still missing critical details. Were the detection limits of HONO determined for this system for this set of field observations? One detection limit determined in another work does not apply to every group using a similar instrument. It is a measure of signal to noise during the observation period.
When the solvent blanks for the MARGA were analyzed, was the system set up in the field location or in the lab? If they were collected in the lab, how can the Authors rule out contamination of the solvent or instrument components as the source of the discrepancy in the subsequent intercomparison? This is not clearly stated and confounds the quality of the intercomparison work reported.
Lastly, the Authors report 0.00 ug/L of nitrite in their blanks, which is misleading. The instrument has a detection limit somewhere in the neighborhood of 0.02 ug/m3, which is a non-zero value. The detection limit is derived from the instrument signal to noise, which means nitrite could be present below the detection limits in these blanks. Consider instead a statement that ‘no discernable peaks above the instrument detection limits were possible to identify both in the gas and particle channels’
Page 5, Lines 164-168: The LOPAP detection limit is given, but the duration of the measurement it applies to does not. Since the comparison between the MARGA and the LOPAP is only valid at the hourly timesecale, this detection limit is required. For the remainder of the data analysis in the manuscript, which uses the LOPAP observations, the detection limit that applies to the time resolution of the dataset needs to be provided – and depicted on figures where appropriate.
The LOPAP accuracy is stated to be derived from a ‘relative standard deviation’ of 10 %, but how this was determined is unclear. Please clarify how this was calculated using the principles of analytical instrumentation (i.e. multiple evaluations of the calibration response, evaluation of an injected check standard). Has the propagated error from subtraction of signal in the second channel of the instrument been considered? It seems that this may not be a conservative estimate of the instrument performance and the Authors should be careful not to overstate this.
Figure 1: The intercomparison regression has been performed using an orthogonal regression, yet it is unclear whether the measurement error has been included in this assessment. There are no error bars depicted on the datapoints, which suggests that appropriate regression using error-based weighting has not been applied. The lack of such consideratio could be valid if both instruments are subject to similar accuracy and precision metrics, but this information for the MARGA is not presented (see first comment in this section) makes the validity impossible to determine.
In panel (a) of the figure, it is clear that there is a major lag issue in the MARGA (i.e. the decay constant in the HONO measurement from a local maximum appears to be constant across many days, following the nocturnal maximum). The same observation is not seen in the LOPAP and also appears to be absent during M1. Clearly there is an inlet effect that is developing over time with HONO partitioning into the Teflon tubing, which would be expected if the inlet is not heated and can retain significant surface water. Further to this, particles will be depositing to the inlet material, as it is not conductive, resulting in chemical reactions between the gas flow and surface-deposited material. Both of these issues should be discussed as confounding factors in the relevant section of the manuscript.
The Authors concluded that inlet production and denuder artefacts generating HONO from other atmospheric constituents are the source of the systematic bias observed. The argument for inlet HONO production is convincing from the results of M1 and should be possible to correct the MARGA dataset for based on tubing length and the atmospheric sample residence time, including the remaining inlet surface area upstream of the y-split fitting. However, the denuder artefacts argument is not convincing as the correlations between the two measurements are quite strong (perhaps stronger than currently depicted due to the absence of error-weighted regression). If interferences in the denuder were driving variability, they would depend on atmospheric composition that is decoupled from HONO chemistry, resulting in random error instead of systematic error. The second channel of the LOPAP should track interferences that could arise in the MARGA denuder quite well, although the magnitude would differ. The Authors could investigate the relative magnitude of the second channel signal of the LOPAP compared to the primary channel to discern the potential for additional interferences in the MARGA denuder.
Finally, the ratio of MARGA to LOPAP observations from M1 are extremely close to the ratio of molecular weights between NaNO2 and the nitrite ion. The primary standard used for calibration of each instrument should be re-evaluated to ensure that these measurements are sound. There is a break in the data between M1 and M2, presumably due to instrument maintenance, as stated in the manuscript. Were both instruments recalibrated during this time? Were the same standard solutions used to calibrate both instruments? In the Reviewer’s prior experience, these two considerations have been major sources of error in intercomparing instruments that rely on their calibration from aqueous nitrite standards, resulting in exactly this type of systematic error.
Page 2, Line 40: Consider denoting reactions as ‘(R1)’ with increasing numbers for subsequent reactions. This will simplify the notation used throughout the manuscript and create greater clarity.
Page 2, Line 62: Should ‘homogeneous nucleation’ be ‘homogenous reaction’?
Page 2, Line 64: ‘However’ is not necessary here and is used incorrectly in many places throughout the manuscript. Its use should denote a contrasting result or statement to the preceding sentence. Similar adverbs are used where they are not required throughout the manuscript, sometimes resulting in unclear meaning of the scientific results.
Also in this sentence, the Authors should be careful to state that multiple mechanisms may contribute a significant proportion of produced HONO. The way this is currently written suggests that only one will dominate a given dataset, yet the balance of the literature is clearly demonstrating to us that there are many mechanisms at work that can vary in their importance depending on the time of day. For example, dew!
Pages 2-3, Lines 66-73: The reactions presented should increase by one and be in order. The reasoning behind denoting reactions ‘2a’ and ‘2b’ or ‘3a’ and ‘3b’ are not given and are, frankly, very distracting to keep track of given the exploration of these reactions throughout the manuscript. Further to this, there is no ‘reaction 8’ presented. Is there one that is used in the model? If not, please revise the numbering of the reactions and consider using the shorthand notation of ‘R1, R2, R3…’ throughout the manuscript.
Also, HA, A, and X are not defined in reaction 3b and need to be stated somewhere in the preceding manuscript.
Page 3, Line 82: This is an example where more organized reaction notation will improve clarity. ‘according to reaction 1’ can be replaced with (R1).
Pages 3-4, Lines 104-109: This addition to the manuscript is difficult to follow. The direction and magnitude of the discrepancy in the intercomparison are more important than the correlation coefficient. Which method was biased high? How can that be rationalized by considering the methodology used and the chemistry it (or the sample matrix) can promote? For example, if the denuders were always lower, there are many instances in the literature that show nitrite is oxidized to nitrate in the condensed phase in the presence of ozone. If the denuder results are systematically lower than those from the MARGA, then a chemical loss can be hypothesized. Overall, this addition to the paper needs to draw of a key result to motivate the intercomparison in this work. As it currently stands, the reasoning is not possible to follow without reading the Stieger paper in detail.
Page 4, Line 117: The final point of this work would be better described as ‘the relative importance of dew as a sink and source of HONO’.
Page 4, Line 136: Injection of 25 mL of sample onto an IC is not possible. Clarify the volume of the collected hourly samples for the MARGA and clearly describe the volume used to quantify the atmospheric analytes (e.g. 25 uL injection loop or 10 mL treated with a preconcentration column).
Page 5, Line 170: This is confusing. Should ‘settled’ be another word here, such as ‘selected’? This is an example of unclear writing that is present throughout the manuscript.
Page 6, Lines 178 and 193-195: There is a concerning contradiction here on how the dew samples were analyzed. First the Authors state that they were stored for a year, which would be highly problematic as nitrite is known to be unstable in aqueous solution. Later, the Authors state that samples were analyzed within 6 hours on the MARGA. Which is it? Clarify.
Page 6, Line 188: How do the volumes of the blanks that were collected compare to the samples? Or was the F(NO2-) from the blanks used to correct the dew water samples? Please clarify by specifically stating how these were used towards the work-up of the dew water dataset.
Pages 7-8, Section 3.1: This entire section needs to be revisited in light of addressing the major revision above. The Authors use misleading statements in the current revisions, such as at Line 237 where they state their results are ‘in excellent agreement’ with prior findings that the MARGA instrument compares terribly with another technique. This gives the wrong impression. Consider using ‘this result is consistent with’ instead.
The added section on offline batch denuders here is very confusing. The relevance to the intercomparison, and evidence to back up the length of the discourse, are vague. Why is the offline batch denuder something to consider when the second channel of the LOPAP should be a direct measure of potential interfering species from the atmospheric matrix? Why does the pH matter if the Authors conclude that it was not an issue in this work? How does this relate to the systematic bias between the MARGA and LOPAP HONO measurements?
Page 8, Lines 279-280: The daytime sample population of HONO mixing ratios is not normally distributed, the Authors should consider reporting the median values here instead of the means.
Page 9, Line 311: ‘led’ should be ‘lead’
Page 10, Line 340: Why are units in the denominator of this equation?
Page 11, Line 381: Should be ‘least squares regression’
Page 11, Lines 384-385: Give the ratio or the percentage only. No need for both.
Page 12, Lines 397-399: The added statement is too speculative on a single specific mechanism. Consider stepping back from a single mechanism to explain the differences and instead emphasize the increasing need for more diverse environmental observations to understand HONO production chemistry in the nocturnal atmosphere.
Page 12, Line 405: This equation should be inserted as all the other equations in the manuscript have been. All equation numbers through the manuscript will need to be revised following this.
Page 12, Lines 412-421: This section is a contradiction of itself between the first sentence and the last. First the Authors state that all the HONO could be produced on the aerosol, using reactive uptake coefficients consistent with the literature from lab studies, but inconsistent with observations of the full nocturnal boundary layer profiles of HONO and its precursors. The Authors then state that the correlation between [HONO]/[NO2] versus Sa is sufficient evidence that this mechanism is not important, yet the correlation is moderate in magnitude. This is a weak argument that needs to be reworked. The Authors are encouraged to visit the recent work from the group of Jochen Stutz to better place their findings and reasoning regarding aerosol-mediated NO2-to-HONO conversion into the context of our current understanding.
Page 14, Lines 484-493: The Authors need to be careful here with their reasoning. Bulk water pH collected from a glass surface into a bottle does not directly translate into bulk water pH for dew found on soil or vegetated surfaces. The chemical nature of the material, with which the water is in contact, can influence the effective pH. A cautionary statement should be made here.
At Line 490 there are too many significant digits. Correct this.
Page 16, Line 531: Clarity can be improved. Consider ‘used are’ instead of ‘are referred to’
Page 16, Line 542: Clarity can be improved. Consider ‘expect that the reaction between NO and OH’ instead of ‘indicate that the reaction 3a’.
Page 16, Line 544-546: This sentence is unclear. Rephrase.
Page 19, Line 630: Why is mixing height spelled out here instead of using the variable ‘H’ in this equation?
Page 20, Lines 664-665: Reference formatting is incorrect. Ensure this is to journal guidelines throughout the manuscript prior to resubmission.
Page 30, Table 2: The nitrite detection limits for the MARGA in terms of concentration need to be considered. Some of the entries here are very small and may be below the instrument detection limits. In such cases, appropriate notation should be given for this and the detection limit given in the footnotes.
Page 35, Figure 3: The scales for the different axes overlap on the left and right sides of the figure and space between them should be added. The particulate nitrite measurements look like they are often below the detection limit and either the LOD needs to be depicted on that figure or those points should have different symbols/be excluded. Why is the presence of PM10 nitrite not discussed in the manuscript, particularly for the intercomparison? Could deposition of this PM10 nitrite in the inlet lines or denuder contribute to the systematic difference between the MARGA and LOPAP observations?
Page 38, Figure 6: The note on the reactions should be removed from the caption after re-assigning appropriate reaction numbers throughout the manuscript. Then R1, R3, or R9 in the figure legend are easily cross-referenced and several lines of caption are no longer necessary.
The figure and associated legend are too complicated and can be easily simplified The Authors state in the discussion that their base case is Model 3, and therefore, should not depict Models 1 or 2 on this figure. The composition of ‘Model 3’ should be given in the caption and then only the additional terms would need to be stated for the subsequent model runs. The runs, as they currently are described, are confusing.
Panel a from this figure is not necessary. It can either be removed entirely or relocated to the SI.
Page 39, Figure 7: The terms on the axis labels for this figure are confusing and should be revised. It is not clear why ‘internal time’ is necessary to retain, when UTC can be used to better effect.
Supporting Information, Page 2, Figure S1: The three photos should be denoted as panels a, b, and c. The inlet photo should be panel a, followed by the M1 setup in panel b, and then the M2 setup in panel c. Delete the ‘Figure S1a’ and ‘Figure S1b’ from below the photos.
Supporting Information, Page 5, Figure S4: Fix axis overlap to improve clarity.
Supporting Information, Page 9, Figure S8: What range of dates does the depicted data originate from?
Supporting Information, Page 12, Figure S10: This figure is very confusing to follow, mostly due to the presentation of panel b. Why is it noted before panel a? Why is RH on the horizontal axis? Fix this. Place the panel b to the right panel a, and depict HONO_unknown increasing as a function of RH if anything. Since this plot includes all the different days of the observations, it is very hard to see the utility of it. Presumably, this is to show that rapid humidity changes result in the release of HONO. Either a case study showing this from a single day should be given, or a more robust analysis performed to demonstrate this phenomenon.
Supporting Information, Page 13, Figure S11: Delete the ‘Note’ since the reaction notation throughout the manuscript should be fixed anyways, making this no longer necessary.