|This paper includes some valuable calculations and the authors have made some good progress toward clarifying their methodology. However, some of the methodological details remain difficult to follow. It also appears that the authors have used only a small subset of the available pN2O data in MEMENTO, for reasons that aren’t clear. This subset appears to have a summertime bias in the South Atlantic and a wintertime bias in the North Atlantic. Since pN2O is the one of the main constraints used in the model optimization, these seasonal biases may affect the results. |
I support publication of this paper in principle, but before it is ready I still think there are a number of details that should be clarified and sections of text that could be written more clearly.
p2L32. The text here mentions 4 methods. Are these 4 enumerated in the lines that follow? For example, is the following method 1,"We extend the global ocean biogeochemistry model PlankTOM10 with additional N cycle processes."? Or are these additional approaches (it seems like there are more than 4 total)? Please start the sentences with transition words like, "First", "Second" instead of just "We" to make this clear. Is the sentence on line35 starting with "Then" approach # 3? If so, please use "Third" This is a key paragraph where the authors lay out what the paper will do, yet I am already lost.
P3L30/P4L1. There are a lot more deep and surface N2O data in MEMENTO than the reported n=8047 and n=6136 mentioned here. Were only a subset of the available data selected and why? Also, were the surface data generally in units of ppb and the deep data in nmol/L?
P4L4 Are deep data converted to delta_pN2O or just the surface data? If deep data are converted too, please mention that this necessitates first converting the deep nM data to ppb using a solubility function. This is a large uncertainty. The statement “we have taken the database at face value” is inadequate for conveying the extent of the uncertainties involved in combining nM and ppb data in equation 3.
P.4 Section 2.4 (Nitrification) and p5 Section 2.5 (NH4+ uptake) appear to be databases 1 and 2 of the 4 mentioned. I was expecting 2.6 then to be MEMENTO deep N2O and 2.7 to be MEMENTO surface data. Instead, we jump to 2.6 N2O production. This is an example of why it remains challenging to follow the methodology of this paper.
P4L27. Does Yool provide an actual data base, or simply assume a constant rate of 0.2 /day everywhere (as written, the sentence implies the latter)?
Section 2.4. Given that a cost function of 2 means that, on average, the model deviates from the observations by a factor 2 (Section 2.3), does this paragraph suggest that the model deviates from observations by a factor of > 4 on average? In other words, it provides no real constraint. Shouldn’t that be stated explicitly somewhere in this section? Is there any meaningful difference between a cost function of 4.22 and 4.16?
P5L34. So, the diagnostic model dN2O/AOU ratios are not optimized against the MEMENTO database using the cost function? The intro and the mention of the 4 datasets had led me to expect they would be.
P6L11-12 “We indenpendently optimised the ratios of N2O production and consumption from denitrification” These are minor terms in the budget compared to N2O production from nitrification. Why wasn’t that ratio/coefficient optimized?
P6L14 “The ratios of both submodels were optimized using the databases of observed N2O concentration and pN2O” Is this referring to the deep N2O or the surface N2O data or both?
Figure 3 caption. Please specify where these values are from (model, MEMENTO, etc).
Figure 3 annotation shows 0.0815 + 2.7551, but this is reported on p3L15 as 81.5 ± 1.4 μmol/mol. First, the error has completely changed. Second, there is a switch from nmol N2O and umol AOU in the figure to units of umol/mol in the text, with a factor of 1000 thrown in to add to the confusion. Better to be consistent across figure and text.
Figure 6/Section 2.5. The model is credited with reproducing “the large scale pattern of surface NH+4 concentration (which) shows an increase with latitude.” However, the performance seems pretty poor and the pattern could equally well be described as high in the Southern Ocean (where nutrient utilization is known to be low) and around continental boundaries where there is nutrient input from land. Similar to Section 2.4, the cost function of 3 seems quite large and suggests there’s no real constraint here.
Figure 9 and 11 caption and Equation 5 on p. 6. What does MSE stand for? Please spell it out in all these places. (The captions should be understandable without referring back to the text.)
Section 2.8 Should Equation 6 be presented before 5? It seems like 5 builds upon the definition of MSE introduced in Eq 6.
P7L26. I’m confused by the use of “even though” here. Given that the prognostic model represents N2O consumption in the OMZ, why would that be expected to improve (i.e., increase) the concentration of N2O between 200 and 1500m?
P8L5-7. These sentences belong in the methodology. Also, as mentioned above, why isn’t the nitrification N2O/NO3 ratio optimized too? That seems like the most important term in the model.
P8L14 I would suggest writing as 0.183 mmol N2O, to avoid switching units, which is confusing for the reader.
P8L16-17. “pN2O provided a better constraint than the N2O concentration distribution” Back in Section 2.2, deep N2O is mentioned as dataset number 3 used to optimize the fluxes. The sentence just cited suggests that deep N2O is not actually used. Please clarify.
Figure 10a) This figure represents only 6136 data points (I think there are a factor of 10 more surface N2O data in MEMENTO than that), yet the figure suggests extensive coverage of the global ocean, and near complete coverage in the Atlantic. Have the data been binned and gridded and if so how? It seems like a single data point has been expanded as a ~5x5 pixel, which implies much better coverage than may really exist. Also, what is the seasonal distribution of the data? I suspect the South Atlantic data are all from austral summer, while the North Atlantic data are mainly from fall/winter. I don’t think the Atlantic is undersaturated to the extent implied by this figure on an annual basis. Can the data be binned by season and plotted in a 4 panel plot?
Figure 10b) Following on the above comment, it is confusing to plot the “same months where there are observations, and annual averages everywhere else,” especially if there are strong seasonal biases that differ by region. I would again suggest 4 panel seasonal plot for the model results.
Figure 10b) Also, the red contours appear to smear over the black continents.
Figure 10d) Please make green and red lines thicker. They are illegible.
Figure 11. Are the different symbols (0.06, 0.11, 0.17, 0.34 for A) and (0.07, 0.12, 0.17, 0.34 for B) essentially the same thing or is the similarity in numbers pure coincidence? If the former, why the slight differences (e.g., 0.06 v. 0.07) and inconsistent shapes? Also, why does A say “different symbols indicate different low O2 ratios” while B says “points with the same symbols have different N2O ratios for nitrification.” I thought the nitrification ratio was not optimized.
P8L23-24. This implies that the oxic nitrification ratio is optimized, at least in the prognostic model. Yet as far as I can tell, the Methods only describe optimizing the denitrification and suboxic N2O parameters.
P8L23 and L32. Are deep offshore and near-shore non-coastal synonymous? Please use consistent terminology to avoid confusion.
P9L8 Again, it is important to know whether the surface pN2O data used were originally in ppb units or in nM. If the latter, these measurements likely have an uncertainty of +/- 0.5 nM or more, such that the 3% cited here (which may be accurate for the solubility function per se) greatly understates the actually uncertainty in ppb due to the solubility conversion.
P9L19-20 “In light of this, we decided to recalculate the N-cycle-based N2O production based on currently available data. We find that we can estimate all the relevant steps in the N cycle with observational data” These seem like non sequiturs. The logic here is difficult to follow.
P9L26 NO3- is the electron ACCEPTOR in denitrification. Organic C is the electron donor.
P9L28. “this estimate” Better to restate the numbers, e.g., 4.6 +/- 3.1 vs. 2.5 +/- 0.8. More importantly, these two errors are estimated in very different ways, such that it is not clear they are directly comparable. The authors are comparing a global back-of-the-envelope calculation, with large global errors on any given parameter, to a more grid-specific calculation. Of course the latter will have smaller uncertainty.
P9L30. Please use “further” only once in this sentence and perhaps break into 2 sentences.
P9L30 What does “their” refer to?
Discussion section in general. The writing and organization of thoughts could use improvement. Please avoid starting sentences with “This” unless the antecedent is clear.