Review on manuscript amt-2024-43

Title: An easy-to-use water vapor sampling approach for stable isotope analysis using affordable membrane valve multi-foil bags

Author(s): Adrian Dahlmann et al.

General comments:

In the manuscript by Dahlmann et al. the authors present an alternative approach (in comparison to Herbstritt et al., 2023, Magh et al., 2022 and Havranek et al., 2021) of obtaining water vapor samples for analyzing the isotopic composition of soil water obtained with gas-permeable membranes (GPM) and storing them in multi-layer foil bags. They performed different experiments to test maximum storage time, potential memory effects and reusability as well as field applicability. The authors conclude that their approach is a simple, cost-effective, and versatile approach.

The paper is nicely written and well structured.

However, I have two main concerns:

As approach and concept in this manuscript are very similar to the paper by Herbstritt et al. (2023) - only a different type of GPM and a different (commercially available) bag type were used - the results should be compared point by point and discussed accordingly. Here some revisions and additional considerations are needed.

Moreover, the cited study by Jiménez-Rodríguez is only available as preprint in HESSD. It was under review in 2019 but not accepted due to substantial issues. No revision was provided by the authors afterwards, thus no accepted peer-reviewed version exists. This pre-print can therefore not be cited in your manuscript.

Specific Comments:

1. 28: what you call ‘precision’ is the deviation from the true value, isn’t it?
2. 30: (and throughout the manuscript) ‘water stable isotopes’ not ‘stable water isotopes’
3. 34: ‘suitable for many applications’ was already mentioned in L. 31. Please rephrase
4. 61: ’which can cost anywhere from’ sounds a bit sloppy
5. 67: ‘…provide new insights in research’ please be a bit more specific or delete.
6. 98: wrong! ‰ is not “million”
7. 112: P/N of bags? Fitting of bags? Volume of bags?
8. 115: use the manufacturer as reference for material properties such as ‘Water Vapor Transmission Rate’
9. 125: Fig. 1: Where is the bag? On the right side in Fig. 1b?
10. 130: how much standard water was added to the 100 mL bottle? And what was the size of the GPM inside the bottle? This is crucial for the isotopic equilibrium
11. 275: Jiménez-Rodríguez can’t be cited (see general comments)
12. 289: Fig. 3: would be nice if you could add “L22” and “M22” directly to the figure, there is enough white space. The Legend is also a bit small.
13. 305: if you assume an error during the measurement, I’m wondering why do you present these data and didn’t repeat the measurement?
14. 325 and 332: Jiménez-Rodríguez can’t be cited (see general comments)
15. 324 to 342: please rephrase and take into account the findings of Herbstritt et al. (2023)
16. 354: Fig. 4: see comments to Fig. 3
17. 378: “...has not been described in the literature before” please rephrase or delete.
    
    I would say it is comparable to Herbstritt et al. (2023), Fig. B1.
18. 378-400: Please discuss and compare the results of Herbstritt et al. (2023) on memory effects and conditioning here.
19. 504 ff: Please check the reference list thoroughly, e.g. Millar et al., (2018) or Orlowski et al., (2016b) are in the reference list but not cited in the text.

Technical correction:

1. 11: ‘water stable’ not ‘water-stable’
2. 16: ‘easy-to-perform’ instead of ‘easy to perform’
3. 30: I suggest “…using GPM combined with…” instead of “…using GPM and…”
4. 34: ‘suitable’ instead of ‘suited’
5. 83: please use different bullet points for the field experiments e.g. i) ii) iii) or (a) (b) (c)
6. 98: wrong wording: detailed in
7. 108: Majoube is from 1971
8. 215: ‘procedure similar to’ instead of ‘similar procedure to’
9. 247: please delete line break
10. 284: I suggest ‘On average’ instead of ‘In average’
11. 396: ‘especially in’ instead of ‘in especially’
12. 447: please add dates

General comments

The authors present a new method of collecting discrete vapor samples for water vapor stable isotope analysis using inflatable multi-foil bags. The presented method contributes to a new, currently evolving field of stable isotope analysis still lacking an agreed-upon standard procedure suitable and attractive for many users interested in performing in situ isotope assays without field-access to an analyzer. Therefore, any reported experience in this regard is highly welcome and I recommend publication after proper revision.

The manuscript describes the use of bags, which differ only in valves (which do not seem to have an effect) from the ones used in a previous study (Herbstritt et al., 2023, doi: 10.5194/hess-27-3701-2023). I therefore suggest a more thorough discussion emphasizing how this work expands the findings of the previous study. Moreover, I don’t understand how the proposed treatment of previously used bags would help to get meaningful results if reused for unknown samples. I have a feeling that the tested treatment to remove memory effects does not account for the potential conditions faced by researchers interested in using the proposed method regarding, e.g., feasible or necessary storage time and range of previously observed isotope values.

Formally, the authors decided to combine results and discussion. Unfortunately, this often leads to a limited description of the results. I believe the manuscript would benefit from a better distinction between description and interpretation of the presented findings. Also, some additional technical details (flow measurement devices, new or reused bags for the field test, rinsing atmosphere prior to reuse, etc.) should be added to the method section. Finally, a detailed SOP listing suggested settings and potential pitfalls may be helpful for future users of the proposed method.

I provide a list of specific comments below.

Specific comments

L10: “water stable”, not “water-stable”

L16: “easy-to-perform, in situ”, not “easy to perform, in-situ” (also elsewhere in the MS: “in situ” without hyphen)

L22: insert “spectrometer” or equivalent after “laser”

L25: “can lead” seems too weak, as there will always be influence of previous samples. I suggest “does lead” or “will lead”

L26: Consider rephrasing to: “…showed that the memory effect increases with duration of storage.”

L28: You state the precision, which describes the scattering of repeated or replicate measurements. What is the accuracy, i.e. the deviation from the target value?

L30 (and elsewhere): “Water stable”, not “stable water”

L38f: I do not see why hydrology and meteorology would focus on the biosphere. Consider rephrasing.

L59 (and elsewhere): Do not cite preprints like Jimenez-Rodriguez et al. (2019). It is against AMT guidelines and it devalues the work of reviewers. Especially, do not call such work “successful” (L70) when in fact it has not successfully passed a peer-review process.

L61: “less than 50 euros” is quite vague. Can you be more specific?

L63: in what aspect is the lab environment stable? – Temperature?

L63: What do you mean by “configuration”?

L72: Again, be more specific about pricing. This helps other researchers considering using your method. In addition, the numbers never appeared in the manuscript again, i.e. they were not discussed. Nonetheless, you refer to them prominently in the manuscript’s title. How do they compare to the per-sample costs of the containers used by Magh et al., (2022, doi: 10.5194/hess-26-3573-2022) and Herbstritt et al. (2023)?

L95f: Please be more specific. How was vapor from standards produced? Was it in equilibrium with the liquid phase (resulting in temperature-dependent isotope fractionation) or flash-evaporated (with no fractionation)?

L98: per mil, not parts per million (I wonder how this went unnoticed by five co-authors…)

L108: The Majoube paper is from 1971, not 1961.

L112f: Would it be possible to state a part number for these bags as well? I am unable to find this product in a web query. In addition, how does a membrane-based valve work? Does the sample have to pass through a membrane?

L115: This number seems to be huge! Assuming that the sample bags (front and back) have an area of roughly one tenth of a square meter, more than half of a sample (which comprises ~17µL or 17mg of water per 1 L air at room temperature when saturated) would be exchanged per day. Can this be true? Please, also state the conditions (temperature, relative humidity, vapor pressure gradient), under which the water vapor transmission rate was determined (without citing a preprint). Otherwise, this number is meaningless. Or disturbingly high.

L121: Did you test a version without electrical isolation tape that did not work? I am wondering if the tape really makes a difference regarding proper sealing.

L127: What was the length of the GPM?

L133: How was the flow rate measured? And what would have been the maximum possible flow ensuring equilibrium given the GPM length you selected?

L135: Under non-EQ conditions, the vapor isotopic composition would also depend on water isotopic composition and surrounding temperature. But not exclusively.

L140: By “outgoing”, do you mean the flow going out of the sample vessel or the flow going out of the open outlet?

L163: How dry was the air after passage through the desiccant? Was this value tested and constant over the course of the experiment?

L166f: What would happen, if the bags were filled to more than 90% capacity? And why isn’t a lower filling capacity stated in the first place? How about filling only to the minimum volume necessary to reach a plateau on the analyzer during analysis? Did you play with that variable as well? How would that impact feasible sample throughput? How would the reduced sample volume affect its vulnerability, e.g., regarding memory effects?

L173ff: This statement is a repetition of L141f. Consider deleting.

L180: 100 mL bottle volume minus 60 mL of water leaves 40 mL headspace volume which is exchanged in < 1 min(?). Is this sufficient for establishing equilibrium given the applied flow rates? Were the tubes submerged?

L193: Was this the observed temperature range during sampling? Then 25°C (L197) may not be enough to prevent condensation.

L218: Why did you test only the effect of one-day storage when you intended to store natural samples for up to seven days? Did you refill them with L22 before you “then proceeded” (L219) with H22? Why? Did you also assess the memory effect on samples stored in re-used bags for seven days after the previous samples had also been stored for that period? From your experience, what kind of preparation would be necessary in that case to still obtain meaningful isotope measurements from unknown samples stored in re-used bags?

L220: What do you mean by “usual steps”? Did you refill with H22 and measure/empty immediately? How are the obtained findings transferable to a setting where, e.g., L22 was the first sample collected with a new bag and H22 was the sample collected with the (now reused) bag – also stored for 1 day, or 3 days, or 7 days? I am afraid, this is the weak point of the entire reusability test. By emptying the bags overnight (L223), you avoided exactly the effects that samples in reused bags may be subjected to. The point of reusing bags for unknown samples collected remotely should be to NOT have to refill/empty them repeatedly with the sample of interest and then measure them immediately. Can you propose a preparation routine for to-be-reused bags that ensures the isotopic composition of any unknown sample to be reproducible with sufficient accuracy after typical storage times? If not, I am afraid, the combined storage and memory test is not very exhaustive. (Later, you suggest rinsing 10 times with dry air but you do not present data proving the usefulness of that procedure.)

L229: Please state here already, if you used new or reused bags for this part of the study.

L234f: This sentence sounds odd. Either insert “samples” after “45 cm” or delete “for” and change “taken” to “sampled”

L239: Equilibrium is not indicated by stable values. Steady-state conditions are indicated by stable values. One way to test for equilibrium conditions is to vary the flow rate around the target value and see if this has an effect on readings of vapor mixing ratio and isotope signatures. Was this done?

L241: What was the time per in situ measurement (as compared to 15 min of bag filling)?

L242: The logger only records the readings from an attached sensor. What sensor was connected to the logger to obtain temperature measurements?

L243: Please, also state here the durations of the individual steps. Most importantly, how long were samples stored in the reused(?) bags prior to measurements? How does this compare to the combined storage and memory test? And how is this transferable to a setting with no field-access to an analyzer? (I understood that bag measurements were conducted in the field shortly after filling.)

L245: This statement is a repetition of L231f. Consider deleting.

L282 (and elsewhere): For consistency, delete quotation marks for the names of the standards (here: L22 and M22).

L290ff: This seems to be a repetition of the previous statement. Rephrase or delete

L302f: “increased deviation” translates to high inaccuracy, not “imprecision”. Accuracy describes the deviation from the target value and is not synonymous with precision, which describes the scatter of repeated or replicate measurements around a common mean.

L303: insert “samples from” after “as”.

L305: please elaborate on the “error during measurement”. What went wrong and how can users of your method avoid this error?

L312: I don’t think it is fair to compare the accuracy of two methods that used totally different storage times (1-7 days vs. 30 days).

L321: Given that Magh et al. (2022) used off-the-shelf components, I tend to say that their method is not more difficult to handle than yours. Further, the “static properties of the glass vials” (L322f) make overfilling impossible during sampling (as compared to a mandatory maximum of 90% in the case of the gas sampling bags used in this study) and allow for simply letting dry air flow in during measurement with no need of extra pumping. Apart from potential breaking, glass vials may also be more robust relative to the thin plastic and aluminum layers of sampling bags in many typical field settings (you report damaged bags yourself (L407)).

L329ff: Personally, I find it alarming when the standard closest to ambient air delivers the best results as it points to an unaccounted-for influence of ambient air. The question must be how you can ensure that your method delivers meaningful results regardless of the isotopic composition of standards or samples. And how does this impact the measurement of unknown field samples when collected using newly prepared, equally pre-treated bags?

L337: No. Flushing with dry air in the case of Herbstritt et al. (2023) did not cause the scattering. Rather, it was unsuitable to remove the scattering caused by previously collected, diverse samples as efficiently as flushing with moist air did.

L353: The connection between storage time and memory effect has already been shown in the Herbstritt study.

L356f: Insert “target” or equivalent before “standard deviation” (2x).

L363: I don’t know which part of the Herbstritt study you are referring to but as I understand they used ambient, non-saturated air of arbitrary isotopic composition to pre-condition their bags.

L377f: Clearly, the magnitude is a function of the isotopic spread between the standards used here. The exponential decrease – expressed in the standard deviation of an entire batch of to-be-reused bags – was also shown before (Herbstritt et al., 2022, Fig. 5b).

L379f (and elsewhere): I think it is not necessary to repeat the isotopic composition of the standards so often. Ideally, the outcome of your method should be independent of these values anyway.

L382: Why did you stop at H7? It would also be important to confirm that the readings stay in that range.

L397ff: You advise to reuse bags but you did not show how the isotopic signature of unknown samples can be obtained in the foreseen application, i.e. remote sampling followed by lab-based analysis on a different day. In the storage and memory test you repeatedly flushed the reused bags with standard vapor until the readings were acceptable (after irrelevantly short “storage” times). The proposed procedure (filling and emptying at least seven times (L400) and promptly measuring) is certainly not desirable (or feasible) when collecting unknown samples in remote locations. What would be the achievable sampling frequency in that case? And would that still be an advantage compared to direct in situ measurements performed with an analyzer that has been brought to the field?

L400: With what and for how long should re-used bags be filled? I am sure this has in impact on feasible sample storage time. Can you also comment on a quantitative relationship between the ranges in isotopic compositions of previous samples and the necessary number of pre-sampling filling cycles?

L404: Did you compare in situ measurements and bag measurement only during two or during all 18 campaigns? If two, then how were conditions different, especially regarding elapsed time between sampling and measurement and relative to the sample storage time tested in the combined experiment? Please specify in the method section.

L407: To make life easier for potential users of your method, please specify “filling errors”. In addition, how did you identify condensation? Where did you see it?

L409: This is important and should appear in the method section already: What did you use for rinsing the bags and where was this step performed? Standard-derived vapor in the lab or the to-be-collected, unknown sample vapor in the field?  If the latter, what was the required per-sample time required for this step? 10 x 15 min = 150 min?

L432: On what kind of analyzers do co-extracted organic compounds interfere with water stable isotope measurements?

L444: After what?

L446: Please specify “wide”

L447: The period needs to be specified.

L455: For additional plausibility, can you compare the nature of the scatter, e.g., by comparing the linearity (R²) of the dataset, with that of precipitation data and other datasets of soil water isotopes? Is there a difference in linearity between the two campaigns with field-access to the analyzer and the other 16 without (if that was the difference)? How were standards produced and treated in these two different cases? How many validation standards were co-measured and what was their precision and accuracy?

L458: transpiration rather does not cause enrichment. Evaporation does. Please change “evapotranspiration” to “evaporative”

L462f: Where do I find the seasonal variability you are referring to?

L465f: This seems to be a bit off. Usually, the lower boundary of the plow layer is around 20 cm, not 45 cm. Was it different in your case? Can you also comment on the large range of isotope values observed for 150 cm depth (yellow symbols in Fig. 6)? I would expect to see a less pronounced variation at that depth.

L468: Why does soil compaction flaw the measurements? In situ measurements have been conducted successfully in boreholes of (I would say: rather compact) trees by one of the co-authors. So why wouldn’t they work in compacted soil? And why would that be an issue at 45 cm but not at 150 cm depth?

L475: I think, “appropriate” is inappropriate here. You did not test the effect on samples stored in reused bags for more than 1 hour. (Or you forgot to mention that.) Consequently, I do not see how reliable measurements of unknown samples stored for typical time periods in reused bags can be performed based on the findings of this study.

L476: rinsing with dry air does not match the procedure described in the combined memory and storage experiment. Please explain (before the conclusion), why rinsing with dry air – previously suspected to increase scatter – does (or should do) the same trick that flushing with moist air does.

L485: are these numbers based on two or on 18 campaigns?

L490: Not “can” but “will”

S1: AMT is a European Journal. I suggest using the metric system and SI units.

S2 & S3: What depths are you referring to? Weren’t these measurements performed on standard vapor sampled in the lab?