the Creative Commons Attribution 4.0 License.
the Creative Commons Attribution 4.0 License.
| 28 Oct 2022
Lessons from and best practices for the deployment of the Soil Water Isotope Storage System
Abstract. Soil water isotope datasets are useful for understanding connections between the hydrosphere, atmosphere, biosphere, and geosphere. However, they have been underproduced because of technical challenges associated with collecting those datasets. Here, we present the full testing and automation of the Soil Water Isotope Storage System (SWISS). The unique innovation of the SWISS is that we are able to automatically collect water vapor from the critical zone at a regular time interval and then store that water vapor until it can be measured back in a laboratory setting. Through a series of quality assurance and quality control tests, we rigorously tested that the SWISS is resistant to both atmospheric intrusion and leaking in both laboratory and field settings. We assessed the accuracy and precision of the SWISS through a series of experiments where water vapor of known composition was introduced into the flasks, stored for 14 days, and then measured. From these experiments, after applying an offset correction, we assess the precision of the SWISS at 0.9 ‰ and 3.7 ‰ for δ18O and δ2H, respectively. We deployed three SWISS units to three different field sites to demonstrate that the SWISS stores water vapor reliably enough that we are able to differentiate dynamics both between the sites as well within a single soil column. Overall, we demonstrate that the SWISS is able to faithfully retain the stable isotope composition of soil water vapor for long enough to allow researchers to address a wide range of ecohydrologic questions.
-
Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
-
Preprint
(3180 KB)
-
Supplement
(24118 KB)
-
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3180 KB) - Metadata XML
-
Supplement
(24118 KB) - BibTeX
- EndNote
- Final revised paper
Journal article(s) based on this preprint
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1170', Stefan Seeger, 21 Dec 2022
The manuscript "Lessons from and best practices for the deployment of the Soil Water Isotope Storage System" presents an important step towards the automated aquisition of in-situ soil water vapor samples from remote sites that are not suited for the field deployment of stable water isotope analyzers. The authors have submitted a very well structured paper that I enjoyed reading.
1. My biggest concern is of formal nature, because in my opinion this paper is more a technical note than a research paper - I do not see any research question asked or answered, nor do I see any scientifically meaningful findings. Nevertheless I really do appriciate the effort that has been made to thoroughly develop and test such a setup and I think the manuscript is worthy for publication.
2. Concerning the actual content of the manuscript, I would appreciate a more detailed discussion of the outliers and oberved system failures.
2.1. In Line 240, you describe that you repeated your dry air tests "until the majority (at least 13/15) flasks" had a sufficiently low moisture content. Why do not all flasks have to pass your threshold criterion. How do you treat measurements of the somewhat leaky flasks?
2.2 During your testing with known isotopic compositions it is easy to spot faulty flasks, but on what basis would you treat actual field measurements? Is there a reliable way to spot outliers?
2.3. During the field application you had to discard 10 of 45 measurements. That is a failure rate of over 22%. Do you have any suggestions on how to reduce this failure rate?
Apart from that, I only have minor points to comment on.
Line 73: Your list of successfull field deployments of laser-based instruments for in situ measurements is missing some notable cases that either are older or produced much more measurements than your given examples:
Volkmann2016 https://doi.org/10.1111/nph.13868
Gaj2016 https://doi.org/10.5194/hess-20-715-2016
Seeger2021 https://doi.org/10.5194/bg-18-4603-2021
Gessler2022 https://doi.org/10.1111/nph.17767Line 267-268: Could you elaborate on why and how the heating does "create a longer and more stable window of measurement time". I kind of get a vague idea about this within the follwoing paragraph, but I think a more explicit explanation of the effect of this heating procedure right at this point would be suiting.
Lines 285-286: "For some flasks, using either a later portion of the measurement period, or slightly earlier offered a more stable signal."
Do you treat those bottles differently during normal operation of the system, or do you treat all bottles equally, even though they might perform better if treated more individually?Lines 318-319: You refer to a "SWISS unit offest correction (detailed below)". Then there is one last method subsection and the results begin. You kind of describe an offset correction in the results subsection "5.1.2. Water vapor test", but maybe the description should happen within the methods part of the paper? Or you could place a reference to where the the offset correction is actually described.
Lines 349-370: In that section you describe 18O and 2H completely separately. Do the outliers of 18O coincide with the outliers of 2H, or are they completely independent? How do the outliers look like in the dual isotope plot? Could that help to diagnose why these outliers are outliers?
Lines 374-380:
I think the long term field storage tests look a lot like 43 days of storage are not any worse than 34 days. Do you expect that there is actually a critical time limit somewehre between 43 and 52 days which can explain the difference between Fondue and Toblerone? Or has Toblerone issues that also would affect its results after a shorter storage period?
Repeated tests with one and the same SWISS unit over different time spans might be more informative than the comparison of three separate units after different storage times. (But of course that would take a lot of time...)Line 524: "Samples were taken approximately every five days...". Why is it "approximately" five days? I suppose your automation procedure does not involve a random number generator? Shouldn't you be able to state the precise sampling interval? Maybe just drop the "approximately" (even if - for whatever reasons - it weren't perfect 5 day intervals).
Line 218: Could you specify the type and manufacturer of the "helium leak detector".
Line 247: "...did not lead by the time..." maybe you meant "leak" instead of "lead"?
Citation: https://doi.org/10.5194/egusphere-2022-1170-RC1 -
AC1: 'Reply on RC1', Rachel Havranek, 22 Feb 2023
The authors thank Stefan Seeger for a constructive and helpful review. Below we detail our response to his comments:
1. My biggest concern is of formal nature, because in my opinion this paper is more a technical note than a research paper - I do not see any research question asked or answered, nor do I see any scientifically meaningful findings. Nevertheless I really do appriciate the effort that has been made to thoroughly develop and test such a setup and I think the manuscript is worthy for publication.
Author Response: We agree that this paper is more of a technical note rather than research paper and will change the submitted format appropriately.
2. Concerning the actual content of the manuscript, I would appreciate a more detailed discussion of the outliers and observed system failures.
2.1. In Line 240, you describe that you repeated your dry air tests "until the majority (at least 13/15) flasks" had a sufficiently low moisture content. Why do not all flasks have to pass your threshold criterion. How do you treat measurements of the somewhat leaky flasks?
2.2 During your testing with known isotopic compositions it is easy to spot faulty flasks, but on what basis would you treat actual field measurements? Is there a reliable way to spot outliers?
2.3. During the field application you had to discard 10 of 45 measurements. That is a failure rate of over 22%. Do you have any suggestions on how to reduce this failure rate?
Author Response: Because the goal of this paper is to make the system as usable by the greater community as possible, and to be as transparent as possible about the system’s strengths and weaknesses, we are happy to add further descriptions of the outliers and system failures.
We appreciate comment 2.1 and will more clearly articulate our philosophy about how to move through QA/QC efficiently; our threshold of at least 13/15 flasks holding dry-air was to be able to keep moving through the process so that SWISS units could be ready for field deployment. This discussion will be included in section “6.1.4 Lessons learned and recommendations from the QA/QC and field suitability tests.” We will also note ways in which we would likely do QA/QC differently in the future, so that other users can adopt a QA/QC process that works for their uses.
We will use the ‘mock’ field deployments to discuss In greater detail ways to identify if condensation is affecting isotope values during measurement based on peak shape, and ways we identified suspected condensation during vapor collection. We will also use the mock field test as a way to discuss more thoroughly how we treat flasks that we suspect are slightly leaky. Lastly, we will also discuss the importance of sequential data collection to identify data that are spurious, and our approach to designing sequence timing to be able to both identify bad samples and answer the question at hand.
A 22% failure rate is quite high, and I think our field deployment points to ways that either the automation system could be improved and/or when the SWISS is likely to fail. First, it’s not entirely clear what soil water vapor samples collected from very dry soils using the vapor permeable tubing represent, and a targeted study looking at dry soil conditions would be a really helpful next step. Second, some updates to the automation schema that make it more remotely controllable would help researchers avoid sampling on days when condensation in the field might be likely.
Citation: https://doi.org/10.5194/egusphere-2022-1170-AC1
-
AC1: 'Reply on RC1', Rachel Havranek, 22 Feb 2023
-
RC2: 'Comment on egusphere-2022-1170', Anonymous Referee #2, 24 Dec 2022
General comments:
In their manuscript entitled “Lessons from and best practices for the deployment of the Soil Water isotope Storage System”, Havranek et al. describe their experience with automating and operating a self-developed soil water vapor sampling system for subsequent, lab-based stable isotope analysis. They present an extensive testing procedure necessary prior to unattended field-deployment. Overall, the manuscript contains a lot of detailed instructions and hints that are definitely helpful for replicating the setup and producing soil water isotope datasets, which are valuable for the ecohydrology community and thus the readers of HESS. Therefore, I appreciate the effort of the authors and I recommend publication after major revision following the comments I specified below.
In my opinion, the manuscript would benefit from a better structuring as currently some important method details are described in the result section, part of the data is presented not before the discussion section etc. (see details in the specific comments). Also, the discussion section currently comprises a lot of repetitions of method and result details. Instead, it should be focused more on the critical evaluation of the presented data and other findings. Given the problems encountered in the field and the corresponding failure rate, I find it unfortunate to call it a “full” testing as obviously not the full spectrum of potentially relevant environmental impacts could be accounted for during those tests.
Further, I would have liked to see an independent validation of the presented soil vapor isotope data by established means in order to have a profound accuracy assessment. In my understanding, precision is the standard deviation of replicate measurements, which have not been performed on soil samples. Therefore, I am not sure whether the precision and accuracy assessments from the lab tests can be assumed for the natural soil samples as well.
Formally, i would recommend using italics (once introduced) for the different methods throughout manuscript, clearly distincting between method, results and discussion statements, and using past tense in method and results.
Specific comments:
Line 12: “full” may be too strong given the unaccounted-for circumstances encountered during the unattended field-deployment
L19: You are stating the precision. What about the accuracy? Do you assume perfect accuracy even for non-lab conditions after performing the offset correction?
L22: “faithfully” may be too strong given that soil water isotope data were not validated by established means.
L31: root WATER uptake
L40: You might consider mentioning here also the work of Wassenaar and colleagues (doi: 10.1021/es802065s) who invented the principle of using laser-based instrument to measure vapor for liquid water isotope assessment. All in situ isotope sampling approaches are now based on this principle.
L46: The expansion OF in situ…
L50-54: I am sure this is a good example to prove your point why a setup like yours is helpful/needed. Unfortunately, I do not understand what it means. What are br-GDGTs?
L62-63: I would expect such recommendations to appear in the discussion/conclusion but not in the introduction.
L68-96: This reads like a second introduction where you identify a problem you are intending to solve. Why not merge this information with section 1?
L79: vapor CONCENTRATION gradient
L89: you already defined the abbreviation “SWISS” in line 61/62
L103: Do you have indications/references that this time frame is sufficient for the soil to return to natural conditions? Can this be tested? How?
L121: Wouldn’t you want to prevent condensation instead of only limiting it?
L140: This reference is missing in the Works Cited section
L144: Wouldn’t vapor bound to the flask walls cause a memory effect that needs to be re-assessed and corrected for prior to every new deployment of each single flask? If so, how would this impact the universal offset you applied if not all flask had the same filling history in terms of isotopic composition prior to common deployment?
L147 and throughout MS: Please do not use different formats for stating the tubing diameter (1/8 inch vs. 1/8” vs. 1/8th inch etc.)
L151: please define the material of the foam and the nature of the insulation (thermal?)
L160: Do “offsets” mean that there is no equilibrium at shorter (or greater?) lengths?
L162: Was it mandatory as well to cut the Bev-A-Line connections to identical lengths?
L186: Please define VDC
L200ff: In my opinion, such statements discussing the suitability of and potential alternatives for certain components are better placed in the discussion
L218: Please define the helium detector model and manufacturer.
L219: To me, this is a result statement
L230: what quantity was measured on each flask? I understand that the dead-end pull method applies a vacuum to the flasks. Were the custom-made flasks tested for vacuum suitability? Are extra safety measures necessary for this step? Would a vacuum test be able to identify large, medium, or small leaks all at once?
L231: please move the parentheses to line 227 and use italics consistently throughout the manuscript when naming the different methods
L235: “we found…” sounds like a result statement
L236: ”…would likely…” sounds like a discussion statement
L240: Why did the dry air tests have to be repeated? Did you encounter gradual decrease of vapor adsorption at the flasks’ walls or did you change anything in the setup (e.g. tightening of connections) between those tests?
L247: I think it must be “leak”, not “lead”
L253: How far and for how long was the tubing immersed in water?
L270: How was dryness verified?
L279: Picarro claims that the measurement range for their L2130-i model is 1000-50000 ppmv. How come that you have condensation issues at vapor concentrations below 50000 ppmv?
L281: Why does changing pressure cause isotope fractionation?
L283: what signal was supposed to be stable? Isotope? Vapor concentration?
L287ff: why was it necessary to use two different methods to assess instrument stability? And how stable was the instrument compared to the SWISS analytical uncertainty?
L295: In what aspects were field conditions different from lab conditions and why wasn’t it sufficient to perform these tests just outside the lab building? Are the results of these tests not transferable to other field sites?
L314: Using two different methods within one batch of samples and standards contradicts the principle of identical treatment (Werner & Brand, 2001, doi: 10.1002/rcm.258). Why is it justified to do so in this case?
L332/338: It would be nice to find the 500 ppmv threshold on the vertical axis of Figure 3. Also, I would prefer to have the horizontal axis start with “2”.
L346: These method details should appear in the method section, not in the results.
L349: “we expect” sounds like a discussion statement (also in L362). What are the uncertainty limits and are they really transferable given the quite different setups (Oerter et al. vs. SWISS)?
L379f: This is already an interpretation of the results. Why is there an inter-unit variability? Or why do you expect one? And why is it (expected to be) different from the intra-unit variability?
Figure 4, 5, and 6: Please add the permil symbol to the vertical axes labels when presenting isotope data. Please add the 500 ppmv to the axis if this is the concentration threshold.
L508: This is misleading as all flasks contain water vapor. Maybe you want to make the distinction between FLASH-EVAPORATED water vapor and atmospheric vapor?
L513 and elsewhere: I think table captions should be placed above the tables
L533ff: Why were those data excluded? 13000 ppmv is well within the instrument’s measurement range. If the soil had been very dry, aren’t the obtained data then representative for this setting and can be interpreted accordingly? Do you mean atmosphere via leaks or by being present in the soil despite the huge gas-liquid interfacial area present even in quite dry soils and facilitating quasi-instant exchange between the liquid phase and a potentially intruding atmosphere?
L541f: How was condensation detected and why were these data excluded? Presenting (and later discussing) even spurious data in this rather technical manuscript would help future users of your setup to identify similar problems.
L548: The discussion of the helium test is missing
L551-556: Why is a seven-day test sufficient when the intended field-deployment is on a much longer timescale?
L557-561: How is it possible that not all of the atmospheric water vapor has been flushed given the achieved turnover rate? How can this be tested/identified? What solution would you suggest to prevent this from happening?
L564-570: This is a repetition form previous sections. Unfortunately, many paragraphs in the discussion start with a repetition of the results which should be avoided.
L580: These data must be presented in the result section. And discussed here. I for one find it very remarkable that there is absolutely no variation in temperature with depth AND time for two of the three field sites. Any error in this variable would have an impact on the trustworthiness of vapor concentration and thus isotope readings, right? Also please add “(°C)” to the table heading of the temperature column.
L586-601: This is not a proper discussion but mainly a repetition from previous sections.
L599: What is the minimum acceptable concentration for this method? Given the overall analytical uncertainty, the additional uncertainty resulting from lower-than-optimum vapor concentrations should not make that much of a difference for quite a wide range, right?
L607: are you referring to air or below-ground temperatures?
L614: “quite” is too vague. Is the system sufficiently resistant to facilitate a specific uncertainty that allows for deciphering natural variations which – as you state elsewhere – become smaller with depth?
L618: how was condensation noticed? This is valuable information for unexperienced users considering to employ your system.
L620: “oxygen isotope” is redundant.
L624: Why is this surprising?
L633: I think you didn’t mention “replacing” in the method section. Only “tightening”.
L639: the observed “greater exchange and leaking” is based on the measurement of only two flasks (#10 is missing in the seven-day data, Figure SI 5) whereas three of the stainless steel fitted flasks (#3, #14, #15) reveal HIGHER vapor concentration than the PTFE fitted flasks after seven days and LOWER-than-before vapor concentrations after 27 days. How is the latter possible?
L645: At what depth does the magnitude of seasonal variability have to be expected to fall below the system’s analytical uncertainty? And what is the accuracy here?
L648: In my understanding, vapor bound to the flask walls would introduce a memory effect which you excluded (L651). Did you really test for this by using flask with intentionally different filling histories for sampling identical vapor sources?
L653-655: Why is the concentration benchmark of multiple dry air tests relevant and not the duration of the test itself?
L657: I’m struggling with “faithfully” because obviously there were non-tested factors causing the high failure rates during the actual application. Also, how would you identify outliers without an independent validation of soil water isotope data?
L663: “Helpful”? Isn’t it even mandatory in order to prevent condensation of the soil’s vapor-saturated gas sample? Would a second Valco valve for the application of a constant zero-air dilution stream do the trick more reliably (similar to Volkmann & Weiler, doi: 10.5194/hess-18-1819-2014)? Or would this go beyond the budget?
L673: Please specify that the uncertainty refers to the hydrogen isotope values
L682: All data should be presented in the results section
L686: How much is that in mm? Can you exclude convective precipitation with limited spatial extent or why was it “likely” (L687) indicative for precipitation at your field site?
L691: If any such precipitation event homogenizes the soil water isotope depth profile then how can you still assume only small seasonal variability at those depths? Wouldn’t you then rather have to expect a significant shift in the soil water isotope depth profile after every precipitation event of similar magnitude? What other factors (soil moisture?, pore size distribution?) could be relevant in this context?
L717: Results. Please move to and describe in the result section.
L721f: Please also show discarded data in order to give readers of this rather technical paper the opportunity to learn from your experience. Then explain in detail what aspect(s) made you exclude these data from further consideration. What do you mean by “sufficiently”? According to Picarro’s instrument specifications, even at 2500 ppmv the uncertainty induced by low vapor concentration is still below the SWISS overall uncertainty. If such low vapor concentrations were representative for very dry soil, wouldn’t the SWISS still be able to deliver meaningful results?
L737: Can you provide details of this evidence? Which conditions caused condensation and how can it be avoided or otherwise dealt with?
L742: What does “CoAgMet” mean?
L748: The low d-excess value is not helpful in this context because it is not an independent measure but is calculated directly from the aforementioned oxygen and hydrogen isotope values.
L749: Again, please report such information in the results section.
L766: Why does uneven heating have an adverse effect? Wouldn’t ANY heating be helpful as long as the induced temperature exceeds the sample’s dew point, no matter by how much?
L768: Do you mean consistent amounts of heat or stable temperature?
L775: What problems would an IoT cellular router solve?
L778: Why is the inherent thermal structure of the soil important? How does it affect isotope measurements?
L783: “can be” sounds a little weak. Previously, you “strongly recommended” this procedure to be applied.
Supplement: Please use similar formatting for the different figures.
Table SI 1: You emphasized the option to reproduce the SWISS setup (L169). Can you please list the monetary costs for different components or the entire SWISS setup as well?
Table SI 2: Are you endorsing Amazon?
Table SI 3: Are the high standard deviation values for Toblerone after 52 days a result of noise or trends during measurements?
Supplement section 1: Build-out description: How many flasks were ultimately broken before the desired setup was achieved?
Citation: https://doi.org/10.5194/egusphere-2022-1170-RC2 -
AC2: 'Reply on RC2', Rachel Havranek, 22 Feb 2023
The authors thank this reviewer for their thorough and constructive review. Their comments will significantly improve this manuscript. Below we detail our response to their comments.
RC: In my opinion, the manuscript would benefit from a better structuring as currently some important method details are described in the result section, part of the data is presented not before the discussion section etc. (see details in the specific comments). Also, the discussion section currently comprises a lot of repetitions of method and result details. Instead, it should be focused more on the critical evaluation of the presented data and other findings.
Author Response: We appreciate their concerns about the structure of the paper – we have found it challenging to structure the paper in such a way that clearly articulates the sequential development steps of the SWISS, while maintaining a traditional scientific paper format. In revisions, we will more clearly separate results from the discussion, and appreciate that the reviewer helped to identify specific places in the discussion where we can write more efficiently.
RC: Given the problems encountered in the field and the corresponding failure rate, I find it unfortunate to call it a “full” testing as obviously not the full spectrum of potentially relevant environmental impacts could be accounted for during those tests. Further, I would have liked to see an independent validation of the presented soil vapor isotope data by established means in order to have a profound accuracy assessment. In my understanding, precision is the standard deviation of replicate measurements, which have not been performed on soil samples. Therefore, I am not sure whether the precision and accuracy assessments from the lab tests can be assumed for the natural soil samples as well.
Author Response: We appreciate that the data we have presented is not truly a ‘full’ testing of the system, and we did not design these initial experiments to test the SWISS against other methods of extraction soil water isotope data, nor is this paper intended to address that stage of development and testing. Our goal with this paper is to get the SWISS out quickly to other potential users so that they may be able to build and test the SWISS for their scientific questions. We agree that testing the SWISS against other methods of soil water extraction is an important next step for testing the SWISS, but requires a more focused experiment that fits the needs of multiple stakeholders in the ecohydrology community. Further, the first-order tests that this paper presents were a necessary first step to be confident that the system works as expected, and how to identify problems, so that later experiments will be reliable. That level of testing should be done to complement the testing that has already been done on the vapor permeable tubing and to address concerns about the reliability of the tubing across different soil textures and environments. It is therefore outside the scope of this paper to complete that level of testing, though we agree that is an important and obvious next step. In revisions, we will be careful in how we describe the level of testing that has been done on the SWISS and advocate more clearly for rigorous testing of the system against other traditional methods of soil water extraction methods in our future work section.
RC: Formally, i would recommend using italics (once introduced) for the different methods throughout manuscript, clearly distincting between method, results and discussion statements, and using past tense in method and results.
Author Response:We appreciate this comment and will slightly reorganize the methods sections so that the different methods are more clearly defined. We will also carefully edit the manuscript to correct tense and grammar.
Citation: https://doi.org/10.5194/egusphere-2022-1170-AC2
-
AC2: 'Reply on RC2', Rachel Havranek, 22 Feb 2023
Interactive discussion
Status: closed
-
RC1: 'Comment on egusphere-2022-1170', Stefan Seeger, 21 Dec 2022
The manuscript "Lessons from and best practices for the deployment of the Soil Water Isotope Storage System" presents an important step towards the automated aquisition of in-situ soil water vapor samples from remote sites that are not suited for the field deployment of stable water isotope analyzers. The authors have submitted a very well structured paper that I enjoyed reading.
1. My biggest concern is of formal nature, because in my opinion this paper is more a technical note than a research paper - I do not see any research question asked or answered, nor do I see any scientifically meaningful findings. Nevertheless I really do appriciate the effort that has been made to thoroughly develop and test such a setup and I think the manuscript is worthy for publication.
2. Concerning the actual content of the manuscript, I would appreciate a more detailed discussion of the outliers and oberved system failures.
2.1. In Line 240, you describe that you repeated your dry air tests "until the majority (at least 13/15) flasks" had a sufficiently low moisture content. Why do not all flasks have to pass your threshold criterion. How do you treat measurements of the somewhat leaky flasks?
2.2 During your testing with known isotopic compositions it is easy to spot faulty flasks, but on what basis would you treat actual field measurements? Is there a reliable way to spot outliers?
2.3. During the field application you had to discard 10 of 45 measurements. That is a failure rate of over 22%. Do you have any suggestions on how to reduce this failure rate?
Apart from that, I only have minor points to comment on.
Line 73: Your list of successfull field deployments of laser-based instruments for in situ measurements is missing some notable cases that either are older or produced much more measurements than your given examples:
Volkmann2016 https://doi.org/10.1111/nph.13868
Gaj2016 https://doi.org/10.5194/hess-20-715-2016
Seeger2021 https://doi.org/10.5194/bg-18-4603-2021
Gessler2022 https://doi.org/10.1111/nph.17767Line 267-268: Could you elaborate on why and how the heating does "create a longer and more stable window of measurement time". I kind of get a vague idea about this within the follwoing paragraph, but I think a more explicit explanation of the effect of this heating procedure right at this point would be suiting.
Lines 285-286: "For some flasks, using either a later portion of the measurement period, or slightly earlier offered a more stable signal."
Do you treat those bottles differently during normal operation of the system, or do you treat all bottles equally, even though they might perform better if treated more individually?Lines 318-319: You refer to a "SWISS unit offest correction (detailed below)". Then there is one last method subsection and the results begin. You kind of describe an offset correction in the results subsection "5.1.2. Water vapor test", but maybe the description should happen within the methods part of the paper? Or you could place a reference to where the the offset correction is actually described.
Lines 349-370: In that section you describe 18O and 2H completely separately. Do the outliers of 18O coincide with the outliers of 2H, or are they completely independent? How do the outliers look like in the dual isotope plot? Could that help to diagnose why these outliers are outliers?
Lines 374-380:
I think the long term field storage tests look a lot like 43 days of storage are not any worse than 34 days. Do you expect that there is actually a critical time limit somewehre between 43 and 52 days which can explain the difference between Fondue and Toblerone? Or has Toblerone issues that also would affect its results after a shorter storage period?
Repeated tests with one and the same SWISS unit over different time spans might be more informative than the comparison of three separate units after different storage times. (But of course that would take a lot of time...)Line 524: "Samples were taken approximately every five days...". Why is it "approximately" five days? I suppose your automation procedure does not involve a random number generator? Shouldn't you be able to state the precise sampling interval? Maybe just drop the "approximately" (even if - for whatever reasons - it weren't perfect 5 day intervals).
Line 218: Could you specify the type and manufacturer of the "helium leak detector".
Line 247: "...did not lead by the time..." maybe you meant "leak" instead of "lead"?
Citation: https://doi.org/10.5194/egusphere-2022-1170-RC1 -
AC1: 'Reply on RC1', Rachel Havranek, 22 Feb 2023
The authors thank Stefan Seeger for a constructive and helpful review. Below we detail our response to his comments:
1. My biggest concern is of formal nature, because in my opinion this paper is more a technical note than a research paper - I do not see any research question asked or answered, nor do I see any scientifically meaningful findings. Nevertheless I really do appriciate the effort that has been made to thoroughly develop and test such a setup and I think the manuscript is worthy for publication.
Author Response: We agree that this paper is more of a technical note rather than research paper and will change the submitted format appropriately.
2. Concerning the actual content of the manuscript, I would appreciate a more detailed discussion of the outliers and observed system failures.
2.1. In Line 240, you describe that you repeated your dry air tests "until the majority (at least 13/15) flasks" had a sufficiently low moisture content. Why do not all flasks have to pass your threshold criterion. How do you treat measurements of the somewhat leaky flasks?
2.2 During your testing with known isotopic compositions it is easy to spot faulty flasks, but on what basis would you treat actual field measurements? Is there a reliable way to spot outliers?
2.3. During the field application you had to discard 10 of 45 measurements. That is a failure rate of over 22%. Do you have any suggestions on how to reduce this failure rate?
Author Response: Because the goal of this paper is to make the system as usable by the greater community as possible, and to be as transparent as possible about the system’s strengths and weaknesses, we are happy to add further descriptions of the outliers and system failures.
We appreciate comment 2.1 and will more clearly articulate our philosophy about how to move through QA/QC efficiently; our threshold of at least 13/15 flasks holding dry-air was to be able to keep moving through the process so that SWISS units could be ready for field deployment. This discussion will be included in section “6.1.4 Lessons learned and recommendations from the QA/QC and field suitability tests.” We will also note ways in which we would likely do QA/QC differently in the future, so that other users can adopt a QA/QC process that works for their uses.
We will use the ‘mock’ field deployments to discuss In greater detail ways to identify if condensation is affecting isotope values during measurement based on peak shape, and ways we identified suspected condensation during vapor collection. We will also use the mock field test as a way to discuss more thoroughly how we treat flasks that we suspect are slightly leaky. Lastly, we will also discuss the importance of sequential data collection to identify data that are spurious, and our approach to designing sequence timing to be able to both identify bad samples and answer the question at hand.
A 22% failure rate is quite high, and I think our field deployment points to ways that either the automation system could be improved and/or when the SWISS is likely to fail. First, it’s not entirely clear what soil water vapor samples collected from very dry soils using the vapor permeable tubing represent, and a targeted study looking at dry soil conditions would be a really helpful next step. Second, some updates to the automation schema that make it more remotely controllable would help researchers avoid sampling on days when condensation in the field might be likely.
Citation: https://doi.org/10.5194/egusphere-2022-1170-AC1
-
AC1: 'Reply on RC1', Rachel Havranek, 22 Feb 2023
-
RC2: 'Comment on egusphere-2022-1170', Anonymous Referee #2, 24 Dec 2022
General comments:
In their manuscript entitled “Lessons from and best practices for the deployment of the Soil Water isotope Storage System”, Havranek et al. describe their experience with automating and operating a self-developed soil water vapor sampling system for subsequent, lab-based stable isotope analysis. They present an extensive testing procedure necessary prior to unattended field-deployment. Overall, the manuscript contains a lot of detailed instructions and hints that are definitely helpful for replicating the setup and producing soil water isotope datasets, which are valuable for the ecohydrology community and thus the readers of HESS. Therefore, I appreciate the effort of the authors and I recommend publication after major revision following the comments I specified below.
In my opinion, the manuscript would benefit from a better structuring as currently some important method details are described in the result section, part of the data is presented not before the discussion section etc. (see details in the specific comments). Also, the discussion section currently comprises a lot of repetitions of method and result details. Instead, it should be focused more on the critical evaluation of the presented data and other findings. Given the problems encountered in the field and the corresponding failure rate, I find it unfortunate to call it a “full” testing as obviously not the full spectrum of potentially relevant environmental impacts could be accounted for during those tests.
Further, I would have liked to see an independent validation of the presented soil vapor isotope data by established means in order to have a profound accuracy assessment. In my understanding, precision is the standard deviation of replicate measurements, which have not been performed on soil samples. Therefore, I am not sure whether the precision and accuracy assessments from the lab tests can be assumed for the natural soil samples as well.
Formally, i would recommend using italics (once introduced) for the different methods throughout manuscript, clearly distincting between method, results and discussion statements, and using past tense in method and results.
Specific comments:
Line 12: “full” may be too strong given the unaccounted-for circumstances encountered during the unattended field-deployment
L19: You are stating the precision. What about the accuracy? Do you assume perfect accuracy even for non-lab conditions after performing the offset correction?
L22: “faithfully” may be too strong given that soil water isotope data were not validated by established means.
L31: root WATER uptake
L40: You might consider mentioning here also the work of Wassenaar and colleagues (doi: 10.1021/es802065s) who invented the principle of using laser-based instrument to measure vapor for liquid water isotope assessment. All in situ isotope sampling approaches are now based on this principle.
L46: The expansion OF in situ…
L50-54: I am sure this is a good example to prove your point why a setup like yours is helpful/needed. Unfortunately, I do not understand what it means. What are br-GDGTs?
L62-63: I would expect such recommendations to appear in the discussion/conclusion but not in the introduction.
L68-96: This reads like a second introduction where you identify a problem you are intending to solve. Why not merge this information with section 1?
L79: vapor CONCENTRATION gradient
L89: you already defined the abbreviation “SWISS” in line 61/62
L103: Do you have indications/references that this time frame is sufficient for the soil to return to natural conditions? Can this be tested? How?
L121: Wouldn’t you want to prevent condensation instead of only limiting it?
L140: This reference is missing in the Works Cited section
L144: Wouldn’t vapor bound to the flask walls cause a memory effect that needs to be re-assessed and corrected for prior to every new deployment of each single flask? If so, how would this impact the universal offset you applied if not all flask had the same filling history in terms of isotopic composition prior to common deployment?
L147 and throughout MS: Please do not use different formats for stating the tubing diameter (1/8 inch vs. 1/8” vs. 1/8th inch etc.)
L151: please define the material of the foam and the nature of the insulation (thermal?)
L160: Do “offsets” mean that there is no equilibrium at shorter (or greater?) lengths?
L162: Was it mandatory as well to cut the Bev-A-Line connections to identical lengths?
L186: Please define VDC
L200ff: In my opinion, such statements discussing the suitability of and potential alternatives for certain components are better placed in the discussion
L218: Please define the helium detector model and manufacturer.
L219: To me, this is a result statement
L230: what quantity was measured on each flask? I understand that the dead-end pull method applies a vacuum to the flasks. Were the custom-made flasks tested for vacuum suitability? Are extra safety measures necessary for this step? Would a vacuum test be able to identify large, medium, or small leaks all at once?
L231: please move the parentheses to line 227 and use italics consistently throughout the manuscript when naming the different methods
L235: “we found…” sounds like a result statement
L236: ”…would likely…” sounds like a discussion statement
L240: Why did the dry air tests have to be repeated? Did you encounter gradual decrease of vapor adsorption at the flasks’ walls or did you change anything in the setup (e.g. tightening of connections) between those tests?
L247: I think it must be “leak”, not “lead”
L253: How far and for how long was the tubing immersed in water?
L270: How was dryness verified?
L279: Picarro claims that the measurement range for their L2130-i model is 1000-50000 ppmv. How come that you have condensation issues at vapor concentrations below 50000 ppmv?
L281: Why does changing pressure cause isotope fractionation?
L283: what signal was supposed to be stable? Isotope? Vapor concentration?
L287ff: why was it necessary to use two different methods to assess instrument stability? And how stable was the instrument compared to the SWISS analytical uncertainty?
L295: In what aspects were field conditions different from lab conditions and why wasn’t it sufficient to perform these tests just outside the lab building? Are the results of these tests not transferable to other field sites?
L314: Using two different methods within one batch of samples and standards contradicts the principle of identical treatment (Werner & Brand, 2001, doi: 10.1002/rcm.258). Why is it justified to do so in this case?
L332/338: It would be nice to find the 500 ppmv threshold on the vertical axis of Figure 3. Also, I would prefer to have the horizontal axis start with “2”.
L346: These method details should appear in the method section, not in the results.
L349: “we expect” sounds like a discussion statement (also in L362). What are the uncertainty limits and are they really transferable given the quite different setups (Oerter et al. vs. SWISS)?
L379f: This is already an interpretation of the results. Why is there an inter-unit variability? Or why do you expect one? And why is it (expected to be) different from the intra-unit variability?
Figure 4, 5, and 6: Please add the permil symbol to the vertical axes labels when presenting isotope data. Please add the 500 ppmv to the axis if this is the concentration threshold.
L508: This is misleading as all flasks contain water vapor. Maybe you want to make the distinction between FLASH-EVAPORATED water vapor and atmospheric vapor?
L513 and elsewhere: I think table captions should be placed above the tables
L533ff: Why were those data excluded? 13000 ppmv is well within the instrument’s measurement range. If the soil had been very dry, aren’t the obtained data then representative for this setting and can be interpreted accordingly? Do you mean atmosphere via leaks or by being present in the soil despite the huge gas-liquid interfacial area present even in quite dry soils and facilitating quasi-instant exchange between the liquid phase and a potentially intruding atmosphere?
L541f: How was condensation detected and why were these data excluded? Presenting (and later discussing) even spurious data in this rather technical manuscript would help future users of your setup to identify similar problems.
L548: The discussion of the helium test is missing
L551-556: Why is a seven-day test sufficient when the intended field-deployment is on a much longer timescale?
L557-561: How is it possible that not all of the atmospheric water vapor has been flushed given the achieved turnover rate? How can this be tested/identified? What solution would you suggest to prevent this from happening?
L564-570: This is a repetition form previous sections. Unfortunately, many paragraphs in the discussion start with a repetition of the results which should be avoided.
L580: These data must be presented in the result section. And discussed here. I for one find it very remarkable that there is absolutely no variation in temperature with depth AND time for two of the three field sites. Any error in this variable would have an impact on the trustworthiness of vapor concentration and thus isotope readings, right? Also please add “(°C)” to the table heading of the temperature column.
L586-601: This is not a proper discussion but mainly a repetition from previous sections.
L599: What is the minimum acceptable concentration for this method? Given the overall analytical uncertainty, the additional uncertainty resulting from lower-than-optimum vapor concentrations should not make that much of a difference for quite a wide range, right?
L607: are you referring to air or below-ground temperatures?
L614: “quite” is too vague. Is the system sufficiently resistant to facilitate a specific uncertainty that allows for deciphering natural variations which – as you state elsewhere – become smaller with depth?
L618: how was condensation noticed? This is valuable information for unexperienced users considering to employ your system.
L620: “oxygen isotope” is redundant.
L624: Why is this surprising?
L633: I think you didn’t mention “replacing” in the method section. Only “tightening”.
L639: the observed “greater exchange and leaking” is based on the measurement of only two flasks (#10 is missing in the seven-day data, Figure SI 5) whereas three of the stainless steel fitted flasks (#3, #14, #15) reveal HIGHER vapor concentration than the PTFE fitted flasks after seven days and LOWER-than-before vapor concentrations after 27 days. How is the latter possible?
L645: At what depth does the magnitude of seasonal variability have to be expected to fall below the system’s analytical uncertainty? And what is the accuracy here?
L648: In my understanding, vapor bound to the flask walls would introduce a memory effect which you excluded (L651). Did you really test for this by using flask with intentionally different filling histories for sampling identical vapor sources?
L653-655: Why is the concentration benchmark of multiple dry air tests relevant and not the duration of the test itself?
L657: I’m struggling with “faithfully” because obviously there were non-tested factors causing the high failure rates during the actual application. Also, how would you identify outliers without an independent validation of soil water isotope data?
L663: “Helpful”? Isn’t it even mandatory in order to prevent condensation of the soil’s vapor-saturated gas sample? Would a second Valco valve for the application of a constant zero-air dilution stream do the trick more reliably (similar to Volkmann & Weiler, doi: 10.5194/hess-18-1819-2014)? Or would this go beyond the budget?
L673: Please specify that the uncertainty refers to the hydrogen isotope values
L682: All data should be presented in the results section
L686: How much is that in mm? Can you exclude convective precipitation with limited spatial extent or why was it “likely” (L687) indicative for precipitation at your field site?
L691: If any such precipitation event homogenizes the soil water isotope depth profile then how can you still assume only small seasonal variability at those depths? Wouldn’t you then rather have to expect a significant shift in the soil water isotope depth profile after every precipitation event of similar magnitude? What other factors (soil moisture?, pore size distribution?) could be relevant in this context?
L717: Results. Please move to and describe in the result section.
L721f: Please also show discarded data in order to give readers of this rather technical paper the opportunity to learn from your experience. Then explain in detail what aspect(s) made you exclude these data from further consideration. What do you mean by “sufficiently”? According to Picarro’s instrument specifications, even at 2500 ppmv the uncertainty induced by low vapor concentration is still below the SWISS overall uncertainty. If such low vapor concentrations were representative for very dry soil, wouldn’t the SWISS still be able to deliver meaningful results?
L737: Can you provide details of this evidence? Which conditions caused condensation and how can it be avoided or otherwise dealt with?
L742: What does “CoAgMet” mean?
L748: The low d-excess value is not helpful in this context because it is not an independent measure but is calculated directly from the aforementioned oxygen and hydrogen isotope values.
L749: Again, please report such information in the results section.
L766: Why does uneven heating have an adverse effect? Wouldn’t ANY heating be helpful as long as the induced temperature exceeds the sample’s dew point, no matter by how much?
L768: Do you mean consistent amounts of heat or stable temperature?
L775: What problems would an IoT cellular router solve?
L778: Why is the inherent thermal structure of the soil important? How does it affect isotope measurements?
L783: “can be” sounds a little weak. Previously, you “strongly recommended” this procedure to be applied.
Supplement: Please use similar formatting for the different figures.
Table SI 1: You emphasized the option to reproduce the SWISS setup (L169). Can you please list the monetary costs for different components or the entire SWISS setup as well?
Table SI 2: Are you endorsing Amazon?
Table SI 3: Are the high standard deviation values for Toblerone after 52 days a result of noise or trends during measurements?
Supplement section 1: Build-out description: How many flasks were ultimately broken before the desired setup was achieved?
Citation: https://doi.org/10.5194/egusphere-2022-1170-RC2 -
AC2: 'Reply on RC2', Rachel Havranek, 22 Feb 2023
The authors thank this reviewer for their thorough and constructive review. Their comments will significantly improve this manuscript. Below we detail our response to their comments.
RC: In my opinion, the manuscript would benefit from a better structuring as currently some important method details are described in the result section, part of the data is presented not before the discussion section etc. (see details in the specific comments). Also, the discussion section currently comprises a lot of repetitions of method and result details. Instead, it should be focused more on the critical evaluation of the presented data and other findings.
Author Response: We appreciate their concerns about the structure of the paper – we have found it challenging to structure the paper in such a way that clearly articulates the sequential development steps of the SWISS, while maintaining a traditional scientific paper format. In revisions, we will more clearly separate results from the discussion, and appreciate that the reviewer helped to identify specific places in the discussion where we can write more efficiently.
RC: Given the problems encountered in the field and the corresponding failure rate, I find it unfortunate to call it a “full” testing as obviously not the full spectrum of potentially relevant environmental impacts could be accounted for during those tests. Further, I would have liked to see an independent validation of the presented soil vapor isotope data by established means in order to have a profound accuracy assessment. In my understanding, precision is the standard deviation of replicate measurements, which have not been performed on soil samples. Therefore, I am not sure whether the precision and accuracy assessments from the lab tests can be assumed for the natural soil samples as well.
Author Response: We appreciate that the data we have presented is not truly a ‘full’ testing of the system, and we did not design these initial experiments to test the SWISS against other methods of extraction soil water isotope data, nor is this paper intended to address that stage of development and testing. Our goal with this paper is to get the SWISS out quickly to other potential users so that they may be able to build and test the SWISS for their scientific questions. We agree that testing the SWISS against other methods of soil water extraction is an important next step for testing the SWISS, but requires a more focused experiment that fits the needs of multiple stakeholders in the ecohydrology community. Further, the first-order tests that this paper presents were a necessary first step to be confident that the system works as expected, and how to identify problems, so that later experiments will be reliable. That level of testing should be done to complement the testing that has already been done on the vapor permeable tubing and to address concerns about the reliability of the tubing across different soil textures and environments. It is therefore outside the scope of this paper to complete that level of testing, though we agree that is an important and obvious next step. In revisions, we will be careful in how we describe the level of testing that has been done on the SWISS and advocate more clearly for rigorous testing of the system against other traditional methods of soil water extraction methods in our future work section.
RC: Formally, i would recommend using italics (once introduced) for the different methods throughout manuscript, clearly distincting between method, results and discussion statements, and using past tense in method and results.
Author Response:We appreciate this comment and will slightly reorganize the methods sections so that the different methods are more clearly defined. We will also carefully edit the manuscript to correct tense and grammar.
Citation: https://doi.org/10.5194/egusphere-2022-1170-AC2
-
AC2: 'Reply on RC2', Rachel Havranek, 22 Feb 2023
Peer review completion
Journal article(s) based on this preprint
Viewed
| HTML | XML | Total | Supplement | BibTeX | EndNote | |
|---|---|---|---|---|---|---|
| 292 | 139 | 26 | 457 | 64 | 9 | 10 |
- HTML: 292
- PDF: 139
- XML: 26
- Total: 457
- Supplement: 64
- BibTeX: 9
- EndNote: 10
Viewed (geographical distribution)
| Country | # | Views | % |
|---|
| Total: | 0 |
| HTML: | 0 |
| PDF: | 0 |
| XML: | 0 |
- 1
Rachel Elizabeth Havranek
Kathryn E. Snell
Sebastian H. Kopf
Brett Davidheiser-Kroll
Valerie Morris
Bruce Vaughn
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
- Preprint
(3180 KB) - Metadata XML
-
Supplement
(24118 KB) - BibTeX
- EndNote
- Final revised paper