Prediction of biopore- and matrix-dominated flow from X-ray CT-derived
macropore network characteristics

Naveed, Muhammad; Moldrup, Per; Schaap, Marcel G.; Tuller, Markus; Kulkarni, Ramaprasad; Vogel, Hans-Jörg; Wollesen de Jonge, Lis

doi:https://doi.org/10.5194/hess-20-4017-2016

Articles | Volume 20, issue 10

https://doi.org/10.5194/hess-20-4017-2016

© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.

https://doi.org/10.5194/hess-20-4017-2016

© Author(s) 2016. This work is distributed under
the Creative Commons Attribution 3.0 License.

Articles | Volume 20, issue 10

Research article

|

06 Oct 2016

Research article |

| 06 Oct 2016

Prediction of biopore- and matrix-dominated flow from X-ray CT-derived macropore network characteristics

Muhammad Naveed, Per Moldrup, Marcel G. Schaap, Markus Tuller, Ramaprasad Kulkarni, Hans-Jörg Vogel, and Lis Wollesen de Jonge

Download

Final revised paper (published on 06 Oct 2016)
Supplement to the final revised paper
Preprint (discussion started on 20 Nov 2015)

Interactive discussion

Status: closed

AC: Author comment | RC: Referee comment | SC: Short comment | EC: Editor comment

- Printer-friendly version

- Supplement

RC C5627: 'What is new?', Anonymous Referee #1, 14 Dec 2015
- AC C6715: 'Reply to comments (Referee # 1)', Muhammad Naveed, 13 Feb 2016
RC C5632: 'Report of an interesting experimental data set', Anonymous Referee #2, 15 Dec 2015
- AC C6718: 'Reply to comments (Referee # 2)', Muhammad Naveed, 13 Feb 2016
SC C5713: 'review', John Koestel, 19 Dec 2015
- RC C5804: 'An example of previous papers', Anonymous Referee #2, 23 Dec 2015
- AC C6722: 'Reply to short comment', Muhammad Naveed, 13 Feb 2016
RC C5794: 'Referee comments', Anonymous Referee #3, 22 Dec 2015
- AC C6727: 'Reply to comments (Referee # 3)', Muhammad Naveed, 13 Feb 2016
RC C5811: 'Reviewer comments', Anonymous Referee #4, 24 Dec 2015
- AC C6732: 'Reply to comments (Referee # 4)', Muhammad Naveed, 13 Feb 2016

Peer-review completion

AR: Author's response | RR: Referee report | ED: Editor decision

ED: Reconsider after major revisions (14 Feb 2016) by Alberto Guadagnini

AR by Muhammad Naveed on behalf of the Authors (19 Apr 2016)

ED: Referee Nomination & Report Request started (03 May 2016) by Alberto Guadagnini

RR by Anonymous Referee #2 (03 Jun 2016)

Suggestions for revision or reasons for rejection

The authors revised the paper and provided an improved version. Nevertheless, I am sorry, but I think that after these changes the manuscript still does not fully satisfies the quality standard for publication on HESS.
The data processing yields a set of one-to-one functional relationships between CT-based indices and physical properties of the samples (saturated permeability, air permeability and gas diffusion at two different values of suction). My personal opinion is that this is still far from a "predicting model", which can be used to derive physical properties from structural data, including a description of the macropore region.

I provide below a few comments and I suggest some corrections, which I hope can be useful if the paper is accepted for publication.
(1) Please, check the style for expressing values of quantities, in particular when a range of values is indicated (I suggest to conform to the recommendations of section 7.7 of the NIST guide to the SI; see http://www.nist.gov/pml/pubs/sp811/sec07.cfm).
(2) I agree with comment #4 by Referee #3 of the first version: saturated permeability is independt of the fluid, so that "water" should be erased from the expression "saturated water permeability" throughout the whole paper. Along the same line, I suggest to denote this quantity with "ksat" instead of "kw", not only in the text, but also in the tables and figures. Notice that "saturated permeability" could be substituted with the symbol "ksat" at several places in the text.
(3) Page 7, lines 21 to 24. Isn't it possible to provide a quantitative index to separate and label columns with biopore- or matrix-dominated flow?
(4) Page 9, line 10. Correct the dimensions of the pressure difference.
(5) Page 11.
a) Lines 7 to 9. The sentences "Although... distributions." can be erased, without loss of information.
b) Lines 20 to 22. The sentences "Large... distribution." can be shortened.
(6) Page 13, line 20. Substitute "pores" with "pore diameters". Similar expressions are used in other sentences and should be changed.
(7) Sorry, but in my view the conclusions are not so exciting. In particular the first conclusion is strictly related to the study site and seems to be of moderate relevance for the goal of the paper.
(8) Table 1.
a) Correct the measurement units.
b) Are the significant digits used for each measured quantity coherent with the measurement resolution and accuracy? For instance, I am afraid that hydraulic conductivity cannot be measured with 5 significant digits. The same comment applies to the supplementary material.
c) Please, check whether the number of significant digits for mean values is coherent with the standard deviation.

Hide

RR by Anonymous Referee #4 (07 Jun 2016)

Suggestions for revision or reasons for rejection

I have read the new version of the manuscript as well as the authors’ response to earlier comments received from reviewers or volunteers.
I can say that the manuscript is undoubtedly in better shape than it earlier was. To me it reads much more smoothly. As of the factor of novelty, I don’t think that we are already at the stage in this field when this study and this data set cannot provide sufficiently new information. Novel information has now been pointed out better, and the publication of the data set is of great additional value. I have two points of significance and a few small remaining comments that I’m asking the authors and editors to consider before accepting the paper’s final version.

1. The authors received comments on the matter of ‘field scale’. They agreed to remove this term from the title, but the question points beyond the title. The study still suggests that this is field scale, while e.g. earlier studies (PTF studies) were concerned of a much smaller scale. The fact is that both types of studies are concerned of the same scale. Indeed, many PTF studies stop at testing findings at the same sample scale, but many others attempt to conclude about information or phenomena at a larger scale, even above field scale. Both those PTF studies and this study are supported by samples of the same size and scale. In fact, pF rings are typically even taller than those used in this study, i.e. lots of PTF studies rely on larger samples. This needs to be resolved, preferably by removing the field scale notion from the paper entirely – unless that aspect is really well supported.

2. I feel that the authors push to position themselves away from PTFs. This seems to be apparent by the sum of comments made about earlier PTF works, e.g. on how much data is needed, what ‘scale’ those are developed for, and most importantly those they cannot predict macropore flow. Latter is true, but the authors need to remain objective and modest about what this paper really is. This is just as much a PTF study as all those referred earlier studies are. It carries all those characters, all the strength and all the weaknesses that those earlier studies do. The only substantial difference is in what the inputs and outputs are, and the methodology how those were obtained. Otherwise, the sample size/scale is the same, the study is based on a collection of 65 local samples (vs. limited data – see minor comment below), and the authors present regression equations between soil properties (Table 2, Figure 8). Please also see additional comment below on P12, L23. Please revise the manuscript to remove the ‘us’ and ‘them’ attitude and present this study by what it is: a PTF work with novel data support. This means revising the text everywhere where there is commenting on PTF data support, scale, etc.

3. Some of the earlier reviewer comments given were responded to as ‘Done’, while it is apparent from the manuscript’s new version that they were not done. I assume they were forgotten, e.g. comments 5, 8, 29 of reviewer 4. They all should be checked again.

Minor/editorials:
P2, L4: due to its inherently
P3, L2: for the partitioning
P3, L23: ‘higher’ should preferably be replaced with ‘greater’ in this context. This is applicable to the whole manuscript, please check consistently.
P4, L21.23: here I have a problem with the ‘field scale’
P4, L24: development of new imaging techniques or depelopments in imaging
P5, L12-13: neither is your study conducted at the field scale!
P5, L12-13: how many samples mean ‘limited’, and how many don’t? Better avoid such statements I think, because with the technology developing, 65 samples will soon be referred to as ‘limited data’.
P5, L22-26: again, the ‘limited number’ statement would be good to avoid.
P5, L22.26: To me this sentence is misplaced, and should better be built in somewhere before the ‘second and novel part’ sentence (or even earlier, maybe in the previous paragraph) as something identified from the literature search, which you will then address in this study.
P6, L15: How does the 15 by 15 grid cited here compare to the grid presented in Figure 2? Or is 15m the spacing, between 2 points in Fig 2? Then state that you had a 12x5 rectangle with 15m spacing + 5 points in between, apparently.
P8, L22-23: Soil texture was determined from disturbed soil samples using a combination of wet sieving and the hydrometer method, after passing the sample through a 2-mm sieve.
P10, L9-10: this does not answer the original comment – still no statement similar to what has been in the response file.
P11, L4: from 14 to 19%
P11, L26: out of the 4 marked
P12, L11: still incorrect, those studies did not predict kw, you converted them from Ksat.
P12, L15-16: narrow range of saturated water permeability was predicted.
P12, L16: …a wide range
P12, L17: The primary reason for the failure of the…
P12, L23: PTFs are not developed for the scale of a horizon necessarily, just samples are taken from horizons for some practical considerations. In the same sense, all of these samples from 5-8.5 depth are representative of the A horizon, so there is just as much reason to classify your study under the same umbrella with those PTF studies. This is again a reason why I don’t see this study to be so far from PTF studies that the authors try to distance it from.
P12, L24-25: …developing empirical models for the prediction…
P12, L26: From those, we have tested …
P13, L5: instead of ‘past century’ use ‘previous century’, or ‘20th century’
P13, L5: models for the prediction
P15, L2: in the soil column
P15, L6: CT-derived macroporosity (Figure 8).
P15, L6-7: I still think that there should be a better term found for “two-branch system data trend”
P15, L8: by ‘upper branch’, are you referring to the branch ‘showing greater permeability’ (or whatever applies)? Use more precise references to the data.
P15, L10 and L12: they are first marked in Figure 2, aren’t they? That is where you introduce them.
P15, L10-12 (two instances): replace ‘fall under this branch’ with e.g. ‘are members of this branch’ – or something similar in meaning.
P15, L13: for the two categories
P15, L14: replace ‘reflected’ with ‘suggests’
P16, L4-5: Significant strong power regressions… this sentence should be revised/replaced for various reasons (also for two-branch system)
P16, L5: Similarly
P16, L5-7: This significant increase is relative to what? Include in the sentence.
P16, L19: What is an ‘independent significant power regression’? Rephrase.
P17, L21: was found to be the key predictor that yielded the highest adjusted R2 value…. among all macropore flow parameters….
P18, L1: predictors of macropore flow in the future
P19, L6-7: Existing empirical models overpredicted saturated water permeability in case of …. And underpredicted it in case of…. (However, I need to add that your models are empirical too, so a different term may be better).
P19, L16: Total macroporosity
P19, L23: replace ‘data’ with ‘image’
Figure headings: are Figure 4 an 5 headings mixed up?

Hide

ED: Reconsider after major revisions (13 Jun 2016) by Alberto Guadagnini

The manuscript has been reviewed by two of the reviewers who were involved in the previous round of reviews. One of them is still very negative, his/her main point being linked with the apparent lack of predictive capability of the type of correlations they come up with. While this might be true, I urge the authors to clearly state this aspect in the body of the manuscript so that there is no ambiguity to it. The second reviewer still lists some major concerns about the PTF-based nature of the work as well as about the appropriate identification of the scale which is of relevance to the study. The reviewer also lists a series of other points which are relevant, even as they can be considered as minor.
After my own reading, I do concur that the data set can be of potential value for HESS, even as the model does not appear to have a convincing predictive capability. This said, I also noted that the details of the variogram analysis underlying the kriging application (for example, the type of variogram considered as well as the way its parameters are estimated, including a quantification of the associated uncertainty) are missing from the manuscript (Incidentally, is the exponent “9” correct in Equation (1)? Just to make sure). All in all, I do concur with the less negative reviewer and I am willing to give the authors a very last opportunity to strengthen their work. Note that I am still recommending major revisions to be implemented, according to the spirit of my assessment above. In doing these revisions, the authors are also encouraged to avoid leaving out amendments which they claim have been undertaken. To me, it has to be clearly stated that the contribution of the work to the advancement of our knowledge is not too large and it is compensated by the presentation of the valuable data-set that the authors analyze. If this is acceptable to the authors, then I suggest they prepare a revised manuscripts along these lines. The latter will then most likely be re-reviewed.

Hide

AR by Muhammad Naveed on behalf of the Authors (25 Jul 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (26 Jul 2016) by Alberto Guadagnini

RR by Anonymous Referee #4 (29 Aug 2016)

Suggestions for revision or reasons for rejection

I have seen the latest version (2nd revision) of the manuscript and the comments of two reviewers and by the editor. I think this last round of revision has taken the manuscript to where it should be; it matured a lot. It reads well, it is organized properly, and I believe the added item of re-parameterizing the Ahuja et al. 1984 model was a good and demonstrative addition. The manuscript now addresses the most significant reviewer concerns, as well as the many small remarks. I saw a very welcome change in positioning the paper in the field of hydraulic property estimation, and an improvement in demonstrating its capabilities. Latter being said, I think that even if ‘improvement’ in estimations per se is not yet achieved, demonstrating very promising capabilities by the use of new technology and obtained data has its very legitimate place as a publication. I list a few remaining items to be addressed before the manuscript is accepted. Most of those are a fallout of the recent revisions.

Comments:

1. I understand that the abbreviation of ‘ksat’ for ’saturated permeability’ was put in place in response to reviewer comment. Now I see a bit of a confusing picture – partly as a result of that. Now Ksat and ksat are both used, and it may be wise to gently remind the reader not to mix them up. I think some will simply think about hydraulic conductivity under ksat as well, when this study is cited later. IN FACT, in this very study, on P15, L8-22, the two are washed together by the authors themselves! The authors converted their data to sat_perm from sat_cond (stated on P11). Sat_perm is identified by ksat (lowercase k) on P15, L8, and existing models that estimate sat_cond are cited on L10-12 on the same page. Then on L12-15 the same models are identified as ksat (lowercase k) models that do not perform well. The case is taken further to the summary of this paragraph (L20-22) where these models are again cited to predict ksat (lowercase k). Please fix this throughout the paper very carefully, and state very clearly which of your data you have been using in each sub-chapter of your study, sat_cond or sat_perm – since you had both available. I don’t expect any notable changes to the outcome of the study, but it should be clear what is what. If any terminology or abbreviation is changed, check also on Tables and Figures to remain consistent. To this end, according to the caption of Figure 4, those existing models predicted permeability, which is not the case. In some way this comment is still connected to the old comment on ‘existing models’ estimating Ksat, rather than Kw (now ksat).

2. On linear/power regressions and the resulting figure (i.e. text on P13, Figure 8):
(a) I don’t see a single linear fit in Figure 8, i.e. wherever a regression was significant, it always ended up being a power model. Perhaps alter the text on P13 to say that while fitting a linear model was tried, a power model was always superior, where a significant regression model was found. Also check the caption of Figure 8.
(b) There are two different reasons why some plots in Figure 8 only show one regression fit, but those are not separated by color or any other indicator. Plots d,g,h,k and l have one curve fitted because the other was not significant, while plots q and r have only one model because the two models did not differ significantly from each other. Consider noting this at least in the figure caption.

3. The introduction of the Ahuja model parameterization is good. I would like to ask to consider three improvements to its presentation:
(a) On P12, under the introduction of the modeling part, it would be good to introduce what effective cylindrical diameter is equivalent to -33 kPa, that is originally used by the Ahuja model, and how that relates to the image resolution and the derived properties. An alternate placement for this can be the discussion part on this subject.
(b) Is data available to calculate effective porosity and then actually present the results by the original Ahuja model for these soils? It would be great to show the comparison of the conventionally parameterized and the newly parameterized models, and discuss the improvement, if any.
(c) Figure 9 – whether expanded by the original Ahuja model’s predictions or not – could present at least the R2, but perhaps also the RMSE, if found meaningful. It is for future reference, so that studies remain quantitatively comparable.

4. P16, L21-23 and Figure 6: Instead of only stating that the two metrics ‘agreed well’, can at least the R2 be provided, or also the RMSE? It is for future reference, so that studies remain quantitatively comparable.

5. P18, L15-17: This sentence appears to be in reference of Figure 8a. There does not appear sufficient data support to say that the two branches actually merge. They do approach, that is clear, and likely merge at some point. I also noticed that now there is only a single use of the ‘two-branch data trend’ phrase - good. With or without that term replaced, I suggest the following alternate sentence here: “Two data branches are observed at lower imaged macroporosity, which approach towards a single branch as imaged macroporosity increases.”

Minor editorials:
P8, L6: …containing 100 to 1000 biopores per m2. (need to state unit twice?)
P15, L24: that -> those and herewith -> thereby
P17, L8: delete ‘absolutely’?
P17, L9: ‘accuracy of the employed’: replace with ‘applicability of the applied’ ?
P18, L13: See comment 2a above. Not a single linear model seems to be shown in Figure 8, so perhaps delete it from the sentence here.
P18, L22: I think it is better to say ‘Distinct significant’ (as on P20, L8) rather than the other way around. If agreed, change it here as well as on P19, L16.
P19, L17: …i.e. those with biopore-dominated flow and with matrix-dominated flow.
P20, L25: of the Kozeny-Carman
P22, L14: i.e. presented biopore-dominated flow
P22, L17: by splitting soil columns into a group of matrix-dominated flow columns and a group of biopore-dominated flow columns.
P22, L18: performance of the Ahuja…model
P22, L22: A and B of the Ahuja…model
Figure 9 caption: performance of the Ahuja…model

Hide

ED: Publish subject to minor revisions (Editor review) (31 Aug 2016) by Alberto Guadagnini

AR by Muhammad Naveed on behalf of the Authors (15 Sep 2016) Author's response Manuscript

ED: Publish as is (17 Sep 2016) by Alberto Guadagnini

AR by Muhammad Naveed on behalf of the Authors (17 Sep 2016)

Short summary

Quantification of rapid flow of water and associated transport of contaminants through large soil pores generated by earthworms or decaying plant roots is of crucial importance for sustaining both soil and water quality. Advanced visualization and analysis techniques based on state-of-the-art X-ray computed tomography have been applied to 65 soil cores extracted from an agricultural field in Silstrup, Denmark, to improve models for the prediction of fast, preferential flow processes in soils.