Testing alternative uses of electromagnetic data to reduce the prediction  error of groundwater models

Christensen, Nikolaj Kruse; Christensen, Steen; Ferre, Ty Paul A.

doi:https://doi.org/10.5194/hess-20-1925-2016

Articles | Volume 20, issue 5

https://doi.org/10.5194/hess-20-1925-2016

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

https://doi.org/10.5194/hess-20-1925-2016

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

Articles | Volume 20, issue 5

Research article

|

13 May 2016

Research article |

| 13 May 2016

Testing alternative uses of electromagnetic data to reduce the prediction error of groundwater models

Nikolaj Kruse Christensen, Steen Christensen, and Ty Paul A. Ferre

Download

Final revised paper (published on 13 May 2016)
Preprint (discussion started on 24 Sep 2015)

Interactive discussion

Status: closed

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

- Printer-friendly version

- Supplement

RC C3932: 'Review', Anonymous Referee #1, 29 Sep 2015
- AC C5086: 'Reply to referee #1', Nikolaj Kruse Christensen, 25 Nov 2015
RC C4878: 'Review of Christensen et al., “A framework for testing the use of electric and electromagnetic data to reduce the prediction error of groundwater models.”', Anonymous Referee #2, 16 Nov 2015
- AC C5101: 'Reply to referee #2', Nikolaj Kruse Christensen, 25 Nov 2015

Peer-review completion

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

ED: Reconsider after major revisions (28 Nov 2015) by Mauro Giudici

AR by Nikolaj Kruse Christensen on behalf of the Authors (08 Jan 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (08 Jan 2016) by Mauro Giudici

RR by Anonymous Referee #1 (11 Jan 2016)

Suggestions for revision or reasons for rejection

As the old reviewer 1, I have now read the new version of the manuscript and the author responses to the two reviews. The new manuscript is a vast improvement and most sweeping statements of the previous version has been removed. It is overall a nice contribution that will be of interest to hydrogeophysicists and hydrogeologists around the world. I don’t agree with all the responses made by the reviewers in their response letter, but I find mostly that the new version of the manuscript is mature and convincing.

I would like to make a few more comments.

1. The response letter state that the HYTEB code is now accessible from a given website. I strongly suggest that this information is also made in the paper, for example, in the acknowledgements.

2. Note that data are plural, so it should not be “this” data, but “these” data. For example, page 2, line 3. There are a few cases like this (e.g., page 8, line 14).

3. Page 6, line 13: Either keep the sentence as it is, but remove “objectively”, or keep “objectively” and state that you can investigate the worth of geophysical data for a specific data integration/calibration/joint inversion procedure. The approach provide objective results for a given workflow, it does not provide objective insights about the use of geophysics in hydrogeology. For example, I would most likely address the problem very differently, but if I would use HYTEB, then I would get the same results as the authors.

4. Page 6, line 28: Consider to remove “electric data” as they are not discussed in the paper.

5. Page 9, first few lines. Clarify that if a positive or negative log-log relationship exist as a site it would be expected to have a significant scatter.

6. Section 2.6. I assume that you refer here to data that are not used in the calibration. It would be good to make this very clear. This was at least not clear in my reading. If it would be the data used in calibration, then there would be no need for a separate section as this comparison is made during calibration/inversion.

7. Page 12, line 13. You don’t make measurements of “electrical resistivity”. You use either inferred electrical resistivities or use data sensitive to electrical resistivity.

8. Page 14, lines 11 and 18. Please decide if you have 6 or 7 predictions of interest.

9. Page 15, line 9. I suggest replacing “system” with “property”.

10. Page 17, line 5 and 8. Your TEM data are primarily sensitive to porosity, so the sentence reads strangely to a geophysicist. Here, the authors have chosen to not account for comments made by reviewer 2. This is a bit unfortunate. It would be better to say that electrical resistivity is sensitive to porosity, but that this relationship is not incorporated within the present work.

11. I have some problems to understand the joint inversion. The third term of equation 4 is fitted by varying the hydraulic conductivity field that in turn leads to varying electrical resistivity using equation 1. This is a joint inversion with a perfectly known petrophysical relationship. My confusion arises with equation 5. It seems that Kmf is entirely linked to the model that you are estimating, so how can it stabilise the solution? I understand how this is done for the sequential inversion, in which the previously inverted geophysics is used to regularize the hydrological inversion (see equation 6). It seems to me that something is missing in the description here and I wouldn’t know how to implement this joint inversion using equation 4 and 5. The regularisation term needs to come from other sources of information or assumptions than those used in the data misfit function. This is indeed what is done in the so-called sequential inversion.

12. page 23, line 15: It should be “structural”.

Hide

RR by Anonymous Referee #3 (03 Feb 2016)

ED: Reconsider after major revisions (10 Feb 2016) by Mauro Giudici

AR by Nikolaj Kruse Christensen on behalf of the Authors (09 Mar 2016) Author's response Manuscript

ED: Referee Nomination & Report Request started (13 Mar 2016) by Mauro Giudici

RR by Anonymous Referee #1 (14 Mar 2016)

RR by Anonymous Referee #3 (17 Mar 2016)

Suggestions for revision or reasons for rejection

Dear authors,

I think that most of my comments are tackled in the revised version of the manuscript. I have a few minor comments related to this updated version:

1. P5L5-7. For clarity, I would switch the sentence as "the simulated response of one model (geophysical model) is used as input to constrain the other model (hydrologic model)" because it is generally the hydrologic model which is the objective (the one which is used for predictions).

2. P6L15-16. It is true that the specific scheme of Ruggeri et al. has not been extended to groundwater modelling, but other stochastic Bayesian schemes have. For example, the study of Hermans et al. (2015) uses multiple-point statistics and probability perturbation method, it integrates geophysical data to the groundwater model in a Bayesian way, with a link to hydrofacies (indirectly linked to hydraulic conductivity) in a field case.
I would also remove the sentence in parenthesis ("Such extension will probably...") as it is speculative. Other stochastic schemes have already been used, it could probably be the case for this one, too.
Hermans et al. 2015. Uncertainty in training image-based inversion of hydraulic head data constrained to ERT data: Workflow and case study. Water Resources Research, 51, 5332-5352.

3. P6LL20-21. "Such data set is rare for investigation of the systems and scale considered here". This sentence is a little-bit strong as it suggests that it could not be applied in practice. What is needed are co-located boreholes and EM soundings, which is not that rare if the geophysical survey is well-designed. Especially in Denmark where a huge database is available, some work is currently done to link resistivity to lithology (and therefore hydraulic conductivity). I thus expect that Bayesian method will be extended in the future and should not be presented as impractical.

4. P6L25. The paper by Irving and Singha does not have a small number of parameters (more than 500) but it is simplified in term of lithology.

5. P19L9. I think it should be equation 2 instead of 4.

6. P19L16. I would give here the reference to Herckenrath et al. (2013) so that it is clear which paper the reader should read to understand the method in details (for example, how the parameters are updated). I think some of the confusion arises from the fact that it is not specified here that resistivity is also considered as unkown in JHI and inverted together with K using Pest.

7. P21L3-7. Why in parenthesis? Note that it is not case-specific as some authors found improvement by using a filtering scheme. (reference to Beaujean et al., 2014).
Beaujean et al. 2014. Calibration of seawater intrusion models: Inverse parameter estimation using surface electrical resistivity tomography and borehole data, Water Resources Research, 50, 6828-6849.

8. P24L16-17. I would replace "may not be unique" by "is non-unique". This problem is well-known in geophysics for underdetermined, ill-posed problems. Changing the parameters in Pest doesn't make the minimization unique, it would just be a specific case (if it exists) for which the solutions tend to the same result. So I would remove the last sentence.

9. P26L2-4. This affirmation is somehow in contradiction with the fact that the filtering scheme did not improve the results.

10. P26L23-24. "... show wider horizontal scatter...". I would rephrase. As it is, the sentence sounds negative, whereas the wider scatter enables to encompass the 1:1 line to some extent, and is thus an improvement (at least, a few hydraulic conductivities are correctly estimated).

11. P27L5. Deepest instead of deeper?

12. P29L26. Do you mean column 5 of Figure 7, because there are only 5 columns, not 6.

13. The reference list should be updated with the new references.

Hide

ED: Publish subject to minor revisions (Editor review) (25 Mar 2016) by Mauro Giudici

AR by Nikolaj Kruse Christensen on behalf of the Authors (03 Apr 2016) Author's response Manuscript

ED: Publish subject to minor revisions (Editor review) (12 Apr 2016) by Mauro Giudici

AR by Nikolaj Kruse Christensen on behalf of the Authors (14 Apr 2016) Author's response Manuscript

ED: Publish as is (16 Apr 2016) by Mauro Giudici

AR by Nikolaj Kruse Christensen on behalf of the Authors (25 Apr 2016)

Short summary

Our primary objective in this study is to provide a virtual environment that allows users to determine the value of geophysical data and, furthermore, to investigate how best to use those data to develop groundwater models and to reduce their prediction errors. When this has been carried through for alternative data sampling, parameterization and inversion approaches, the best alternative can be chosen by comparison of prediction results between the alternatives.