Modeling the Impact of Groundwater Pumping on Karst Geotechnical Risks in Sete Lagoas (MG), Brazil

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Review of the paper:

Modeling the Impact of Groundwater Pumping on Karst Geotechnical Risks in Sete Lagoas (MG), Brazil 

The papers deal with groundwater numerical modeling of a catchment in Sete Lagos and the connected risk due to groundwater exploitation. The paper is divided into two parts. In the first part, the Authors set up a groundwater flow model in the FEFLOW environment to model the water table from 1940 to 2100. In the second one, the Authors evaluate the karst geotechnical risk connected to pumping actions. The work provides several insights, particularly regarding the GW model, like the hydrogeological setting, the information about recharge, and the model calibration and its application to the evaluation to the subsidence risk.

The paper employs a very well-defined geological framework and implements it within FEFLOW. I’m a groundwater hydrologist so I can give a contribution to my field of expertise, and not about geotechnical risk. Thus, I think that substantial adjustments would be enough to make the paper suitable for publication. Please follow the suggestions below. There are some works, at least two (see references suggested below) that can help the Authors in fixing several points of the manuscript and can serve as benchmarks for presenting their results more suitably, although these works are related to the groundwater modeling of the Venetian Plain Aquifers (Northern Italy). Bets regards

Major points

-        The Methodology Section is fairly terse and needs to be improved. There are no equations, neither for the GW flow model nor for the risk assessment part of the Methodology. Please provide both, with the former consisting of a brief description of which equations FEFLOW solves.

-        The same must be done for GW recharge, here clarifying if you consider also the unsaturated part of the domain (in this case, present the van Genuchten equation), otherwise by specifying that the GW recharge is directly injected into the water table.

-        How are the drawdowns for the yearly reconstructed water table calculated? Specify in the Methodology.

-        Row 270-272: this statement should be supported by a proper reference, it cannot be always taken as valid.

-        How did you detect the bedrock surface? Please explain adequately, eventually referring to Gatto et al. (2023), where the role of the bedrock formation in modeling GW in the FEFLOW environment is widely discussed. See e.g. Gatto et al. (2023)’s Figure 10.

-        I would modify Figure 3, for example, the second panel, by highlighting the boundary condition type along the domain border upon different colors. See e.g. Gatto et al. (2024)’s Figure 4. The equations for different types of boundary conditions should be written as well.

-        It is not clear how you divided your dataset into simulation and validation wells. Can you specify it? I suggest highlighting these two populations in Figure 4 (first upper left panel). Otherwise, if you employ all the wells and you are not able to retrieve zero residuals, we have a problem. It is suggested to take e.g. 20% of the well for ex-post validation of the numerical model, as done in Schiavo (2024), Figure 1.

-        Is it possible to relate the drawdown magnitude with the local geology types of the study area? This would be interesting to assess if this issue is related to the local hydrogeological setting, hence that is possible to identify it in similar geological areas. Eventually, refer to Schiavo (2024).

-        The discretization mesh is not offered to the reader, but it is important to assess the spatial discretization and the areas where mesh refinement is high. Please fix this, for example seeing the work of Gatto et al. (2023), Figure 3.

-        Figure 4 A reports the calibration for all the yearly information. I would like to be informed about each yearly R correlation coefficient between the observed and modeled water table. You can provide a table for this, or a boxplot for assessing its historical variability.

-        The manuscript is very hard to read when you are listing the parameters and their values within the lines. Please provide a Table listing all of them. The same must be done for equations, not in the text, but highlighted and numbered each. If all of these parameters are reported in Figure 4 (but a Table would be a more proper choice), please cite the location/Figure/Table and remove them from the manuscript’s lines.

-        The future trends of GW levels are not provided. This goal seems to me a little bit ambitious, also because of the aforementioned working points of the manuscript. I would suggest removing the ‘future projection’ part of the manuscript to focus on the validation of the results and to shed light on the relationship between withdrawals, subsidence, and local geological features. You can better focus your attention on future projections in another work, where trends can be clearly presented. The general impression that the paper gave me is a sort of hurry in presenting the Methodology and the results and their compression in a few paragraphs, while they might be presented and visually offered in a more relaxed way.

Minor points

-        The upper right panel in Figure 1 seems to be unnecessary. I’d suggest moving it, as well as the enclosed photo on the left, to a supplementary information file, and let the reader be able to focus on the main panel.

-        In Figure 4 c, an error of 10 m is not negligible. Hence, specify at least another level in the scale for quantifying residual heads, e.g. 0-2 m, 2-10 m, and so on.

-        Figure 5: It seems important to me to distinguish areas that in the future are susceptible to different GW patterns, such as switching from drainage to recharge behavior. I’d underline these changes for each yearly map, for example, offering in dashed or solid filling colors the water table variations. This would help to describe GW circulation patterns, which are hard to understand only upon flow arrows as in Figure 5.

-        Figure 5: I’d put water table contours in different line styles or colors, with bold ones for hundreds.

-        I’d add another Figure, mirroring Figure 5, depicting only water table variation concerning the estimated water table at present (2020 of yours). This way, the colored palette would show increments or decrements from year to year. I would do this by taking Schiavo's (2024) Figure 3 as an example.

From my side, these points must be addressed, and then the paper is solid and can be published. Best regards

References:

Gatto, B., Furlanetto, Camporese, M., Trentin, T., Salandin, P., 2023. Quantifying groundwater recharge in the Venetian high plain between the Brenta and Piave Rivers through integrated surface–subsurface hydrological modeling. J. Hydrol.: Reg. Stud. 50 (2023), 101550 https://doi.org/10.1016/j.ejrh.2023.101550

Schiavo, M., 2024. Spatial modeling of the water table and its historical variations in Northeastern Italy via a geostatistical approach. Groundwater for Sustainable Development, https://doi.org/10.1016/j.gsd.2024.101186

Author Response

Dear Reviewer 1,

The authors would like to thank you for the considerations and suggestions brought for the improvement of the submitted manuscript. After discussing them together, we present the new version of the manuscript with red indications showing where changes were made compared to the initial manuscript. We also highlight that the reference citation format is now correct, with reference numbers in square brackets, as requested. Please find attached the Reviewer 1 reply file in PDF.

 

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

The manuscript by Galvão et al, is a very interesting and important groundwater modelling study addressing geologic and hydrologic risks from over pumping in a karstic environment in Brazil.

 

I have only few suggested minor revisions for this very interesting study:

 

L82: “This central area is over a graben filled with karstified limestone rocks covered by 82nconsolidated sediments, hosting two main solutionally enlarged bedding planes and a large groundwater reservoir.” This sentence should be rephrased. It is not clear what is meant by : “This central area is over a graben…”

 

L 206: “the limestone matrix and epikarst were compiled from other correlated karst aquifers” It is not clear what the authors mean by “other correlated karst aquifers”. I suggest rephrasing or to reword the sentence.

 

L228: “However, from the 1990s, there is a progressive and consistent

alteration the equipotential curves within the central urban area attributed to elevated

pumping rates.” This sentence requires rephrasing.

 

Figure 2: Profile A to A’: At a distance of around 15000 m there is a very sharp transition for a depth of around 30 m from Karst to Basement rock. Geologically the way the units are present could only be explained by a fault? However, there is no fault shown in the figure.

 

L277: “To ensure the reliability of the numerical model in terms if discharge calibration,

values measured in situ from the conceptual model were used to validate the calibration” This sentence needs to be rephrased.

 

Hydrological risk factors: It would be interesting for the reader to explain why that specific level of <720 m asl is high risk. Why is set at 720 masl?

 

L309: “were evaluated, expecting difference” I suggest rephrasing, if it was evaluated, the word “expecting” should not be there.

 

Future Scenarios: Two future scenarios were investigated. Why was there no future scenario investigated in which the pumping increased, because the water demand and population is increasing ? Some more background could be useful here. Are alternative water sources being considered in the future?

 

L385: “about lithological and televiewer well logs is available.” This should be “well logs are available”

 

L 392: When normalized to the highest hydraulic gradient of 195 m, NRMSE results are between 2% and 12%, being effective for regional-scale models (Anderson et 393 al., 2015) (Fig. 4D).” I suggest to reword from “being effective” to “being acceptable”.

 

L398: “east limits does not” Should be “..do not..”

Comments on the Quality of English Language

The manuscript by Galvão et al, is a very interesting and important groundwater modelling study addressing geologic and hydrologic risks from over pumping in a karstic environment in Brazil.

 

I have only few suggested minor revisions for this very interesting study:

 

L82: “This central area is over a graben filled with karstified limestone rocks covered by 82nconsolidated sediments, hosting two main solutionally enlarged bedding planes and a large groundwater reservoir.” This sentence should be rephrased. It is not clear what is meant by : “This central area is over a graben…”

 

L 206: “the limestone matrix and epikarst were compiled from other correlated karst aquifers” It is not clear what the authors mean by “other correlated karst aquifers”. I suggest rephrasing or to reword the sentence.

 

L228: “However, from the 1990s, there is a progressive and consistent

alteration the equipotential curves within the central urban area attributed to elevated

pumping rates.” This sentence requires rephrasing.

 

Figure 2: Profile A to A’: At a distance of around 15000 m there is a very sharp transition for a depth of around 30 m from Karst to Basement rock. Geologically the way the units are present could only be explained by a fault? However there is no fault shown in the figure.

 

L277: “To ensure the reliability of the numerical model in terms if discharge calibration,

values measured in situ from the conceptual model were used to validate the calibration” This sentence needs to be rephrased.

 

Hydrological risk factors: It would be interesting for the reader to explain why that specific level of <720 m asl is high risk. Why is set at 720 masl?

 

L309: “were evaluated, expecting difference” I suggest to rephrase, if it was evaluated, the word “expecting” should not be there.

 

Future Scenarios: Two future scenarios were investigated. Why was there no future scenario investigated in which the pumping increased, because the water demand and population is increasing ? Some more background could be useful here. Are alternative water sources being considered in the future?

 

L385: “about lithological and televiewer well logs is available.” This should be “well logs are available”

 

L 392: When normalized to the highest hydraulic gradient of 195 m, NRMSE results are between 2% and 12%, being effective for regional-scale models (Anderson et 393 al., 2015) (Fig. 4D).” I suggest to reword from “being effective” to “being acceptable”.

 

L398: “east limits does not” Should be “..do not..”

Author Response

Dear Reviewer 2,

The authors would like to thank you for the considerations and suggestions brought for the improvement of the submitted manuscript. After discussing them together, we present the new version of the manuscript with red indications showing where changes were made compared to the initial manuscript. We also highlight that the reference citation format is now correct, with reference numbers in square brackets, as requested. Please find attached the Reviewer 2 reply file in PDF.

 

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Authors,

 

thank you for your cover letter. I think there is more room for improving the paper in the direction of its novelty. I'm itemizing below some further revision points, some of which are the same as the previous revision, please try to see if a revision is feasible. Bests

 

- I'm sticking to the point that GW depletions as differences in respect to 1940 ones (piezometric map) should be highly informative about which area experienced the highest change in the water table at a very low cost, by adding another Figure. If you think the information gain is not that much, you can put that as a supplementary information Figure.

- boundary conditions should be highlighted more, as done in Gatto et al. (2023).

- the GW model is deterministic for each year analyzed. How to deal with sources of uncertainty coming from different geological heterogeneous setups? Do you think that your framework in FEFLOW may benefit of a stochastic way of resolving the numerical model, as done in Schiavo (2024)? Please comment properly, this shift towards stochastic GW modeling would be scientifically sounding and quite novel. It needs at least to be touched in the Introduction and the Discussion part.

- what may be the impact of stochastic FEFLOW modeling on the geotechnical risk assessment? Can you comment on this?

 

Suggested references:

- Gatto, B., Furlanetto, Camporese, M., Trentin, T., Salandin, P., 2023. Quantifying groundwater recharge in the Venetian high plain between the Brenta and Piave Rivers through integrated surface–subsurface hydrological modeling. J. Hydrol.: Reg. Stud. 50 (2023), 101550 https://doi.org/10.1016/j.ejrh.2023.101550

- Schiavo, M. (2024). Numerical impact of variable volumes of Monte Carlo simulations of heterogeneous conductivity fields in groundwater flow models. J. Hydrol. (634), 131072, https://doi.org/10.1016/j.jhydrol.2024.131072

Author Response

Dear Reviewer 1,

 

The authors would like to thank again for the considerations and suggestions in this round 2. We present the second version of the manuscript with red indications showing where the new changes were made compared to the last version:

 

Major points

1. I'm sticking to the point that GW depletions as differences in respect to 1940 ones (piezometric map) should be highly informative about which area experienced the highest change in the water table at a very low cost, by adding another Figure. If you think the information gain is not that much, you can put that as a supplementary information Figure.

R: The authors thank the suggestions. A new figure was made and added to the manuscript with information of past evolutions of water table variation until 2020 based on the first potentiometric surface from 1940 and future scenarios up to 2100 comparing to 2020.

 

2. Boundary conditions should be highlighted more, as done in Gatto et al. (2023).

R: The authors agree, and complements were added, and Figure 2 was also updated for additional information.

 

3. The GW model is deterministic for each year analyzed. How to deal with sources of uncertainty coming from different geological heterogeneous setups? Do you think that your framework in FEFLOW may benefit of a stochastic way of resolving the numerical model, as done in Schiavo (2024)? Please comment properly, this shift towards stochastic GW modeling would be scientifically sounding and quite novel. It needs at least to be touched in the Introduction and the Discussion part.
4. What may be the impact of stochastic FEFLOW modeling on the geotechnical risk assessment? Can you comment on this?

R: The authors agree, and complements were added in the introduction as required. Also, a new subitem (5.4. Limitations and considerations) was added at the end of the discussion.

 

5. Suggested references:

- Gatto, B., Furlanetto, Camporese, M., Trentin, T., Salandin, P., 2023. Quantifying groundwater recharge in the Venetian high plain between the Brenta and Piave Rivers through integrated surface–subsurface hydrological modeling. J. Hydrol.: Reg. Stud. 50 (2023), 101550 https://doi.org/10.1016/j.ejrh.2023.101550

 

- Schiavo, M. (2024). Numerical impact of variable volumes of Monte Carlo simulations of heterogeneous conductivity fields in groundwater flow models. J. Hydrol. (634), 131072, https://doi.org/10.1016/j.jhydrol.2024.131072.

R: The authors agree, and both references have been included in the manuscript. We took care to reorder all citation numbers throughout the manuscript.