Exploiting Earth Observations to Enable Groundwater Modeling in the Data-Sparse Region of Goulbi Maradi, Niger

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Dear Editor,

the paper deals with hydrogeological model of a transboundary porous aquifer between Niger and Nigeria. The manuscript shows detailed description of data and methods; the language is very good, it is easy to read. Some suggestions and considerations have been added throughout the text pdf.

Therefore, I suggest minor revisions before publication.

Sincerely

Comments for author File: Comments.pdf

Author Response

the paper deals with hydrogeological model of a transboundary porous aquifer between Niger and Nigeria. The manuscript shows detailed description of data and methods; the language is very good, it is easy to read. Some suggestions and considerations have been added throughout the text pdf.

Therefore, I suggest minor revisions before publication.

Thank you for this summary.

Please, insert a geological and hidrogeological map of the study area. (comment on page 2 of manuscript)

We have added a geological map as Figure 2. This is a modification of a map we had originally as Figure 4 to highlight boundary conditions. The boundary condition figure (now Figure 6) has been improved to more clearly highlight each boundary. Furthermore, we added a new figure illustrating the hydrogeology in the form of an idealized north-south cross section (Figure 3).

I suggest to insert the names of Niger and Nigeria along the red international border (comment on page 4 of manuscript)

Done.

revise the format of paragraph and the figure displacement throughout the text (comment on page 5 of manuscript)

We are not sure how this figure got inserted in the middle of the PDF as it was not this way in the MS Word document. For some reason it inserted the entire figure in the cross reference in the text. We believe we have fixed the issue that was causing this to occur (wrong style on the paragraph containing the figure).

not clear, please rephrase (comment on page 10 of manuscript)

We have reworded this paragraph and updated the legend on the figure to make it more clear.

 

Reviewer 2 Report

Comments and Suggestions for Authors

I have reviewed the paper and find it very interesting. 

The introduction has been well written. The literature review is detailed and very relevant to the study.

Data used for the research is sound, as well as the analysis conducted.

Are there works in Niger that have estimated recharge in similar geology? If so, how does your work compare with such research works?

Check first one under section 5.1; "Through" or "Though"

For a study of this nature, the geological map of the study area should have been added. This can aid in establishing a good relationship between the aquifer properties (ie hydraulic conductivity) and the underlying geology.

What recommendations would you suggest for future research?

Author Response

I have reviewed the paper and find it very interesting.

The introduction has been well written. The literature review is detailed and very relevant to the study.

Data used for the research is sound, as well as the analysis conducted.

Thank you for these comments.

Are there works in Niger that have estimated recharge in similar geology? If so, how does your work compare with such research works?

Yes, we have updated the text in section 4.2 to discuss comparisons with other studies in the region.

Check first one under section 5.1; "Through" or "Though"

We have corrected the word.

For a study of this nature, the geological map of the study area should have been added. This can aid in establishing a good relationship between the aquifer properties (ie hydraulic conductivity) and the underlying geology.

A geologic map and an idealized hydrogeolical cross section have been inserted on page 2 and 3 of the manuscript.

What recommendations would you suggest for future research?

We have incorporated future research directions into the conclusion of the paper. It is suggested that future studies could focus on iteratively refining the groundwater model and enhancing it as investments are made in the region. This could involve improvements in monitoring well development, data collection, and the exploration of additional methods for estimating recharge. We also added a note about using the model for stochastic simulations.

Reviewer 3 Report

Comments and Suggestions for Authors

The paper is well-written and tells a full story of groundwater model development and application. However, at its present stage it resembles a government technical report and lacks the key content that makes it a scientific journal paper.

 

Comment 1: Section 1.4 does not provide enough explanation on the scientific novelty of the study. Groundwater models have been developed and calibrated worldwide, groundwater modeling has been done in data-scarce areas, and combination of remote sensing and in-situ data has been applied and many past studies as well. It is suggested that the author narrow the objective down to the niche this study fits into. For example, how is the approach different from other groundwater modeling in the same area, or other combination of remote sensing and in-situ data in the same or different areas. Another example could be how the data scarcity in the study area compares to data scarcity in other studies, such that there is uniqueness in this study. Note that those are just examples that a reviewer can imagine; it would be the responsibility of the authors, who know the study best, to come up with the best niche for the study.

 

Comment 2: Given that the study emphasizes the groundwater modeling was done in a data-scarce area, it is reasonable to expect a stochastic groundwater modeling approach, and understanding uncertainty will be key. The study does not apply a stochastic approach and does not justify such a decision, and furthermore there is no quantitative evaluation of modeling uncertainties. Without these key components, it is challenging for the reader to understand how data scarcity and the approach of data assimilation affect the groundwater model, and how this study is different from a regular groundwater modeling practice.

 

Comment 3: The study does not provide enough explanation on how the addition of storage volume from GRACE affects the groundwater model. Based on sections 5.1.2 and 5.1.3, it appears that the model was fit against the head observation, and the GRACE is used merely as a source of confirmation. It appears that one would obtain the same groundwater model with or without GRACE data following the approach in the study. Given that the emphasis of the study is the combination of remote sensing and in-situ data, the study should demonstrate how the addition of GRACE improves the model. This could be done, for example, by comparing the model calibrated to head alone and the model calibrated to head plus GRACE. Alternatively, if a stochastic approach is taken one could show how GRACE data reduces the parameter uncertainty of the model calibrated to observed head. There could be other ways as well.

Author Response

The paper is well-written and tells a full story of groundwater model development and application. However, at its present stage it resembles a government technical report and lacks the key content that makes it a scientific journal paper.

We will respond to this in our responses to the comments below. For the sake of clarity, we will change the order of the comments.

Comment 1: Section 1.4 does not provide enough explanation on the scientific novelty of the study. Groundwater models have been developed and calibrated worldwide, groundwater modeling has been done in data-scarce areas, and combination of remote sensing and in-situ data has been applied and many past studies as well. It is suggested that the author narrow the objective down to the niche this study fits into. For example, how is the approach different from other groundwater modeling in the same area, or other combination of remote sensing and in-situ data in the same or different areas. Another example could be how the data scarcity in the study area compares to data scarcity in other studies, such that there is uniqueness in this study. Note that those are just examples that a reviewer can imagine; it would be the responsibility of the authors, who know the study best, to come up with the best niche for the study.

We have modified section 1.4 to more clearly indicate the novel contributions of this research. As mentioned in our modified text, we believe this differs from most groundwater modeling studies using GRACE in terms of how extensively the GRACE data was used not just for volume change calibration, but also for recharge estimation and to back-calculate estimates of groundwater withdrawals via pumping. This is the first model developed in West Africa using GRACE data and serves as an template for future studies in this data sparse region (validated via extensive Google Scholar searches).

Comment 3: The study does not provide enough explanation on how the addition of storage volume from GRACE affects the groundwater model. Based on sections 5.1.2 and 5.1.3, it appears that the model was fit against the head observation, and the GRACE is used merely as a source of confirmation. It appears that one would obtain the same groundwater model with or without GRACE data following the approach in the study. Given that the emphasis of the study is the combination of remote sensing and in-situ data, the study should demonstrate how the addition of GRACE improves the model. This could be done, for example, by comparing the model calibrated to head alone and the model calibrated to head plus GRACE. Alternatively, if a stochastic approach is taken one could show how GRACE data reduces the parameter uncertainty of the model calibrated to observed head. There could be other ways as well.

No, this is not how the GRACE data were used. We faced the following data limitations when building the model:

• Lack of reliable recharge data
• Lack of reliable pumping data or estimates
• Highly limited historical water level data

So our strategy was as follows: We used the water table fluctuation method on GRACE-derived groundwater storage anomalies to develop a time-varying upper and lower estimate of recharge in the region. We then went through an iterative calibration process where we adjusted the overall pumping rates over regions of the model until we matched the heads for the limited time range where we had water level data and then we matched volumetric changes over the entire range.

This process is described both in the Flow Budget section (Section 3.1.3) and the Model Development section (Section 4), the Calibration Results section (Section 5.1), and discussed again in the first paragraph of the Conclusions section (Section 6).

Comment 2: Given that the study emphasizes the groundwater modeling was done in a data-scarce area, it is reasonable to expect a stochastic groundwater modeling approach, and understanding uncertainty will be key. The study does not apply a stochastic approach and does not justify such a decision, and furthermore there is no quantitative evaluation of modeling uncertainties. Without these key components, it is challenging for the reader to understand how data scarcity and the approach of data assimilation affect the groundwater model, and how this study is different from a regular groundwater modeling practice.

The objective of this research study funded by NASA was to build and calibrate a groundwater model in Southern Niger and use it to compare model response to recharge assumptions. As described above, the unique contributions of this work are the manner in which we built the model in a data sparse region and the fact that it is the first project of this kind in West Africa.

The immediate priority in using the model was to evaluate how much additional groundwater could be extracted in the future while still maintaining sustainable conditions based on a series of assumptions about groundwater recharge. Now that the model is constructed, it could be iteratively improved as additional data becomes available in the future.

We agree that stochastic analysis can be useful if data are uncertain or sparse. However, the objective of this research was to use satellite data to build a model in a data scarce region, then use that model to estimate recharge. Performing an in-depth analysis of potential management options was not a goal of this paper and is not in the scope of this paper.  That said, evaluating the three different scenarios with included multiple recharge and pumping assumptions is a stochastic analysis. Rather than presenting these as statistics, we presented each set of parameters in detail. A more complete stochastic analysis is one of several potential applications of the model now that it is built and calibrated.  However, as our goal was to identify the expected withdrawal rates given different recharge assumptions, a full stochastic analysis is not required. Stochastic analyses are useful depending on the application.

Round 2

Reviewer 3 Report

Comments and Suggestions for Authors

The authors addressed the comments well. 

I could not agree with the authors' point of view on seeing stochastic analyses and uncertainties as a "potential application", rather than an integral part of model development in data-scarce situations.

However, the authors addressed this comment by explaining the limited scope of the study, which is acceptable. Given that, I would suggest a minor revision to include the explanation provided in the authors' response, perhaps by including a new paragraph in section 1.4 or wherever the authors see fit.

Author Response

The authors addressed the comments well.

Thank you for your help making this a better manuscript

I could not agree with the authors' point of view on seeing stochastic analyses and uncertainties as a "potential application", rather than an integral part of model development in data-scarce situations.

We understand. It was not a “stochastic” analysis, but we did vary the important parameters over a reasonable range. Rather than randomly choosing values, we choose select values. We do not claim to have performed a stochastic analysis in the manuscript.

However, the authors addressed this comment by explaining the limited scope of the study, which is acceptable. Given that, I would suggest a minor revision to include the explanation provided in the authors' response, perhaps by including a new paragraph in section 1.4 or wherever the authors see fit.

We added a modified version of the stochastic discussion as a paragraph at the end of Section 1.4 to address this comment.

Author Response File: Author Response.pdf