A Stepwise Modelling Approach to Identifying Structural Features That Control Groundwater Flow in a Folded Carbonate Aquifer System
, Nicolas Guyennon 1, Anna Bruna Petrangeli 1, Emanuele Romano 1 and Cristina Di Salvo 2,*
William B. White
Round 1
Reviewer 1 Report
Thank for the opportunity to review you manuscript regarding a step-wise modeling approach to identify structural features controlling groundwater flow. I appreciate the approach you took and find merit to your work, however I have concerns regarding the conceptual model development and input information as well as the model results. The attached file provides numerous comments that need to be addressed.
Comments for author File: 
Comments.pdf
Author Response
Reviewer 1
Comments and Suggestions for Authors
Thank you for the opportunity to review you manuscript regarding a step-wise modeling approach to identify structural features controlling groundwater flow. I appreciate the approach you took and find merit to your work, however I have concerns regarding the conceptual model development and input information as well as the model results. The attached file provides numerous comments that need to be addressed.
ANSWER:Thanks for your very nice comments and for the time you spent for these in-depth review. Your observations are interesting and reveal a profound knowledge of this topic, we hope we will be able to address your comments appropriately.
LINE 11 carbonate
DONE
LINE 11 Mountains
ANSWER:Thanks for the suggestion; however, we think that Apennine Range (instead of chain) suits better to the Apennines. Same correction was done at Line 100..
LINE 12 move this phrase to after the word aquifer
DONE
LINE 25 add faults and folds
DONE
LINE 35 large
DONE
LINE 63 This, the generalities can be modeled, but not the specifics. As such it may not be suitable for deterministically modeling flow along faults or other local features
ANSWER We completely agree, thanks for this thought, we added this point in the Introduction. (SEE LINES 64-65)
LINE 91 simpler model
DONE
LINE 102 what about topography?
ANSWER: this question asks about the relations of the groundwater table and the topography. The area where the recharge is most effective is the high elevation plateaus of Monte Coscerno, Monte Aspra and other summits which are above 1000 m asl exceeding 1670 m asl at Monte Aspra and Monte Coscerno. We added a new figure 2 with the topography to show the orography of the area. However, there is no direct relation between the topography and the potentiometric surface.
LINE 104 layers?
YES, DONE
LINE 105 system
DONE
LINE 106 are
DONE
LINE 111 structural or topographic?
ANSWER: structural, according to the bedding attitudes. Thanks for noticing that it was not clear, we modified the text accordingly (LINES 118-120)
LINE 112 from where?
ANSWER: from east to west, we modified the text to avoid misunderstanding (LINE 120)
LINE 115 seems like the anticline is asymmetrical
ANSWER: Yes it is a NNW-SSE trending anticline belonging to the frontal thrust ramp of a regionally curved thrust known as the Mount Coscerno-Rivodutri thrust). It displays a box-shaped profile with a vertical-to-overturned forelimb. Additional explanation and relevant references were added in the text (SEE LINES 101-105)
LINE 119 (Figure 1) linear stream valleys to the NE and NW - do these follow faults?
ANSWER: yes, most of the linear streams in the study area are driven by tectonics (direct/inverse faults, thrusts, ramps…). We added faults, bedding attitudes and the axis of the anticline to show the structural control of the tectonics on the hydrological network.
LINE 120 (FIG 1) Using geologic base layers will better clarify the geologic setting. Geologic structure including faults, folds and bedding attitudes should be shown
ANSWER: Yes we agree, we modified Figure 1 to include the structural elements. Thanks for this suggestion.
LINE 129 installed
DONE
LINE 130 is incised into the
DONE
LINE 144 not sure I understand this. Is it seasonal or based on precipitation trends?
ANSWER: We refer to the seasonal variation of the discharge of the river, variation that is very low. We better specified in the text (LINE 154)
LINE 147-150 don't the springs represent zones of preferred flow? What is their origin?
ANSWER: this aquifer feeds the streambed springs for several kilometers along the Nera river. There is no concentration of flow towards springs, with the exception of the springs Lupa, Pacce and Peschiera. The location of the reaches where the stream is gaining could be related to tectonics in the sense that the Nera river flows parallel to a regional overthrust (Valnerina thrust) as shown in Figure 1 and therefore it represents the no-flow boundary of the aquifer at the lowest topographic elevation. Explanation added in the text (LINES 159-163)
LINE 153 (FIG.2) need more description of the layers in the top panel. the blue in second panel is the potentiometric surface, I presume? is it really that flat? That does not fit very well conceptually with diffuse flow, which implies some resistance to flow. What are the dashed horizontal lines? What is the depth of incision of the NE and NW-trending valleys - perhaps you could should that schematically
ANSWER: Explanation added in the caption referring to the legend in Figure 1.
The blue filling both in first and second panel represents the saturated area and the line represent the (hypothetic) potentiometric surface. In the figure there is no vertical exaggeration and a gradient of 3-5 permill as usual in carbonate aquifers is not visible at this scale. The dashed horizontal lines represent the continuation of the potentiometric surface of the lower and intermediate aquifer where they are confined.
To answer to the last question, some valleys are real canyons and reach an important depth. However, they are usually dry, showing that the potentiometric surface is below that elevation and it outcrops only along the Nera river. At lines 274-276 we explain that we used the topography elevation in the deep valleys as an additional bound for the potentiometric surface control.
The depth of the incision of the NE and NW-trending valleys is shown in the newly made Figure 2 (bottom panel). Some of these valleys are very deep and profoundly incised and they demonstrate the low gradient of the sub-aquifers.
LINE 163 i would think one would have to consider discharge as well. How do you avoid considering that?
ANSWER: In the hydrological balance we considered the traditional equation:
P = Q + AET + GW + DS
where P is precipitation, Q is runoff, AET is actual evapotranspiration, GW is recharge to the aquifer aquifer and DS is change in soil storage. Once soil moisture is filled (100 mm) the runoff (Q) +infiltration to groundwater (GW) processes start. We considered GW as a fraction (ranging from 5 to 90%) of water surplus (we amended the text, there was a mistake) and the remaining share (95% to 10%) is runoff. This calculation was performed at the daily time step on a grid of 1 km2.
Text was modified accordingly (LINES 182-189)
LINE 166 I think one needs some validation of this based on actual measurements
ANSWER: the infiltration coefficients have been estimated on the grounds of existing technical reports on the study area and on an expert judgment basis, assuming that limestone formations have higher potential infiltration coefficients than marl formations. The validation is based on the comparison of the calculated annual infiltration with the observed aquifer discharge. Text was modified accordingly (LINES 191-193)
LINE 210: spelling?
ANSWER: DONE, Preconjugated
LINE 216 are these defined?
ANSWER: we added the explicit name of the packages BCF and LPF
LINE 254 (Table 2) very limited number of data points for calibration!
ANSWER: You are right the head data are extremely limited. Our approach is based on the presence of a nearly continuous draining line along the river, representing a bond for the model, and the challenge was to correctly reproduce the outflow distribution. The very few number of data points for the head IS the real limitation of modeling in mountainous areas, because the orography of these areas, together with the richness of water along the stream all year round, discourages the development of well fields.
LINE 257 You have not mentioned these parameters. Is the Nera an alluvial or bedrock stream? What basis do you have for estimated input riverbed K values?
ANSWER: Nera river is a bedrock stream with a limited thickness of loose sediments, up to few meters of permeable gravels and sands. The valley is mostly very narrow and the bedrock is often outcropping very close to the river. We do not have any values for the riverbed Kv but we know that it consists mainly of alluvial gravel and sands with very high permeability. (see lines 302-309)
LINE 262 capital or another parameter?
ANSWER: capital, thanks. DONE
LINE 267 hmmm...sounds like you are imposing a high-K fault zone. Is there basis for this?
ANSWER: Yes, the reviewer is right, we are imposing a high-K fault zone. As explained further in the discussion (see LINES 362-363) this artifact was needed to “push” the groundwater to flow towards north and feed the springs in the upstream reaches of the river. We do not have any defined element for the existence of this high-k strip but it might be justified by a fractured zone along the axis of the anticline. We only have a portion of anticline axis observed in the surface, which iss mapped on the published geological maps for the northern part of the study area (SEE FIGURE 1 WHERE THE ANTICLINE AXIS HAS NOW BEEN REPORTED). However, the general direction of the structure is meridian, so we assume that the anticline axis is nearly meridian (we added additional information about geology and structure of the anticline at lines 101-105)
LINE 279 (Fig.5) where is high-K strip?
ANSWER: It is in the layer 5 in Fig.5 (now Fig.6). We assumed that the deepest layer gets a large share of the recharge from the rainfall infiltration in the southern and central part of the structure (the high altitude part of the study area, above 1000 m asl) and directs it towards the northern part, feeding the intermediate and the upper aquifers (layers 3 and 1). We added a label “High K strip” in the far-right top left of the figure
LINE 280 (Table 4) Again, indicating whether this is a bedrock or alluvial stream is important to know whether these are reasonable. the calibrated streambed thickness makes no sense for an alluvial stream. The Nera River has to flow directly on bedrock for this to make any sense. If that is the case the Kv for riverbed is practically meaningless
ANSWER : Thanks for this, we see your point. As explained before the Nera river has an alluvial bed of loose and highly permeable sediments of limited thickness. In the model the calculation of the exchange stream-aquifer requests the “conductance term” which is the ratio of the riverbed Kv values and the riverbed thickness, times the cell area (SEE LINES 302-303). What really matters is the ratio Kv/thickness. During calibration, in order to obtain the observed high outflow to the river, we needed to increase this ratio from 10-5 m/s to 10-4 m/s to obtain a minimal resistance to the groundwater flow from aquifer to the river. We obtain the same value with e.g. 3 meters of riverbed thickness and 10-3 Kv, which is actually more consistent with the high permeability of the gravels and the presence of a few meters of riverbed above the bedrock. We agree that the riverbed thickness we postulated is meaningless and we changed our model with the afore mentioned values that seem to have a more physical sense. Thanks for suggesting this, we missed this point! We inserted in the paper LINES 303-309 an explanation for this.
LINE 284 (Table 2) really too limited for the calibration to be meaningful
ANSWER: as explained before, our challenge was to reproduce the distribution of the outflow to the springs and to the river (Flux target in Table 2). The two available wells are actually too few to be meaningful, we agree with the reviewer. But this was the only information available for the potentiometric surface in addition to the elevation of the springs. In our opinion, at the regional scale the simulation of the outflow to the river makes sense also with few data on the potentiometric surface. As you suggest in your comment, the evaluation of the calculated potentiometric surface is very doubtful because there are no data. However, we are confident about the evaluation of the general aquifer discharge because the simulation reproduces the spatial distribution of the outflow fairly correctly. Perhaps the most interesting aspect is that to get to this calibration we had to insert an element that favored the outflow in the basal layer from south to north as explained in the Discussion.
LINE 325 326 is this the N-S line in the far right panel? If so, please label it.
ANSWER: yes it is, We added a label as requested in Fig.5 (now Fig.6)
LINE 327 I agree, but probably vastly oversimplified
ANSWER: We agree, this reflects a simplified view of the geological setting of the aquifer, we added a comment on this (SEE LINES 363-365)
LINE 347-350 I agree, it is a place to start, but I would like it to reflect the actual geology more. That is why I request that the actual distribution of faults, folds and bedding attitudes be represented on figure 1.
ANSWER: We agree with the reviewer. At lines 363-369 we explain the links between the model and the geological, stratigraphical and structural setting. As requested we added faults, folds and bedding attitudes in Fig.1 to facilitate the understanding of our hypothesis.
LINE 367 complexly folded and faulted
ANSWER: We replaced as suggested (LINES 404)
LINE 372 high?
ANSWER:Not necessarily high. In the first phase, the number of zones was increased, with variable k values. Then, in order to match observations for R1 and R2, high K zones were added.
Author Response File: Author Response.pdf
Reviewer 2 Report
Overall, an interesting study and well-presented manuscript except some minor typos. The authors had incorporated most of the relevant research in the field including some new ones. Yet, it is advised that authors expand their literature review (line 77) to include new approaches for groundwater prediction such as integrated modeling by Asadi, A. and Adhikari, A. 2021 (https://doi.org/10.3390/w14030426) and others.
Author Response
Reviewer 2
Comments and Suggestions for Authors
Overall, an interesting study and well-presented manuscript except some minor typos. The authors had incorporated most of the relevant research in the field including some new ones. Yet, it is advised that authors expand their literature review (line 77) to include new approaches for groundwater prediction such as integrated modeling by Asadi, A. and Adhikari, A. 2021 (https://doi.org/10.3390/w14030426) and others.
We wish to thank the reviewer for the positive evaluation. We expanded our literature review as recommended to include new approaches for groundwater prediction (SEE LINES 58-60).
Reviewer 3 Report
The authors have made a detailed evaluation of one of the methods for modeling karst aquifers, the equivalent porous medium model. It seems to have worked quite well and demonstrated that the more geology that can be put into the model, the better the results. The authors found no field evidence for karstic conduits which may be why this approach worked so well.
Modeling karst aquifers continues to be a challenge and it is helpful to have the various approaches tested.
Author Response
Thank you for appreciating our research.
Round 2
Reviewer 1 Report
Thank you for improving the figures and clarifying the recharge conceptualization and model results. I would still prefer to see additional model runs that investigate the role of the NW- and NE-trending red fault zones. I will leave it to the editors to decide if such work is needed. The manuscript reads much better, but does require some minor editorial improvement. Very nice work!
Comments for author File: Comments.pdf
Author Response
Reviewer 1 (2nd round)
Comments and Suggestions for Authors
Thank you for improving the figures and clarifying the recharge conceptualization and model results. I would still prefer to see additional model runs that investigate the role of the NW- and NE-trending red fault zones. I will leave it to the editors to decide if such work is needed. The manuscript reads much better, but does require some minor editorial improvement. Very nice work!
ANSWER
We are deeply grateful to the reviewer for this second review and for the positive judgement of our work. Please find in the following the answers to comments. All new amendments are with track-changes on.
LINE 118 leakage?
ANSWER: Thank you, It was a mistake, and we recognize that in this case, the correct term is leakage. We replaced “leakance” with “leakage”.
LINE 133 define strike and dip symbol. What is a direct fault? Do you mean vertical or normal or?
ANSWER: We added the symbols of strike and dip in the legend of Figure 1A. With the definition “direct fault” we mean a normal fault. According to the suggestion, we modified “direct” with “normal”.
LINE 378 ok, the high hydraulic conductivity strip roughly follows the hinge of the anticline and as such
would likely be highly fractured. I think it would be useful to do additional model runs with the red faults as high hydraulic conductivity zones as well. If the results do not match the observed heads then you can argue that the red faults are not significant in regard to structural control on recharge or discharge.
ANSWER: Thank you for the correct observation. We omitted to describe the model runs which were made to include faults as high hydraulic conductivity zones. As you might notice, in Fig.6, there are thin strips of relatively high K , in the sub-aquifer 2 and 3 (the black line following approximately one of the NW-SE faults). However, despite adding local zone or strips with high K, we were never able to drive enough water to the northern boundary of the aquifer. The only way to reach the best results matching observations was by adding the high k strip in layer 5. Of course, there might be other possibilities, bringing to similar results, but we found this one satisfactory and (remotely) inspired by the local structural setting. However, thanks for reminding us this other high K strip, we added a few lines commenting on this (LINES 363-366).
LINE 382 reach R2 not labeled in figure 6
We added the reaches from R1 to R4 in the figure, with labels. We also added the reference to the reaches in the caption.
LINE 415 S and R are not defined
ANSWER Thanks for noticing this. “S” and “R” represent the location of the two wells “Scheggino” and “Renari” (see Tab.2 and Tab.5). An explanation was added to the caption of figure 8. Also Table 5 was modified to explicit definitions.
Author Response File: Author Response.pdf
