Electromagnetic Induction Is a Fast and Non-Destructive Approach to Estimate the Influence of Subsurface Heterogeneity on Forest Canopy Structure

Round 1

Reviewer 1 Report

Dear Authors, 

Please find my comments in the attached documents.

Best,

Comments for author File: Comments.pdf

Author Response

# Reviewer 1:

 The study by Carriere et al. illustrates the use of geophysical methods to image forest soils. Although not widespread in this context, the advantages and potential of ERT and EM methods over traditional methods for soil characterization are clearly demonstrated. The complex interplay between soil-water and vegetation is also investigated thanks to remote measurements. Overall, the manuscript is well written and reads well for a large community including geophysicists. The methodologies used are well described and their combination relatively novel. The scientific question raised i.e. how much can we use correlation between PAI to estimate soil texture or the other way around can we use EM/ERT to estimate plant behavior is of major interest.

Figures are good quality (see some possible improvements in the comments below).

My main concern is about the processing and interpretation of the data. While the correlation between EM and PAI seems obvious, in the current state I’m not convinced that:

• ERT fully support the EM interpretation. The statistical analysis is not convincing, authors should either moderate their conclusions or redo data processing i.e. at least extract 2d transects from EM maps (at the maximum sensitivity depth) at the position of the ERT profiles and compare it quantitatively with ERT conductivity values. At best, I suggest to add an additional processing step including a calibration of EM data using ERT (see Lavoué et al 2010). Ideally authors should also exploit the ratio between vertical and horizontal dipole orientation when both measurements are available (to this end the SI1 figure seems really informative but is not fully exploited)***.

 

We understand your comment. We had discussed these issues among us when preparing the article. However, it was decided to present the results in a homogeneous way between the two sites. Indeed, it would have been relevant to relate ERT with PAI or ERT and EMI, however was not done for two reasons :

• The limited number of common points between the PAI and ERT measures at Font-blanche (9 points). This limited number of points does not permit a robust statistical analysis.
• The complexity of relating the EMI conductivities to the ERT resistivities. The two techniques do not provide the same absolute values of resistivity (or conductivity) because the two geophysical techniques are based on different physical principles. This raises a major geophysical challenge that we do not aim to answer in this article. This is particularly true because we are working in a resistant medium (karst) in LIN mode. Many authors working in comparable geological contexts also choose to not qualitatively compare apparent conductivity values obtained by EMI and apparent resistivity values obtained by ERT (Al-Fares 2011; Khaldaoui et al. 2020 ; Pavoni et al. 2021 ; Valois et al. 2010). Because ERT is better known , we use ERT as a benchmark for the reader.

 

References :

Al-Fares, W. Contribution of the Geophysical Methods in Characterizing the Water Leakage in Afamia B Dam, Syria. Journal of Applied Geophysics 2011, 75, 464–471,.

Khaldaoui, F.; Djediat, Y.; Baker, H.; Ydri, A.; Djeddi, M.; Hamadou, K.; Bouzar, A. Use of Electrical Resistivity Tomography (ERT) and Electromagnetic Induction (EMI) Methods to Characterize Karst Hazards in North-Eastern of Algeria. Arabian Journal of Geosciences 2020, 13, 1–9.

Pavoni, M.; Sirch, F.; Boaga, J. Electrical and Electromagnetic Geophysical Prospecting for the Monitoring of Rock Glaciers in the Dolomites, Northeast Italy. Sensors 2021, 21, 1294.

Valois, R.; Bermejo, L.; Guerin, R.; Hinguant, S.; Pigeaud, R.; Rodet, J. Karstic Morphologies Identified with Geophysics around Saulges Caves (Mayenne, France). Archaeological Prospection 2010, 17, 151–160.

 

• the EM variability is explained by the soil depth variations. Considering that the soil depth estimates ranges from 0 to 60cm, and that the EM31 sensitivity in horizontal model to this layer is low.

The soil depth estimated by the soil pits and with the mining bar range between 0 to 60cm. Indeed, the EM31 sensitivity, even in horizontal mode, is not optimal to characterize a 60cm layer. This is why the EMI vs. mining bar comparison is only presented in the supplementary information because it is debatable. However, the development of tree root zones can extend beyond 60cm into the weathered rock zone (down to 3 to 5m). Some rare roots even go further 5m but the volumes of soil and roots involved are probably small and less detectable by surface geophysics.

 

We added a sentence to the new lines 172-173 to clarify this.

“These investigation depths are consistent with the expected development of root zones for the three species.”

 

Below some minor comments:

- In all the manuscript: replace (for the EM method) “electrical conductivity” by “apparent electrical conductivity (Eca)”

 

We agree and changed “electrical conductivity” by “apparent electrical conductivity” or “Eca” in all manuscript (new lines 167, 174, 186, 188, 263, 267, 276, 299, 314, 322, 337, 340, 554)

 

- Not sure I fully understand the notion of weathered rock? As it is mentioned several times in the manuscript can you define it?

You are right we clarified this notion at new lines 109-111 :

“We assume that the water resource available to plants is contained in the soil and the weathered rock (fracture, fault, conduits) also called epikarst”.

 

- Title: “a fast and convenient” sounds more like an observation. Would suggest to rephrase to something more neutral. Moreover, does the title really reflect the content considering all the methodologies deployed?

 

We rephrase the title as follow (new line 1)

“Electromagnetic induction is a fast and non-destructive approach to estimate the influence of underground heterogeneity on forest canopy structure”

Indeed, we could talk about ERT in the title, but we want to keep this article quite concise and oriented towards EMI because ERT has already been used several times in forest ecology studies (e.g. Robinson et al. 2012; Carrière et al. 2020; Homolák et al. 2020). ERT just reinforces the geophysical interpretation and will allow the non-specialist reader unfamiliar with EMI to cling onto a more familiar technique (which is ERT).

 

References :

 

Carrière, S.D.; Ruffault, J.; Pimont, F.; Doussan, C.; Simioni, G.; Chalikakis, K.; Limousin, J.-M.; Scotti, I.; Courdier, F.; Cakpo, C.-B. Impact of Local Soil and Subsoil Conditions on Inter-Individual Variations in Tree Responses to Drought: Insights from Electrical Resistivity Tomography. Science of the Total Environment 2020, 698, 134247.

 

Homolák, M.; Gömöryová, E.; Pichler, V. Can Soil Electrical Resistivity Measurements Aid the Identification of Forest Areas Prone to Windthrow Disturbance? Forests 2020, 11.

 

Robinson, J.L.; Slater, L.D.; Schäfer, K.V.R. Evidence for Spatial Variability in Hydraulic Redistribution within an Oak–Pine Forest from Resistivity Imaging. Journal of Hydrology 2012, 430–431, 69–79, doi:10.1016/j.jhydrol.2012.02.002.

 

 

Abstract

- L.17: Brand? (copyright) Geonics? (EM31)

We agreed and we added “Geonics” at the new line 19.

- L.22: Could you define the geological concept of soil/subsoil ? Subsoil = here bedrock/limestone ?

You are right.

First, we deleted “soil/subsoil” concept from abstract because it is too long define it within the abstract. We replace “soil/subsoil” by “ground properties” at the new line 24.

Second, we detailed the concept of soil/subsoil in the introduction at new lines 54-58.

“In this article we define soil in the pedological sense (soil horizons above the weathering rock) and the subsoil represents what lies beneath the soil, i.e. the intact and weathered rock (fractures, faults, contuits). We use in this article the notion of soil/subsoil because geophysical measurements, as implemented in this work, do not allow to clearly distinguish the two entities.”

 

 

Introduction

- L.86: an emerging trend in connecting soil and plant with geophysics is occurring in the field of agrogeophysics. As such I recommend to mentioned it and add references.

Garré, S., Hyndman, D., Mary, B., and Werban, U. (2021). Geophysics conquering new territories: The rise of “agrogeophysics.” Vadose zone j. 20. doi:10.1002/vzj2.20115.

Thank you for this reference update. We added the reference at new line 83.

“Garré et al [23] define the term agrogeophysics.”

 

- L.94: add references to recent studies:

- Blanchy, G., Watts, C. W., Ashton, R. W., Webster, C. P., Hawkesford, M. J., Whalley, W. R., et al. (2020). Accounting for heterogeneity in the θ–σ relationship: Application to wheat phenotyping using EMI. Vadose zone j. 19. doi:10.1002/vzj2.20037.

- Pavoni, M., Sirch, F., and Boaga, J. (2021). Electrical and Electromagnetic Geophysical Prospecting for the Monitoring of Rock Glaciers in the Dolomites, Northeast Italy. Sensors 21, 1294. doi:10.3390/s21041294.

Thank you for these recent reference. We added a sentence (new lines 82-83) to cite Blanchy et al. (2020) :

“In agronomy and soil sciences, geophysics has entered the list of commonly used tools [20–22].”

Pavoni et al. (2021), was now cited at new line 94.

 

2.1.2: Electromagnetic induction

- L. 147 to 149: Soil pits excavated along the profile revealed soil depths ranging from 0 to 60 cm with a rock fraction ranging between 50 % in the near surface and 90 % below 50 cm. Can you describe more in details the position of the soil pits.

 We agree and we give more information about soil pit for both sites (new lines 127-128 and 148-149)

 

“Seven soil pits distributed over the experimental site to highlight the spatial heterogeneity of the site revealed a stony rendzina (calcisol) whose thickness ranged between 15 and 70 cm according to pedologic pit exploration.”

 

“Four soil pits distributed over the experimental site to highlight the spatial heterogeneity of the site revealed soil depths ranging from 0 to 60 cm with a rock fraction ranging between 50 % in the near surface and 90 % below 50 cm.“

 

2.2.1: Electromagnetic induction

L.169: “The investigation depth varies slightly according to the underground electrical conductivity” ??

add references about the relationship between depth-specific sigma and measured sigma, for a given coil orientation i.e:

- McNeill JD (1992) Rapid, accurate mapping of soil salinity by electromagnetic ground conductivity meters. In: Advances in measurement of soil physical properties: bringing theory into practice. ASA, CSSA, SSSA, Madison, WI, USA, pp 209–229

- McNeill, J. D. (1980). Electromagnetic terrain conductivity measurement at low induction numbers. Mississauga, ON, Canada: GeonicsLimited

Right, we added both references at the new line 171.

Add the operating frequency of the instrument I.e. 9.8 kHz.

We agreed and we add this information at new lines 170.

You should say some word about the Low Induction Number assumption in a Karstic environement. The prevailing Low induction number LIN condition may require operation in environments restricted to very low (<12 mS/m) conductivities.

You are right, we added a sentence about LIN condition (new lines 170-171).

“The field conditions (low electrical conductivity) make the device used in low induction number (LIN) conditions [35,36].”

 

About the data acquisition.

You should say some words about the data processing:

• Interpolation/smoothing type between points, number of extreme values removed.

You are right. We added the two following sentences at the new lines 185-188.

“We removed 57 points out of 256 due to these metal objects which disturb EMI signal. The ECa maps shown in Figures 2 and 3 were calculated with Surfer software (Golden Software®) using the kriging method.”

• Justify why inversion has not been used.

Why you did not use both Vertical and Horizontal dipole orientation to refine the interpretation?

Combining both provides data estimating the conductivity and thickness.

We did not perform an inversion because the acquisition was performed only in horizontal mode at Valliguières. We wanted to maintain a homogeneous presentation of the results between the two sites.

 

• Instrument calibration, correction of temperature effect on Eca

We agreed and we give more information at new lines 188-190.

“We did not apply any temperature correction on ECa because the temperature was quite stable during the acquisition. Moreover, we checked that there was no device drift by returning to the starting point at the end of each survey”.

 

2.2.2: Electrical resistivity tomography

L.191: 2 years later then … does it change something in absolute?

You are right. We discussed this point at new lines 201-205.

“The ERT and EMI acquisitions were not performed in the same year for logistical reasons, but we assume that this time lag is not limiting for a relative comparison of geophysical properties between the two techniques. Indeed, in these forest contexts where the topography is very flat, most of the variations in geophysical properties are related to seasonal variations (i.e. temperature and water content)”.

 

L.194: “using three different complementary protocols including Wenner-Schlumberger, Dipole-

Dipole, and Gradient array [31]”. Did you merge them before inversion?

Yes, we give this information at new line 208.

“The three protocols were merged before inversion”.

 

 

L.196: I guess here you wanted to say that the mesh resolution adopted was … instead of the ERT 4 model resolution. If so please rephrase.

You are right. We rephrased this sentence (new line 214).

“The mesh was refined and the resulting ERT models have a 1 m lateral resolution and a vertical resolution that ranges gradually from 0.4 m near the surface to 2.5 m at a depth of 20 m.”

 

Same as for 2.2.1. Say a word about data filtering (reciprocal analysis,...), number of excluded data, number of iterations to reach the convergence criteria, etc...

We agreed and we give more information at new lines 208-214.

“We only kept data with a repetition error of less than 1% for processing. Up to ten additional points could be removed per section using Res2Dinv's "exterminate bad data points" function which identifies incoherent resistivity values[40]. Apparent resistivities were processed using the Res2Dinv. The models presented below were all obtained after 3 iterations. This limited number of iterations is a good compromise that allows to reach acceptable RMS without generating extreme resistivity values to reach a mathematical optimum that would risk to move away from the field reality.”

 

2.3 Ecophysiological measurements

L237: maybe relocate these references to the introduction

Throughout the review, the line numbers mentioned by the reviewer 1 do not match any numbering on the documents we have access to. Here, we do not understand what reference the reviewer is talking about. All the references cited in the "Ecophysiological measurements" section are very specific to hemispheric photo processing.

 

3.1 Font-Blanche: a naturally regenerated forest

For both 3.1 and 3.2 the comparison between EM and ERT is not straightforward. You should consider to express all the methods with the same units (see+) and extract 2d transect at the depth/position. I understand than some ERT lines were not covered by EM (M30, PM30, P30) because of iron structures but what about the others ?

- +Blanchy, G., Watts, C. W., Ashton, R. W., Webster, C. P., Hawkesford, M. J., Whalley, W. R., et al. (2020). Accounting for heterogeneity in the θ–σ relationship: Application to wheat phenotyping using EMI. Vadose zone j. 19. doi:10.1002/vzj2.20037.

We understand this very pertinent remark. We would also like to relate the values of ERT and EMI. However, this raises a major geophysical challenge that we do not aim to answer in this article. Indeed, the two geophysical techniques are based on different physical principles. They do not provide the same absolute values of resistivity. We can therefore appreciate only the relative variations of resistivity without raising this major geophysical problem. Probably this is particularly true because we are working in a globally resistant medium (karst) in LIN mode. Many authors working in comparable geological contexts also choose to not qualitatively compare apparent conductivity values obtained by EMI and apparent resistivity values obtained by ERT (Al-Fares 2011; Khaldaoui et al. 2020 ; Pavoni et al. 2021 ; Valois et al. 2010).

References :

Al-Fares, W. Contribution of the Geophysical Methods in Characterizing the Water Leakage in Afamia B Dam, Syria. Journal of Applied Geophysics 2011, 75, 464–471, doi:10.1016/j.jappgeo.2011.07.014.

Khaldaoui, F.; Djediat, Y.; Baker, H.; Ydri, A.; Djeddi, M.; Hamadou, K.; Bouzar, A. Use of Electrical Resistivity Tomography (ERT) and Electromagnetic Induction (EMI) Methods to Characterize Karst Hazards in North-Eastern of Algeria. Arabian Journal of Geosciences 2020, 13, 1–9.

Pavoni, M.; Sirch, F.; Boaga, J. Electrical and Electromagnetic Geophysical Prospecting for the Monitoring of Rock Glaciers in the Dolomites, Northeast Italy. Sensors 2021, 21, 1294.

Valois, R.; Bermejo, L.; Guerin, R.; Hinguant, S.; Pigeaud, R.; Rodet, J. Karstic Morphologies Identified with Geophysics around Saulges Caves (Mayenne, France). Archaeological Prospection 2010, 17, 151–160, doi:10.1002/arp.385.

 

3.2 The Valliguières: Atlas cedar experimental plantation

1. I couldn't find any reference to plot 51/52 and 21/22 in Fig 3A/B. It is not obvious to me that two are more homogeneous, …

We understand this comment but it seems difficult to add the parcel numbers to Figure 3, as Figure 3 is already full of information. We have explicitly added a reference to Figure 1B in the new version of the text (new lines 284-286).

“Some plots are fairly homogeneous (Fig. 3A), such as plots 51 and 52 (northernmost plot, see position on Fig. 1B), while others are more heterogeneous, such as plots 21 and 22 (southernmost plot, see position on Fig. 1B).”

… neither that the contrast are oriented west/east. Justify clearly your observations.

It is not the contrast that is oriented east/west but the geological structure of anticline. The contrast is perpendicular to the structure (i.e. north/south). We rephrased this sentence at new line 286.

“EMI allow to identify the general geological structure directed East/West, which agreed with the syncline general orientation [45].”

 

L.273: again the correlation between ERT and EM is not clear to me. You might try to plot a 1:1 line between 2d extracted EM (at the maximum sensitivity depth) and ERT to demonstrate it.

As with the comment in 3.1 and general comment i), we would like to remind that this paper is focused on EMI, which is a geophysical approach that has never been used for forest ecology studies. ERT is a better known and more widely used technique. ERT is used here as a benchmark for the reader.

The relationship between EMI conductivity and ERT resistivity raises a real geophysical challenge (especially in low-conductivity media). The exploration of this scientific issue is not the purpose of this article.

 

3.3. Combined interpretation

You should describe the statistical analysis adopted. Here I guess the p-value is the pearson coefficient?. Might be indicated in the figure legends (?). Moreover as the Pearson statistical analysis is subject to bias depending on outliers values you should describe more accurately the process of excluding them or not. This goes back to my previous comment about the data processing of EMI.

Right, we give a more detailed legend (new lines 323-325).

 

“n is the number of value, R is pearson correlation, p-value is calculated on Pearson correlation, r_s is the Spearman correlation.”

 

We also calculated the nonparametric Spearman correlation (r_s) which is not sensitive to outliers. The calculated r_s values confirm our interpretation.

 

L289: “ERT and EMI…” this conclusion in the current state of data observation should be mitigated to me.

We clarified this statement with a new sentence at new lines 305-308.

“For example, the ERT-11/12 cross-section (Fig. 3B), performed in the most conductive area of the EMI map (Fig. 3A), reveals a significantly higher conductivity than the ERT-41/42 cross-section which is performed in a more resistant area.”

 

L290: “At the Font-Blanche site, the EMI signal was closely related to thickness estimates”. Same comment as for L289. Difficult to conclude on how much EMI reflects soil thickness with such a low correlation coefficient.

Indeed, the correlation is limited between the EMI signal in horizontal dipole and the soil depth (R = 0.64). However, this correlation is largely significant (p-value = 3.10-4). In addition, it should be taken into consideration that EM31 is not the most suitable to characterize the first meter of soil. Initially this device was chosen to characterize the root zone of oak and pine trees which is about 5m (agreed with EM31 in vertical dipole). The correlation we found with the EM31 is significant and could be strengthened with a smaller device (type CMD-mini explorer).

 

L.295: not only figure 4 but also figure 2 illustrate similitude between PAI and EMI.

We agreed and we modified the sentence at new line 312.

“The results shown in figure 2 and 4 highlight that,…”

 

L309/310 and Figure 5: I suspect that forming groups not only removed influencing parameters but also outliers points which ultimately gives you a better correlation. Indeed, Eca ranges from 5 to 5.7 in Fig5a and 4.6 to 4.8 in Fig5b after considering subgroups. Compare to figure 4B it seems that forming groups also remove outliers values.

I would better conclude that EM method is not enough convenient and wouldn't grant it automatically to study soil/subsoil in this particular forest area (Valliguières forest).

 

We fully agree that the correlations are weak at the plot scale. Especially since the number of points is very limited (n = 22 and n = 9). We interpret this just as a trend that is consistent with the general interpretation. Finally, we caution that the number of points is limited and that interpretation at the plot scale should be done with caution. We rephrased le new lines 336-337.

“However, the limited number of observation points (n = 20 and n = 9) and the smaller range of ECa (Fig. 5) lead us to interpret plot scale result with caution.”

 

Conclusion

- L 353: define “stand scale” or rephrase.

We change “stand” for “experimental site”.

- L 357: I wouldn't put any references in the conclusion. Relocate them in the introduction/interpretation section or rephrase.

It does not appear to be recommended to exclude references from the conclusion for WATER. As an illustration, please find three article below with references in conclusion

References :

Liu, Y.; Fang, Y.; Hu, H.; Tian, F.; Dong, Z.; Khan, M.Y.A. Ecohydrological Separation Hypothesis: Review and Prospect. Water 2020, 12, 2077.

Mémin, A.; Ghienne, J.-F.; Hinderer, J.; Roquin, C.; Schuster, M. The Hydro-Isostatic Rebound Related to Megalake Chad (Holocene, Africa): First Numerical Modelling and Significance for Paleo-Shorelines Elevation. Water 2020, 12, 3180.

Shen, H.; Leblanc, M.; Frappart, F.; Seoane, L.; O’Grady, D.; Olioso, A.; Tweed, S. A Comparative Study of GRACE with Continental Evapotranspiration Estimates in Australian Semi-Arid and Arid Basins: Sensitivity to Climate Variability and Extremes. Water 2017, 9, 614, doi:10.3390/w9090614.

 

Instead of opening up the article on climate change considerations (which I consider to be a shortcut here), I would suggest to conclude on how to improve the understanding of such functioning i.e. using time lapse seasonal variation to see effects of water distribution into the different soil/subsoil.

The objective of this article is to show to the forest ecology community that EMI is an interesting technique that can partially answer some of their questions. For this reason, our conclusion is oriented towards forest ecology. However, we have added a sentence that goes in the direction you propose (new lines 388-390).

“Moreover, it would be relevant to characterize a forest site at various depths by combining several EMI devices. The exploration of the seasonal variation of EMI signal could also be relevant to characterize the spatial variability of forest soils and subsoils.”

 

Figure 3

• Put the orientation of the ERT lines (with increasing x towards East direction?) - exactly what you’ve done for figure 2 using aa, bb, cc, dd labels.

You are right. We give the direction of ERT cross-section on the new version of the figure.

• Link numbering and reference to ERT lines (51/52, 21/22) directly on the figure. In other words, in Fig3. which is ERT51, ERT52, ERT21, ERT22?

This is a good idea that we apply in the new version of the figure.

• Change “electrical conductivity” to “apparent conductivity or ECa” to be consistent with fig 2 legend.

Yes, it was done.

• Figures 3 legend. “Electrical conductivity/resistivity of the substrate at Valliguières” is too vague, you should link each physical variable to its method i.e. (A) apparent conductivity ECa for EM (B) inverted resistivity for ERT

You are right. The caption was not clear and we rephrased this sentence (new line 298).

“Geophysical results at Valliguières experimental site.”

Looking at the figure I’d have been interesting to consider ERT profiles in the perpendicular position since the gradient since more visible, won’t you?

Totally agreed. This was proposed by one of the co-authors during the project but it could not be done for practical reasons.

 

Supplementary Information

- Its not clear to me how this figure is build. Did you extract EM values at the exact same positions than the soil pits ?

You are right, this SI was not very clear. We provided more information at new lines 546-549.

“The miner's bar survey is a method frequently used in the field in forest ecology to variability in soil thickness [15]. The surveys consist of sinking a miner's bar to the maximum depth to estimate the soil depth. The surveys were repeated 3 to 6 times depending on the variability at 50cm around the center of each EMI measurement.”

 

- Why the values here are in microS/m but of the same range than those in mS/m ?

You are right. It was an error that we corrected in the new version of the figure SI1.

 

- This figure is really interesting but not enough exploited (only one mention of it in the manuscript). It highlights differences between HD and VD which are sensitive to different depth of investigation. This should be discuss versus soil depth.

You are right. The difference in correlation between HD and VD is interesting and consistent with what might be expected. In VD mode the correlation is less good since the depth of investigation is greater.

We have chosen not to discuss this in detail in the main text because the number of measurement points is limited. This SI only supports our general interpretation.

Author Response File: Author Response.docx

Reviewer 2 Report

This paper assesses the use of electromagnetic induction to estimate the influence of underground heterogeneity on forest canopy structure. I consider you clearly state the novelty and value of this study and you have achieved significant results using the approach in two different and well-selected sites. Moreover, you have done an extensive discussion of the previous literature, including relevant references in the field, and clearly define the studied area characteristics and challenges to deal with to develop your research. Finally, the conclusions drawn have been supported by the data and data relationships. Furthermore, I would like to suggest few changes to improve the reading of the text and its understanding:

Section 2.2.2: Consider rewritten lines 194-196 and justify why it is necessary to invert ERT data using RES2DINV. Please also provide the software version and the inversion method selected.

Please explain and justify the interpolation procedure and or the algorithm used for obtaining the EMI (Fig. 3) and PAI maps (Fig. 2C). Moreover, include the number of points EMI points acquired.

Below are more specific and minor comments:

Lines 86, 88 and 89: Please introduce before the meaning of the acronyms TDR, ERT and EMI.

Figure 1: Considering includes the units of the distances scale in Fig 1A and Fig. 1B and a distance scale in Fig 1A.

Line 118: “plot TC” is written in the Figure caption and “TD” in Fig 1C. Revise it.

Line 194, 265 and 286: Consider using the term “array/s” (instead of protocol) when referring Wenner-Schlumberger, Dipole-Dipole…

Line 213 and 214: Please provide the version of the software used and consider also add provider information.

In some parts of the text, you use Figure X and some others the abbreviation (Fig. X). I suggest using a unified format all along with the manuscript.

Author Response

# Reviewer 2 :

This paper assesses the use of electromagnetic induction to estimate the influence of underground heterogeneity on forest canopy structure. I consider you clearly state the novelty and value of this study and you have achieved significant results using the approach in two different and well-selected sites. Moreover, you have done an extensive discussion of the previous literature, including relevant references in the field, and clearly define the studied area characteristics and challenges to deal with to develop your research. Finally, the conclusions drawn have been supported by the data and data relationships. Furthermore, I would like to suggest few changes to improve the reading of the text and its understanding:

 

Thank you for this general comment.

 

Section 2.2.2: Consider rewritten lines 194-196 and justify why it is necessary to invert ERT data using RES2DINV. Please also provide the software version and the inversion method selected.

You are right. We give the version number at new line 211.

 

Please explain and justify the interpolation procedure and or the algorithm used for obtaining the EMI (Fig. 3) and PAI maps (Fig. 2C). Moreover, include the number of points EMI points acquired.

You are right. As recommended by reviewer 1, we explain how are calculated EMI maps (new lines 185-188).

“The ECa maps shown in Figures 2 and 3 were calculated with Surfer software (Golden Software®) using the kriging method.”

We give the number of EMI points acquired in Valliguières (new lines 182-183).

“We acquired 22213 points in Valliguières”.

 

Below are more specific and minor comments:

Lines 86, 88 and 89: Please introduce before the meaning of the acronyms TDR, ERT and EMI.

We agree therefore we developed the acronyms of TDR (new line 86), ERT (new lines 88-89), EMI (new line 90)

 

Figure 1: Considering includes the units of the distances scale in Fig 1A and Fig. 1B and a distance scale in Fig 1A.

You are right. We corrected all scales on Fig.1.

 

Line 118: “plot TC” is written in the Figure caption and “TD” in Fig 1C. Revise it.

Right. This was corrected on the new Fig.1.

 

Line 194, 265 and 286: Consider using the term “array/s” (instead of protocol) when referring Wenner-Schlumberger, Dipole-Dipole…

We changed “protocol” for “array” (at new lines 207 and 207).

 

Line 213 and 214: Please provide the version of the software used and consider also add provider information.

You are right. According to reviewer 1 comment, we give more information about ERT inversion (new lines 211-214).

“Apparent resistivities were processed using the Res2Dinv 3.57.36. The models presented below were all obtained after 3 iterations. This limited number of iterations is a good compromise that allows to reach acceptable RMS without generating extreme resistivity values to reach a mathematical optimum that would risk to move away from the field reality. The mesh was refined and…”

 

In some parts of the text, you use Figure X and some others the abbreviation (Fig. X). I suggest using a unified format all along with the manuscript.

You are right we unified the text for “Fig. X” (new lines 196, 268, 309).

 

 

 

 

 

Round 2

Reviewer 1 Report

Dear Authors, 

Dear Authors, 

Thanks for your revisions. I really appreciated the detailed answers to all my questions. 
I feel like I'm not completely in line with the interpretation but now that the methods are adequately described and some conclusions were mitigated, yet there is no scientific lack preventing the reader to make its own opinion.
Anyway, on its own, the article holds many promises and is very original. As the authors kindly remind me, this is the first application of such methods for forest ecology. 
I really hope this article will pave the way for more studies and I'm looking forward to seeing it online.

Sincerely.

Author Response

Dear Editor,

We are pleased to re-submit our paper entitled "Electromagnetic induction is a fast and non-destructive approach to estimate the influence of subsurface heterogeneity on forest canopy structure" for consideration in WATER.

We were pleased that the new version was appreciated reviewers. We would like to thank you and the reviewers for the time spent on our manuscript. Please find in this document (in yellow) our answers to all comments and suggestions.

We hope you will consider this version suitable for publication in WATER journal.

Yours sincerely

Simon Carrière (on behalf of the authors)

 

# Reviewer 1:

Dear Authors, 

Thanks for your revisions. I really appreciated the detailed answers to all my questions. 
I feel like I'm not completely in line with the interpretation but now that the methods are adequately described and some conclusions were mitigated, yet there is no scientific lack preventing the reader to make its own opinion.
Anyway, on its own, the article holds many promises and is very original. As the authors kindly remind me, this is the first application of such methods for forest ecology. 
I really hope this article will pave the way for more studies and I'm looking forward to seeing it online.

Sincerely.

Thank you for your encouraging comments.