Ground Truth Validation of Sentinel-2 Data Using Mobile Wireless Ad Hoc Sensor Networks (MWSN) in Vegetation Stands

Round 1

Reviewer 1 Report (New Reviewer)

The remote sensing data of Sentinel-2 is verified by mobile wireless Ad-hoc sensor networks, and the experimental results also provide support for its validity. Your article is very substantial, leaving me with few critical remarks to offer, except for the following.

(1) If I understand correctly, from Figure 2, the installation location of MWSN nodes looks more like northeast to southwest, please clarify.

(2) In the abstract, the position where "MWSN" appears for the first time should give the full spelling, which is conducive to attracting readers.

(3) Please give more details on the inter-calibration of the close-range sensors.

Minor editing of English language required.

Author Response

Please see the attachment

Author Response File: Author Response.pdf

Reviewer 2 Report (New Reviewer)

The authors present on the information gained from a mobile wireless ad-hoc sensor network deployed in Germany. I commend the authors on their advanced work to bridge the gap between remote sensing and close-range ‘ground truth’ data by developing and deploying a means to collect close-range data at a regional scale simultaneous to the Sentinel satellite passes. It is clear they have succeeded in their efforts, and I hope that others will follow this path as well potentially towards a global network of such sensors. I commend the authors for their work which I found very compelling.

General Comments:

I found the level of detail to be excellent and the information presented was at a very high-quality of presentation.

The paragraphs in the discussion lacked a narrative flow which could be improved upon. I would suggest focusing on the results in terms of the notable categories, be it incidence angle effects, topographic, sub-pixel, etc and summarize among stations rather than picking things out one by one for a given station.

I found the first part of the conclusion to be somewhat repetitive from the discussion – I appreciated the clear synthesis which came later on.

Line Comments:

32. aims towards

33. this sentence is convoluted and could be simplified/clarified

35. English phrasing in the sentence beginning with ‘Especially’ is not correct

41. the meaning is a bit obscured by the list of acronyms – I would suggest providing a more literate explanation of the difference between the German program and these others

52. clarify what is meant by super-positions of plant processes and signals

55. this seems recursive, the ground truth is required to meaningfully infer conditions given the heterogeneity in the signals examined

66. I think you have done well to establish the need for ground/ close-range sensor networks, however, you do not clearly explain what a mobile ad-hoc sensor network is explicitly.

94. I think the phrasing in this sentence is awkward

98. leaf area index?

122. this feels a bit like a comparison against nothing. Certainly, other sites would not be selected with your specific objectives in mind…

164. perhaps an explanation of these different protocols in natural language would benefit the reader

271. these data

273. is Gitelson and Merzlyak the best citation for this?

351. replace found with ‘find’

351. the sentence beginning with ‘Since canopy’ should be clarified

356. this sentence has poor phrasing as well as should be clarified

370. what does it mean ‘the more or less share’? very confusing sentence

405. I didn’t see an analysis which supports this statement

422. This sentence is not clear – what is meant by ‘synergistically-effective function sharing’?

425. you interchangeably refer to RS or remote-sensing throughout the manuscript

451. it is clear that careful interpretation of the data from both space-borne and close-range sensors will be critical

Bravo!

My main initial concerns were with what I perceived to be awkward English-language phrasing, and, in some areas, I perceived the acronyms and ‘jargon’ or high-level details effected the readability to a notable degree. These could be improved with minor edits.

Author Response

Please see the attachment.

Reviewer 3 Report (New Reviewer)

I liked the paper and will recommend it for publication after minor revisions.

My major concern is that the manuscript has too many abbreviations. At some point I got lost and found myself going back and forth between the pages. Please, try to be less wordy and more specific.

The same could be said regarding the figures. I assume that some of it could be dropped, incorporated in another as (a) and (b) or moved to supplementary (For example, Figs. 3 and 4).

Some specifics.

Section 2.3.1 Is too long and has too many details. Please, try rewriting it.

Table 2 in my opinion could be incorporated into Fig. 5

Fig. 6 does not bring any additional information. I would recommend not presenting it at all.

Fig. 9. Presenting a range of NDVI and explaining it would benefit

Same for Fig. 10

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

Reviewer 4 Report (New Reviewer)

 this paper presents an interesting approach to addressing the challenges in validating and enhancing remote sensing data with close-range measurements. The combination of wireless sensors and temporal compression techniques offers potential benefits for monitoring environmental changes and understanding dynamic vegetation behaviors throughout the day. However, a comprehensive evaluation of the MWSN's accuracy, precision, and reliability would be crucial to establish its scientific significance and practical utility in improving remote sensing-derived information products.

This article falls under the category of satellite data ground validation, showcasing the author's extensive work. However, there are some details that need improvement.

1.It is suggested to include an appendix for the abbreviation list of professional terms in the article.

2.The abstract elaborates too much on the technical background, without highlighting the author's work, main findings, and the technological value of the research. It is suggested to rewrite the abstract.

3.The current article reads more like a technical report, and it is hoped that the author can make revisions from the perspective of a scientific paper.

4.Native English speakers' help is needed for proofreading and improvement of the English version.

Native English speakers' help is needed for proofreading and improvement of the English version.

Author Response

Please see the attachment.

Author Response File: Author Response.pdf

This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.

Round 1

Reviewer 1 Report

This paper by Mollenhauer et al. concerns the validation of Sentinel-2 Data using ground truth spectral data acquired with Mobile Wireless Ad-hoc Sensor Networks (MWSN) in vegetation stands. Even though the idea of the paper is interesting there are several concerns. The most important one lays on the fact that only one Sentinel image (for one date) is used on the paper, which may raise questions about the validity and repeatability of the conclusions. Sentinel-2 acquires images every 5 days, so using a larger data set, especially during different vegetation phenological phases, is easily feasible and would be more appropriate. Additionally, in several parts of the results and discussions sections, findings are discussed in relation to environmental or atmospheric parameters without presenting any kind or relevant data or correlations to explanatory parameters (details may be found below).

Accordingly, the paper is not suitable for publication in the present form.

Comments

L29. Please put MWSN in the next line.

L174. Please provide details about the MWSN sensors (manufacturer, model, etc). What is the measurement frequency?

L180-181. In these lines, authors denote that “The advantage of this design is that the spectral characteristics of the sensors can be adjusted to specific satellite missions or vegetation indices, among others”. However, by comparing the spectral characteristics of the MWSN sensors to S2 (Tables 2 and 3), even though this is valid for bands’ central wavelength, there are differences in bandwidths. Could you please explain why were the MWSN sensors not adjusted exactly to S2 characteristics?

L197. ESA already provides geo- radiometric and atmospherically corrected products (L2A). Why did you choose to manually apply atmospheric corrections with Sen2Cor instead of using the L2A product? How does your correction compare to the L2A product provided by ESA? What software did you use for the Sen2Cor application?

L243. Please provide the source of the data (spectral response factors) for both S2 and MWSN shown in Fig. 8. Additionally, the data shown in Fig. 8 are not always consistent with the data presented in Tables 2 and 3. For example, response for S2 band 4 ranges from about 646 to 684 nm, resulting in a bandwidth of 38 nm instead of 30 nm.

L252. It is not clear what MWSN data are compared to S2 data in Fig.10. Are the box plots for MWSN data produced from all measurements during the day or from a certain period around the S2 acquisition time, as mentioned in Fig. 10 legend?

L259-262. It is not clear what MWSN data are compared to S2 data in Fig.11. Is the MWSN measurement closest to the S2 acquisition time used? If that is the case, why do you use a different protocol compared to Fig. 10?

L263-265. I suppose this paragraph concerns Fig. 12 (not mentioned). However, as shown in Fig. 9, NDVI is rather stable between 09:15 to 12:15 and no data concerning “unstable weather conditions” are provided.

L267-277. The NDVI differences between stations are attributed to environmental and pedological differences between stations. However, no correlation diagrams between any environmental or pedological parameter and NDVI is shown.

L278-284. The temporal fluctuations of NDVI are attributed to BRDF effects. How is this justified? Any correlations e.g. between solar angles and NDVI? Why are there large differences during the course of the day for some stations (e.g. stations 1 and 2 in the morning) while no differences exist for other stations (e.g. stations 7 and 9)?

L285-289. Depressions in NDVI during the day are attributed to environmental conditions or atmospheric phenomena, but no data provided to support this explanation.

L290. For both bands’ reflectances (Fig. 10) and VIs (Fig. 11, 12), correlation graphs between S2 and MWSN data would better depict the comparisons.

L290-344. The discussion of the results in these paragraphs is based only on speculations without any connection / correlation to any kind of explanatory data. Even though the theoretical basis of these explanations may be correct, it would be more appropriate to connect the results with measured parameters and present solid explanatory arguments. For example, vegetation cover, or water status between stations could have been measured and correlated with VI data to support the results.

Author Response

"Please see the attachment."

Author Response File: Author Response.pdf

Reviewer 2 Report

This is a really well written manuscript with high-quality graphics. When I read the abstract, my initial questions were why optimize for S2 when there are other sensors (which they answered) and how stable the calibrations are (which is not really answered). Overall I think it's a solid contribution highlighting a novel field validation method, but they also included a really interesting site so there are relevant scientific findings as well. While BRDF is known to be key, I appreciate them showing how dramatic the impact could be. This is very relevant to non-satellite datasets such as airborne or UAV radiances.

In general I think this could be published as-is, but I have two comments.

1) Very minor, but the last figure introduces new analysis that is not described in the results. I am OK with that but in general you shouldn't be introducing new data/graphics in the discussion, you should be discussing what was already presented.

2) While I'm sure the field calibration is effective, I would really like to see some plot of calibration stability (drift). How often do the sensors need to be calibrated? How much do the spectra degrade during deployment due to (e.g.) getting dirty, etc?  These are key questions for actually deploying the network. If the goal is to look at spatial gradients, probably not so important, but if the goal is to provide calibration/validation for S2 products then there should be some discussion of how stable the calibration is with time, how often it needs to be redone, and what you do if the it's drifting (i.e. is it a step function, or a linear degradation?)

Author Response

Please see the attachment.

Author Response File: Author Response.pdf