Review Reports

Round 1

Reviewer 1 Report

The authors utilized a tethered balloon system (TBS) equipped with 4 separate cameras ranging in spectral range from 380 nm to a maximum of 14,000 nm to estimate the ability to passively monitor marine wildlife. The TBS was anchored along the shore while three separate wildlife models were towed behind a powered watercraft within the visible range. Images were taken to assess the detectability distance of each of the 4 camera systems under varying conditions and at three separate TBS altitudes, 50 m, 150 m, and 250 m. The ICI Mirage performed the best in regards to the maximum distance for a confirmed observation of targets at 724 m, while both the ICI Mirage and Sony RGB cameras attained the maximum distance of detection at 1243 m. The authors conclude that RGB cameras paired with thermal imaging systems may be the best way to outfit TBS as these sensors complement each other while also reducing the shortcomings inherent in each system. Furthermore, they demonstrate that the reduced cost of the ICI 8640 LWIR had comparable detection accuracy and efficiency which may reduce the cost of TBS surveillance if sensing beyond RGB is required. Lastly, the authors recommend performing traditional survey methods (human observation) with TBS camera surveys. While they acknowledge that future steps are needed to fully refine TBS as a marine mammal surveillance system, they conclude that similar systems may be useful in understanding the presence of marine mammals in a growing marine energy industry.

The authors have presented a well-written, albeit unfocused paper highlighting the need to assess how improvements in passive, visual monitoring technology (high-resolution RGB, infrared, and multi-spectral cameras) might aid in assessing the impacts of the ME industry on marine mammal populations. They also make clear the differences in detection distance between all the systems tested within the study. However, there is a distinct lack of presented data and statistical analyses that leave a reader unconvinced about the author’s conclusions in this study. Furthermore, there are some experimental design problems that further bring into question the validity of the conclusions in the study.

The major issues are as follows. First, and probably the least of concern, many studies have been conducted where targets/surrogates were used to study detection of marine animals using aerial surveys (see Robbins, peddemors, Kennely, and Ives, 2014; Colefax, Kelaher, walsh et al., 2021). However, many of these are done with the camera perpendicular to the water. Without a vertical silhouette for the marine mammal targets presented in this study, there is a reduced ability to detect the targets. Targets with a 3-dimensional silhouette would be a better representation of true marine mammal subjects that would be present during real surveillance in regards to the angle of the cameras onboard the TBS. Without this, the detection probability within this study is likely biased towards lower rates of observations. This would likely alter the results of this study and should be addressed. Furthermore, marine mammal skin temperature is much higher than ambient water temperature, which was not addressed within this study. As a result, we do not know how comparable the surrogates may be in detectability compared that of a marine mammal, may represent a bias towards lower detectability within this study.

A major problem within this study is also the tow patterns for the targets and a lack of clearly stating not just whether detectability was a success but how this success rate differs between all the environments and conditions included within the study. According to Fig. 3, the targets were towed haphazardly through the survey area. A much better experimental design would be to sequentially tow the targets from near distances to far distances to better identify the rate of successful detection for each camera system based on distance. A simple “farthest detection distance” is much less useful if the authors do not present data on how reliable the detection rate is on those distances. This also brings into question the lack of controls. Some of the targets (small in particular) only had 7 pixels worth of footprint within the frame selected. Without comparing the difference in true detectability, the maximum distances may be overstated within this study if marine flotsam is indistinguishable from small marine mammals at greater distances from the TBS. The authors also include multiple altitudes for the TBS but do not discuss how altitude may affect detectability. Lastly, the environmental conditions in which the data was collected really only reflect perfect conditions. This is likely the vast minority of conditions that a ME site would experience. The authors should re-perform data collection under a greater variety of environmental conditions or explicitly discuss the limitations of using TBS under harsher conditions which may limit the applicability of TBS as a surveillance tool.

Due to these considerations, I cannot recommend this study for publication without major and significant revisions, and potential inclusion of more robust flights or re-analysis of the data collected.

Below is a detailed, line-by-line review of my questions and concerns.

Line 12. Although specified, don’t use acronyms within the abstract.

Line 17. How many images were collected for each system? You may not have enough room for the abstract but this should be added somewhere within the paper

Line 31. I would remove hydrophones/acoustic telemetry. While these are used to monitor animal movement, I would not call them observational technology.

Line 32-35. This isn’t true? UAVs have RGB sensors that can collect data within the entire visible light spectrum, not just 30 nm spectrum of light.

Line 45-47. This needs to be reframed. Sun glare is a factor for all observations for marine animals at the surface, not just UAVs.

Line 50-52. This is also questionable the way this is stated. Images can be taken from quite a large distance away with advanced zoom capabilities in modern. I would focus more on the ability to survey at greater altitudes with aerial vehicles.

Line 70-71. Is there a citation for UAS altering marine animal behavior? Most studies indicate there is no adverse effect of flying over marine animals.

Line 73-80. This is very vaguely worded. While it becomes clear what is being measured later in the paper, please change this paragraph to clearly outline the goals of the study/what is being done and why.

Line 93-95. This doesn’t have to be included in the manuscript here. If absolutely needed, refer to this in the discussion, but as stated it seems like a non sequitur.

Line 97-102. If you measured all of these variables, why are they not included in the paper? It would be very interesting to see if any of these have effects on the detection distance. If none of these data were included in the study, please remove this.

Line 118-120. This can all be one sentence.

Line 124. It would be good to add what the limitation of the TBS actually is in this sentence, so the reader can understand how much better the coverage may be from, say, a UAS.

Line 125. Wait, I thought the images analyzed were taken at altitudes of 50 m, 150 m, and 250 m. Is this a typo?

Line 124-128. Were these the conditions that were present during flights, or were they the specific goals for the range of environmental conditions to test detection distance? This is worded vaguely.

Line 142. It is never specified within the paper whether distances measured in this study account for the altitude of TBS. Please specify whether these are Euclidean measurements, or the direct distance measurements, e.g. the hypotenuse between the TBS and the target.

Line 159-160. Why only these two specifications - weren’t more than two variables measured for each flight? Also, what are the specific classifications? Low Beaufort to measure farthest detection under the best conditions?

Line 164-165. Again, why such a narrow band for these classifications is you measured up to 7.4 m/s?

Line 187-188. Why? Again, without a clear indication of how many approximate distances this is too vague to really mean anything.

Line 187-191. Run-on sentence.

Line 191-193. What percentage of the total dataset was used? Still unclear why the decision to trim the dataset was made.

Line 195-198. Is this true? Wouldn’t you want to find an optimum in the least number of TBS but the highest accuracy in identification? So, if the identification rate is low at maximum distance you will need more TBS to accurately monitor the entire area.

Line 207. With a measurement of 7 pixels, this is nearly indistinguishable. That makes the argument that identification is inaccurate at this distance for a large class of the target species (seals, sea lions etc).

Line 214-218. Something the author’s should think about is establishing a minimum size and distance curve within the study. This would be much more useful to monitoring agencies, as they can alter their monitoring design based on local fauna. If the most common animals are pinnipeds, this sensor is not sufficient for the area at this distance. However, if the most common species are blue whales, then yes the sensor would be able to be used to identify target species.

Line 229-231. Did any RGB levels get measured within this study? Hard to make this conclusion with no data. I would rephrase to indicate this.

Line 233-234. This statement can’t really be made without some statistical backing. Again, lacking any stats of any kind this is not a convincing argument.

Line 245-246. How did you measure correlation? With no stats or correlation matrix, this really can’t be stated.

Line 247-249. How was this measured or statistically tested?

Line 249-252. But the surrogates in this study also lacked a heat source. How was this determined? Without the foreknowledge that every object in frame would be the surrogate and/or knowing the surrogate would be behind the largest heat-signature (e.g. the boat), would the authors have been to tell the difference between ocean flotsam or the surrogates?

Line 252-255. The resolution is so low on the images in Fig. 7 that they do not have a marine mammal silhouette. If the authors didn’t know that the surrogates were “whales” would they be able to identify the object as such?

Line 269-271. Isn’t that the case for all of these sensors, and not limited to just the RGB camera? Also, this can’t be claimed because depth is not a factor in regards to the data collected in this study. Either re-word to be less of a definitive statement or include some citations to back up this claim.

Line 273-278. Run-on sentence.

Line 284-287. None of the data collected are real-time data. Every image had to be downloaded from the camera and post-processed, so this can’t be claimed about TBS without significant manipulation of the onboard systems. For example, this would be a streaming of images immediately after capture to a human or A.I. post-processing unit to identify any objects in the frame as data was captured.

Line 287-291. This can’t be claimed without backing by other means or further explanation. TBS images as presented do not measure any of these factors and thus cannot assess a spike or gradual increase in them. Nor can they measure the physiological stress levels of animals observed in the images. Perhaps they can be used to conduct long-term studies to identify a potential decrease in marine mammal presence in regards to these stressors, but as written this can’t be stated.

Line 291-293. How? The statement needs elaboration to not be superfluous.

Line 297-299. I’m unclear how this applies to this paper without elaboration. Also, this needs to be supported in some way. Was there an increase in engagement due to these campaigns? Did the Triton Initiative alter their projects in some way as a result of these campaigns? Simply putting someone on social media doesn’t necessarily mean it was effective.

Line 320-321. This statement isn’t supported by the data presented within the study. If the smallest surrogate was routinely unidentifiable or only consisted of < 30 pixels, is it really plausible that these sensors would be able to identify a sea lions fin at the surface?

Line 328-332. Again, is this plausible? With the low resolution and amorphous silhouettes for surrogates discussed within this paper, is it really likely that humans or A.I. would be able to identify a marine mammal to species level at the maximum distance?

Line 342-348. This isn’t supported by the data presented in this study. By using the same identical surrogates without different coloration or form, the authors do not provide sufficient data to conclude that they would be able to identify different species at extended ranges with RGB. While different sizes were accurately identified, it would be hard to distinguish species of different sizes and similar coloration from younger, smaller individuals, and older, larger individuals of the same species.

Line 370-374. Is there any data that supports increased thermal absorption for the surrogates? Were images at the start of flights colder than at the end of the flights to measure an increased thermal temperature?

Line 387-389. Don’t the sensors only capture surface interactions as well?

Line 408. Wasn’t one of the problems in this study that the anterior section of the large surrogate being submerged decreased visibility? How would this particular TBS system be used to study submerged subjects?

Line 410-413. I would remove this. None of these points were specifically measured not described how one would collect data for any of these situations.

Line 418-423. There is no clear link between the data presented in this study to these claims.

Line 425-431. One particular point that needs to be addressed in the discussion of this study is the relative cost of TBS flights. While TBS may or may not be effective, if they are expensive to purchase and operate, agencies will turn to cheaper alternatives. This is also compounded on the fact that many agencies may need multiple TBS units to survey their operation site, personnel to sort and quality check data, and other personnel to disseminate data to the public/regulation agencies.

The writing is fine, but lacking sufficient detail for methods and results.

Author Response

We are grateful for your comments on our report and providing constructive comments. We found your comments to be helpful and led to team discussions about future efforts with our research project. Additionally, we incorporated the changes you suggested and this made the article more robust and comprehensive. Your insights provided valuable perspectives that enhanced the overall quality of our work. We are grateful for your thoughtful review and the time you took to carefully analyze our article.

We have provided a table responding to each of your comments.

Thank you for being an integral part of the peer review process and for your dedication to maintaining the rigor and excellence of academic research.

Sincerely, 

Alicia Amerson

 

Reviewer 1 Comments:	Responses
Line 12. Although specified, don’t use acronyms within the abstract.	Removed acronym
Line 17. How many images were collected for each system? You may not have enough room for the abstract but this should be added somewhere within the paper	An appendix with all the data is provided as a supplement
Line 31. I would remove hydrophones/acoustic telemetry. While these are used to monitor animal movement, I would not call them observational technology.	hydrophones by themselves can tell you presence/ absence in an area, and in an array can be used to track a vocalizing animal
Line 32-35. This isn’t true? UAVs have RGB sensors that can collect data within the entire visible light spectrum, not just 30 nm spectrum of light.	The spectral range has been corrected.
Line 45-47. This needs to be reframed. Sun glare is a factor for all observations for marine animals at the surface, not just UAVs.	Revised
Line 50-52. This is also questionable the way this is stated. Images can be taken from quite a large distance away with advanced zoom capabilities in modern. I would focus more on the ability to survey at greater altitudes with aerial vehicles.	Addressed
Line 70-71. Is there a citation for UAS altering marine animal behavior? Most studies indicate there is no adverse effect of flying over marine animals.	We decided to remove this comparison to UAS, as TBS disturbance to marine mammals needs to be reviewed more thoroughly, similar to UAS.
Line 73-80. This is very vaguely worded. While it becomes clear what is being measured later in the paper, please change this paragraph to clearly outline the goals of the study/what is being done and why.	Rewrote this paragraph.
Line 93-95. This doesn’t have to be included in the manuscript here. If absolutely needed, refer to this in the discussion, but as stated it seems like a non sequitur.	Deleted
Line 97-102. If you measured all of these variables, why are they not included in the paper? It would be very interesting to see if any of these have effects on the detection distance. If none of these data were included in the study, please remove this.	The GPS data is included in the study insofar as it was used to calculate distance from sensor to towboat, as explained below. Additionally, windspeed was used to classify Beaufort scale (also stated below). Although we list other variables here not explicitly used in the analysis conducted for this study, a motivating aspect of this work was to explore the viability of the TBS technology for these and similar applications. Therefore listing all environmental variables measured within the system could be informative for readers and does not seem gratuitous to us.
Line 118-120. This can all be one sentence.	We prefer two sentences
Line 124. It would be good to add what the limitation of the TBS actually is in this sentence, so the reader can understand how much better the coverage may be from, say, a UAS.	, rather than ending due to a limitation of the TBS. TBS flight approvals for this campaign required ground visibility of at least 3 statute miles and for the balloon to maintain 500' of separation below the base of any cloud. The TBS operational criteria for this campaign limited flights to when wind speeds were below 11 m/s at the surface or when lightning was not within a 16 km radius.
Line 125. Wait, I thought the images analyzed were taken at altitudes of 50 m, 150 m, and 250 m. Is this a typo?	This was corrected, it was a typo
Line 124-128. Were these the conditions that were present during flights, or were they the specific goals for the range of environmental conditions to test detection distance? This is worded vaguely.	Flights occurred during daylight and collected images at 50, 150, and 250 m above ground level for durations of 1 to 4 hours in clear sky, broken to overcast clouds, temperatures from 16 °C to 30 °C, relative humidities from 54% to 100%, and wind speeds from 0 to 7.4 m/s at the surface and 0 to 9.6 m/s aloft. Six flights were conducted from 2 December to 7 December 2022, as detailed in Table 2.
Line 142. It is never specified within the paper whether distances measured in this study account for the altitude of TBS. Please specify whether these are Euclidean measurements, or the direct distance measurements, e.g. the hypotenuse between the TBS and the target.	Added word "direct" in this sentence to clarify. Will explicitly state again where it is discussed in depth below.
Line 159-160. Why only these two specifications - weren’t more than two variables measured for each flight? Also, what are the specific classifications? Low Beaufort to measure farthest detection under the best conditions?	These two specification were chosen because we consider them to be salient and relevant collection parameters/conditions that could impact detection in any similar data acquisition campaign. We have added language in the text to say so. The specific classifications are detailed in the same paragraph below. Low Beaufort is included to account for the range of wind conditions that could be present under viable collection scenarios.
Line 164-165. Again, why such a narrow band for these classifications is you measured up to 7.4 m/s?	For this study we wanted to focus on the lower end of the scale when operating conditions are favorable and when ocean surface roughness is not extreme, in order to first test detection feasibility under calmer but still representative scenarios before possibly testing more extreme scenarios in future work.
Line 187-188. Why? Again, without a clear indication of how many approximate distances this is too vague to really mean anything.	Unclear what the reviewer is getting at here.
Line 187-191. Run-on sentence.	Fixed.
Line 191-193. What percentage of the total dataset was used? Still unclear why the decision to trim the dataset was made.	The following language was added to section 2.4. "The use of a representative conditional matrix in this study aligns with the broader aim of exploring the feasibility of the TBS-imager system for marine animal detection rather than conducting an exhaustive statistical analysis of the collected data."  Also - We made this choice because of time/resource constraints.
Line 195-198. Is this true? Wouldn’t you want to find an optimum in the least number of TBS but the highest accuracy in identification? So, if the identification rate is low at maximum distance you will need more TBS to accurately monitor the entire area.	Included the need to have the highest accuracy of identification which is very important.
Line 207. With a measurement of 7 pixels, this is nearly indistinguishable. That makes the argument that identification is inaccurate at this distance for a large class of the target species (seals, sea lions etc).	Stronger language has been added to emphasize that point.
Line 214-218. Something the author’s should think about is establishing a minimum size and distance curve within the study. This would be much more useful to monitoring agencies, as they can alter their monitoring design based on local fauna. If the most common animals are pinnipeds, this sensor is not sufficient for the area at this distance. However, if the most common species are blue whales, then yes the sensor would be able to be used to identify target species.	When we do live target study we will incorporate this min size distance curve - this is a fantastic idea and good way to communicate with the regulator - thank you for this recommendation. For this paper we are not going to include this.
Line 229-231. Did any RGB levels get measured within this study? Hard to make this conclusion with no data. I would rephrase to indicate this.	We believe our statement here is appropriately qualified as pointing only to a possible use of spectral signatures. The visual hue discrimination possible from a simple RGB rendering is suggestive of such potential. We make no firm conclusions in that regard.
Line 233-234. This statement can’t really be made without some statistical backing. Again, lacking any stats of any kind this is not a convincing argument.	We have added language to this sentence and the following to qualify these observation and make clear that they are qualitative in nature.
Line 245-246. How did you measure correlation? With no stats or correlation matrix, this really can’t be stated.	The language has been modified to remove the possible implication of a statistical correlation.
Line 247-249. How was this measured or statistically tested?	Sentence removed.
Line 249-252. But the surrogates in this study also lacked a heat source. How was this determined? Without the foreknowledge that every object in frame would be the surrogate and/or knowing the surrogate would be behind the largest heat-signature (e.g. the boat), would the authors have been to tell the difference between ocean flotsam or the surrogates?	Some clarifying language was added at the end of the sentence. The point we are making here is a fairly general one. We are stating that a strong thermal signal presents a potential detection irrespective of shape. Accurate positive identification would likely still depend on complementary non-thermal images in cases of amorphousness (e.g. a true color image at higher resolution might be able to distinguish whether the object were heated flotsam, yet may not have formed the basis for detection on its own). But it does not seem unreasonable to us that a surfacing mammal should carry a higher heat signature than inert debris. In contrast, high digital number values in non-thermal imaging bands do not by themselves carry the same significance.
Line 252-255. The resolution is so low on the images in Fig. 7 that they do not have a marine mammal silhouette. If the authors didn’t know that the surrogates were “whales” would they be able to identify the object as such?	We have added a sentence to make the point more emphatically.
Line 269-271. Isn’t that the case for all of these sensors, and not limited to just the RGB camera? Also, this can’t be claimed because depth is not a factor in regards to the data collected in this study. Either re-word to be less of a definitive statement or include some citations to back up this claim.	reworded the last sentence to include all sensors and emphasize accurate identification
Line 273-278. Run-on sentence.	period added
Line 284-287. None of the data collected are real-time data. Every image had to be downloaded from the camera and post-processed, so this can’t be claimed about TBS without significant manipulation of the onboard systems. For example, this would be a streaming of images immediately after capture to a human or A.I. post-processing unit to identify any objects in the frame as data was captured.	This is a future development, included those words.
Line 287-291. This can’t be claimed without backing by other means or further explanation. TBS images as presented do not measure any of these factors and thus cannot assess a spike or gradual increase in them. Nor can they measure the physiological stress levels of animals observed in the images. Perhaps they can be used to conduct long-term studies to identify a potential decrease in marine mammal presence in regards to these stressors, but as written this can’t be stated.	Rephrased this to emphasize the work done in this area
Line 291-293. How? The statement needs elaboration to not be superfluous.	Added an example
Line 297-299. I’m unclear how this applies to this paper without elaboration. Also, this needs to be supported in some way. Was there an increase in engagement due to these campaigns? Did the Triton Initiative alter their projects in some way as a result of these campaigns? Simply putting someone on social media doesn’t necessarily mean it was effective.	added current statistics on various communications regarding this research project
Line 320-321. This statement isn’t supported by the data presented within the study. If the smallest surrogate was routinely unidentifiable or only consisted of < 30 pixels, is it really plausible that these sensors would be able to identify a sea lions fin at the surface?	Switched example to logging whale, larger target
Line 328-332. Again, is this plausible? With the low resolution and amorphous silhouettes for surrogates discussed within this paper, is it really likely that humans or A.I. would be able to identify a marine mammal to species level at the maximum distance?	We do not believe we make any claims that marine mammal identification would be necessarily (or even likely) successful at the maximum distance. We are pointing here to an avenue for further capability development. And added this into the section for clarity.
Line 342-348. This isn’t supported by the data presented in this study. By using the same identical surrogates without different coloration or form, the authors do not provide sufficient data to conclude that they would be able to identify different species at extended ranges with RGB. While different sizes were accurately identified, it would be hard to distinguish species of different sizes and similar coloration from younger, smaller individuals, and older, larger individuals of the same species.	Removing paragraph, we did not perform any of this analysis, but we do see this as a future direction.
Line 370-374. Is there any data that supports increased thermal absorption for the surrogates? Were images at the start of flights colder than at the end of the flights to measure an increased thermal temperature?	Removed.
Line 387-389. Don’t the sensors only capture surface interactions as well?	The coastal-blue band (395 - 450 nm) of the MAIA sensor has the potential of seeing through the water column to some extent. Such sensors are used for coastal bathymetry. Thus, it remains work for a future study to examine whether such a sensor could be used to see sufficiently through the water column for near-surface but submerged animal detection.
Line 408. Wasn’t one of the problems in this study that the anterior section of the large surrogate being submerged decreased visibility? How would this particular TBS system be used to study submerged subjects?	We need to further test this application to determine if live targets that are submerged or underwater can be detected. Since we did not have live targets we were not able to validate the technology and sensor package.
Line 410-413. I would remove this. None of these points were specifically measured not described how one would collect data for any of these situations.	As stated in future applications, we are talking about ways the TBS could potentially be used in other blue economy sectors, this bullet supports that idea for others to consider . We are not removing the bullet. Moving this topic down below to follow along with other blue economy bullet. Made a new section for blue economy
Line 418-423. There is no clear link between the data presented in this study to these claims.	We express these are considerations for other blue economy sectors
Line 425-431. One particular point that needs to be addressed in the discussion of this study is the relative cost of TBS flights. While TBS may or may not be effective, if they are expensive to purchase and operate, agencies will turn to cheaper alternatives. This is also compounded on the fact that many agencies may need multiple TBS units to survey their operation site, personnel to sort and quality check data, and other personnel to disseminate data to the public/regulation agencies.	Added a new section to address costs in the conclusion

Author Response File: Author Response.pdf

Reviewer 2 Report

Dear authors, thanks for inviting me to review your work.

I found it very well designed and elaborated within the manuscript. I believe that, as you have reported, this system could contribute to the conservation of marine wildlife.

Please, find below my only suggestion:

Fig. 3: if possible, include an image with a better resolution.

 

Author Response

We are grateful for your comments on our report and providing constructive comments. We found your comments to be helpful and led to team discussions about future efforts with our research project. Additionally, we incorporated the changes you suggested and this made the article more robust and comprehensive. Your insights provided valuable perspectives that enhanced the overall quality of our work. We are grateful for your thoughtful review and the time you took to carefully analyze our article.

We have provided a table responding to each of your comments.

Thank you for being an integral part of the peer review process and for your dedication to maintaining the rigor and excellence of academic research.

Sincerely, 

Alicia Amerson

 

Reviewer 2 Comments	Responses
Fig. 3: if possible, include an image with a better resolution.	Figure replaced

Author Response File: Author Response.pdf

Reviewer 3 Report

In the present work "Validating a tethered balloon system and optical technologies for marine wildlife detection and tracking", authors describe e the potential for using TBSs and imaging sensors for marine wildlife observations. 

The topic is very interesting due to the possible application of new technology in different fields of researches. 

The manuscript is well written and if authors will follow the suggestion given I certainly recommand the paper for the pubblication.

Line 50-54: please add a references about these affermations. 

Line 63-72: here, you show the potentiality of use this technology in understanding marine mammals behaviour. I suggest you also to introduce the implications from an health point of view. It's worthing to note that the comprehension of animal ecology is fundamental in the prevision and management of diseases. Not only animal diseases, but also zoonoses, with a directly involvment of humans. Please see here:  https://doi.org/10.1079/PAVSNNR201813052  https://doi.org/10.1071/WR22136

Moreover, other new technologies can help researcher in this question, please see:  https://doi.org/10.3390/rs12213542.

Line 73-80: I advice you to better explain this part showing the real aim of the study that is not explicitly written. 

Please, explain better figure 3. I suggest you to use a different hatch to distinguish the different track of red and blue line. 

I suggest you to create a separate chapter for conclusions. In particular, you can start this one with "4.5 Future Applications of the TBS and Sensors" where you speak about the future applications giving also some implications that this technology could have for further investigations. I suggest you to remove 4.5 and report this part in the conclusions. 

Author Response

We are grateful for your comments on our report and providing constructive comments. We found your comments to be helpful and led to team discussions about future efforts with our research project. Additionally, we incorporated the changes you suggested and this made the article more robust and comprehensive. Your insights provided valuable perspectives that enhanced the overall quality of our work. We are grateful for your thoughtful review and the time you took to carefully analyze our article.

We have provided a table responding to each of your comments.

Thank you for being an integral part of the peer review process and for your dedication to maintaining the rigor and excellence of academic research.

Sincerely, 

Alicia Amerson

 

Reviewer 3 Comments	Responses
Line 50-54: please add a references about these affermations.	Added citations
Line 63-72: here, you show the potentiality of use this technology in understanding marine mammals behaviour. I suggest you also to introduce the implications from an health point of view. It's worthing to note that the comprehension of animal ecology is fundamental in the prevision and management of diseases. Not only animal diseases, but also zoonoses, with a directly involvment of humans. Please see here: https://doi.org/10.1079/PAVSNNR201813052  https://doi.org/10.1071/WR22136	This is a good point, but this study focuses on behavior and displacement. I think for future studies, the inclusion of health as it relates to disease would be beneficial to include and perhaps we can find a research partner more versed in this area to help develop the research and following articles.
Moreover, other new technologies can help researcher in this question, please see:  https://doi.org/10.3390/rs12213542.
Line 73-80: I advice you to better explain this part showing the real aim of the study that is not explicitly written.	Rewrote this paragraph
Please, explain better figure 3. I suggest you to use a different hatch to distinguish the different track of red and blue line.	Added more clarification.
I suggest you to create a separate chapter for conclusions. In particular, you can start this one with "4.5 Future Applications of the TBS and Sensors" where you speak about the future applications giving also some implications that this technology could have for further investigations. I suggest you to remove 4.5 and report this part in the conclusions.	Made the change / inclusion

Author Response File: Author Response.pdf

Round 2

Reviewer 1 Report

I would like to commend the authors for their timely response to the reviews for this technical note. The authors have toned down certain assertive language to make conclusions and observations relevant while also pointing out limitations to the technology and methods being proposed in this manuscript. They have also made it more abundantly clear that this is a stepping-stone study with insights for future research instead of an end-all, be-all technology/method that should be adopted across the board. Furthermore, they have made it clear why they included or removed data from the sample size which was previously unclear. Lastly, they have tightened up the manuscript by adding citations that were previously missing to add evidence to their claims or statements. I am personally interested in the differences in detection between these optical technologies so this was an interesting manuscript to review. However, I cannot recommend this manuscript for publication because some of my major concerns with this manuscript were not addressed.

            I had a lot of concerns about the methodology of this project that remained unanswered. For a methodological study of this kind, it is even more important that the methodology and presentation data is clear and shows whether these technologies can be successful in real-world applications. While the methods as written are clear, the study design is severely lacking in rigor that would create a fair comparison between technology. The haphazard dragging of the surrogates (Fig 3) creates a bias for the detection of said surrogates in close proximity but has a relatively small sample size for detection at further distances. This may be why the authors present no data on accuracy drop-off at a distance within the manuscript, which is, in my opinion, essential for this kind of study. There are no data presented that even indicates how many images were included in the data set. Yes, 5,454 images were collected, but due to the limitation of the Beaufort conditions, the sample size was reduced. Therefore, the reader has no idea what the actual sample size analyzed was, what the rate of image capture was, how many images were analyzed at varying distances etc. These were concerns voiced in my previous review. As a result, the conclusions the authors reached in this manuscript remain largely qualitative without much quantification of how each of these camera systems performed. This renders the conclusions somewhat meaningless for ME companies were they to consider using tether balloon systems.

            Unfortunately, I find that the lack of data provided within this manuscript makes the claims made by the authors unsubstantiated. There is not a clear comparison between the cameras and potential trade-offs for each system, e.g., RGB cameras can detect farther than infrared cameras but have lower accuracy at near distances etc, which significantly detracts from the purpose of the paper. In the end, the authors provide evidence that TBS can be used, but not that they should be used. Additionally, the authors show that these systems can detect surrogates at high distances from the cameras, but no indication of the accuracy of identification of the target. This could lead to an over-representation and actual detrimental conclusions if ME companies were to use these systems for marine monitoring.

Author Response

Dear Reviewer 1,

We would like to express our sincere appreciation for your thoughtful feedback during Round 2 of the review process. Your insights have been invaluable in improving the quality of our article, and we are pleased to inform you of the steps we have taken in response to your concerns.

Firstly, we have diligently addressed the specific aspects of your feedback, and we are pleased to report that another reviewer consolidated your concerns, leading to the creation of new sections in the discussion segment of our article. This comprehensive approach ensures that we have thoroughly covered the issues you raised.

Regarding the applicability of our technologies in real-world applications, we completely understand your point. To gain a deeper understanding, we are planning to conduct the next series of tests with marine mammals in early 2024. These tests will provide us with crucial data to gauge the practicality of our technologies in real-world scenarios.

Furthermore, we appreciate your understanding of the challenges we faced during the testing phase. The limitations imposed by the surrogates' tendency to sink if we motored too fast in the boat forced us to navigate cautiously and at the mercy of the natural elements like wind, tides, and currents moving the boat. As a result, Figure 3's lines may not appear as straight as ideal. We have incorporated this into our discussion, emphasizing the importance of working with real-world conditions and their inherent unpredictability.

We believe these additional details and adjustments will strengthen our case for publishing this article. The hurdles we've overcome to reach this point have been significant, and our ultimate goal is to advance the understanding and implementation of these technologies. We are committed to conducting more specialized research in the near future, and we hope that our work will not only spark interest but also foster collaboration among those who can benefit from this technology.

Additionally, we are actively exploring the potential applications of our technology beyond marine mammal detection and tracking, including surveillance and security. These exploratory steps have demanded tremendous effort from our team, and we are eager to gather more data through our upcoming study, which will address many of the discussion points you have raised.

Your continued engagement and valuable comments have played a pivotal role in shaping our research and future direction. We genuinely hope to engage in further discussions with you as we progress with our work. Your expertise and insights are highly regarded and greatly appreciated.

Once again, thank you for your time and dedication to reviewing our article. We look forward to the possibility of further collaboration and discussion.

Warm regards,

Alicia Amerson

Marine Biologist and Team Leader

Pacific Northwest National Laboratory