A Process-Oriented Exploration of the Evolutionary Structures of Ocean Dynamics with Time Series of a Remote Sensing Dataset
Round 1
Reviewer 1 Report (New Reviewer)
This paper proposes a novel approach (PoEXES) to explore the evolutionary structures of ocean dynamics with a time-series remote sensing dataset based on the combined perspective of GIS and RS. This study is interesting and significant from a process-oriented exploration of ocean dynamic evolution, which can provide a new way to design a spatiotemporal analysis model for dealing with multiscale ocean dynamics. This manuscript can be considered for publication subject to some revisions and clarifications.
Section 2 for the PoEXES method is too long to follow, I recommend the authors simplify it and make the method section more reasonable and readable. The basis for some methods selection (such as the DcSTCA for clustering) should be clarified. And the SLOA algorithm should be introduced a little bit with some details.
There should be a comparative experiment designed for approach performance contrast, and the performance should be well measured. This study lacks method comparison and is hard to determine the advantage and superiority of the proposed technique framework for exploring the evolutionary structures of ocean dynamics. Moreover, the limitation of the approach should be directly pointed out in the conclusion.
This study only adopts one case (SSTA and ENSO event) to address the evolutionary structures of ocean surface temperature dynamics, I suggest the authors employ more cases about ocean surface dynamics to further demonstrate the evolution processes of ocean dynamic phenomena, such as the mesoscale eddy which is highly dynamic and can be detected based on SSHA from satellite altimetry. The significant evolution process regarding the highly dynamic processes (e.g. mesoscale eddy) should be further examined and tracked by the proposed technique PoEXES.
Author Response
Response to Reviewer #1:
This paper proposes a novel approach (PoEXES) to explore the evolutionary structures of ocean dynamics with a time-series remote sensing dataset based on the combined perspective of GIS and RS. This study is interesting and significant from a process-oriented exploration of ocean dynamic evolution, which can provide a new way to design a spatiotemporal analysis model for dealing with multiscale ocean dynamics. This manuscript can be considered for publication subject to some revisions and clarifications.
Comments 1: Section 2 for the PoEXES method is too long to follow, I recommend the authors simplify it and make the method section more reasonable and readable. The basis for some methods selection (such as the DcSTCA for clustering) should be clarified. And the SLOA algorithm should be introduced a little bit with some details.
Reply: Thanks for the reviewer’s suggestions. Parts of Section 2 are rewritten, and some subsections are reorganized, which will make it more reasonable and readable. And the selection of DcSTCA to extract marine snapshot object is given in more details. Also, the SLOA algorithm is rewritten with more details.
Comments 2: There should be a comparative experiment designed for approach performance contrast, and the performance should be well measured. This study lacks method comparison and is hard to determine the advantage and superiority of the proposed technique framework for exploring the evolutionary structures of ocean dynamics. Moreover, the limitation of the approach should be directly pointed out in the conclusion.
Reply: Thanks for the reviewer’s suggestions, which are invaluable for improving the quality of our manuscript. Honestly, this manuscript lacks of method comparison, which lies in two reasons. The first is that, to the best known of our knowledge, there are no published methods of exploring oceanic/geographical evolutionary structure from time series of remote sensing datasets. And the second is that our previous study has proven the advantage and superiority of the method of extracting oceanic dynamic objects (Process-oriented SSTA Object), which has been published in the journal of Big Earth Data (Xue et al., 2022, in Big Earth Data, https://doi.org/10.1080/20964471.2021.1988426). And the method in XUE et al.(2022) is the DcSTCA-based oceanic dynamic object extraction, which is the basis of extracting the evolutionary structures of ocean dynamics, i.e. PoEXES.
In addition, according to the reviewer’s suggestions, the limitation of the PoEXES are directly pointed out in Conclusion Section.
Comments 3: This study only adopts one case (SSTA and ENSO event) to address the evolutionary structures of ocean surface temperature dynamics, I suggest the authors employ more cases about ocean surface dynamics to further demonstrate the evolution processes of ocean dynamic phenomena, such as the mesoscale eddy which is highly dynamic and can be detected based on SSHA from satellite altimetry. The significant evolution process regarding the highly dynamic processes (e.g. mesoscale eddy) should be further examined and tracked by the proposed technique PoEXES.
Reply: The reviewer’s comment is right about that this manuscript only adopts one case (SSTA and ENSO event) to address the evolutionary structures of ocean surface temperature dynamics, and that the mesoscale eddy with highly dynamic can be detected based on SSHA from satellite altimetry. However, the revised manuscript still doesn’t add the mesoscale eddy as another case. The main reasons lie in two folds. The first is that the identification of mesoscale eddy from time series of SSHA dataset is obviously different from the identification of marine snapshot object from SSTA dataset, we need to design a new algorithm to identify mesoscale eddy from SSHA dataset. The second is that the identification and track of mesoscale eddy from time series of SSHA dataset is a scientific issue, which needs another manuscript to address it. That is to say, the identification and track of mesoscale eddy from time series of SSHA dataset is out of the scope of our manuscript.
Although, as aforementioned above, we still believe that the suggestion “The significant evolution process regarding the highly dynamic processes (e.g. mesoscale eddy) should be further examined and tracked by the proposed technique PoEXES” is invaluable for improving our further study. We will adopt the core idea of PoEXES that “Taking the evolutionary scale as a whole unit” to design a novel algorithm to identify and track mesoscale eddies from time series of SSHA dataset in the future, and then explore their dynamic structures. Also, the revised manuscript discusses this issue as a limitation of PoEXES in Conclusion Section.
Reviewer 2 Report (New Reviewer)
This manuscript describes the development of a tool that "explores" evolutionary structures in the ocean. The method, called PoEXES, which uses the SLOA graph construction algorithm. The paper is well written. The methods are clearly described and the paper manuscript is bay to follow. The apply the process on a SSTA dataset from the NOAA OIST between 1982 and 2021. The process what able to identify the scale of evolutionary processes of ENSO events. There are some broad question I have about the research.
How does Table 2 match to events spatially. Can the OIDs objects be identified on the map in figure 7? If so, are you certain that the SSTA process evolutions at all related to ENSO? Please provide some clarification. Perhaps I have missed something there?
Also, the applicability of this approach in terms of a contribution to remote sensing remains unclear. What you you doing here that is unique and represents a contribution to the field. Please be explicit and describing this. Also, there seems to be a lack of justifying the applicability of this approach. What are the benefits of using this approach aside from simply defining thresholds and defining SSTA areas?
There is a need to more clearly stat how PoEXES describes evolutionary structures of ocean dynamics. What is meant by, "ensures their evolutionary relationships are optimum as much as possible"? How is this different from a scale of "data observation"?
Please provide more information about the novelty of you results. It isn't clear how this system identifies when and where marine environments change but also how they change. Please reference back to your SSTA example. As currently written, It is difficult to see how your work does this.
Overall, the paper is well written and clearly documents a process and approach, but the applicability or novelty of the approach remains somewhat unclear. If you provide more information around this, that would be great. Thank you.
Author Response
Comments 1: This manuscript describes the development of a tool that "explores" evolutionary structures in the ocean. The method, called PoEXES, which uses the SLOA graph construction algorithm. The paper is well written. The methods are clearly described and the paper manuscript is bay to follow. The apply the process on a SSTA dataset from the NOAA OIST between 1982 and 2021. The process what able to identify the scale of evolutionary processes of ENSO events.
Reply: Thank for the reviewer’s significant comments for our manuscript.
Comments 2: How does Table 2 match to events spatially. Can the OIDs objects be identified on the map in figure 7? If so, are you certain that the SSTA process evolutions at all related to ENSO? Please provide some clarification. Perhaps I have missed something there?
Reply: Table 2 is the statistical information table about the evolutionary structure types of the SSTA dynamics. Figure 7 only shows the spatial information of the Type I of SSTA dynamics, that is to say, all the OIDs objects of Type I can be identified on the map in Figure 7.
Although, most of the SSTA process evolutions are related to ENSO events, we are not sure all the SSTA process evolutions are related to ENSO events, Table 3 shows the corresponding relationships between process objects of SSTA of Type II and ENSO events. Regarding to the new findings about their relationship, we need more studies in future.
Comments 3: Also, the applicability of this approach in terms of a contribution to remote sensing remains unclear. What you you doing here that is unique and represents a contribution to the field. Please be explicit and describing this. Also, there seems to be a lack of justifying the applicability of this approach. What are the benefits of using this approach aside from simply defining thresholds and defining SSTA areas?
Reply: Thanks for the reviewer’s comments, and these comments are invaluable for improving our manuscript. This manuscript concentrates on a method of processing RS data, thus, the contribution to RS is to help deal with time series of RS images to find dynamic patterns. And the revised manuscript clarifies these contributions in Conclusion Section.
As for the benefits of our proposed approach aside from simply defining thresholds and defining SSTA areas, we believe that the simply thresholds only find the marine abnormal variation in space, and the continuity of the spatial variations between the successive time snapshots is lost. The PoEXES takes an evolutionary scale to ensure the continuity of marine snapshot objects in successive time snapshots, also the merging and splitting relationships among them are obtained.
Comments 4: There is a need to more clearly stat how PoEXES describes evolutionary structures of ocean dynamics. What is meant by, "ensures their evolutionary relationships are optimum as much as possible"? How is this different from a scale of "data observation"?
Reply: Thanks for the reviewer’s comment, and apology for making it unclear about the advantage of PoEXES. Taking the evolutionary scale to design the PoEXES is our motivation, thus, PoEXES can obtained the evolutionary object from origination through development to dissipation, and the evolutionary relationships as well, i.e. merging and splitting relationships.
Comparisons with the scale of data observation, the evolutionary scale has two advantages. The first is that the evolutionary scale ensures the continuity of spatial variations in time, just as discussions in Comment 3.
And the second is that the evolutionary scale will ensures their evolutionary relationships are optimum as much as possible. As known, the scale of data observation is a revisited cycle of satellite remote sensing or the temporal resolution of the dataset, which is generally inconsistent with the scale of natural environmental dynamic. Thus, the evolutionary relationships among time snapshots are too numerous to identify evolutionary structures when the evolutionary scale is greater than data observation scale; when the evolutionary scale is smaller than data observation scale, the key evolutionary relationships will be lost, and the evolutionary structure is incomplete.
Comments 5: Please provide more information about the novelty of you results. It isn't clear how this system identifies when and where marine environments change but also how they change. Please reference back to your SSTA example. As currently written, It is difficult to see how your work does this.
Reply: Thanks for the reviewer’s comment, and apology for making it unclear about the when, where and how marine environments change. In our manuscript, the happing time and place of marine snapshot object/marine sequence object mean when and where marine environments change, and the time and place of splitting or merging among marine snapshot objects mean how marine environments change. Our proposed PoEXES uses the clustering model to extract the time and place when and where marine snapshot objects occur, uses the SLOA algorithm to track the sequence object, and then identifies the time and place when and where marine snapshot objects split or merge. Fig.8 to Fig.12 show when and where SSTA change, and Fig.13 to Fig.14 show when, where and how SSTA change.
To better clarify when, where and how marine environments change, we still follow the reviewer’s comments to rewrite the SSTA example in Section 3.3, shown in Line 386-Line 392, and Line 409-Line 412.
Comments 6: Overall, the paper is well written and clearly documents a process and approach, but the applicability or novelty of the approach remains somewhat unclear. If you provide more information around this, that would be great.
Reply: Thanks for the reviewer’s meaningful and significant comments. The revised manuscript rewrites the Discussion (Section 3.3) and Conclusion Section to clearly describe the applicability or novelty of the our proposed PoEXES.
Reviewer 3 Report (New Reviewer)
Comments for author File: 
Comments.pdf
Author Response
Comments 1: This article is clearly written, scientifically sound and very interesting for the scientific community.
Reply: Thanks for the reviewer’s positive and significant comments.
Comments 2: The reference list should be improved. Although the introduction gives a good coverage of what has been done so far, there are some critical additions:
Within a sentence a more general usage of SAR sensing should be given. See and include for example:
Chen, Principles of Synthetic Aperture Radar Imaging: A System Simulation Approach (Signal and Image Processing of Earth Observations), CRC Press, 2017.
For SAR-Ocean Environment:
Synthetic aperture radar: Marine user's manual, 2004.
For algorithmic details of SAR change detection please include the following studies in the references list:
https://ieeexplore.ieee.org!abstract/document/7954605
https://www.tandfonIine.com/doi/full/10.1080/22797254.2018.14825 23.
Reply: Thanks for the reviewer’s comments, and new references are appended.
Comments 3: When the DOA method is described in Eq.1, the correct term is not 'overlapped area', remote sensing community refers as 'co-registration'. Please update the corresponding sections properly.
Reply: Thanks for the reviewer’s comments. All the ‘overlapped area’ and its representation were updated.
Comments 4: Please discuss how PoEXES relates with the change detection techniques. More specifically, the second stage of PoEXES relies on identifying marine linked objects. How does it perform compared to the coherent and non-coherent change detection?
Reply: The reviewer is right. The second stage of PoEXES is to identify marine linked objects. Generally, this step is little related with the change detection, however, the previous step about clustering-based method to extract marine snapshot objects is more related with the change detection.
The clustering-based method takes the lifespan of a marine anomaly as a whole, and simultaneously considers the spatial, temporal and thematic characteristics of SSTA to extract marine snapshot objects, which performs well on spatiotemporal clustering patterns of SSTA in the Pacific Ocean. These related descriptions are discussed in Section 2.3.
Comments 5: Please discuss if the method proposed and the results in Fig.7 identifies Gulf-Stream and similar circulation patterns?
Reply: Thanks for the reviewer’s comments. Results in Fig.7 show the origination and dissipation location of SSTA dynamics, and the arrow means the direction from the origination and dissipation. Thus, using the results shown in Fig.7, we can represent the Gulf-Stream and similar circulation patterns
Comments 6: What are the other uses of your method except SST? For example, how does it behave in the case of an oil spill or tsunami?
Reply: Thanks for the reviewer’s comments, which are invaluable for improving the quality of our manuscript. Although, only a real SST dataset is used as a case study to evaluate the PoEXES performances in the manuscript, the PoEXES can achieve the similar capabilities when dealing with the sea surface salinity, sea level anomaly, sea surface current, and other marine environmental parameters. In addition, we have done on the sea surface salinity and sea level anomaly to obtain their dynamic structures. Up to now, these results have not been published in open journal. According to the reviewer’s comments, the revised manuscript discusses the capabilities of PoEXES when dealing with other marine environmental parameters in Conclusion section.
Regrading to an oil spill or tsunami, honestly, we have never dealt with it with our proposed PoEXES. Considering the force of driving the oil spill to move on sea surface, which is mainly driven by sea surface current, we think the PoEXES can identify and track the movement of an oil spill in space. However, many more experiments need to be done in future.
Round 2
Reviewer 1 Report (New Reviewer)
After the major revision and well clarification, this manuscript can be accepted for publication at the present form in Remote Sensing.
Reviewer 3 Report (New Reviewer)
The authors did the majority of the additions.
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
I didn't go through all the technical details of the manuscript, but in general, it is nicely written and may attract some good research community attention.
I only have some minor comments.
The description of how the marine snapshot objects are detected from time series data is very brief. Maybe some examples can be presented to illustrate that step.
Section 4: it is a surprise that only 400+ process objects are detected over 40 years. Maybe there are some criteria to filter out the weak/short events.
Is there a database to archive the detected features, so that readers can verify the results?
Author Response
Response to Reviewer #1:
Comments 1: I didn't go through all the technical details of the manuscript, but in general, it is nicely written and may attract some good research community attention.
Reply: Thank for the reviewer’s significant comments for our manuscript.
Comments 2: The description of how the marine snapshot objects are detected from time series data is very brief. Maybe some examples can be presented to illustrate that step.
Reply: Thank for the reviewer’s comments. The revised manuscript adds a new Figure (Figure 4) and an example to show how to the marine snapshot objects from time series data, the more details are in Section 2.3.
Comments 3: Section 4: it is a surprise that only 400+ process objects are detected over 40 years. Maybe there are some criteria to filter out the weak/short events.
Reply: The reviewer is right. We set the spatiotemporal criteria to filter out the weak/short events. The details are as follows: Firstly, the snapshot objects are filtered by anomaly value of SST greater than k times standard deviation of time series of SST, and the snapshot objects need to meet the spatiotemporal topology. Also, the duration of process object is greater than 5 months.
Generally, each process object contains multiple sequence objects, and each sequence object contains multiple snapshot objects. Although, there are a total of 417 process objects, 1180 sequence objects and 3687 snapshot objects are obtained in the past 40 years.
Comments 4: Is there a database to archive the detected features, so that readers can verify the results?
Reply: The reviewer is right. We have built a graph-based database to store marine process objects and their evolutionary relationships. The graph-based database and related works have been published in another journal, we also referred in [9].
Reviewer 2 Report
Review report on “A process-oriented exploring for evolutionary structures of oceanic dynamics with time series of remote sensing images” by Cunjin Xue et al.
1. Lines 147-152: These two definitions are not clear, please give a detailed case.
2. Lines 164-169, Figure 1 and 2: The Caption is too short and the description is too concise. These two important architecture diagrams should be written in detail for readers to understand.
3. Section 3.2: The description is very unclear. I would suggest that the authors improve the illustration of Figure 3 using an actual ocean dynamic problem (eg internal waves, wakes, upwelling, etc.). Since then, the narrative of this section can also be rewritten accordingly.
4. Figure 4: I'm surprised that the parameters in the picture are not defined at all, I can't find the full names of SOD, LOD, TSOD, SO, TSO, LO.
5. Table 1: The reader will be completely incomprehensible about the message this table is trying to present.
6. Figure 5: The same problem occurs again. The description of the figure is too succinct, but the image contains too much information for the reader to understand.
7. Lines 308-324: I had a hard time understanding the message the author was trying to convey. As previously suggested, the authors are asked to give a "practical case of ocean dynamics (eg internal waves, wakes, upwelling, etc." to explain.
8. Table 2: The author does not explain the nouns within the chart at all. What does Warm process object mean? What is Cold process? What exactly does Type I to Type V mean?
9. Lines 346-356: Please rewrite, I can't understand the content.
10. Figure 6: Why is this Type I? How are these arrows calculated? Why are they defined as warm and cold?
11. Table 3: This table refers to Type II events? But what are Type II? How are these years and durations calculated? And what are these IDs?
12. Lines 370-402, Figure 7 to Figure 13: I can't understand what the author is trying to tell the reader.
13. Introduction and Related works need to be combined into one chapter. Authors need to think about what message they want to convey to their readers. What kind of people the author expects to cite this article. Introduction needs to be completely rewritten, the current structure and storytelling is terrible.
14. Looking back at the title, I did not find any "remote sensing images" in this manuscript, only "remote sensing data"
Author Response
Response to Reviewer #2:
Comments 1: Lines 147-152: These two definitions are not clear, please give a detailed case.
Reply: Thank for the reviewer’s comments. The revised manuscript gives detailed cases using an oceanic eddy.
Comments 2: Lines 164-169, Figure 1 and 2: The Caption is too short and the description is too concise. These two important architecture diagrams should be written in detail for readers to understand.
Reply: Thank for the reviewer’s comments. The revised manuscript rewrites the Caption and related works about Figure 1 and 2.
Comments 3: Section 3.2: The description is very unclear. I would suggest that the authors improve the illustration of Figure 3 using an actual ocean dynamic problem (eg internal waves, wakes, upwelling, etc.). Since then, the narrative of this section can also be rewritten accordingly.
Reply: We are sorry for make it unclear to describe the workflow of exploring evolutionary structures of oceanic dynamics. We rewrite the Section 3.2.
We appreciate the reviewer’s comments. However, Figure 3 tries to give the generally technological workflow of exploring evolutionary structures of oceanic dynamics, not a specified ocean dynamic object or phenomenon. Thus, we don’t redraw Figure 3 using an actual ocean dynamic problem.
Comments 4: Figure 4: I'm surprised that the parameters in the picture are not defined at all, I can't find the full names of SOD, LOD, TSOD, SO, TSO, LO.
Reply: Thank for the reviewer’s comments. Figure 5 (Figure 4 in original manuscript) caption gives the full name of the SOD, LOD, TSOD, SO, TSO, LO. Also, the revised manuscript describes these abbreviations in details from Step1 to Step 7.
Comments 5: Table 1: The reader will be completely incomprehensible about the message this table is trying to present.
Reply: Thank for the reviewer’s comments. The title of Table 1 is rewritten.
Comments 6: Figure 5: The same problem occurs again. The description of the figure is too succinct, but the image contains too much information for the reader to understand.
Reply: The reviewer is right. Figure 6 gives too much information, and the title is succinct. The revised manuscript rewrites the caption of Figure 6, and adds legend to represent nodes and relationships. Also, the descriptions of Evolutionary structural type I to Type V under Figure 6 give details of the Figure 6.
Comments 7: Lines 308-324: I had a hard time understanding the message the author was trying to convey. As previously suggested, the authors are asked to give a "practical case of ocean dynamics (eg internal waves, wakes, upwelling, etc." to explain.
Reply: We are sorry to make it unclear information about the descriptions of Evolutionary structural type I to Type V. Lines316-332 in original manuscript try to give the explanations in detail about Evolutionary structural type I to Type V. Figure 6 gives the sketch of evolutionary relationships, and Lines 316-332 give their detailed explanations. Thus, we should read the Figure 6 and the descriptions from 316-332 as a whole.
We appreciate the reviewer’s comments. And Figure 6 caption is rewritten. Taking into account of the page limitation, we would not give a practical case of ocean dynamics as an example in the revised manuscript. There are three reasons, one is that the process object evolves a specified time duration, and it would take much more pages/words to represent its evolutionary structure, the second is that we need a real dataset to prove our algorithm, which also includes real evolutionary structure need to be addressed and discussed. Finally, Figure 8 to Figure 15 in Section Case studies give the real evolutionary structure of one process object of SSTA, which proves our algorithm, and addresses evolutionary structural Type II in details as well.
Comments 8: Table 2: The author does not explain the nouns within the chart at all. What does Warm process object mean? What is Cold process? What exactly does Type I to Type V mean?
Reply: Warm process indicates that process object of SSTA, in which the sea surface temperature is higher than the average temperature, and the cold process indicates that the sea surface temperature is lower than the average temperature, which have been addressed in the first paragraph of Section 3.2.
Type I to Type V indicate five types of evolutionary structures of process object, which have also been defined in Section 2.5.
Comments 9: Lines 346-356: Please rewrite, I can't understand the content.
Reply: Thank for the reviewer’s comments. The revised manuscript rewrites the Section 3.2, including Table 2, 3 and Figure 7.
Comments 10: Figure 6: Why is this Type I? How are these arrows calculated? Why are they defined as warm and cold?
Reply: Thank for the reviewer’s comments. Figure7 only shows the Type I, the main reason is Type I structure is the simplest and the number is the largest (85%). We think the Type I structure is representative for illustration.
An arrow in Figure 7 means a moving direction from origination to dissipation of SSTA dynamic. And the definition of warm or cold process object only considers the value of SST. Generally, the description of SST is warm or cold.
Comments 11: Table 3: This table refers to Type II events? But what are Type II? How are these years and durations calculated? And what are these IDs?
Reply: We are sorry make it unclear for understanding the Table 3. Table 3 only gives an example of interrelationships between Type II of SSTA dynamics with ENSO events. The duration of process object of SSTA is calculated using the originating time and the dissipating time of SSTA. And the IDs means that the object identifier in database. All these descriptions are rewritten in the revised manuscript.
Comments 12: Lines 370-402, Figure 7 to Figure 13: I can't understand what the author is trying to tell the reader.
Reply: We are sorry to make it misunderstand Figure 8 to Figure 14. Figure 8 to Figure 14 show the details of one process object of SSTA, which comes from the real dataset. We try to use Figure 8 to Figure 14 to illustrate two objectives. One is to prove our proposed algorithm, PoEXEs, how to explore evolutionary structure of SSTA dynamics from remote sensing dataset, and the other is to show the evolutionary structure as an example in details, which includes marine snapshot objects, sequence objects, linked objects, and their evolutionary relationships (Figure 15). The second objective tries to answer the reviewer’s Comment 7.
Comments 13: Introduction and Related works need to be combined into one chapter. Authors need to think about what message they want to convey to their readers. What kind of people the author expects to cite this article. Introduction needs to be completely rewritten, the current structure and storytelling is terrible.
Reply: Thank for the reviewer’s comments. This comment is invaluable and helps to improve the quality of our manuscript. The revised manuscript combines Introduction and Related works sections, and rewrites them.
Comments 14: Looking back at the title, I did not find any "remote sensing images" in this manuscript, only "remote sensing data"
Reply: The reviewer is right. This paper focuses on using the remote sensing dataset. According to the reviewer’s suggestion, the title changes to “A process-oriented exploration of the evolutionary structures of ocean dynamics with time series of a remote sensing dataset”.
Reviewer 3 Report
The authors present a method to track and “causally” connect ocean surface anomalies, in order to describe evolutionary processes. Their method is based on the clustering algorithm DcSTCA, introduced by some of the authors in reference [22], and the SLOA graph construction algorithm, which connects the previously identified structures into a single- or multiple-linked evolutionary graph, showing the development, merging, splitting and split-merging processes. This leads to a classification of evolutionary structures, based solely on graph characterization. The manuscript finishes with an easy to see example, studying the El Niño/La Niña system (1+2 area, next to the Peruvian coast), and showing its graph (and spatial!) evolution in the period 1997-1998. The manuscript follows the publication path of references [7], [9], [10], [22], etc. all aimed at the automatic identification of graph-process evolution. A general discusion of these previous approaches to the process-oriented spatiotemporal clustering of complex trajectories would benefit greatly the manuscript, and show more clearly its novelty compared to the previous approaches. I encourage the authors to include this review in a small section prior to the conclusions. Another improvement that would highlight the potential of this method would be including, after the ENSO case study, another less known (or less evident) spatiotemporal evolutionary process, as the ENSO has been so prolifically described in the literature. The manuscript is well structured and contains a potentially novel methodology (although it is far from being self-contained, as the algorithms are either referenced to the bibliography or sketched in words, with no pseudocode). And after including some of the iprovements suggested before, and after a thorough revision of the English writing, it will be ready for publication in Remote Sensing, in my opinion. In the following lines I include some corrections (English and otherwise), suggestions and questions to the authors: L13 (abstract): Advanced Earth *observation* technologies provide a *tool for the study of ocean* dynamics *either* in a basin or *in the* global ocean. L14-15 (abstract): how does *ocean dynamics* evolve in space and time *is* still a challenge. L16 (abstract): time series of *raster datasets*/*a raster dataset* L41: plays significant roles on *regional* and global climate changes L61: there may exist too many evolutionary relationships, *vice versa*, some key evolutionary relationships will be lost. L62: the abovementioned methods *consist* in obtaining the evolutionary relationships L105: overlapping threshold-based method is widely used for *tracking* a rainstorm L121: the observation scale-based methods will generate too many or too *few* evolutionary relationships L130: *From* an event perspective, an event-tracking method based on … L138: *Definitions* (plural) L140: Please, clarify the meaning of “this paper takes an evolutionary scale of such dynamic to design PoEXES”; what does “evolutionary scale” mean here? Also in L425, in the conclusions, where it says: “PoEXES was designed at a scale of dynamic evolution to explore evolutionary structures of oceanic dynamics”. L141: some concepts about *ocean* dynamics and evolution *need* to be addressed from the perspective of algorithm *design*. L195-197: Please, define “DcSTCA” before using the acronym, provide reference (reference [22]), and clarify how it works, and how it fits within the PoEXES workflow. L198-199: “DcSTCA performs well on spatiotemporal clustering patterns of SSTA in Pacific Ocean”; please, clarify this statement: according to [22] DcSTCA detects known spatiotemporal SSTA patterns in Pacific Ocean associated with ENSO, and also other unreported patterns wich share the same properties. All other good properties of DcSTCA are simply “more reasonable” spatiotemporal patterns “than those obtained from ST-DBSCAN”. L215: Please, specify the threshold used on DOA; is it the same threshold for both DOA<t,t+1> and DOA<t,t-1>? L224: the identification of marine sequence objects and marine linked objects is as *follows:* L226-230 (Step 1): Please, clarify the notation; what is the difference between ONt and Oit, etc. L248: Please, define here the labels “Meet” and “MetBy” used below (in L277 and ff.). L291: “As known, where and when the oceanic dynamic generates and disappears play significant roles in reginal and global climate changes …”; please, provide references reporting this important role of spatiotemporal patterns for the study of climate change. L320: Please, clarify what is meant here (and in the next paragraph) by “ring pattern”. L327-328: *The* SST remote sensing dataset *used for case study is the NOAA Optimum Interpolation Sea Surface Temperature V2.0 provided by the 329 NOAA/OAR/ESRL Physical Sciences Division, Boulder, Colorado, USA, and available at 330 http://www.esrl.noaa.gov/psd/ [45]*. *The time period ranges* from January 1982 to December 2021 with a spatial resolution of 1° and a temporal resolution of 1 month. *The standard monthly average anomaly* … L341: According to the SSTA value *bein* greater than zero or not Figures 7-13: Please, include coast-lines for clarity, and make coordinate labels visible (they are currently too small). Also, include the legend appearing in Fig. 11 at the begining of the figure series. What does “Marine state object” mean here? (white color corresponds to continents in theese maps!). Figures 14 (a-b): Please, use larger size typography in the legends or, better, improve image resolution; if possible, use a vector format. L425 and ff.: This point would benefit from a review and discussion of previous methods such as DcSTCA, PoSCM, PoAIES, PoTGM, etc. all of them aimed at finding the same graph pattern evolution descriptions. L433 and ff.: It should be clarified that this discussion/conclusion about the SLOA structure classification (into Types I, II, III, IV and V) not only refers to the theoretically possible structures, but also to those detected when analyzing the data (please, refer to Table 2). L441 and ff.: This conclusion is not fully supported by the results presented in the manuscript. Section 4.2 only relates Type II evolutionary structures with ENSO, not with global climate change, and says nothing about Type III and Type V structures (which are only reported in Table 2). A Type IV structure is described as the very ENSO spatiotemporal pattern during the two-year period 1997-1998. L475 (reference 4): There is a final “2” missing in the DOI; it should be: DOI:10.1145/3161602
Author Response
Response to Reviewer #3:
Comments 1: The authors present a method to track and “causally” connect ocean surface anomalies, in order to describe evolutionary processes. Their method is based on the clustering algorithm DcSTCA, introduced by some of the authors in reference [22], and the SLOA graph construction algorithm, which connects the previously identified structures into a single- or multiple-linked evolutionary graph, showing the development, merging, splitting and split-merging processes. This leads to a classification of evolutionary structures, based solely on graph characterization. The manuscript finishes with an easy to see example, studying the El Niño/La Niña system (1+2 area, next to the Peruvian coast), and showing its graph (and spatial!) evolution in the period 1997-1998. The manuscript follows the publication path of references [7], [9], [10], [22], etc. all aimed at the automatic identification of graph-process evolution. A general discusion of these previous approaches to the process-oriented spatiotemporal clustering of complex trajectories would benefit greatly the manuscript, and show more clearly its novelty compared to the previous approaches. I encourage the authors to include this review in a small section prior to the conclusions.
Reply: Thank for the reviewer’s comments. Your suggestions help us more to improve the quality of our manuscript. In Conclusion Section, we add the first paragraph to discuss our previous approaches and the relationship with the core idea of this paper, which clearly shows its novelty compared to the previous approaches.
Comments 2: Another improvement that would highlight the potential of this method would be including, after the ENSO case study, another less known (or less evident) spatiotemporal evolutionary process, as the ENSO has been so prolifically described in the literature.
Reply: Thank for the reviewer’s comments. The reviewer is right. This manuscript has obtained five types of evolutionary structure, and only Type II are discussed in details, and analyzed the relationships with ENSO. Exception of ENSO, the evolutionary structure of SSTA dynamics also closely relate to other evolutionary process, e.g. IOD (Indian Ocean Dipole), however, this is less known, we need further study. Even though, we still appreciate the reviewer for your suggestions, and the we discuss them as a new further study in Conclusion Section.
Comments 3: The manuscript is well structured and contains a potentially novel methodology (although it is far from being self-contained, as the algorithms are either referenced to the bibliography or sketched in words, with no pseudocode). And after including some of the iprovements suggested before, and after a thorough revision of the English writing, it will be ready for publication in Remote Sensing, in my opinion.
Reply: Thank for the reviewer’s comments. Your suggestions are invaluable for improving our manuscript. We deal with each suggestion and revise our manuscript seriously in order to gain you and the chief-in-editor approval. Regarding to English writing, this manuscript has been reviewed by a native-English-speaking scientific editor.
Comments 4: In the following lines I include some corrections (English and otherwise), suggestions and questions to the authors:
L13 (abstract): Advanced Earth *observation* technologies provide a *tool for the study of ocean* dynamics *either* in a basin or *in the* global ocean.
L14-15 (abstract): how does *ocean dynamics* evolve in space and time *is* still a challenge.
L16 (abstract): time series of *raster datasets*/*a raster dataset*.
L41: plays significant roles on *regional* and global climate changes.
L61: there may exist too many evolutionary relationships, *vice versa*, some key evolutionary relationships will be lost.
L62: the abovementioned methods *consist* in obtaining the evolutionary relationships.
L105: overlapping threshold-based method is widely used for *tracking* a rainstorm.
L121: the observation scale-based methods will generate too many or too *few* evolutionary relationships.
L130: *From* an event perspective, an event-tracking method based on …
L138: *Definitions* (plural).
L141: some concepts about *ocean* dynamics and evolution *need* to be addressed from the perspective of algorithm *design*.
L224: the identification of marine sequence objects and marine linked objects is as *follows:*.
L327-328: *The* SST remote sensing dataset *used for case study is the NOAA Optimum Interpolation Sea Surface Temperature V2.0 provided by the 329 NOAA/OAR/ESRL Physical Sciences Division, Boulder, Colorado, USA, and available at 330 http://www.esrl.noaa.gov/psd/ [45]*. *The time period ranges* from January 1982 to December 2021 with a spatial resolution of 1° and a temporal resolution of 1 month. *The standard monthly average anomaly* …
L341: According to the SSTA value *bein* greater than zero or not.
L475 (reference 4): There is a final “2” missing in the DOI; it should be: DOI:10.1145/3161602
Reply: Thank for the reviewer’s careful comments. All the comments have been corrected.
Comments 5: L140: Please, clarify the meaning of “this paper takes an evolutionary scale of such dynamic to design PoEXES”; what does “evolutionary scale” mean here? Also in L425, in the conclusions, where it says: “PoEXES was designed at a scale of dynamic evolution to explore evolutionary structures of oceanic dynamics”.
Reply: Thank for the reviewer’s comments. The evolutionary scales in Line 122 and Line 450 have the same meaning. For clearly address this concept and its meaning, Section 2.1 gives its definition in detail.
Comments 6: L195-197: Please, define “DcSTCA” before using the acronym, provide reference (reference [22]), and clarify how it works, and how it fits within the PoEXES workflow.
Reply: Thank for the reviewer’s comments. The revised manuscript describes the workflow of DsSTCA in details.
Comments 7: L198-199: “DcSTCA performs well on spatiotemporal clustering patterns of SSTA in Pacific Ocean”; please, clarify this statement: according to [22] DcSTCA detects known spatiotemporal SSTA patterns in Pacific Ocean associated with ENSO, and also other unreported patterns wich share the same properties. All other good properties of DcSTCA are simply “more reasonable” spatiotemporal patterns “than those obtained from ST-DBSCAN”.
Reply: Thank for the reviewer’s comments. The comment has been addressed.
Comments 8: L215: Please, specify the threshold used on DOA; is it the same threshold for both DOA<t,t+1> and DOA<t,t-1>?
Reply: The reviewer is right. The thresholds of DOA<t,t+1> and DOA<t,t-1> are the same. Generally, the threshold is an empirical value, and the revised manuscript address it.
Comments 9: L226-230 (Step 1): Please, clarify the notation; what is the difference between and , etc.
Reply: Thank for the reviewer’s comments. is a dataset of snapshot objects, which contains N snapshot objects, , , … . in stands for 1,2,3... n.
Comments 10:L248: Please, define here the labels “Meet” and “MetBy” used below (in L277 and ff.).
Reply: Thank for the reviewer’s comments. “Meet” and “MetBy” are the temporal topologies meaning that two snapshot objects are continuous in time. These definitions and the spatial topology of Intersection are addressed in revised manuscript.
Comments 11:L291: “As known, where and when the oceanic dynamic generates and disappears play significant roles in reginal and global climate changes …”; please, provide references reporting this important role of spatiotemporal patterns for the study of climate change.
Reply: Thank for the reviewer’s comments. The references are appended.
Comments 12:L320: Please, clarify what is meant here (and in the next paragraph) by “ring pattern”.
Reply: Thank for the reviewer’s comments. It is our carelessness. The correct should be that it is a ring structure not a ring pattern. The revised manuscript rewrites these sentences.
Comments 13: Figures 7-13: Please, include coast-lines for clarity, and make coordinate labels visible (they are currently too small). Also, include the legend appearing in Fig. 11 at the begining of the figure series. What does “Marine state object” mean here? (white color corresponds to continents in theese maps!).
Reply: Thank for the reviewer’s comments. As the spatial resolution of remote sensing dataset is coarse, the coast-lines are not suitable for overlapping these maps. Beside of this issue, all the comments have been corrected in the revised manuscript including coordinate labels, legend, and Marine state object. Also, the original pictures of Figures 8-14 clearly show the coordinate labels. Once this manuscript has been accepted, all the original pictures are uploaded.
Comments 14:Figures 14 (a-b): Please, use larger size typography in the legends or, better, improve image resolution; if possible, use a vector format.
Reply: Thank for the reviewer’s comments. The revised manuscript replaces this Figure with the original figure, drawn by Visio. Once this manuscript has been accepted, all the original pictures are uploaded.
Comments 15: L425 and ff.: This point would benefit from a review and discussion of previous methods such as DcSTCA, PoSCM, PoAIES, PoTGM, etc. all of them aimed at finding the same graph pattern evolution descriptions.
Reply: This comment is invaluable for improving the quality of this manuscript, thank you very much. We append a new paragraph in Section Conclusion to address and discuss them in details.
Comments 16: L433 and ff.: It should be clarified that this discussion/conclusion about the SLOA structure classification (into Types I, II, III, IV and V) not only refers to the theoretically possible structures, but also to those detected when analyzing the data (please, refer to Table 2).
Reply: Thank for the reviewer’s comments. The revised manuscript discusses this issue in Conclusion Section.
Comments 17:L441 and ff.: This conclusion is not fully supported by the results presented in the manuscript. Section 4.2 only relates Type II evolutionary structures with ENSO, not with global climate change, and says nothing about Type III and Type V structures (which are only reported in Table 2). A Type IV structure is described as the very ENSO spatiotemporal pattern during the two-year period 1997-1998.
Reply: The reviewer’ comments are right that Section 4.2 only relates Type II evolutionary structures with ENSO, not with global climate change, and says nothing about Type III and Type V structures. And Section 4.3 addresses Type IV structure as the very ENSO spatiotemporal pattern during the two-year period 1997-1998.
Generally, this manuscript takes Type I, Type II and Type IV to prove our proposed algorithm, PoEXEs. The main reasons are as follows.
1) The main motivation of this manuscript is to design a novel algorithm to explore evolutionary structure of ocean dynamics.
2) Just as the Comment 10 addressed that Type 1 to Type V not only refers to the theoretically possible structures, also the real structure of SSTA dynamics. Some of them closely related with ENSO, the typical signal of climate change, are addressed to prove our algorithm. Maybe, ENSO equals to global climate change is not suitable, the revised manuscript also discusses the other signal of climate change, e.g. IOD, PDO etc.
3) Type I, Type II and Type IV are used only to prove our proposed algorithm, PoEXEs. The mechanisms behinds these types are out of our manuscript, at least, in our opinions.
Reviewer 4 Report
General comments
The paper entitled “A process-oriented exploring for evolutionary structures of oceanic dynamics with time series of remote sensing images” treats about a topic of the highest interest in oceanographic studies scope. Authors introduce a new algorithm called PoEXES to perform a time-space analysis of oceanic areas. The algorithm is composed of 3 stages and at the end; five types of evolutionary structure can be gained. The analysis is carried out on remotely sensed images; therefore the manuscript falls under the journal scope without any doubt.
The declared objective of the manuscript is to assess when, where and how important change in oceans take place.
Concerning the manuscript, the typographical outline is satisfactory. The used language is not always fluent apart some typos spread over the manuscript (see the section specific comments for some examples). Abstract is a little bit unclear. Introduction introduces the reader into the treated topic well, clarifying well the aims of the manuscript. Keywords are pertinent to paper content and appropriate. The highlights are missing. Materials and Methods are should be clearer. The References section is rich with recent citations. Graphic representations are not always fine.
Entering in the very merit of the paper is opinion of this reviewer that the manuscript is sufficient and the experimental part is well described and conducted. The manuscript is lacking of comparative analysis to evidence the need of a new method. Anyway, the evidences proved the Authors’ thesis. In conclusion, this reviewer recommends considering the paper after “major revisions”.
Specific comments
Line 15: maybe “is still a challenge” is more appropriate.
Lines 18 – 20: the sentence sounds awkward. Please, improve it.
Line 61: please, replace “verse visa” with an equivalent term.
Etc etc
Author Response
Response to Reviewer #4:
General comments:
Comments 1: The paper entitled “A process-oriented exploring for evolutionary structures of oceanic dynamics with time series of remote sensing images” treats about a topic of the highest interest in oceanographic studies scope. Authors introduce a new algorithm called PoEXES to perform a time-space analysis of oceanic areas. The algorithm is composed of 3 stages and at the end; five types of evolutionary structure can be gained. The analysis is carried out on remotely sensed images; therefore the manuscript falls under the journal scope without any doubt.
The declared objective of the manuscript is to assess when, where and how important change in oceans take place.
Reply: Thank for the reviewer’s significant comments for our manuscript.
Comments 2: Concerning the manuscript, the typographical outline is satisfactory. The used language is not always fluent apart some typos spread over the manuscript (see the section specific comments for some examples). Abstract is a little bit unclear. Introduction introduces the reader into the treated topic well, clarifying well the aims of the manuscript. Keywords are pertinent to paper content and appropriate. The highlights are missing. Materials and Methods are should be clearer. The References section is rich with recent citations. Graphic representations are not always fine.
Reply: Thank for the reviewer’s comments. We reorganize this manuscript by rewriting Abstract, combining the Introduction and Related Works Section into one section Introduction, redrawing some Figures and discussing the highlights. Also, a native-English-speaking scientific editor has reviewed our manuscript.
Comments 3: Entering in the very merit of the paper is opinion of this reviewer that the manuscript is sufficient and the experimental part is well described and conducted. The manuscript is lacking of comparative analysis to evidence the need of a new method. Anyway, the evidences proved the Authors’ thesis. In conclusion, this reviewer recommends considering the paper after “major revisions”.
Reply: Thank for the reviewer’s positive comments.
Specific comments:
Line 15: maybe “is still a challenge” is more appropriate.
Lines 18 – 20: the sentence sounds awkward. Please, improve it.
Line 61: please, replace “verse visa” with an equivalent term.
Etc etc
Reply: Thanks. All the comments are corrected, and a full review of the English of this manuscript was improved by a native-English-speaking scientific editor.
Round 2
Reviewer 2 Report
Thanks to the author for replying to my comments one by one.
Regarding comments 3 and 7, I don't think I got an answer, and the current manuscript has not received substantial revisions with respect to these two important comments.
Why I want the author to give an example of actual ocean dynamics, because it makes it very clear to the reader to understand the real core value of the author's present large number of algorithms, especially the Evolutionary structural type I to Type V is still described very non-specifically.
It is not acceptable for authors to consider page limits without giving practical examples. The authors claim that this is a general technical procedure, so every example of ocean dynamics can use this method regardless of its spatial and temporal resolution?
The biggest problem with this manuscript is that it has no potential readers. In addition, after reading this article, readers have no or new knowledge and worthy of citation, which will make readers wonder why they should read this article. This manuscript may be helpful to the authors themselves, but once it is accepted for publication, it must be helpful to satellite oceanographers and physical oceanographers.
Author Response
Comments 3: Why I want the author to give an example of actual ocean dynamics, because it makes it very clear to the reader to understand the real core value of the author's present large number of algorithms, especially the Evolutionary structural type I to Type V is still described very non-specifically.
It is not acceptable for authors to consider page limits without giving practical examples. The authors claim that this is a general technical procedure, so every example of ocean dynamics can use this method regardless of its spatial and temporal resolution?
Reply: We appreciate and agree to the reviewer’s comment about “an example of actual ocean dynamics makes it very clear to the reader to understand the real core value of the author's present large number of algorithms”. Actually, this manuscript uses Figure 8 to Figure 15 to give the real evolutionary structure of one process object of SSTA. Figure 8 to Figure 15 not only give a real example of marine process, also prove our algorithm. According to this comment, we redefine the definition of “Marine evolutionary process” using the real example of SST-based eddy.
Regarding to Figure 3, the general workflow of our algorithm, we still believe that the workflow is a general technical procedure, so every example of ocean dynamics can use this method regardless of its spatial and temporal resolution. If a specified example is used to show the workflow, which will reduce the scientific merit of this manuscript. Thus, we insist on not redrawing Figure 3 with an actual ocean dynamic example.
Comments 7: The biggest problem with this manuscript is that it has no potential readers. In addition, after reading this article, readers have no or new knowledge and worthy of citation, which will make readers wonder why they should read this article. This manuscript may be helpful to the authors themselves, but once it is accepted for publication, it must be helpful to satellite oceanographers and physical oceanographers.
Reply: We regret that we don’t agree with this comment that this manuscript has no potential readers. The main reason are follows. 1) Advanced Earth observation technologies make it possible to explore ocean dynamics. And in a comparison of when and where, how does ocean dynamics evolve in space and time is still a challenge. 2) PoEXES takes an evolutionary scale to analyze ocean dynamics, which ensures to obtain the optimum evolutionary relationships of ocean dynamics as much as possible. 3) The discovered evolution structure may help better understand global climate change. The second reason play a foundation for dealing with time series of images, which will benefit satellite oceanographers, and the third reason will be helpful for physical oceanographers.
Reviewer 4 Report
Authors made efforts to improve significantly their work. Is opinion of this reviewer that the present manuscript can be accepted for publication.
Author Response
We would give our faithful appreciations for our significant comments. Your suggestions are invaluable for improving the quality of our manuscript.
Regarding to the English language and style, we invited a native-English-speaking scientific editor to review this manuscript.
