A Pre-Pruning Quadtree Isolation Method with Changing Threshold for ICESat-2 Bathymetric Photon Extraction
Round 1
Reviewer 1 Report
Interesting paper on a very popular topic. A quick search in Google scholar yields many papers published over the past 4 years. The authors present their take on how one might re-interpret ICESat2 observations to generate near-shore bathymetry. The methodology seems reasonable and the results are better than the one dataset they compared to. The authors note that "there is no validation data available for the study area" which may be true but if that is the case then they need to put their results in context with the many other algorithms that have already been published rather than just the one method that they did reference. There is also little discussion of the limitations of the method. It seems that all methods that rely on light penetrating the water will be subject to water conditions and clarity in terms of their effectiveness. There needs to be some explanation of where the authors feel that this method might be applicable and where it likely wouldn't. Since there are so many other works in this area more work needs to be done to place this work in the context of the other methods that are available. Perhaps start with the ATL13 product which is produced by the ICESat2 mission and generates bathymetry data. If they have data over the study area it should be straightforward to do a comparison.
Line Comment
103 "Coastal"
114 The figure is not very effective because it doesn't show the tracks over the actual study site. Or else it is a very confusing figure that I don't understand. If the bottom right is supposed to give overall context then it needs to be separated from the rest of the image with a different background or something because it looks like all part of one image as presented.
217 should be e1lb, correct?
254 unclear, how are you "visually" determining the depth of the water using the photon? As stated it sounds like you are, somehow, getting the "in situ" depth by looking at the photon which clearly isn't possible. I assume you mean that you are looking at the graph but that doesn't come through in the wording here.
401 line 400 and 401 both sentences start with "Compared with surface water photons". Maybe try to find a way to vary the language to avoid verbatim repetition.
425 repetition again "namely 0-10 m and 0-10 m"
Author Response
- The authors note that "there is no validation data available for the study area" which may be true but if that is the case, then they need to put their results in context with the many other algorithms that have already been published rather than just the one method that they did reference.
Answer: Thank you for your question. I hope to answer your question from two aspects.
Firstly, using visually interpreted bathymetric photons is the most accurate method so far. In reference [1], this method was first introduced into the performance verification of the noise removal algorithm. Visual interpretation is considered to be equally effective in verifying coastal environmental data.
Secondly, we have supplemented CONF and AMNP according to your requirements. AVEBM and AMNP, respectively, correspond to two different ideas for improving DBSCAN: increasing neighborhood area and reducing density threshold. As far as we know, this paper is also the first to compare the two ideas' advantages and disadvantages in related research.
- There is also little discussion of the limitations of the method. It seems that all methods that rely on light penetrating the water will be subject to water conditions and clarity in terms of their effectiveness. There needs to be some explanation of where the authors feel that this method might be applicable and where it likely wouldn't.
Answer: Thank you for your advice. The quality of water affects the ability of photons to penetrate water. The results in Figure. 6 and 7 show that when the water is more turbid, the backscattering effect of the water is more prominent, the laser pulse decays faster, and the extraction results become more inaccurate. This will lead to a conclusion that even if the laser pulse cannot reach the same depth in a relatively turbid water body as in a clear water body, it is more challenging to extract sounding photons in a turbid water body because of the faster attenuation of laser pulse energy and more backscattering noise.
- Since there are so many other works in this area, more work needs to be done to place this work in the context of the other methods that are available.
Answer: Thank you for your advice. We supplemented the related results in the experiment and discussion.
- Perhaps start with the ATL13 product which is produced by the ICESat2 mission and generates bathymetry data. If they have data over the study area it should be straightforward to do a comparison.
Answer: You are right, but you may be referring to ATL12 ocean products, not ATL13 inland aquatic products. However, considering the vast difference in resolution between the ATL12 product and ATL03, we did not directly add the ATL12 product to the results. However, the official confidence label of the sea surface signal is provided in ATL03. The sea surface photons labeled as signals are the basis for the production of ATL12 product, so we directly add the photons with official confidence labels greater than or equal to 2 to compare the results. The results show that although the official confidence label may perform well on a larger scale, its performance must only be strengthened when extracting nearshore bathymetric photons.
103 "Coastal"
Answer: Thank you for your comments, which have been revised as required.
114 The figure is not very effective because it doesn't show the tracks over the actual study site. Or else it is a very confusing figure that I don't understand. If the bottom right is supposed to give overall context, then it needs to be separated from the rest of the image with a different background or something because it looks like all part of one image as presented.
Answer: Thank you for your advice. We had hoped to put the legend in the image to make readers feel more immersed but ignored whether the legend was clear or not. According to your suggestion, we have moved the legend and other elements outside the image to ensure that all readers can understand the contents of the figure immediately.
217 should be e1lb, correct?
Answer: Thank you for your comments, which have been revised as required.
254 unclear, how are you "visually" determining the depth of the water using the photon? As stated, it sounds like you are, somehow, getting the "in situ" depth by looking at the photon which clearly isn't possible. I assume you mean that you are looking at the graph but that doesn't come through in the wording here.
Answer: Thank you for your comments, which have been revised as required.
401 line 400 and 401 both sentences start with "Compared with surface water photons". Maybe try to find a way to vary the language to avoid verbatim repetition.
Answer: Thank you for your comments, which have been revised as required.
425 repetition again "namely 0-10 m and 0-10 m"
Answer: Thank you for your comments, which have been revised as required.
Reviewer 2 Report
Please see attachment
Comments for author File: 
Comments.pdf
Author Response
Technical issues:
- Due to the large density and obvious characteristics of water surface photons, I think it is possible to remove water surface photons by histogram method, and then perform PQI on the remaining photons. This operation may have high accuracy and efficiency. What do the authors think? (JUST DISCUSS IT WITH THE AUTHORS)
Answer: You are quite right. When underwater bathymetric photons are known to exist, it is more beneficial to remove noise photons by distinguishing sea surface photons from seabed photons first by the histogram method. However, in most cases, researchers need to gain knowledge of the coastal environment (including water quality, data, etc.), and underwater bathymetric photons do not always exist. We treat the data obtained in the coastal environment as a whole and realize noise removal through the designed workflow, regardless of whether there are underwater bathymetric photons. It is more effective when dealing with massive space-borne data.
- What are the residual noise photons of this method, where is the residual noise photons located, and how much impact on underwater terrain inversion? (JUST DISCUSS IT WITH THE AUTHORS)
Answer: Your question needs to be considered, but it may need to be more relevant to this manuscript’s noise removal content. According to your suggestion, we will add these discussions in future research on photon refraction correction and water depth inversion to make the article's content fuller.
- ATLAS is very sensitive to the signal reflected by the water. Some studies have also obtained high-precision water parameters using data from weak beam channels. Does the author test the data under the weak beam channel?
Answer: The performance of weak beam channels does not be discussed in this manuscript. This is mainly because this paper hopes to verify the effectiveness of the proposed method first, then compare the performance difference with the proposed method, and try to analyze the factors that affect the quadtree isolation method. As you said, we have considered adding the comparison of the extraction effect of solid and weak beams, but this will make the manuscript's content more like a comparison than an introduction to innovative methods. Considering the manuscript's length, we hope to discuss the difference in signal extraction performance between solid and weak orbits in future research.
- How is the operation efficiency of this method?
Answer: When the quadtree isolation algorithm was first proposed, we verified the efficiency of the algorithm. The results show that when the data is divided into 20m segments, QI is more efficient than DBSCAN. This is because the number of quad space segmentation in the 20m data segment is less, and the DBSCAN filter kernel needs to repeatedly calculate the photon density in different directions to achieve directional adaptability.
Xie Huan of Tongji University thinks that increasing the length of the data segment properly will make quadtree isolation have better performance. This statement is true after our verification, but it will lead to an exponential increase in the number of spatial divisions and a decline in the algorithm's efficiency.
In this paper, we have not calculated the algorithm's efficiency because the PQI algorithm is available. In addition, as far as our operational experience is concerned, it is possible to divide the data of the whole scene into 500m, 1km, or 2km data segments as needed, and the algorithm performance and processing efficiency can satisfy users at this time. Directing tens of kilometers of data could be more efficient and efficient.
specific suggestions:
- Abstract writing needs improvement. In the abstract, the author should use objective numbers to describe the experimental results, rather than directly commenting on his own results.
Answer: Thank you for your comments, which have been revised as required.
- Table 1. Where is the data source in Table 1?
Answer: Thank you for your question. The acquisition date and reference ground track in the table can be queried in OpenAltimetry, while the maximum detection depth is obtained by visual interpretation. In order to distinguish the data from the photon density, we also calculated the average photon density of the data according to the reference [1] so that we can understand the algorithm's performance when dealing with data with different densities.
- The flowchart (Figure 2) seems too simple, please improve it.
Answer: Your suggestion is necessary to improve the quality of the paper. Considering that the workflow in this paper does not involve complex logical judgment, the flow chart is modified into a double-column structure. The critical steps of the process are listed on the left side of the newly revised flowchart, while on the right side of the flowchart, significant results, including pre-pruned quadtree, are displayed. Doing so can make readers feel and understand the processing flow more intuitively without knowing the technical details.
- 3.3 performance verification. The earliest use of visual interpretation should not be literature 17, please check and correct.
Answer: Thank you for your advice. We supplemented the literature [2] in the manuscript. If you know the earlier literature, please tell us. Thanks.
- There are some grammatical errors in the manuscript that should be carefully checked and improved.
Answer: Thank you for your suggestion. The grammar and expression of the article have been re-checked.
Reference
- Chen, Y.; Le, Y.; Zhang, D.; Wang, Y.; Qiu, Z.; Wang, L. A photon-counting LiDAR bathymetric method based on adaptive variable ellipse filtering. Remote Sensing of Environment 2021, 256, doi:10.1016/j.rse.2021.112326.
- Chen, B.; Pang, Y.; Li, Z.; Lu, H.; Liu, L.; North, P.R.J.; Rosette, J.A.B. Ground and Top of Canopy Extraction From Photon-Counting LiDAR Data Using Local Outlier Factor With Ellipse Searching Area. Ieee Geoscience and Remote Sensing Letters 2019, 16, 1447-1451, doi:10.1109/lgrs.2019.2899011.
Round 2
Reviewer 2 Report
My concerns have been solved. This work is quite good. It can be published as is.
