Performance and Uncertainty of Satellite-Derived Bathymetry Empirical Approaches in an Energetic Coastal Environment
Round 1
Reviewer 1 Report
Summary
Metrics used to describe the SDB uncertainty are most often computed for all data without differentiating between the different ranges of depth and without verifying the normality assumption for error distribution. In addition, uncertainty associated to SDB is a key information for data interpretation, that must be quantified when interpreting bathymetric changes. Therefore, the work of this paper is of great significance.
This study aimed to evaluate the performance and robustness of well-established SDB empirical models and to propose and validate an uncertainty model based on inferential statistics and using a multi-scene approach for a mixed energy coastal environment. The structure is logical, the figures are of good quality and the historical background has given credit as is relatively appropriate. One note that the matching and gridding method for satellite retrievals and in-situ data should be more clarified, because the time range and spatial resolution is not the same, which is the basis of evaluation. No good matching strategy means that estimates cannot be trusted. The Arcachon lagoon inlet is characterized by a high spatial and temporal variability of water column optical properties mainly controlled by hydrodynamics and seasonal conditions, which means the inshore bathymetry changes rapidly. Hence the effect of inshore bathymetry changes should be more clarified. Overall, I think this manuscript can be considered if the author could adequately address the comments below.
Specific/Detailed Comments
- Section 2.2: it’s better to add a table to detailed describing the ground reference bathymetry data, including the date, coverage, depth range and so on.
- Section 2.3: a table is needed to describe the satellite data.
- The matching and gridding method for satellite retrievals and in-situ data should be more clarified, which is the basis of evaluation.
- The Arcachon lagoon inlet is characterized by a high spatial and temporal variability of water column optical properties mainly controlled by hydrodynamics and seasonal conditions, which means the inshore bathymetry changes rapidly. The effect of inshore bathymetry variation (from river input and the other environment changes and so on…) on the proposed multi-scene approach need more clarified.
- SDB methods need reference bathymetry data, how the reference bathymetry data was used for your method, you should describe.
- Figure 4: lack the labels for the ticks.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
for the more frequent development in the coastal area during recent years, the author proposes an effective method to monitor the spatio-temporal morpho-dynamics change of the sensitive tidal zone. Besides, the whole procession of the building of the model is very clear and rigorous, including the selection of the methods to build the model, the organization of the dataset, and especially the final validation of the model. Moreover, despite the remote sensing data used in the process, there is reliable situ data to calibrate and validate the model. This totally raises its accuracy.
there are some problems existing in the paper. First, of the high dependency of the optical properties of the research goal, the results of the Atmospheric correction are supposed to be displayed in the paper. I suggest that the author should pay more attention to this procession. Second, the selection of the sensitive bands should be more specific i.e. the computation of the sensitivity of all bands in LAT 8 and Sentinal-1. Third, the procession of the classifying of the CBR method needs to be more persuasive like how to affirm whether there is hydro-sedimentary or not. Last of all, I haven’t noticed enough seasonal factors in the building and validation of the model, which is supposed to be one of the most important factors in the study. I suggest that the author should write more about the seasonal differences in the paper.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
Lubac et al. use the sandy Arcachon Bay inlet (SW France) as a case study to evaluate the accuracy and robustness of different algorithms used for extracting water depth from satellite imagery. They bring together an impressive dataset consisting of nearly 1 million echosounder points (in situ water depth measurements) and 89 satellite images. Next, they apply three widely-used approaches—the linear regression model (LGM), switching model (SM), and cluster based regression (CBR) model—to estimate water depth from the satellite imagery. Lubac et al. show that the CBR algorithm performs the best in their study area. Finally, Lubac et al. demonstrate how to use their stack of satellite images to estimate the total uncertainty associated with each pixelwise depth estimate. My only major point of confusion from the manuscript is that I was left wondering how Lubac et al. handled the changing tidal cycle when they created their dataset of echosounder points and satellite images. Since the tidal amplitude at Arcachon Bay is as large as the differences in water depth between different parts of their study area (see Figures 1-2), it becomes essential that the tides are properly accounted for. So, for example, each of the 915,326 echosounder points was acquired at a particular phase in the tidal cycle. And that phase changed over the course of each echosounder survey. Did the authors use a nearby tide gauge to correct each echosounder depth point to the LAT datum (or any constant datum)? Likewise, the Landsat and Sentinel satellite image acquisitions are labeled as being from low tide, high tide, etc. (Table 1), but it wasn’t clear whether the quantitative metric of meters above LAT was tagged with each satellite image (using the timestamp of the image acquisition and the tide reading from a nearby tide gauge). If each satellite image is not corrected for its phase in the tidal cycle, then it seems a considerable amount of spread in the water depths extracted from the multi-scene stack will come from this tidal signal, rather than an intrinsic aspect of the SDB algorithm. For an example of how to handle tides in multi-scene image stacks, you might refer to Geyman & Maloof (2020) “Deriving tidal structure from satellite image time series.” Overall, I think this manuscript represents careful work, and I recommend publication after major revisions. Finally, I hope the specific comments and technical edits in the following sections will help the authors improve clarity in the manuscript.
Specific comments
Lines 78-79: “It is expressed as positive value and assume that error follows a normal distribution.” -> Be careful here, since the “uncertainty” as a statistical parameter does not necessarily assume that the error follows a normal distribution. It is fine for you to define it that way here, but make it clear that normality is not an intrinsic or fundamental aspect of the concept of uncertainty. You could re-phrase the text to say: “Here, we define uncertainty as the statistical parameter “characterizing the range of values within which the true value of a measurement is expected to lie as defined within a particular confidence level” [25]. We express the uncertainty as a positive value and assume that error follows a normal distribution.”
Line 58: maybe you can add a little more detail to help the reader understand this sentence better. What specifically do you mean by “mixed energy environments”? Do you mean mixed wave and tidal current energy?
Lines 109-111: What do you mean by “fraction” in this sentence? Perhaps the authors can rephrase this sentence since it is difficult to figure out what the intended meaning is.
Lines 146: “Accuracy on Zsitu is assumed less than 15 cm.” -> This might be interpreted by some readers to mean that the Z_situ measurements are less accurate than 15 cm. Perhaps you should rephrase to say, “Accuracy on Zsitu is assumed to be within 15 cm.”
Lines 205-206: Mis-citation of the cluster-based regression model – should be [23] (Geyman & Maloof, 2019) rather than [22] (Caballero & Stumpf, 2020).
Line 252: incorrect reference—should be [23], not [22]
Lines 252-254: “The choice of an unsupervised learning technique is motivated by their relevance for classification of optically contrasted environments” -> what do you mean by the “relevance for classification of optically contrasted environments”? Perhaps you can rephrase this sentence to make the meaning more clear.
Figure 4: the x- and y-axes should have labels.
Lines 423-424: “To analyze the sensitivity of bands and ratios to bathymetry…” -> what do you mean by bands being sensitive to bathymetry? Perhaps you can rephrase.
Lines 435-436: “This confirms a comprehensive sensitivity analysis carried out prior to this study…” -> Should there be a citation associated with this statement?
Line edits and technical corrections
Line 53: exploit -> exploits
Line 54: exploit -> exploits
Line 78: assume -> assumes
Lines 86-87: “Finally, uncertainty associated to SDB is a key information for data interpretation, that must be quantified when interpreting bathymetric changes.” -> “Finally, the uncertainty associated with SDB is essential information for data interpretation, and therefore must be quantified when analyzing and interpreting bathymetric changes.”
Lines 107-108: “Their morphology exhibits a strong dynamic with timescales from months to years and decades [30,31].” What do you mean by “exhibits a strong dynamic”? Perhaps you should say “Their morphology changes dynamically on timescales from…”?
Line 114: “river discharges, biological…” -> “river discharges and biological…”
Line 122: “moderate” -> “moderately”
Line 138: “partially” -> “part of”
Line 139: “extension” -> “lateral extent”
Line 141: “tools explaining…” -> “tools, which explains”
Line 150: “Landsat‐8 satellite mission” -> “The Landsat‐8 satellite mission”
Line 153: “formers” -> “former”
Line 156: “for panchromatic” -> “for the panchromatic”
Line 177: “[11] demonstrated that…” -> “For example, Capo et al. [11] demonstrated that…”
Line 180: “30-days” -> “30-day”
Lines 183-184: “images, among which 49 of them are Landsat-8 images and the other 40 are Sentinel-2A/B images.” -> “images, among which 49 are Landsat-8 images and 40 are Sentinel-2A/B images”
Line 185: “is function” -> “is a function”
Line 220: “images” –> “image”
Line 221: “comprised between” -> “bounded by”
Line 223: “For each bins” -> “For each bin”
Line 231: “Step 1” - > “In Step 1”
Line 232: “Step 2” - > “In Step 2”
Line 234: “Step 3” - > “In Step 3”
Line 235: “Step 4” - > “In Step 4”
Line 239: “no more” -> “no longer”
Lines 239-240: “and algorithm” -> “and the algorithm”
Line 240: “LRM” -> “the LRM”
Line 241: “and algorithm” -> “and the algorithm”
Line 242: “one LRM are” -> “one LRM is”
Lines 249-250: “This procedure for the switching method allows to automatically adapt the switching points (???−, ???+) to the water column optical properties changes.” -> “This procedure for the switching method allows for the switching points (???−, ???+) to automatically adapt when the water column optical properties change.”
Line 256: “classes” -> “class”
Line 263: “of distribution” -> “of the distribution”
Line 300: “acquisition” -> “acquisitions”
Line 302: “images” -> “image”
Line 423: “with hydrodynamics” -> “with the hydrodynamic”
Line 425: “each images” -> “each image”
Line 430: “higher than 9 m-depth” -> “deeper than 9 m”
Lines 430-431: “On the opposite” -> “On the other hand”
Line 452: “allows to” -> “allows us to”
Line 455: “allows reducing” -> “allows us to reduce”
Line 464: “different of 0” -> do you mean “different than 0”?
Line 467: another occurrence of “different of 0”
Line 469: “allows to” -> “allows us to”
Line 471: “7.62% that” -> “7.62%, which”
Line 482: “of SDB” -> “of the SDB”
Line 527: “allows to” -> “allows us to”
Line 534: “of spatial distribution of sounding point” -> “of the spatial distribution of sounding points”
Line 555: “To have a better illustration of the” -> “To better illustrate the”
Line 615: “surveys” -> “survey”
Line 627: “provide” -> “provides”
Line 628: “informations” -> “information”
Line 629: “this type of map” -> “these types of maps”
Line 633: “ranges” -> “range”
Line 634: “a better” -> “better”
Line 639: “allows to” -> “allows us to”
Line 641: “consequently on” -> “consequently to”
Line 644: “hydrodynamics” -> “hydrodynamic”
Line 646: “of water column” -> “of the water column”
Line 666: Add “)” after “images”
References
Caballero, I. and Stumpf, R. Towards routine mapping of shallow bathymetry in environments with variable turbidity: contribution of Sentinel-2A/B satellites mission. Remote Sens., 2020, 12, 451-474.
Geyman, E.C. and Maloof, A.C., 2020. Deriving tidal structure from satellite image time series. Earth and Space Science, 7(2), p.e2019EA000958.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 1 Report
The author has addressed all my comments.
Reviewer 3 Report
The authors have addressed my concerns from the previous version.
