Monitoring Shoreline Changes along the Southwestern Coast of South Africa from 1937 to 2020 Using Varied Remote Sensing Data and Approaches

Round 1

Reviewer 1 Report

This is a commendable effort that draws on a thorough analysis of remote sensing data to determine multi-decadal changes on the shoreline of a part of the coast of South Africa.

 

The theme fits perfectly with the journal scope and the authors show clear mastery of the techniques, databases and analytical procedures that enable them to attain the objectives set in the paper. The paper is well written, clearly organized and pleasant to read. The illustrations are clear and well done. Figure 2 is excellent, and overall the methodology section is great. 

 

My only concern regards the fact that the authors do not seem to be beyond a simple determination of shoreline retreat, without actually evoking, even in the briefest of terms, the conditions that explain the shoreline trends other than allusions to longshore drift, wave energy dissipation gradients and changes in longshore grain size without the reader knowing how these conditions are associated with the basic wave conditions and sediment supply and redistribution, etc. The implementation of “proactive strategies for coastal resource protection and management” (lines 60-61) requires at least some insight into the reasons why shorelines are vulnerable. Unfortunately, using remote sensing data to simply reveal trends without attempting to provide explanations of these trends is tantamount to just an expression of the use of a methodological tool, and has not much value in terms of utility for the proclaimed “proactive strategies”. We know nothing of the sediment supply conditions or wave climate of the study area although the authors evoke longshore transport. Ironing out some of these shortcomings will give the paper better added value and more impact. These items are outlined below:

 

Abstract

What is meant by further studies integrating non-remote sensing data to improve the results? What type of data? The underlying uncertainties of remote sensing data are specific to remote sensing and are covered by addressing aspects such as spatial and temporal image resolution, errors, etc, so these uncertainties are not at all related to processes that generate coastal erosion!

 

Introduction

Line 33: the massive sediment flux changes currently seen in the Anthropocene are not due to naturally-induced mechanisms (see for instance: Besset et al., 2019; Syvitski et al., 2022).

Lines 44-46: The claim reported from Vousdoukas et al. (2020) of beaches disappearing with sea-level rise is highly controversial and does not reflect the true morphodynamic resilience of beaches. This statement needs to be balanced by referring to a more cautious interpretation, as embodied in Cooper et al. (2021).

The “other factors” vaguely referred to in line 44 need to be reiterated here, notably anthropogenically-perturbed sediment supply, as evoked in line 33. Furthermore, much of the recent multi-decadal to current erosion of beaches and sandy shorelines is generated by massive decreases in incoming sediment, notably fluvial, outpacing the impact of any current sea-level rise. All of this is needed not just “to understand patterns that pose a global risk to coastal communities” (lines 46-47) but also to understand the causes of coastal erosion.

Lines 68-69: this needs to be stated with great caution. Shoreline change trends are almost never linear and predicting future shoreline changes is still in the domain of wishful thinking. See the example of the 2017 storm you evoke with the sweeping erosion caused. How do we predict future drivers of shoreline mobility such as storms?

Line 82: “on the west coast” – specify South Africa here.

 

Materials and methods

The description of the study site should not just be limited to rainfall. This is not a primary driver of shoreline change. What about the main drivers of shoreline change? The wave regime, tidal range, sediment supply possibilities (from the shoreface? Inherited? Carbonate? Or terrigenous related to streams?), engineereing structures, etc. 

Line 181: How is the High Water Line situated relative to tidal levels and mean sea level in the study area? Provide study site data on tidal range. 

Lines 299-301: So, what degree of caution is needed for the use of the forecasting model? If it needs to be used with caution, what can its utility be in terms of proactive management? How reliable can such predicted shoreline trends be? Are they useful at all? Do they not pose a risk of misguiding management? Future shoreline trends, if you insist on presenting them, should be used at best as simply indicative of potentialshoreline mobility.

  

Discussion

Lines 490-496: At least a presentation of basic wave characteristics affecting this part of the South African coast (heights, directions, periods) could be helpful in further gauging to a first order this variability.

 

References:

 

Besset, M. et al., Multi-decadal variations in delta shorelines and their relationship to river sediment supply: An assessment and review. Earth-Science Reviews, 193-199-219 (2019). https://doi.org/10.1016/j.earscirev.2019.04.018

 

Cooper, J.A.G. et al. Sandy beaches can survive sea-level rise. Nat. Clim. Chang. 10, 993–995 (2020). https://doi.org/10.1038/s41558-020-00934-2

 

Syvitski, J. et al., Earth’s Sediment Budget during the Anthropocene. Nature Reviews Earth & Environment 3, 179–196 (2022). https://www.nature.com/articles/s43017-021-00253-w 

 

 

Author Response

Reviewer 1

We would like to thank Reviewer 1 for the helpful feedback and comments to improve the quality of this article manuscript.

This is a commendable effort that draws on a thorough analysis of remote sensing data to determine multi-decadal changes on the shoreline of a part of the coast of South Africa.

The theme fits perfectly with the journal scope and the authors show clear mastery of the techniques, databases and analytical procedures that enable them to attain the objectives set in the paper. The paper is well written, clearly organized and pleasant to read. The illustrations are clear and well done. Figure 2 is excellent, and overall the methodology section is great.

My only concern regards the fact that the authors do not seem to be beyond a simple determination of shoreline retreat, without actually evoking, even in the briefest of terms, the conditions that explain the shoreline trends other than allusions to longshore drift, wave energy dissipation gradients and changes in longshore grain size without the reader knowing how these conditions are associated with the basic wave conditions and sediment supply and redistribution, etc. The implementation of “proactive strategies for coastal resource protection and management” (lines 60-61) requires at least some insight into the reasons why shorelines are vulnerable. Unfortunately, using remote sensing data to simply reveal trends without attempting to provide explanations of these trends is tantamount to just an expression of the use of a methodological tool, and has not much value in terms of utility for the proclaimed “proactive strategies”. We know nothing of the sediment supply conditions or wave climate of the study area although the authors evoke longshore transport. Ironing out some of these shortcomings will give the paper better added value and more impact. These items are outlined below:

Abstract

What is meant by further studies integrating non-remote sensing data to improve the results? What type of data? The underlying uncertainties of remote sensing data are specific to remote sensing and are covered by addressing aspects such as spatial and temporal image resolution, errors, etc, so these uncertainties are not at all related to processes that generate coastal erosion!

The sentence has been altered to better explain the data types. Better resolution and ground-truthing/field surveys. Line 23 – 25: Further studies should integrate additional high resolution remote sensing data and non-remote sensing data (e.g., field surveys) to improve our results and provide a more thorough understanding of the coastal environment and overcome some of remotely-sensed data underlying uncertainties.

Introduction

Line 33: the massive sediment flux changes currently seen in the Anthropocene are not due to naturally-induced mechanisms (see for instance: Besset et al., 2019; Syvitski et al., 2022).

Syvitski et al., 2022 reference has been added to better substantiate that “discrete or combined interaction” of natural and human forces lead to shoreline change.

Lines 44-46: The claim reported from Vousdoukas et al. (2020) of beaches disappearing with sea-level rise is highly controversial and does not reflect the true morphodynamic resilience of beaches. This statement needs to be balanced by referring to a more cautious interpretation, as embodied in Cooper et al. (2021).

Yes, Cooper’s paper was included in the original research report for this paper, explaining the debates and complexities of studying coastal environments. This has now also been included.

 Line 48: However, coastal environments are inherently dynamic and modelling sea-level rise risks and coastal morphology needs to be site-specific and substantiated with an appropriate range of data (Cooper et al., 2020)

The “other factors” vaguely referred to in line 44 need to be reiterated here, notably anthropogenically-perturbed sediment supply, as evoked in line 33. Furthermore, much of the recent multi-decadal to current erosion of beaches and sandy shorelines is generated by massive decreases in incoming sediment, notably fluvial, outpacing the impact of any current sea-level rise. All of this is needed not just “to understand patterns that pose a global risk to coastal communities” (lines 46-47) but also to understand the causes of coastal erosion.

Added ü

Lines 68-69: this needs to be stated with great caution. Shoreline change trends are almost never linear and predicting future shoreline changes is still in the domain of wishful thinking. See the example of the 2017 storm you evoke with the sweeping erosion caused. How do we predict future drivers of shoreline mobility such as storms?

More ‘nuance’ has been added to this statement. Line 69: provide a unique view into the past to observe shoreline dynamics over time and to use these historical data to model possible predict future shoreline changes.

Line 82: “on the west coast” – specify South Africa here.

Added ü

Materials and methods

The description of the study site should not just be limited to rainfall. This is not a primary driver of shoreline change. What about the main drivers of shoreline change? The wave regime, tidal range, sediment supply possibilities (from the shoreface? Inherited? Carbonate? Or terrigenous related to streams?), engineereing structures, etc.

Addressed ü.Section 2.1. Line 105-125.

Line 181: How is the High Water Line situated relative to tidal levels and mean sea level in the study area? Provide study site data on tidal range.

Addressed ü. Line 212: The HWL is also the more appropriate indicator to use for micro-tidal environments, such as Yzerfontein.

Lines 299-301: So, what degree of caution is needed for the use of the forecasting model? If it needs to be used with caution, what can its utility be in terms of proactive management? How reliable can such predicted shoreline trends be? Are they useful at all? Do they not pose a risk of misguiding management? Future shoreline trends, if you insist on presenting them, should be used at best as simply indicative of potential shoreline mobility.

Indeed, your point is extremely valid. The tool is very limiting and must not misguide management. I have added an extra sentence onto the paragraph.

Line 323: Thus, these forecasts cannot be the main tool for coastal management and planning.

Discussion

Lines 490-496: At least a presentation of basic wave characteristics affecting this part of the South African coast (heights, directions, periods) could be helpful in further gauging to a first order this variability.

With the added coastal information in 2.1, this should hopefully be clearer. 

References:

1. Besset, M. et al., Multi-decadal variations in delta shorelines and their relationship to river sediment supply: An assessment and review. Earth-Science Reviews, 193-199-219 (2019). https://doi.org/10.1016/j.earscirev.2019.04.018
2. Cooper, J.A.G. et al. Sandy beaches can survive sea-level rise. Nat. Clim. Chang. 10, 993–995 (2020). https://doi.org/10.1038/s41558-020-00934-2
3. Syvitski, J. et al., Earth’s Sediment Budget during the Anthropocene. Nature Reviews Earth & Environment 3, 179–196 (2022). https://www.nature.com/articles/s43017-021-00253-w

Reviewer 2 Report

Review of Monitoring Shoreline Changes along the Southwestern Coast 2 of South Africa from 1937 to 2020 using Varied Remote Sensing 3 Data and Approaches by Murray et al. (remotesensing-2032170)

The manuscript by Murray et al. uses aerial photographs and Landsat satellite imagery from 1927-2020 to examine coastal erosion in southwestern Africa. The authors use automated tools, specifically CoastStat, the Google Earth Engine and the Digital Shoreline Analysis System to extract shoreline positions and calculate the extent of shoreline dynamics. The future position of the shoreline is also predicted. The article is well-written and is clear. In addition, the topic of machine learning, which is not really emphasized enough in the text, is at the forefront of research in the geosciences. The methodology is sound, applying well-tested methods in shoreline analysis such as End Point Rate, Linear Regression Rate, Weighted Linear Regression rate and prediction using the Kalman Filter method. The authors seem aware of the limitations of these methods as well as the errors introduced with the different types of data used and analysis preformed. However, there are a few issues with the manuscript, which needs, in my opinion - moderate revision. However, this is not one of the options available.

There are claims in the Abstract, which are not mentioned or developed in the text. The first is the cumulative erosion of 39 m along the stretch of coast analyzed in the study. The second is that this is the result mainly of the 2017 storm. These are important points and should be somewhere in the text.

Since this article deals with coastal erosion in the introduction I am missing a description of the dynamics of the coast (e.g. source of sediment, sediment transport directions and volumes; waves and currents) as well as a review of coastal construction over the years that may have contributed to the evolution and development of the beaches under investigation. While this is not the focus of the article, it is crucial for the Discussion – how has the dynamics of the shoreline changed with the development of the area? It is not enough to present the results and say that they may be connected to a large backshore or perhaps a dune field. You need a well-developed and though out Discussion addressing the possible causes of what you seen in the Results. An example can be seen on Line 529-531 “Much erosion was seen along the southern stretch of Sixteen Mile Beach within the West Coast National Park. The main beach at Yzerfontein experienced the least change, followed by Pearl Bay, which experienced some erosion between 2015 and 2020” - I expect to see a more developed discussion about the differences between the three beaches and the possible cause for this difference, in relation to coastal dynamics and construction in the area.

What little information you do provide about construction in the area is lacking. For instance, more information is needed about the public access car park that backs Yzerfontein Main beach that is mentioned in Line 121 – when was it constructed? How far from the shore? Is there a concrete wall that can lead to beach erosion? Also, does the harbor mentioned for the first time on line 123 interfere with sediment transport to the beaches? This is all crucial information.

Table 2 - It is a bit strange to compare images from different seasons. I understand why this was done, but it introduces an error since coastlines shift drastically from season to season. How was this accounted for (if at all)?

Please see attached annotated manuscript for more specific comments.

Comments for author File: Comments.pdf

Author Response

Reviewer 2

We would like to thank Reviewer 2 for the helpful feedback and comments to improve the quality of this article manuscript.

Review of Monitoring Shoreline Changes along the Southwestern Coast 2 of South Africa from 1937 to 2020 using Varied Remote Sensing 3 Data and Approaches by Murray et al. (remotesensing-2032170)

The manuscript by Murray et al. uses aerial photographs and Landsat satellite imagery from 1927-2020 to examine coastal erosion in southwestern Africa. The authors use automated tools, specifically CoastStat, the Google Earth Engine and the Digital Shoreline Analysis System to extract shoreline positions and calculate the extent of shoreline dynamics. The future position of the shoreline is also predicted. The article is well-written and is clear. In addition, the topic of machine learning, which is not really emphasized enough in the text, is at the forefront of research in the geosciences. The methodology is sound, applying well-tested methods in shoreline analysis such as End Point Rate, Linear Regression Rate, Weighted Linear Regression rate and prediction using the Kalman Filter method. The authors seem aware of the limitations of these methods as well as the errors introduced with the different types of data used and analysis preformed. However, there are a few issues with the manuscript, which needs, in my opinion - moderate revision. However, this is not one of the options available.

There are claims in the Abstract, which are not mentioned or developed in the text. The first is the cumulative erosion of 39 m along the stretch of coast analyzed in the study. The second is that this is the result mainly of the 2017 storm. These are important points and should be somewhere in the text.

Yes, more has now been added to elaborate on these two points in results section. Unfortunately, the 2017 storm cannot be directly linked to the changes described here as this detailed section had to be omitted from this publication due to length constraints. Abstract now changed to

Line 22: Our results show that the coastline changed dynamically between 1937 and 2020, culminating in an average net erosion of 38 m, with the most extensive erosion occurring between 2015 and 2020.

Since this article deals with coastal erosion in the introduction I am missing a description of the dynamics of the coast (e.g. source of sediment, sediment transport directions and volumes; waves and currents) as well as a review of coastal construction over the years that may have contributed to the evolution and development of the beaches under investigation. While this is not the focus of the article, it is crucial for the Discussion – how has the dynamics of the shoreline changed with the development of the area? It is not enough to present the results and say that they may be connected to a large backshore or perhaps a dune field. You need a well-developed and though out Discussion addressing the possible causes of what you seen in the Results. An example can be seen on Line 529-531 “Much erosion was seen along the southern stretch of Sixteen Mile Beach within the West Coast National Park. The main beach at Yzerfontein experienced the least change, followed by Pearl Bay, which experienced some erosion between 2015 and 2020” - I expect to see a more developed discussion about the differences between the three beaches and the possible cause for this difference, in relation to coastal dynamics and construction in the area.

What little information you do provide about construction in the area is lacking. For instance, more information is needed about the public access car park that backs Yzerfontein Main beach that is mentioned in Line 121 – when was it constructed? How far from the shore? Is there a concrete wall that can lead to beach erosion? Also, does the harbor mentioned for the first time on line 123 interfere with sediment transport to the beaches? This is all crucial information.

Extra study site information has now been added to section 2.1.

Lines 105 -125. The coastline is microtidal, with a monthly range of about 1.8 m. It is noteworthy, that unlike many of the other bays on the western coast, Yzerfontein presently has no local riverine input of sediment. All sandy sediment, marine and aeolian, is derived from the south by a combination of longshore drift and the prevailing strong southerly summer winds moving sand from exposed beaches further south. Dassen Island, the rocky Yzerfontein point and the harbour breakwater, which was extended in the 1980s, shelter the Yzerfontein Main Beach to some extent from the southerly marine swell. The Meeurots islet in the middle of the bay and the rocky point Rooipan se Klippe (or Gabbro Point) at the northern end of Yzerfontein Main Beach complicate the refraction and diffraction of the incoming waves, which in turn may affect the wave energy reaching different parts of the shore. The energy of incoming waves is high, with swells that routinely reach 5-6 m and sometimes up to 12 m in rare extreme storms. Storms and rainfall are more common in winter with this region having a semi-arid Mediterranean climate, with dry, warm summers and cool, wet winters [24, 25]

Table 2 - It is a bit strange to compare images from different seasons. I understand why this was done, but it introduces an error since coastlines shift drastically from season to season. How was this accounted for (if at all)?

Yes, the main limiting factor was cloud cover, meaning only images that had absolutely no obstructions could be run, and that CoastSat ran correctly on these images. Therefore, the aim was to choose the best image from main summer season (Dec-Feb) but this had to be extended to ‘general’ summer months of October to March.

Please see attached annotated manuscript for more specific comments (attached here as remotesensing-2032170-review comments)

Reviewer 3 Report

I send the review in the attachment.

Comments for author File: Comments.pdf

Author Response

Reviewer 3

We would like to thank Reviewer 3 for the helpful feedback and comments to improve the quality of this article manuscript.

Dear Authors,

The article entitled: Monitoring Shoreline Changes along the Southwestern Coast of South Africa from 1937 to 2020 Using Varied Remote Sensing Data and Approaches presents a coastline vulnerability from 1937 to 2020 and predicted its change by 2040 by manually delineating shoreline positions from 1937, 1960, and 1977 from aerial photographs and Landsat products between 1985 and 2020 in an automated fashion using the CoastSat toolkit and Google Earth Engine. It is very important because the threat of coastal erosion is widespread on sandy coasts, as well as the destruction of coastal infrastructure and ecosystems.

All chapters (abstract, introduction, materials and methods, results, discussion, as well as conclusions) are very well described and they do not raise any doubts. In terms of the literature review is complex (46 positions), all of which are papers from recognised scientific journals, such as: ISPRS Journal of Photogrammetry and Remote Sensing, Remote Sensing of Environment, Scientific Reports, and others.

However, I would like to point out that the papers cited are related to the subject of this article (coastal erosion, DSAS, CoastSat, Yzerfontein and Sixteen Mile Beach). Nevertheless, in the publication make the following change:

I propose to extend the literature in the introduction, related to the long-term observations of coastline changes such as, for example: Literature was added in the introduction section using the suggested references. Lines 76-84.

1. Coastline Evolution in Sopot (2008 2018) Based on Landsat Satellite Imagery. J. Mar. Sci. Eng. 2020, 8, 464.

Specht et al. (2020) analyzed the coastal variability in Sopot based on Landsat satellite data and found an average coastline shift of 19.1 m towards the sea between 2008 and 2018

2. Wang, X.; Liu, Y.; Ling, F.; Liu, Y.; Fang, F. Spatio-temporal Change Detection of Ningbo Coastline Using Landsat Time-series Images during 1976 2015. ISPRS Int. J. Geo-Inf. 2017, 6, 68.

Wang et al. (2017) studied the spatio-temporal changes of Ningbo coasts between 1976 and 2015 using Landsat sensors and their results showed an increased mean NSM from 187 m to 298 m, with the mean annual NSM reaching 85 m/year, indicating the progress of the coasts towards the sea.

3. Xu, N. Detecting Coastline Change with All Available Landsat Data over 1986 2015: A Case Study for the State of Texas, USA. Atmosphere 2018, 9, 107.

Xu (2018) used nearly three decades of Landsat data (1986–2015) to analyze the coast of the US State of Texas and found that it endured erosion at a rate of –0.154 ±0.063 km2/year, with 52.58% of the total coastline retreated the land, while a proportion of 47.42% encroached the sea.

 

In the introduction, it is worth summarising what the individual chapters of the article contain.

Addressed ü. Extra sentence added. Line 100-101: In the next section, we describe the coastal area in which the study was conducted and clarify the datasets and methodological framework used to understand coastal change in the area studied. We then present the results based on four statistical variables: Net Shoreline Movement (NSM), Shoreline Change Envelope (SCE), End Point Rate (EPR), and Linear Regression Rate (LRR), Weighted Linear Regression (WLR) and predict the potential shoreline changes in the next 10 and 20 years. After discussion, limitations are identified and the potential for future research is suggested before conclusions are drawn.

To sum up, after taking into account the above amendments (minor revision), I suppose that this article is suitable for publication in Remote Sensing.

Thank you very much for your feedback. 

Reviewer 4 Report

Dear authors, thank you for submitting your work on the crucial topic of historical coastal change.

Below you will find my suggestions for improving your manuscript, especially where some additional details are needed, and where some clarifications should be made.

When citing Voudouskas et al (line 44), please cite Cooper at al (2020, Sandy beaches can survive SLR) to ensure that you are giving 'all' views on this intractable and crucial issue of the future of our coasts.

Material and method:

# Figure 2: Please change NSN to NSM under “shoreline changes” (left).

# Regarding the shoreline proxy used in this study:

I question the relevance of a water limit extracted from old aerial and satellite images as they are taken in different months (March, April, October, November, December) and at different tide levels and wave conditions.

In line 185 you indicate that you have manually digitised a HWL on the old images, but in line 208 you indicate that you are using CoastSat, which extracts a waterline. If I understand correctly, you are using two different indicators and therefore two indicators that cannot be compared. Can you clarify this point? By illustrating for example the position of the indicator used on the old images and the satellite images (with captures and position of the lines on different examples).

# L203

Methods exist to quantitatively assess the uncertainty associated with the manual digitisation of a coastline position indicator from old aerial images. Please refer to them and test these methods to see if the results obtained are comparable to "half the pixel size".

e.g.: Kench et al., 2018. Patterns of island change and persistence offer alternate adaptation pathways for atoll nation.

This comment is only indicative. There is no need to reassess the error related to the digitisation of the HWL.

# L233

Why didn't you realign the satellite images to reduce this error to 0 metres using automatic coregistration algorithms? This would have substantially reduced the margin of error.

# L238

DSAS is a well-known tool, it is not necessary to detail the principle of the tool at this point (L242 to 247). Please refer to the user guide and only indicate your specific parameters for the study.

# L250

Why did you choose to space your transects so widely (100 m)? By using images with a resolution between 3 and 15 m, you could have reduced the spacing of your transects (to 20 m for example) and have a more consistent statistical sample that would have allowed you to capture the longitudinal variability of shoreline behaviour.

# L284

Please be careful and extremely clear when using beta shoreline forecasting from DSAS. As expressed by authors: “The forecasts produced by this tool should always be used with caution. The processes driving shoreline change are complicated and not always available as model inputs: many factors that may be important are not considered in this methodology or accounted for within the uncertainty. This methodology assumes that a linear regression thorough past shoreline positions is a good approximation for future shoreline positions; this assumption will not always be valid.” In my opinion, you should be clearer about the limitations of this tool for projecting the future position of the coastline using historical shoreline positions

Results:

# L320-321

“Largest NSM”, “Largest positive value” : I am not comfortable with these formulations and it is difficult to understand their meaning. You can also use “maximum retreat value”, “maximum advance value” or something similar.

# L323

“Greatest amount of erosion”: please reword.

# Table 5

Why did you calculate an uncertainty if you don't take it into account when presenting your results?

It is essential to have a "stable" or "no significant change" class whose bounds correspond to your uncertainty values. You cannot just indicate "erosion" and "accretion".

# L332-341

I cannot perceive the added value and understand the interpretation of the SCE values?

Can you answer this question: what does the SCE provide for the interpretation/attribution of detected changes?

# L424

What is the added value of using the 5 statistics that can be calculated with DSAS?

When reading the results section, the multiplicity of figures and statistical indicators makes it difficult to understand the results.

There are a lot of different figures and values for the same stretch of coastline: you don't know what to remember at the end of the results section.

Please simplify the result section and make it more understandable. Please also include the comment on uncertainty.

# L449

Here we learn about the Sentinel-2 data: why didn't you mention it in the method section? What did you get out of it?

# L451

Section 3.2 is somewhat sparse and covers only a limited part of the study area. We do not learn much useful information for future management of the coastal zone and these results suggest that the beta projection tool is not recommended for your dataset. 

Discussion

You can subdivide your discussion section to discuss the key points of your study. This will make it easier to read.

# L482-483

This has already been said several times.

# L504

So the higher retreat values would be due to an error in detecting the position of the coastline indicators?

# L519-523

You are comparing results that do not seem comparable.

Your study: detection of shoreline changes over 83 years from transects spaced at 100 m

Their study: detection of shoreline changes over 32 years from transects spaced at 500 m

Your study includes 50 more years of changes than the study you cite.

Conclusions

# L535

I feel like I'm discovering these figures: please simplify your results section to highlight the key values to remember.

# L544-546

Please be careful.

The results obtained automatically with CoastSat should not be considered as reflecting strict accuracy. Like the indicators you have manually digitised, these results also have their associated margin of error, which should be clearly explained in material / method section and integrate to the Table 4.

Author Response

Reviewer 4

We would like to thank Reviewer 4 for the helpful feedback and comments to improve the quality of this article manuscript.

 

Comments and Suggestions for Authors

Dear authors, thank you for submitting your work on the crucial topic of historical coastal change.

Below you will find my suggestions for improving your manuscript, especially where some additional details are needed, and where some clarifications should be made.

When citing Voudouskas et al (line 44), please cite Cooper at al (2020, Sandy beaches can survive SLR) to ensure that you are giving 'all' views on this intractable and crucial issue of the future of our coasts.

Yes, a very important article. We have included it back into the text. Line 48: However, coastal environments are inherently dynamic and modelling sea-level rise risks and coastal morphology needs to be site-specific and substantiated with an appropriate range of data (Cooper et al., 2020)

Material and method:

# Figure 2: Please change NSN to NSM under “shoreline changes” (left).

Addressed ü. Thank you for picking up this typo.

# Regarding the shoreline proxy used in this study:

I question the relevance of a water limit extracted from old aerial and satellite images as they are taken in different months (March, April, October, November, December) and at different tide levels and wave conditions.

Further detail has been given to justify the dates and tidal conditions.

Addressed ü.

In line 185 you indicate that you have manually digitised a HWL on the old images, but in line 208 you indicate that you are using CoastSat, which extracts a waterline. If I understand correctly, you are using two different indicators and therefore two indicators that cannot be compared. Can you clarify this point? By illustrating for example the position of the indicator used on the old images and the satellite images (with captures and position of the lines on different examples).

From the CoastSat creators, Vos et al., we are assuming their sand/water interface to be similar to HWL given the relative ‘micro tidal’ coast.

# L203

Methods exist to quantitatively assess the uncertainty associated with the manual digitisation of a coastline position indicator from old aerial images. Please refer to them and test these methods to see if the results obtained are comparable to "half the pixel size".

The uncertainty calculations followed Niang et al. 2020 (Niang, AJ Monitoring Long-Term Shoreline Changes along Yanbu, Kingdom of Saudi Arabia Using Remote Sensing and GIS Techniques. J. Taibah Univ. Sci. 2020, 14, 762–776). Who suggest approximating the digitization error to half the pixel size.

e.g.: Kench et al., 2018. Patterns of island change and persistence offer alternate adaptation pathways for atoll nation.

This comment is only indicative. There is no need to reassess the error related to the digitisation of the HWL.

# L233

Why didn't you realign the satellite images to reduce this error to 0 metres using automatic coregistration algorithms? This would have substantially reduced the margin of error.

This error comes with the CoastSat metadata. Since the programme is automated, this cannot be reduced. Vos, K.; Splinter, K.D.; Harley, M.D.; Simmons, J.A.; Turner, I.L. CoastSat: A Google Earth Engine-Enabled Python Toolkit to Extract Shorelines from Publicly Available Satellite Imagery. Environ. Model. Softw. 2019, 122, 104528.

# L238

DSAS is a well-known tool, it is not necessary to detail the principle of the tool at this point (L242 to 247). Please refer to the user guide and only indicate your specific parameters for the study.

Addressed ü unnecessary amounts of detail were removed. Line 183.

# L250

Why did you choose to space your transects so widely (100 m)? By using images with a resolution between 3 and 15 m, you could have reduced the spacing of your transects (to 20 m for example) and have a more consistent statistical sample that would have allowed you to capture the longitudinal variability of shoreline behaviour.

The main reason for 100 m was because the study area was 30 km long and 100 m seemed the norm from other DSAS studies.

# L284

Please be careful and extremely clear when using beta shoreline forecasting from DSAS. As expressed by authors: “The forecasts produced by this tool should always be used with caution. The processes driving shoreline change are complicated and not always available as model inputs: many factors that may be important are not considered in this methodology or accounted for within the uncertainty. This methodology assumes that a linear regression thorough past shoreline positions is a good approximation for future shoreline positions; this assumption will not always be valid.” In my opinion, you should be clearer about the limitations of this tool for projecting the future position of the coastline using historical shoreline positions

Yes, you have raised the same point as reviewer 1. The forecasting tool is very limiting and must not misguide management. I have added an extra sentence onto the paragraph. Line 343: Thus these forecasts cannot be the main tool for coastal management and planning.

Results:

# L320-321

“Largest NSM”, “Largest positive value” : I am not comfortable with these formulations and it is difficult to understand their meaning. You can also use “maximum retreat value”, “maximum advance value” or something similar.

Addressed ü

# L323

“Greatest amount of erosion”: please reword.

Addressed ü

# Table 5

Why did you calculate an uncertainty if you don't take it into account when presenting your results?

It is essential to have a "stable" or "no significant change" class whose bounds correspond to your uncertainty values. You cannot just indicate "erosion" and "accretion".

The NSM and LRR do not take into account the uncertainty values which is why the WRL is highlighted as the ‘most important’ result.

# L332-341

I cannot perceive the added value and understand the interpretation of the SCE values?

Can you answer this question: what does the SCE provide for the interpretation/attribution of detected changes?

Shoreline change envelop shows the ‘total’ or maximum distance covered, so if a shoreline accreted one decade and eroded another, this distance would be greater than just the net movement.

# L449

Here we learn about the Sentinel-2 data: why didn't you mention it in the method section? What did you get out of it?

Thank you for picking this out. The original research project used Landsat and Sentinel-2, with Sentinel-2 being used to identify seasonal and short-term trends in the data from 2015-2020. However, this had to be omitted in the article to shorten it’s length and scope. This has now been fixed.

# L451

Section 3.2 is somewhat sparse and covers only a limited part of the study area. We do not learn much useful information for future management of the coastal zone and these results suggest that the beta projection tool is not recommended for your dataset. 

Yes, greater substantiation and warnings have been added to the methodologies and results about this forecasting tool. Line 340: Thus these forecasts cannot be the main tool for coastal management and planning.

And line 517: However, these forecast results are limited to only the historical data provided and do not consider sediment transport processes, infrastructure or climate which impacts the coastline.

Discussion

You can subdivide your discussion section to discuss the key points of your study. This will make it easier to read.

Since the discussion is less than a page long and the points rasised link together no sub-sections have been added.

# L504

So the higher retreat values would be due to an error in detecting the position of the coastline indicators?

It is a possibility, but difficult to substantiate.

# L519-523

You are comparing results that do not seem comparable.

Your study: detection of shoreline changes over 83 years from transects spaced at 100 m

Their study: detection of shoreline changes over 32 years from transects spaced at 500 m

Your study includes 50 more years of changes than the study you cite.

This paragraph has been reworded to be more clear in how these projects have similar aspects to their methodologies and where the statistic of –1.71 ±0.44 m/yr comes from their web map.

Line 561: Additionally, our results can be compared and contrasted  to Luijendijk et al. [6], who conducted a global-scale study of shoreline change by calculating erosion and accretion rates at every 500 m interval of sandy beach by extracting the shoreline positions using Landsat satellite imagery between 1984 and 2016. Their interactive web-map shows an average erosion rate of –1.71 ±0.44 m/yr around our study site (http://shorelinemonitor.deltares.nl/)

 

Conclusions

# L535

I feel like I'm discovering these figures: please simplify your results section to highlight the key values to remember.

Added WLR to the sentence to point the reader to which significant figures we are referring to.

Line 578: added WLR

# L544-546

 

Please be careful
The results obtained automatically with CoastSat should not be considered as reflecting strict accuracy. Like the indicators you have manually digitised, these results also have their associated margin of error, which should be clearly explained in material / method section and integrate to the Table 4.

Reworded a conclusion sentence to further highlight uncertainties.

Line 582: Our study has reaffirmed the value of the combined use of historical aerial imagery and Landsat data, together with CoastSat, Google Earth Engine and DSAS geospatial assets, as they offer a unique opportunity to explore space-time patterns of coasts and their potential evolution in the future – notwithstanding the underlying uncertainties that follow all remotely-sensed data.

 

Round 2

Reviewer 2 Report

I am more-or-less satisfied with the revision made by the authors. There are only a few issues left to resolve:

I am still lacking a little more discussion on the causes of the shoreline trends they observed in the data and their potential connection to coastal infrastructure for example. I think it would add more to the paper, but ultimately, I leave this for the authors to decide.

Line 81 is the first time NSM is used. Please spell it out here before using the abbreviation.

As I stated in the annotated manuscript - it is hard to see the scale bar on Figure 1a. black on black (or dark blue) does not work. Please fix this.

Line 321 - there is a double space between "stated" and "that"

Line 346 - missing space between "advance" and (seaward).

Line 348 - double space between "extensive" and "erosion".

Line 545 - double space between "contrasted" and "to".

Author Response

Reviewer comments and our responses

I am still lacking a little more discussion on the causes of the shoreline trends they observed in the data and their potential connection to coastal infrastructure for example. I think it would add more to the paper, but ultimately, I leave this for the authors to decide.

We agreed with the reviewer regarding more discussion on the causes of the shoreline trends and their potential connection to coastal infrastructure. The factors and impacts of the shoreline spatial and temporal trends will be studied in detail in the following manuscript. Our primary aim, for now, is  to test remote sensing in detecting and monitoring the shoreline.

Line 81 is the first time NSM is used. Please spell it out here before using the abbreviation.

NSM is spelt out “ Net Shoreline Movement.”

As I stated in the annotated manuscript - it is hard to see the scale bar on Figure 1a. black on black (or dark blue) does not work. Please fix this.

The quality of Fig 1a scale has been improved

Line 321 - there is a double space between “stated” and “that”

A single space has been deleted

Line 346 - missing space between “advance” and (seaward).

A single space has been added

Line 348 - double space between “extensive” and “erosion”.

A single space has been deleted

Line 545 - double space between “contrasted” and “to”

A single space has been deleted