UAV approaches for improved mapping of vegetation cover and estimation of carbon storage of small saltmarshes: examples from Loch Fleet, northeast Scotland

Hiles, William; Miller, Lucy Catherine; Smeaton, Craig; Austin, William Edward Newns

doi:https://doi.org/10.5194/egusphere-2023-1185

Preprints

https://doi.org/10.5194/egusphere-2023-1185

Preprints

31 Jul 2023

| 31 Jul 2023

UAV approaches for improved mapping of vegetation cover and estimation of carbon storage of small saltmarshes: examples from Loch Fleet, northeast Scotland

William Hiles, Lucy Catherine Miller, Craig Smeaton, and William Edward Newns Austin

Abstract. Saltmarsh environments are recognised as key components of many biophysical and biochemical processes at the local and global scale. Accurately mapping these environments, and understanding how they are changing over time, is crucial for better understanding these systems. However, traditional surveying techniques are time-consuming and are inadequate for understanding how these dynamic systems may be changing temporally and spatially. The development of Uncrewed Aerial Vehicle (UAV) technology presents an opportunity for efficiently mapping saltmarsh extent. Here we develop a methodology which combines field vegetation surveys with multispectral UAV data collected at two scales to estimate saltmarsh area and organic carbon storage at three saltmarshes in Loch Fleet (Scotland). We find that the Normalised Difference Vegetation Index (NDVI) values for surveyed saltmarsh vegetation communities, in combination with local tidal data, can be used to reliably estimate saltmarsh area. Using these area estimates, together with known plant community and soil organic carbon relationships, saltmarsh soil organic carbon storage is modelled. Based on our most reliable UAV-derived saltmarsh area estimates, we find that organic carbon storage is 15–20 % lower than previous area estimates would indicate. The methodology presented here potentially provides a cheap, affordable, and rapid method for saltmarsh mapping which could be implemented more widely to test and refine existing estimates of saltmarsh extent and is particularly well-suited to the mapping of small areas of saltmarsh habitat.

Received: 01 Jun 2023 – Discussion started: 31 Jul 2023

Publisher's note: Copernicus Publications remains neutral with regard to jurisdictional claims made in the text, published maps, institutional affiliations, or any other geographical representation in this preprint. The responsibility to include appropriate place names lies with the authors.

Download & links

Preprint (PDF, 2175 KB)

Notice on discussion status
The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.
Preprint (2175 KB)

Download & links

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Journal article(s) based on this preprint

20 Feb 2024

UAV approaches for improved mapping of vegetation cover and estimation of carbon storage of small saltmarshes: examples from Loch Fleet, northeast Scotland

William Hiles, Lucy C. Miller, Craig Smeaton, and William E. N. Austin

Biogeosciences, 21, 929–948, https://doi.org/10.5194/bg-21-929-2024,https://doi.org/10.5194/bg-21-929-2024, 2024

Short summary

William Hiles, Lucy Catherine Miller, Craig Smeaton, and William Edward Newns Austin

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2023-1185', Anonymous Referee #1, 15 Aug 2023
General comments:
The paper describes an approach combing vegetation survey, UAV data and tidal data to estimate saltmarsh extent and saltmarsh organic carbon storage. It is certainly intended to be a method development study but has significant amount of work on the effects of areal estimates on organic carbon storage estimates. Do the authors use the estimate of OC storage as a way to assess the reliability of different approaches in extracting saltmarsh extent? As a result, I am less certain of the paper's objective(s). In any case, I see some values of this work but would like to see some improvements.
I found the five parts in Results section confusing and not appropriate to respond to the three objectives the authors raised in Introduction section. Especially for section 3.1, the authors should explain why it was included in Results section. Section 3.2 and 3.3 should be moved to Methods section, as it describe how to classify vegetation communities and estimate areas.
Several major concerns:
How do the authors define the saltmarsh environments? Does it mean saltmarsh ecosystems?

There is a fundamental issue here that the paper did not explore. When classifying saltmarsh vegetation, the authors used the National Vegetation Classification (NVC) scheme. It seems not require training data and validation data as well, so how the mapping accuracy can be achieved? This relates to the following estimation of saltmarsh extent.

In Introduction section, the authors point out the location (elevation) of saltmarshes, that is, saltmarshes form between the high astronomical tide (HAT) and mean tidal level (MTL). In section 2.5.1, the authors stated that the saltmarsh area was estimated by calculating the area that inundated under each tidal condition. In this context, the authors focused on three tidal contexts: High Astronomical Tide (HAT), Mean Low Water Springs (MLWS), Mean High Water Neaps (MHWN). What is the connection between MTL, MLWS, and MHWN? Why the authors recognize the saltmarshes are expected to always be inundated?

Some paragraphs in Section 2.5 and Section 2.6 should be connected.

How to classify the saltmarsh vegetation? By using NVC scheme or NDVI extraction values described in section 2.5.2? The authors should add more details about vegetation classification.

For the first objective of this work, the authors stated to delineate saltmarsh habitats. Vegetation composition is a key component of saltmarsh habitats, however, the related content is not well depicted in this paper. Please change the objective 1 more precisely.

Do the authors used the same OC storage estimation method with Haynes et al., 2016? Section 2.7 need more related descriptions.

Specific & minor concerns:
Line 35: simple or straightforward?

Line 61: saltmarshes are not always small features.

Line 94-98: these three sentences are not suitable for the study area section.

Section 2.3: this section only describes how to design and collect UAV data. Seems better to change to UAV data collection.

Line 136-137: please state what parameter did the internal sunlight sensor process.

Table 1: the tidal ranges are the mean values?

Line 191: typo, rewrite this sentence.

Figure 4: the Wick HAT in middle 2 figures is not identical to the legend color scheme.

Table 2: keep the same decimal place.

Figure 5: abbreviation of SM10,…SM8 should be clarified.

The form of the tables (Table 1-5) are not appropriate.
Citation: https://doi.org/10.5194/egusphere-2023-1185-RC1
- AC2: 'Reply on RC1', William Hiles, 03 Oct 2023
  
  We would like to thank the reviewer for their helpful comments which have highlighted several areas where we have been able to improve the manuscript. We hope we have satisfactorily responded to all suggestions below:
  The paper describes an approach combing vegetation survey, UAV data and tidal data to estimate saltmarsh extent and saltmarsh organic carbon storage. It is certainly intended to be a method development study but has significant amount of work on the effects of areal estimates on organic carbon storage estimates. Do the authors use the estimate of OC storage as a way to assess the reliability of different approaches in extracting saltmarsh extent? As a result, I am less certain of the paper's objective(s). In any case, I see some values of this work but would like to see some improvements.
  We have chosen to use carbon storage as a proxy to show the impact of our proposed method, as blue carbon storage is a major question in studies of near-coastal environments and because it fits within the scope of the special issue. We do not intend to use it as a means of testing the reliability of our approaches, but rather to demonstrate the impact of different areal extent estimates on estimates of carbon storage, which has knock-on effects on issues such as nature-based solution approaches and carbon budgeting. We have adjusted our third objective to better reflect this approach.
  I found the five parts in Results section confusing and not appropriate to respond to the three objectives the authors raised in Introduction section. Especially for section 3.1, the authors should explain why it was included in Results section. Section 3.2 and 3.3 should be moved to Methods section, as it describe how to classify vegetation communities and estimate areas.
  We appreciate that there is a case to be made for including Sections 3.2 and 3.3 in the methods section rather than the results section. However, we believe that each of these sections (3.1 to 3.3) represent the primary results of our field and UAV study and, although they do form the basis for the later mapping results (essentially forming our training dataset), they are best placed in the Results section rather than the Methods section.
  Several major concerns:
  1. How do the authors define the saltmarsh environments? Does it mean saltmarsh ecosystems?
  We appreciate the reviewer for highlighting to us our interchangeable use of environments, vegetation, ecosystems, etc. We have worked through the manuscript and standardised usage so that “saltmarsh vegetation (communities)” refers only to vegetation, whereas “saltmarsh environments” refers to the broader coastal system that includes vegetation characteristics, tidal characteristics, and the underlying saltmarsh sediments.
  2. There is a fundamental issue here that the paper did not explore. When classifying saltmarsh vegetation, the authors used the National Vegetation Classification (NVC) scheme. It seems not require training data and validation data as well, so how the mapping accuracy can be achieved? This relates to the following estimation of saltmarsh extent.
  The NVC system is an established and extensively used system based on an extensive database of vegetation data collected across the United Kingdom, representing almost all communities present in the UK. This classification scheme is based on approximately 35,000 samples, which removes the need for individual studies to develop training datasets. The MAVIS system then uses multivariate methods to statistically assign a newly surveyed vegetation community to an established existing community. We’ve restructured the sentence in section 2.4 explaining this to make it clearer.
  3. In Introduction section, the authors point out the location (elevation) of saltmarshes, that is, saltmarshes form between the high astronomical tide (HAT) and mean tidal level (MTL). In section 2.5.1, the authors stated that the saltmarsh area was estimated by calculating the area that inundated under each tidal condition. In this context, the authors focused on three tidal contexts: High Astronomical Tide (HAT), Mean Low Water Springs (MLWS), Mean High Water Neaps (MHWN). What is the connection between MTL, MLWS, and MHWN? Why the authors recognize the saltmarshes are expected to always be inundated?
  We thank the reviewer for highlighting that we have not fully explained the rationale behind these choices. These metrics do not mean that saltmarshes will always be inundated, but they reflect different periodicities of inundation which are thought to exert control on saltmarsh vegetation formation. We take the HAT because this represents the highest elevation that might be expected to be periodically (but not necessarily annually) inundated by sea water, and we use this to constrain the upper limits of the possible saltmarsh environments. We could not use MTL, as that data isn’t available from the UK Tidal Gauge Network from which we obtained the tidal data, and this isn’t necessarily the most useful value, and we have modified the text to more explicitly focus on the HAT, which we ultimately used for the analyses. We have changed several sections to account for this: the section in the Introduction introducing tidal ranges as key threshold for saltmarsh formation; Table 1 and Figure 4 (we ultimately don’t use MHWS in our later analyses, so we have removed the data); and we have restructured Section 2.5.1 to more explicitly explain our use of tidal data.
  4. Some paragraphs in Section 2.5 and Section 2.6 should be connected.
  We have reworked sections 2.5 and 2.6 to better explain our methods. Information on calculating our NDVI ranges has moved from Section 2.6 into Section 2.5.2, and Section 2.6 now only describes how we combine the data from 2.5.1 (tidal data) and 2.5.2 (NDVI data) to get our final estimates of saltmarsh area – these sections should now be better/more satisfactorily linked together.
  5. How to classify the saltmarsh vegetation? By using NVC scheme or NDVI extraction values described in section 2.5.2? The authors should add more details about vegetation classification.
  We have added some content to the start of section 2.5.2 to explicitly explain the relationship between the vegetation survey data, the subsequent NVC classification, and the use of that data to develop relationships between vegetation community and NDVI signatures. This is later expanded upon in Section 3.2.
  6. For the first objective of this work, the authors stated to delineate saltmarsh habitats. Vegetation composition is a key component of saltmarsh habitats, however, the related content is not well depicted in this paper. Please change the objective 1 more precisely.
  Objective 1 has been refined in response to comments from another reviewer, and in response to the terminology in point one of this review.
  7. Do the authors used the same OC storage estimation method with Haynes et al., 2016? Section 2.7 need more related descriptions.
  Haynes (2016) only mapped saltmarsh vegetation but did not estimate carbon storage; to create a comparison point for our carbon storage case study, we have therefore simulated the carbon storage based on their area estimates, using an identical method to the one we used to simulate carbon storage for our new area estimates. We have added a small paragraph at the end of section 2.7 explicitly stating that we used the same method to estimate the carbon storage that would be implied by the Haynes (2016) vegetation maps.
  Specific & minor concerns:
  1. Line 35: simple or straightforward?
  Amended to straightforward.
  2. Line 61: saltmarshes are not always small features.
  We have amended this sentence to clarify we’re referring to UAV's ability to map small saltmarsh areas.
  3. Line 94-98: these three sentences are not suitable for the study area section.
  We have moved these lines into Section 2.2, before the description of our survey methods.
  4. Section 2.3: this section only describes how to design and collect UAV data. Seems better to change to UAV data collection.
  We have amended this as recommended.
  5. Line 136-137: please state what parameter did the internal sunlight sensor process.
  We have amended this sentence to include the solar irradiance values collected by the UAV.
  6. Table 1: the tidal ranges are the mean values?
  We have added some additional context at the first point where we discuss the tidal data (section 2.5.1).
  7. Line 191: typo, rewrite this sentence.
  We cannot identify any typo around line 191.
  8. Figure 4: the Wick HAT in middle 2 figures is not identical to the legend color scheme.
  We will adjust Figure 4 to show only HAT, as this is the primary data we have chosen to ultimately use in the analyses, and will ensure that the colour correctly matches the legend by making the polygon opaque.
  9. Table 2: keep the same decimal place.
  The Elevation data is only produced to the nearest cm, whereas the NDVI data is produced to three decimal placed. Given they are not directly comparable, we don’t consider it necessary to amend the elevation to an unknown millimeter precision and we wouldn’t like to remove some of the detail in the NDVI data.
  10. Figure 5: abbreviation of SM10,…SM8 should be clarified.
  We have added a sentence in section 2.4, clarifying that they SM-X coding that appears through the paper represents a saltmarsh vegetation community as classified using the NVC. We appreciate the reviewer for highlighting that we had not explicitly stated this.
  11. The form of the tables (Table 1-5) are not appropriate.
  We’d be happy to amend the tables as requested, however we are not sure what form would be preferable. If the reviewer or editor would like to advise us further, we will make the suggested adjustments.
  
  Citation: https://doi.org/10.5194/egusphere-2023-1185-AC2
RC2:
'Comment on egusphere-2023-1185', Anonymous Referee #2, 07 Sep 2023

Overall, the manuscript is well written and an enjoyable read. The approach is novel and of interest to those working in saltmarshes and coastal wetlands. There are, however, a couple of limitations that need to be addressed by the authors. Having read the manuscript, I am left wondering why the authors did not use a classification method, such as the Random Forest approach used by Villoslada et al. (2020)? Further justification for not using an approach such this, especially when the authors have the available data, is required.
There is also a need to consider the wider context and implications of the work. For example, it would be interesting to see some discussion of application of the approach in other systems such as mangroves or even restored saltmarshes including managed realignment sites. There have been a number of studies into the use of UAS approaches and blue carbon in these settings, it would be beneficial to evaluate if the method developed in this study could be beneficial to these investigations. Other systems, beyond coastal environments, could also be evaluated here to increase the application of the work. With these additions, the manuscript would be considerably stronger and of wider appeal to those working in both UAS remote sensing and blue carbon.

Citation: https://doi.org/10.5194/egusphere-2023-1185-RC2
- AC1: 'Reply on RC2', William Hiles, 03 Oct 2023
  
  We would like to thank the reviewer for their positive comments at the beginning of the review. We would also like to thank them for the suggestions to expand the scope of the work to consider other approaches and potential broader implications. We hope the following responses are satisfactory.
  Having read the manuscript, I am left wondering why the authors did not use a classification method, such as the Random Forest approach used by Villoslada et al. (2020)? Further justification for not using an approach such this, especially when the authors have the available data, is required.
  Thank you for this suggestion, which would be a very interesting approach with our data. In general, our main aim in this paper has been to identify and test methods which could be easily and readily implemented for saltmarsh monitoring by a range of stakeholders, either within the research community or government agencies, for example. For this reason, we have opted to try to keep the methodology as simple and as widely applicable as possible, with minimal computing power and with as few variables as possible. We have added a short section in our Methods (Section 2.5) to explain this.
  However, this suggestion is incredibly interesting, and we would be very interested in exploring this further in the future with this dataset. It would be very interesting to test whether we could identify our discrete saltmarsh communities in our data using a wider variety of spectral signals and using machine learning approaches, which would represent a significant step forward in our ability to remotely map saltmarsh communities and carbon storage in the UK. We have added a line in the Conclusions pointing towards this potential.
  There is also a need to consider the wider context and implications of the work. For example, it would be interesting to see some discussion of application of the approach in other systems such as mangroves or even restored saltmarshes including managed realignment sites. There have been a number of studies into the use of UAS approaches and blue carbon in these settings, it would be beneficial to evaluate if the method developed in this study could be beneficial to these investigations. Other systems, beyond coastal environments, could also be evaluated here to increase the application of the work. With these additions, the manuscript would be considerably stronger and of wider appeal to those working in both UAS remote sensing and blue carbon.
  We thank the reviewer for this insightful comment. We have added a section (4.4) highlighting some of the applications for other issues in coastal environments aside from carbon storage (such as monitoring accretion and erosion, and vegetation colonisation), and looking towards other blue carbon communities (principally mangroves). Given the importance of tidal thresholds in our methods, we have chosen not to expand into non-coastal systems as the method risks losing some of its applicability, and it would expand the scope of the manuscript beyond that of the special issue.
  
  Citation: https://doi.org/10.5194/egusphere-2023-1185-AC1

Interactive discussion

Status: closed

RC1:
'Comment on egusphere-2023-1185', Anonymous Referee #1, 15 Aug 2023
General comments:
The paper describes an approach combing vegetation survey, UAV data and tidal data to estimate saltmarsh extent and saltmarsh organic carbon storage. It is certainly intended to be a method development study but has significant amount of work on the effects of areal estimates on organic carbon storage estimates. Do the authors use the estimate of OC storage as a way to assess the reliability of different approaches in extracting saltmarsh extent? As a result, I am less certain of the paper's objective(s). In any case, I see some values of this work but would like to see some improvements.
I found the five parts in Results section confusing and not appropriate to respond to the three objectives the authors raised in Introduction section. Especially for section 3.1, the authors should explain why it was included in Results section. Section 3.2 and 3.3 should be moved to Methods section, as it describe how to classify vegetation communities and estimate areas.
Several major concerns:
How do the authors define the saltmarsh environments? Does it mean saltmarsh ecosystems?

There is a fundamental issue here that the paper did not explore. When classifying saltmarsh vegetation, the authors used the National Vegetation Classification (NVC) scheme. It seems not require training data and validation data as well, so how the mapping accuracy can be achieved? This relates to the following estimation of saltmarsh extent.

In Introduction section, the authors point out the location (elevation) of saltmarshes, that is, saltmarshes form between the high astronomical tide (HAT) and mean tidal level (MTL). In section 2.5.1, the authors stated that the saltmarsh area was estimated by calculating the area that inundated under each tidal condition. In this context, the authors focused on three tidal contexts: High Astronomical Tide (HAT), Mean Low Water Springs (MLWS), Mean High Water Neaps (MHWN). What is the connection between MTL, MLWS, and MHWN? Why the authors recognize the saltmarshes are expected to always be inundated?

Some paragraphs in Section 2.5 and Section 2.6 should be connected.

How to classify the saltmarsh vegetation? By using NVC scheme or NDVI extraction values described in section 2.5.2? The authors should add more details about vegetation classification.

For the first objective of this work, the authors stated to delineate saltmarsh habitats. Vegetation composition is a key component of saltmarsh habitats, however, the related content is not well depicted in this paper. Please change the objective 1 more precisely.

Do the authors used the same OC storage estimation method with Haynes et al., 2016? Section 2.7 need more related descriptions.

Specific & minor concerns:
Line 35: simple or straightforward?

Line 61: saltmarshes are not always small features.

Line 94-98: these three sentences are not suitable for the study area section.

Section 2.3: this section only describes how to design and collect UAV data. Seems better to change to UAV data collection.

Line 136-137: please state what parameter did the internal sunlight sensor process.

Table 1: the tidal ranges are the mean values?

Line 191: typo, rewrite this sentence.

Figure 4: the Wick HAT in middle 2 figures is not identical to the legend color scheme.

Table 2: keep the same decimal place.

Figure 5: abbreviation of SM10,…SM8 should be clarified.

The form of the tables (Table 1-5) are not appropriate.
Citation: https://doi.org/10.5194/egusphere-2023-1185-RC1
- AC2: 'Reply on RC1', William Hiles, 03 Oct 2023
  
  We would like to thank the reviewer for their helpful comments which have highlighted several areas where we have been able to improve the manuscript. We hope we have satisfactorily responded to all suggestions below:
  The paper describes an approach combing vegetation survey, UAV data and tidal data to estimate saltmarsh extent and saltmarsh organic carbon storage. It is certainly intended to be a method development study but has significant amount of work on the effects of areal estimates on organic carbon storage estimates. Do the authors use the estimate of OC storage as a way to assess the reliability of different approaches in extracting saltmarsh extent? As a result, I am less certain of the paper's objective(s). In any case, I see some values of this work but would like to see some improvements.
  We have chosen to use carbon storage as a proxy to show the impact of our proposed method, as blue carbon storage is a major question in studies of near-coastal environments and because it fits within the scope of the special issue. We do not intend to use it as a means of testing the reliability of our approaches, but rather to demonstrate the impact of different areal extent estimates on estimates of carbon storage, which has knock-on effects on issues such as nature-based solution approaches and carbon budgeting. We have adjusted our third objective to better reflect this approach.
  I found the five parts in Results section confusing and not appropriate to respond to the three objectives the authors raised in Introduction section. Especially for section 3.1, the authors should explain why it was included in Results section. Section 3.2 and 3.3 should be moved to Methods section, as it describe how to classify vegetation communities and estimate areas.
  We appreciate that there is a case to be made for including Sections 3.2 and 3.3 in the methods section rather than the results section. However, we believe that each of these sections (3.1 to 3.3) represent the primary results of our field and UAV study and, although they do form the basis for the later mapping results (essentially forming our training dataset), they are best placed in the Results section rather than the Methods section.
  Several major concerns:
  1. How do the authors define the saltmarsh environments? Does it mean saltmarsh ecosystems?
  We appreciate the reviewer for highlighting to us our interchangeable use of environments, vegetation, ecosystems, etc. We have worked through the manuscript and standardised usage so that “saltmarsh vegetation (communities)” refers only to vegetation, whereas “saltmarsh environments” refers to the broader coastal system that includes vegetation characteristics, tidal characteristics, and the underlying saltmarsh sediments.
  2. There is a fundamental issue here that the paper did not explore. When classifying saltmarsh vegetation, the authors used the National Vegetation Classification (NVC) scheme. It seems not require training data and validation data as well, so how the mapping accuracy can be achieved? This relates to the following estimation of saltmarsh extent.
  The NVC system is an established and extensively used system based on an extensive database of vegetation data collected across the United Kingdom, representing almost all communities present in the UK. This classification scheme is based on approximately 35,000 samples, which removes the need for individual studies to develop training datasets. The MAVIS system then uses multivariate methods to statistically assign a newly surveyed vegetation community to an established existing community. We’ve restructured the sentence in section 2.4 explaining this to make it clearer.
  3. In Introduction section, the authors point out the location (elevation) of saltmarshes, that is, saltmarshes form between the high astronomical tide (HAT) and mean tidal level (MTL). In section 2.5.1, the authors stated that the saltmarsh area was estimated by calculating the area that inundated under each tidal condition. In this context, the authors focused on three tidal contexts: High Astronomical Tide (HAT), Mean Low Water Springs (MLWS), Mean High Water Neaps (MHWN). What is the connection between MTL, MLWS, and MHWN? Why the authors recognize the saltmarshes are expected to always be inundated?
  We thank the reviewer for highlighting that we have not fully explained the rationale behind these choices. These metrics do not mean that saltmarshes will always be inundated, but they reflect different periodicities of inundation which are thought to exert control on saltmarsh vegetation formation. We take the HAT because this represents the highest elevation that might be expected to be periodically (but not necessarily annually) inundated by sea water, and we use this to constrain the upper limits of the possible saltmarsh environments. We could not use MTL, as that data isn’t available from the UK Tidal Gauge Network from which we obtained the tidal data, and this isn’t necessarily the most useful value, and we have modified the text to more explicitly focus on the HAT, which we ultimately used for the analyses. We have changed several sections to account for this: the section in the Introduction introducing tidal ranges as key threshold for saltmarsh formation; Table 1 and Figure 4 (we ultimately don’t use MHWS in our later analyses, so we have removed the data); and we have restructured Section 2.5.1 to more explicitly explain our use of tidal data.
  4. Some paragraphs in Section 2.5 and Section 2.6 should be connected.
  We have reworked sections 2.5 and 2.6 to better explain our methods. Information on calculating our NDVI ranges has moved from Section 2.6 into Section 2.5.2, and Section 2.6 now only describes how we combine the data from 2.5.1 (tidal data) and 2.5.2 (NDVI data) to get our final estimates of saltmarsh area – these sections should now be better/more satisfactorily linked together.
  5. How to classify the saltmarsh vegetation? By using NVC scheme or NDVI extraction values described in section 2.5.2? The authors should add more details about vegetation classification.
  We have added some content to the start of section 2.5.2 to explicitly explain the relationship between the vegetation survey data, the subsequent NVC classification, and the use of that data to develop relationships between vegetation community and NDVI signatures. This is later expanded upon in Section 3.2.
  6. For the first objective of this work, the authors stated to delineate saltmarsh habitats. Vegetation composition is a key component of saltmarsh habitats, however, the related content is not well depicted in this paper. Please change the objective 1 more precisely.
  Objective 1 has been refined in response to comments from another reviewer, and in response to the terminology in point one of this review.
  7. Do the authors used the same OC storage estimation method with Haynes et al., 2016? Section 2.7 need more related descriptions.
  Haynes (2016) only mapped saltmarsh vegetation but did not estimate carbon storage; to create a comparison point for our carbon storage case study, we have therefore simulated the carbon storage based on their area estimates, using an identical method to the one we used to simulate carbon storage for our new area estimates. We have added a small paragraph at the end of section 2.7 explicitly stating that we used the same method to estimate the carbon storage that would be implied by the Haynes (2016) vegetation maps.
  Specific & minor concerns:
  1. Line 35: simple or straightforward?
  Amended to straightforward.
  2. Line 61: saltmarshes are not always small features.
  We have amended this sentence to clarify we’re referring to UAV's ability to map small saltmarsh areas.
  3. Line 94-98: these three sentences are not suitable for the study area section.
  We have moved these lines into Section 2.2, before the description of our survey methods.
  4. Section 2.3: this section only describes how to design and collect UAV data. Seems better to change to UAV data collection.
  We have amended this as recommended.
  5. Line 136-137: please state what parameter did the internal sunlight sensor process.
  We have amended this sentence to include the solar irradiance values collected by the UAV.
  6. Table 1: the tidal ranges are the mean values?
  We have added some additional context at the first point where we discuss the tidal data (section 2.5.1).
  7. Line 191: typo, rewrite this sentence.
  We cannot identify any typo around line 191.
  8. Figure 4: the Wick HAT in middle 2 figures is not identical to the legend color scheme.
  We will adjust Figure 4 to show only HAT, as this is the primary data we have chosen to ultimately use in the analyses, and will ensure that the colour correctly matches the legend by making the polygon opaque.
  9. Table 2: keep the same decimal place.
  The Elevation data is only produced to the nearest cm, whereas the NDVI data is produced to three decimal placed. Given they are not directly comparable, we don’t consider it necessary to amend the elevation to an unknown millimeter precision and we wouldn’t like to remove some of the detail in the NDVI data.
  10. Figure 5: abbreviation of SM10,…SM8 should be clarified.
  We have added a sentence in section 2.4, clarifying that they SM-X coding that appears through the paper represents a saltmarsh vegetation community as classified using the NVC. We appreciate the reviewer for highlighting that we had not explicitly stated this.
  11. The form of the tables (Table 1-5) are not appropriate.
  We’d be happy to amend the tables as requested, however we are not sure what form would be preferable. If the reviewer or editor would like to advise us further, we will make the suggested adjustments.
  
  Citation: https://doi.org/10.5194/egusphere-2023-1185-AC2
RC2:
'Comment on egusphere-2023-1185', Anonymous Referee #2, 07 Sep 2023

Overall, the manuscript is well written and an enjoyable read. The approach is novel and of interest to those working in saltmarshes and coastal wetlands. There are, however, a couple of limitations that need to be addressed by the authors. Having read the manuscript, I am left wondering why the authors did not use a classification method, such as the Random Forest approach used by Villoslada et al. (2020)? Further justification for not using an approach such this, especially when the authors have the available data, is required.
There is also a need to consider the wider context and implications of the work. For example, it would be interesting to see some discussion of application of the approach in other systems such as mangroves or even restored saltmarshes including managed realignment sites. There have been a number of studies into the use of UAS approaches and blue carbon in these settings, it would be beneficial to evaluate if the method developed in this study could be beneficial to these investigations. Other systems, beyond coastal environments, could also be evaluated here to increase the application of the work. With these additions, the manuscript would be considerably stronger and of wider appeal to those working in both UAS remote sensing and blue carbon.

Citation: https://doi.org/10.5194/egusphere-2023-1185-RC2
- AC1: 'Reply on RC2', William Hiles, 03 Oct 2023
  
  We would like to thank the reviewer for their positive comments at the beginning of the review. We would also like to thank them for the suggestions to expand the scope of the work to consider other approaches and potential broader implications. We hope the following responses are satisfactory.
  Having read the manuscript, I am left wondering why the authors did not use a classification method, such as the Random Forest approach used by Villoslada et al. (2020)? Further justification for not using an approach such this, especially when the authors have the available data, is required.
  Thank you for this suggestion, which would be a very interesting approach with our data. In general, our main aim in this paper has been to identify and test methods which could be easily and readily implemented for saltmarsh monitoring by a range of stakeholders, either within the research community or government agencies, for example. For this reason, we have opted to try to keep the methodology as simple and as widely applicable as possible, with minimal computing power and with as few variables as possible. We have added a short section in our Methods (Section 2.5) to explain this.
  However, this suggestion is incredibly interesting, and we would be very interested in exploring this further in the future with this dataset. It would be very interesting to test whether we could identify our discrete saltmarsh communities in our data using a wider variety of spectral signals and using machine learning approaches, which would represent a significant step forward in our ability to remotely map saltmarsh communities and carbon storage in the UK. We have added a line in the Conclusions pointing towards this potential.
  There is also a need to consider the wider context and implications of the work. For example, it would be interesting to see some discussion of application of the approach in other systems such as mangroves or even restored saltmarshes including managed realignment sites. There have been a number of studies into the use of UAS approaches and blue carbon in these settings, it would be beneficial to evaluate if the method developed in this study could be beneficial to these investigations. Other systems, beyond coastal environments, could also be evaluated here to increase the application of the work. With these additions, the manuscript would be considerably stronger and of wider appeal to those working in both UAS remote sensing and blue carbon.
  We thank the reviewer for this insightful comment. We have added a section (4.4) highlighting some of the applications for other issues in coastal environments aside from carbon storage (such as monitoring accretion and erosion, and vegetation colonisation), and looking towards other blue carbon communities (principally mangroves). Given the importance of tidal thresholds in our methods, we have chosen not to expand into non-coastal systems as the method risks losing some of its applicability, and it would expand the scope of the manuscript beyond that of the special issue.
  
  Citation: https://doi.org/10.5194/egusphere-2023-1185-AC1

Peer review completion

AR: Author's response | RR: Referee report | ED: Editor decision | EF: Editorial file upload

ED: Reconsider after major revisions (08 Oct 2023) by Sonia Silvestri

ED: Reconsider after major revisions (10 Oct 2023) by Jack Middelburg (Co-editor-in-chief)

AR by Craig Smeaton on behalf of the Authors (12 Nov 2023) Author's response Author's tracked changes Manuscript

ED: Referee Nomination & Report Request started (14 Nov 2023) by Sonia Silvestri

RR by Anonymous Referee #1 (25 Nov 2023)

ED: Publish subject to minor revisions (review by editor) (04 Dec 2023) by Sonia Silvestri

ED: Publish subject to minor revisions (review by editor) (07 Dec 2023) by Jack Middelburg (Co-editor-in-chief)

AR by Craig Smeaton on behalf of the Authors (20 Dec 2023) Author's response Author's tracked changes Manuscript

ED: Publish as is (01 Jan 2024) by Sonia Silvestri

ED: Publish as is (02 Jan 2024) by Jack Middelburg (Co-editor-in-chief)

AR by Craig Smeaton on behalf of the Authors (11 Jan 2024) Manuscript

Journal article(s) based on this preprint

20 Feb 2024

UAV approaches for improved mapping of vegetation cover and estimation of carbon storage of small saltmarshes: examples from Loch Fleet, northeast Scotland

William Hiles, Lucy C. Miller, Craig Smeaton, and William E. N. Austin

Biogeosciences, 21, 929–948, https://doi.org/10.5194/bg-21-929-2024,https://doi.org/10.5194/bg-21-929-2024, 2024

Short summary

William Hiles, Lucy Catherine Miller, Craig Smeaton, and William Edward Newns Austin

Viewed

Total article views: 339 (including HTML, PDF, and XML)

HTML	PDF	XML	Total	BibTeX	EndNote
233	77	29	339	16	14

HTML: 233
PDF: 77
XML: 29
Total: 339
BibTeX: 16
EndNote: 14

Views and downloads (calculated since 31 Jul 2023)

Month	HTML	PDF	XML	Total
Jul 2023	20	1	1	22
Aug 2023	79	21	8	108
Sep 2023	38	16	4	58
Oct 2023	59	14	12	85
Nov 2023	9	2	2	13
Dec 2023	7	7	0	14
Jan 2024	10	8	1	19
Feb 2024	11	8	1	20
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0

Cumulative views and downloads (calculated since 31 Jul 2023)

Month	HTML	PDF	XML	Total
Jul 2023	20	1	1	22
Aug 2023	79	21	8	108
Sep 2023	38	16	4	58
Oct 2023	59	14	12	85
Nov 2023	9	2	2	13
Dec 2023	7	7	0	14
Jan 2024	10	8	1	19
Feb 2024	11	8	1	20
Mar 2024	0
Apr 2024	0
May 2024	0
Jun 2024	0
Jul 2024	0
Aug 2024	0

Viewed (geographical distribution)

Total article views: 335 (including HTML, PDF, and XML) Thereof 335 with geography defined and 0 with unknown origin.

Country	#	Views	%

Latest update: 01 Sep 2024

Download

The requested preprint has a corresponding peer-reviewed final revised paper. You are encouraged to refer to the final revised version.

Preprint (2175 KB)
Metadata XML

Short summary

Saltmarsh soils may help to limit the rate of climate change by storing carbon. To understand their impacts, they must be accurately mapped. We use drone data to estimate the size of three saltmarshes in NE Scotland. We find that drone imagery, combined with tidal data, can reliably inform our understanding of saltmarsh size. When compared with previous work using vegetation communities, we find that our most reliable new estimates of stored carbon are 15–20 % smaller than previously estimated.


Total:	0
HTML:	0
PDF:	0
XML:	0

UAV approaches for improved mapping of vegetation cover and estimation of carbon storage of small saltmarshes: examples from Loch Fleet, northeast Scotland

Journal article(s) based on this preprint

Interactive discussion

Interactive discussion

Peer review completion

Suggestions for revision or reasons for rejection

Journal article(s) based on this preprint

Viewed

Viewed (geographical distribution)