Multiscale Diagnosis of Mangrove Status in Data-Poor Context Using Very High Spatial Resolution Satellite Images: A Case Study in Pichavaram Mangrove Forest, Tamil Nadu, India
Round 1
Reviewer 1 Report
This paper shows how by using very high spatial resolution in the context of Pichavaram mangrove area an improve result is obtain for multiscale diagnosis of the ecosystem situation for the coastal community. This paper improves the understanding of satellite imagery data usage in mangrove with the main focus of mangroves ecosystem monitoring.
Section 3.1 is not using Normalized Difference Vegetation Index, could that be included to strengthen the data in Table 2?
Not clear for High spatial resolution imagery could be a key factor when classifying vegetation communities accurately. Authors should clarify this point.
The paper would be benefit from a clear state of what window size for texture feature extraction was used as it is a very important consideration that is depended on the phenological characteristics of the plant communities of interest. Not clear if authors estimated this by using a semivariogram.
The thematic vegetation map can also serve to better analyse spatial patterns of the studied vegetation communities, identify habitats of interest and help determine wetland areas that are susceptible to change. Therefore, mapping these changes in graminoid community distribution and structure could be used as an indicator to monitor the negative effects of climate, ecological and anthropogenic impacts in the Pichavaram mangrove area and other similar wetlands as well.
I recommend this paper to be followed by a paper where authors purchase ancillary datasets for use in the analysis.
Author Response
Section 3.1 is not using Normalized Difference Vegetation Index, could that be included to strengthen the data in Table 2?
Yes, we confirm that we did not use NDVI in our analysis. Vegetation indices (such as NDVI) computed over wetland vegetation are often influenced by underneath water area to an extent difficult to predict without physical interpretation even in very high spatial resolution images (see references 27 and 43). Further, as long as all bands are used in the supervised classification process integrates information from both red and NIR bans there is no need of duplicating such information through NDVI. To clarify our text, we added at the very beginning of the section 2.3 (L217-220):
“Our main objective was to distinguish mangroves from non-mangrove areas to capture changes in the mangrove extent. For that, we developed our analysis through a supervised classification process based on all, panchromatic and multispectral, bands available for each image, as described below. We did not use any vegetation indices.”
- Taureau, F.; Robin, M.; Proisy, C.; Fromard, F.; Imbert, D.; Debaine, F. Mapping the Mangrove Forest Canopy Using Spectral Unmixing of Very High Spatial Resolution Satellite Images. Remote Sensing 2019, 11, 367, https://doi.org/10.3390/rs11030367
- Viennois, G.; Proisy, C.; Féret, J.-B.; Prosperi, J.; Sidik, F.; Suhardjono; Rahmania, R.; Longépé, N.; Germain, O.; Gaspar, P. Multitemporal analysis of high spatial resolution satellite imagery for mangrove species mapping, Bali, Indonesia. IEEE J. Sel. Top. Appl. Earth Observ. Remote Sens. 2016, 9, 3680-3686, http://dx.doi.org/10.1109/JSTARS.2016.2553170.
Not clear for High spatial resolution imagery could be a key factor when classifying vegetation communities accurately. Authors should clarify this point.
If we do understand well the meaning of this comment, yes, high spatial resolution imagery does improve the ability of mapping dominant species and vegetation communities. We discussed this point in 4.2 (Discussion section, Lines 487-507). Wang et al. 2016 propose a reference article for this point (see ref. 28 and below) in the Introduction and Discussion sections.
- Wang, T.; Zhang, H.; Lin, H.; Fang, C. Textural–spectral feature-based species classification of mangroves in Mai Po Nature Reserve from Worldview-3 imagery. Remote Sensing 2016, 8, 1-15, https://doi.org/10.3390/rs8010024.
The paper would be benefit from a clear state of what window size for texture feature extraction was used as it is a very important consideration that is depended on the phenological characteristics of the plant communities of interest. Not clear if authors estimated this by using a semivariogram.
Actually, we did not conduct any textural analysis on our satellite images. We had mentioned in our manuscript the limitations of conducting a textural or combined spectral-textural analysis for mapping and monitoring species distributions in mangrove ecosystems, especially for the Pichavaram mangroves. Please, see 4.2 (L497-501): “Forest canopy in the interior mangroves is formed by tree crowns of few square meters either densely distributed with uniform tree height or sparsely located in the case of fishbone plantations. In such a scenario, even the use of sophisticated textural or combined spectral-textural approaches [26,28] applied on VHSR images for mapping and monitoring species has its limitations [43]. This aspect could be addressed in future using combination of active sensors, such as LiDAR along with in–situ extensive field observations of species location and canopy structure and size.”
The thematic vegetation map can also serve to better analyse spatial patterns of the studied vegetation communities, identify habitats of interest and help determine wetland areas that are susceptible to change. Therefore, mapping these changes in graminoid community distribution and structure could be used as an indicator to monitor the negative effects of climate, ecological and anthropogenic impacts in the Pichavaram mangrove area and other similar wetlands as well.
We completely agree with this suggestion. Detailed mapping and time-series monitoring of species/community level information is indeed required to assess effects of climate change as well as local anthropogenic impacts on mangrove ecosystems such as Pichavaram. However, as mentioned in our manuscript, in order to map species/communities within the mangrove ecosystem, we do require geo-referenced information on species/communities to conduct spectral or textural classification. Unless such detailed information is available, detailed mapping or time-series analysis is very difficult. We hope to assess this problem in possible future efforts to map the Pichavaram mangroves.
I recommend this paper to be followed by a paper where authors purchase ancillary datasets for use in the analysis.
For sure, we welcome the suggestion for a follow-up study. However, as there are no coordinated efforts to collect data on mangrove species or on other environmental parameters, ancillary datasets are very difficult to acquire. If we are authorized by the Forest Department and financially supported to achieve a follow-up study, we hope to build on our current observations, and present a more comprehensive study on the status of the Pichavaram mangroves.
Reviewer 2 Report
Comment
This manuscript addressed a topic of mangrove in data-poor context using high spatial resolution satellite images in Pichavaram mangrove forest, Tamil Nadu, India. This topic is suitable for remote sensing fields. I reviewed it and provided some special comments as follows.
- Title and Abstract could reflect whole text but Title using “very high spatial resolution satellite images” is not so suitable. If so, authors should define what is “very high spatial resolution satellite images”?
- In Introduction chapter, putting Figure 1 is not suitable because Introduction should describe background information, significance, study purpose. Not show Figure 1 here. If it is an important information in your study, it could move to Materials and Methods.
- The important of Pichavaram mangrove forest should be emphasized in Introduction chapter because international audiences might be not know why it is so important.
- Materials and Methods is suitable. I have no further comment here.
- Result could reflect study purpose. I only have slight suggestion for authors. Figure 9 might have more description.
- I am not sure Figures 12 and 13 are important results of this study? If so, moving to Result chapter might be more suitable. I still could not clearly understand why comparison of mangrove extents as estimated by VHSR images is so important in this study?
- In Conclusion, I suggested combination of two paragraphs as one.
Overall, I feel that this study is interesting in remote sensing field. Therefore, I am pleased to recommend it for publication in the remote sensing after revised.
Comments for author File: Comments.pdf
Author Response
Title and Abstract could reflect whole text but Title using “very high spatial resolution satellite images” is not so suitable. If so, authors should define what is “very high spatial resolution satellite images”?
Actually, there are more than 210 articles in the Web of Science with a title including ‘very high spatial resolution’. So, we let the title as it is. However, we agree that a clear definition is necessary at the earliest in the text and for comparison with MSR (moderate spatial resolution) images defined as images with pixel size > 4m. That is why we introduced in the abstract (L25) and for the first occurrence of the acronym VHSR in the Introduction (L71-72): very high spatial resolution images are images provided with pixel size < 4 m.
In Introduction chapter, putting Figure 1 is not suitable because Introduction should describe background information, significance, study purpose. Not show Figure 1 here. If it is an important information in your study, it could move to Materials and Methods.
Through the Figure 1, we wanted that the readers/audience, even those who are not so familiar with the technical aspects of remote sensing can understand the comparative capabilities of Moderate Spatial Resolution (MSR) and VHSR images (we indicated spatial resolution in each image excerpts of the Figure 1). We think that placing Figure 1 in the Introduction brings the readers to the central idea of our work. Putting it in the Materials and Methods section (also possible) would, to our opinion, suggest that the distinct capabilities of MRS and VHSR images for mangrove studies are already well-quantified to be asserted as factual information: this is not the case. So, we prefer to let the Figure 1 within the introductory section.
The important of Pichavaram mangrove forest should be emphasized in Introduction chapter because international audiences might be not know why it is so important.
Yes. We added the following sentence with a reference to an important paper (Introduction, L97-99).
“This mangrove region takes part in a major restoration programme of wetlands of the east coast of India initiated in the 1990s by national and local authorities [32].”
Result could reflect study purpose. I only have slight suggestion for authors. Figure 9 might have more description.
- We updated both the Figure 9 legend itself and the text above (L365-L370), as underlined below:
The annual expansion rate of the mangroves within the fishbone plots implemented prior to or in 2003 is about 1.7% but could reach 4% for a number of plots where mangrove cover was between 20 to 40 % in 2003. However, a number of plots showed minor loss in mangrove cover and a single fishbone plot, located close to the mouth of the Uppanaru River, did not show expansion of its mangrove cover during this entire study period interval (Figure 9a, bottom evolution profile).
Figure 9. Changes in the mangrove cover (% relative to fishbone area) within individual fishbone plots implemented prior to or in (a) 2003 and (b) between 2005-2011.
I am not sure Figures 12 and 13 are important results of this study? If so, moving to Result chapter might be more suitable. I still could not clearly understand why comparison of mangrove extents as estimated by VHSR images is so important in this study?
Thanks for your comment. One of the important issues that we have attempted to highlight in our study is emphasizing the need for mapping and monitoring changes within the mangrove extents at the finest possible scale. Moderate resolutions often over-estimate mangrove areas, especially in mixed pixels scenario (Figure 12) and detailed information about the nature of changes can be captured in Figure 13. We think that both figures are important to aliment the discussion on the improved potential of VHSR imagery for the study of mangroves.
In Conclusion, I suggested combination of two paragraphs as one.
Yes, we agree. The two paragraphs in the conclusion section are combined and we synthesized the concluding idea (see > L698 in the Conclusion) as follows:
“… where lack of ancillary data might be an impediment. Through the study of Pichavaram mangrove region, we illustrated how VHSR images can provide, in a simple manner, robust and crucial information on fragile mangrove regions, especially where there is an apparent lack of forest data as well as infrequent monitoring. This remains a prerequisite to diagnose the ecosystem status, identify stressful conditions and warn against rapid degradation.”
Reviewer 3 Report
It will be useful to have more information for the reasons that you have used so many different satellite images.
If I understood well the classification shows the mangrove area? Is this correct? You can take any other qualitative measures?
Why you don't expect the results of the VHR satellite images?
Author Response
It will be useful to have more information for the reasons that you have used so many different satellite images.
Yes (if the reviewer’s comment is about the number of images acquired by different satellite sensors). Overall, the number of VHSR images available on the Pichavaram site since 2003 is not large for a given satellite sensor and a number of images were inoperable due to the important cloud cover. Consequently, the five VHSR images selected between 2003 and 2019 and purchased for the analysis correspond to a shortlisted dataset. We added the following text in 2.2.1 (L169-170):
“We searched for cloud-free VHSR satellite images covering the Pichavaram mangroves on the Maxar® web-catalog. We shortlisted five images and purchased them ;…”
If I understood well the classification shows the mangrove area? Is this correct? You can take any other qualitative measures?
Yes, the classification results delineate mangrove areas, non-mangrove areas and water. The LULC time-series was generated based on supervised classification, and some visual adjustment, particularly to remove pixels that were incorrectly classified as mangroves. No additional qualitative measures were taken.
Why you don't expect the results of the VHR satellite images?
We are not sure of the meaning of this question. Anyway, the potential of the VHSR satellite images deserved to be demonstrated for the Pichavaram region where mangrove monitoring was based on only moderate spatial resolution imagery. It is what we stated in the Abstract and in the Conclusion.