Assessing Landsat Images Availability and Its Effects on Phenological Metrics
Round 1
Reviewer 1 Report
The subject of this paper is of a great importance. Case-study is also relevant for modern research agenda.
The text is well-written, clear and concise. Illustrations are rich and interesting to explore.
Though my overall impression is favorable, however there are some major points that should be addressed before the paper is published.
- Literature review is a bit superficial. The remote sensing topic is widely covered by dozens of thousands of research material, so it is too optimistic to review only 31 references. This must be extended.
- The choice of methods is not grounded with some logic. E.g., why only K-means algorithm is used for classification purposes? There are many other options (first of all, K-medoids), so it is needed to refer on them and then explain why do you use particularly K-means.
- Conclusions are formulated in a very general manner and do not reflect the vivid advantages of the paper connected with results of estimates on Brazilian regions.
Some minor issues after reading the text are below.
L40. Missing reference (? sign instead inside parentheses).
L100–101. Double and. Would you consider replacing one with a comma?
Caption of fig. 3. Why don't you simplify the text inside parentheses to just (2014–2019)?
Author Response
We thank the reviewer for her/his relevant comments and suggestions. All the proposed changes have been done. In the following paragraphs, the way we attended each comment is written in blue. We attach the new improved version of the manuscript, and a version with changes highlighted in blue (inserted text) and red (deleted text).
Review:
The subject of this paper is of a great importance. Case-study is also relevant for modern research agenda.
The text is well-written, clear and concise. Illustrations are rich and interesting to explore.
Though my overall impression is favorable, however there are some major points that should be addressed before the paper is published.
-
Literature review is a bit superficial. The remote sensing topic is widely covered by dozens of thousands of research material, so it is too optimistic to review only 31 references. This must be extended.
-
The choice of methods is not grounded with some logic. E.g., why only K-means algorithm is used for classification purposes? There are many other options (first of all, K-medoids), so it is needed to refer on them and then explain why do you use particularly K-means.
-
Conclusions are formulated in a very general manner and do not reflect the vivid advantages of the paper connected with results of estimates on Brazilian regions.
Some minor issues after reading the text are below.
L40. Missing reference (? sign instead inside parentheses). The missing reference has been added
L100–101. Double and. Would you consider replacing one with a comma? done
Caption of fig. 3. Why don't you simplify the text inside parentheses to just (2014–2019)? done
Reviewer 2 Report
Comments for the work entitled "ASSESSING LANDSAT IMAGES AVAILABILITY AND ITS EFFECTS ON PHENOLOGICAL METRICS":
- The abstract must include the objective pursued by this work.
- Line 40: Include the quote in [? ]. Correct the following references.
- Lines 41-42: In both cases, use numbers or letters to indicate the days.
- It is convenient to include in section 1, both the starting hypotheses (once the literature has been reviewed, although I think there are only 15 references) and the structure of the rest of the manuscript (this so that the reader can have an initial guide).
- Why do you think it is better to use Google Earth Engine than other platforms (besides being free)? What limitations does this plant have and how can it affect the results obtained? What bibliography has used it successfully? In which other were the results not adequate?
- It would be interesting to include a graph with the phases of the applied methodology.
- Likewise, you should explain succinctly why you applied the method of Klosterman et al. and Filippa et al.
- Figure 1: What is the red rectangle which represents the study area in Northeastern Brazil?
- It is not clear what each of the clusters refers to.
- The discussion should be more extensive. They should specify how the temporal variation has been examined and how this correlates with the reviewed literature. Are the results similar or are there significant differences? What does the study contribute in this very small time range, that is, can we obtain any relevant conclusion or would it be necessary to expand the study? And if it is compared with what happened in another area of the same Brazil, what happens?
- Along the same lines, the conclusions should be reinforced.
Author Response
We thank the reviewer for her/his relevant comments and suggestions. All the proposed changes have been done. In the following paragraphs, the way we attended each comment is written in blue. We attach the new improved version of the manuscript and a version with changes highlighted in blue (inserted text) and red (deleted text).
Review:
Comments and Suggestions for Authors
- The abstract must include the objective pursued by this work. We inserted the sentence “This paper aims to evaluate whether there is sufficient availability of Landsat 7 and 8 images over Brazil to support the analysis of phenodynamics of vegetation“ into the abstract.
- Line 40: Include the quote in [? ]. Correct the following references. done
- Lines 41-42: In both cases, use numbers or letters to indicate the days. done
- It is convenient to include in section 1, both the starting hypotheses (once the literature has been reviewed, although I think there are only 15 references) and the structure of the rest of the manuscript (this so that the reader can have an initial guide).
We added the hypothesis and the structure of the paper after the objectives:
This study aims to evaluate the availability of Landsat 7 and 8 images over the Brazilian territory, seeking to verify whether there is sufficient availability to support studies of the analysis of phenodynamics of vegetation. We hypothesise that it will be challenging to characterise vegetation phenology in regions where the number of valid (cloud-free) remotely-sensed observation is low or when the observations are unevenly distributed over the year.
This paper is organised as follows. Section 2 describes the materials and the methods to map Landsat imagery availability and the modelling of the effect of valid observations scarcity on phenological metrics estimation. Section 3 presents i) the Spatio-temporal patterns of image availability at the national level and for an area of the Brazilian Northeast semi-arid and, ii) the uncertainty of phenological metrics related with the different patterns of availability in the semi-arid area. We discuss the impacts of remotely-sensed observation scarcity on land studies in section 4. Finally, Section 5 presents our conclusions.
- Why do you think it is better to use Google Earth Engine than other platforms (besides being free)? What limitations does this plant have and how can it affect the results obtained? What bibliography has used it successfully? In which other were the results not adequate?
- It would be interesting to include a graph with the phases of the applied methodology. Google Earth
Yes it is a good idea. A flowchart of the entire analysis was included.
- Likewise, you should explain succinctly why you applied the method of Klosterman et al. and Filippa et al.
We added: We chose the Klosterman method because Zhou (2018) identified it as the most robust among four models to determine the phenological metrics for vegetation, presenting a single growth season.
- Figure 1: What is the red rectangle which represents the study area in Northeastern Brazil?
It represents the boundaries of the study area in Northeastern Brazil. We added this information in the legend of figure 2 and as a mention in the text which introduces the Northeastern Brazil study area.
- It is not clear what each of the clusters refers to.
We explain better the role of K-means clustering: K-means is one of the most common clustering algorithms, it is simple to implement, guarantees convergence and scales to large data sets. It carries out clustering making the intra-cluster observations as similar as possible while keeping the clusters as different as possible. Each resulting cluster represents a different temporal pattern of data availability. We added the mention Each cluster represents a different pattern of valid observations availability over the year into the caption of Figure 5.- The discussion should be more extensive. They should specify how the temporal variation has been examined and how this correlates with the reviewed literature. Are the results similar or are there significant differences? What does the study contribute in this very small time range, that is, can we obtain any relevant conclusion or would it be necessary to expand the study? And if it is compared with what happened in another area of the same Brazil, what happens?
- Along the same lines, the conclusions should be reinforced.
Discussion and conclusions were reinforced with the inclusion of new paragraphs/sentences. The number of references has also been significantly improved (20 news citations). See new version and version with tracking changes.
Round 2
Reviewer 1 Report
The authors have made a substantial progress with my suggestions. The paper has been improved much.
The paper deserves to be processed by the editors and then published.
Wish a good luck to the authors.
Reviewer 2 Report
Comments:
- Place Figure 6 in line 166.
