Downscale MODIS Land Surface Temperature Based on Three Different Models to Analyze Surface Urban Heat Island: A Case Study of Hangzhou
Round 1
Reviewer 1 Report
This a clear and well-written article about downscaling the large-scale images for urban heat island estimation. I particularly liked the flow chart in the Methodology section, which explains the calculation process in non-technical terms. However, there are some issues, which I detailed below. I would recommend the article for publication in the journal after all the issues will have been properly addressed.
General remark: Using “high-scale” and “low-scale” terms is misguiding. I recommend using more common terms such as “high-resolution” and “low-resolution” images/data.
Lines 68-79 represent, in fact, method description and should be moved to the Methodology section.
Equation 1 &2: you need to show what does the SP mean.
Line 93: Should it be “multivariate models” instead of “multivariate model”?
Line 108 and equation 3: Should be “linear”.
Line 146: Should it be “most affect” instead of “most affects”?
Line 157: As long as you talk about R2, I believe you should change “correlation” to “regression” in the phrase.
Line 158: I don’t understand what does it mean “can sensitivity measure”. Please, consider re-phrasing.
Lines 159-161: There is a repetition of the same idea (the consistency of the real imagery and simulation imagery) in one sentence. Please, consider re-phrasing.
Lines 195-196: “to enable users to about imageries”. I don’t understand the expression, please, consider re-phrasing.
Lines 196-201: Instead of presenting in detail the bands not useful for your models, you’d better describe those that are useful and were introduced in the computation.
Data sources and processing: It is not clear whether you used atmospheric and other types of corrections or not. You should specify the exact source of your images and what types of corrections were previously applied (if not made by you). If it was you who corrected the images, then explain how.
Figures 6, 7, and 9: What does the black dashed line mean? Common notation in the figure title is y vs. x.
Figure 6 and Table 1: It is always useful to add p-values for regression models and the coefficients. Thus the reader will have a much clearer image about the significance of your models.
Lines 230-231: Of course, the shape of the cloud shows it very clear, however, p-values serve much better for the justification of the low quality of the models.
Lines 239-241: You should compare p-values as well.
Lines 253-255: You should move this to the place where you first mentioned the sci-kit (line 135).
Line 289: “educed”??
Lines 287-290: P-values should be compared as well.
Line 295: Maybe “seasons” instead of “times” would sound better…
Lines 331-333: Please, provide the reference to this theory.
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The paper examined three models to downscale MODIS land surface temperature. The proposed methods were well explained and discussed. I think the manuscript is good and may be suitable to publish in Remote Sensing. I only have a few minor corrections
1. The paper needs an intensive editing before publication and authors should pay attention to the reference style of the journal, the coherence of paragraphs. Also, definition of equations variables should not be in new paragraph.
2. Also, the manuscript needs to be checked in the writing style. Do not use contractions like don’t, can’t, it’s and so on. They are informal in the scientific writing.
3. Line 49 and elsewhere, the number of the reference should follow the name of the author.
4. Line 141 “n” italic.
5. Figure 8. I think it is better to use a, b, c, letters for each graph and refer to them in the text and figure caption.
6. Line 195 “These remote sensing imageries use the World Reference System to enable users to about imageries over any portion of the world by path and row numbers” I do not understand this sentence. Authors may mean that the data were already rectified to the World Reference System (WGS1984). If so, this sentence is better than the existing sentence.
7. What is the difference between figure 13 and 14? Yes, figure 13 for the daytime and figure 14 for the nighttime but should be clear in the text. “…, during daytime and at night respectively”.
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
After starting reading, I found the following language related problems just on page 1. After that, I stopped reading and just scanned through the paper.
I have major concerns about the validation of results because the paper does not address the validation procedure (at least cross validation is required, testing with independent data is suggested). Moreover, the authors wrote that the problem of overfitting is addressed by parameter tuning.
Typically, parameter tuning results in overfitting! Avoiding the overfitting of models is addressed by the Random Forest algorithm itself and must be validated with Independent or unseen data.
If some kind of proper validation was used, this must be clearly stated. So far, the results seem to be interesting, but proof is a must.
Please revise and resubmit.
P1
L13: spatil -> spatial
L23: that at night was generally WHAT??
L32: and IS widely used
L33: urban heat island ANALYSIS/DETECTION?
L33: other fields
L34: in the fields
L37: single data -> single measurement/single dataset
L39: of about
L40: resolution of
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Round 2
Reviewer 3 Report
Thank you for the fast revision of the manuscript and the clarification regarding my initial concerns.
The English editing Service greatly improved the readability of the paper. From my side, it can be published in the current form.
