Global Land Surface Temperature Influenced by Vegetation Cover and PM_2.5 from 2001 to 2016

Round 1

Reviewer 1 Report

Review paper remotesensing-394583

The authors address the issue of surface temperature evolution from 2001 to 2016 at global scale and its correlation with vegetation index NDVI and PM2.5. They perform a statistical analysis based on Level 3 (?) gridded satellite data. The analysis is somewhat lacking physical insights, although the statistical results are interesting and reasonable. The main problem of the paper is a sloppy English with a lot of awkward sentences.

Here some suggestions and remarks to improve reading and understanding of the paper.
Pag. 1, line 45, change increasing with is increasing
Pag. 2, line 52, change of natural with for natural
Pag. 2, line 60. As authors say there are many studies focused on retrieval algorithms and validation, therefore to summarize the vast literature with two references alone is not appropriate. Add after Ref 13 these two more doi:10.1016/j.rse.2012.12.008, doi:10.5194/amt-6-3613-2013, and add to Ref 14 these further two, doi:10.5194/amt-8-2981-2015, doi:10.3390/rs10060976
Pag. 2, line 67, cancel out also after El Nino
Pag. 2, line 68, change becoming with is becoming
Pag. 3, line 111, rephrase point 3) as to analyze the factors influencing LST.
Pag. 3, line 125-127. Why do you think the scale factor 0.2 is so important to be mentioned? Remove the first sentence and rephrase the second as: The Mean Value Composite (MVC) method has been used to calculate yearly values of LST for the period 2001 to 2016.
Pag. 4, line 143, remove all the sentence After….and rephrase with: The procedure is iterated until all Δ? are within the given threshold.
Page 4, line 153, I was not able to reach the website. Check if it is still active. Also, I would rather do the reverse, instead of interpolating PM2.5 to 0.05, I would have preferred to relax the MODIS data spatial resolution form 0.05 to 0.1. Some more explanations is needed here.
Page 4, line 155, via ArcGIS10.2….. is not needed, cancel out up to the full stop.
Page 4, line 163, Change the complete sentence Time series data…..with Monthly time series were processed to form yearly time series; in this way, for each pixel the time series has a length n=16, spanning from 2001 to 2016.
Page 4, line 169, cancel out the sentence in brackets (in this paper n=16)
Page 5, line 176. At the end of this section it is important to stress that slope and range are not affected by possible constant biases present in the data. For sure, there are biases in LST, NDVI and PM2.5. So its is important that authors stress their analysis is insensitive to constant biases, because it gains more credibility to the analysis.
Page 5, line 200, Pearson not Personal
Page 5, line 202 to 204. Cancel out the whole sentence, it is not needed.
Page 5, line 208, change around the world with at the global scale
Page 6, line 221-222, cancel out the whole sentence after the comma in other words….
Page 6, line 229, change making with has
Page. 8, line 229, Pearson instead of personal
Page 8, Fig. 2, the dark and light green cannot be distinguished in Fig. 2a, consider changing colorbar, the same for Fig. 2b. Caption, line 269, change bans with bins
Page 10 line 327, cancel out obviously, the same as on line 366.
In general, please consider to review the English everywhere. It would be better if the paper could be read and checked a native English speaker before resubmitting it.

Author Response

Response report to reviewer 1

Author Response File: Author Response.docx

Reviewer 2 Report

The work presents a multitemporal study (2001-16) using Terra MODIS LST data to investigate spatiotemporal variations and dynamics of LSTs in relation to NDVI and PM2.5, at a global scale.

The state of the art is poorly addressed, and it is not sufficiently referenced in the result discussion. Some parts are not clear and must be improved. The formatting of the paper must be also improved.

Below, specific comments:

Abstract:

- It is not significant to report three decimal digits for the temperature, especially if compared with the estimation accuracy of MODIS LST (on the order of 1 K).

- Line 26: what is the temporal scale of the reported increase?

Introduction:

- The state of the art concerning the relationships between LST/UHI and aerosol pollutants should be added. Several works studied the effect on UHI of atmospheric pollutants, and their findings are useful to understand the results reported in the manuscript and to improve the Discussion section.

-The state of the art concerning the NDVI and LST correlation analysis should be added. A huge literature exists.

-Line 108: “several parameters”: how many? Or only the two cited?

-Aims of the work: a clear explanation of data, data sources, time period, is missing

2. Materials and Methods

- Since the satellite is sun-synchronous, how is managed the different time of acquisition on a global scale? And what about day and night acquisitions? Since the LST is very different at different daytimes.

-Line 127: what is the Mean Value Composite (MVC) method? How does it work?

-Line 142: “a small percentage of the multiyear average NDVI value for each pixel was set at a threshold (Δ)”. This operation is not clear. How is this threshold selected?

-Lines 145-150: the explanation and syntax must be improved. How is SINDVI computed? Please, provide a more technical and scientific explanation on the basis of the used dataset.

- Line 158: “Because of lower PM2.5 concentrations in the Arctic and Antarctica, these data could meet the requirements of this study”: Why? Is it not clear.

-Section 2.2.1: the authors refer to the slope and range for each pixel. How many pixels are in each annual image? What is the pixel size in km?

-Subsection 2.2.2: the correlation coefficient is a well-known parameter. Eq. 5 is not necessary, and the explanation can be shortened (it is obvious the meaning of negative and positive correlation).

-Line 200 and 259: “Personal” instead of Person.

3.Results.

-Reporting three decimal digits for temperature is not significant (considering the MODIS accuracy).

- Figure 1a: the ROI are referred to single pixels or to larger areas? Please, specify in a quantitative way.

-Figure 2: for a better understanding, the colour of the 4 correlation levels should be chosen differently to improve the contrast.

-Line 276-287: the effect of aerosol and pollutions on LST is not simple or expected positive as pointed out by the authors. As suggested above, literature works on this topic are already published. Based on this state of the art (to be added in the Introduction), the discussion must take into account the literature findings. The same in the discussion between LST and vegetation cover.

-Section 3.4 and figure 4: Are the results a mean over the entire country? Or how large is the ROI? How many pixels? It is not clear, especially with reference to Figure 1. Also, across the years, it does not appear as a significant decrease, since the trend is quite variable.

-line 320: “…but LST and SINDVI grew simultaneously in Saudi Arabia”. It is not clear the sentence, since this area is here explained as an area with LST decreasing trend.

4.Discussion.

-A sub-section listing the limitations (technical and meaning standpoint) of the proposed work should be added.

5.Conclusion

- In the conclusion, future research directions should be highlighted.

Author Response

Response report to reviewer 2

Author Response File: Author Response.docx

Reviewer 3 Report

The manuscript “Global Land Surface Temperature Influenced by Vegetation Cover and PM2.5 from 2001 to 2016” covers a very important topic as it try to find out climate indicators that we can monitor by satellites. The authors propose to integrate NDVI and PM2.5 for each year and see the correlations with LST.

These two parameters are not something really new, at least correaltion with NDVI has been studied for several times. So perhaps add some text on why exactly you have chosenthese two parameters and why do you neglect some other.

I must say that I haven’t seen the use of SINDVI (integrated NDVI per year) and I find it as a very good option to aggregate annually values. However, I miss some discussion on how much of uncertainty you bring with this into analysis. The proccesses you study are not linear!

In addition, I think that you should provide a deeper discussion on the trends of LST, SINDVI and PM2.5. Here I refer to their independent evolution, do not comapre them. I find it quite problematic how you neglect for instance forest fires – these might result in a little bit higher temperature as long they burn, once they are out the area warms quicker and more than the healthy vegetation…

Have you considered the saturation of NDVI_ Some areas with a lot of vegetation show low correlation with LST also, because the NDVI is in those areas saturated (close to 1 and it cannot get higher). Discuss if some other index might be there more appropriate.

Considering all, I suggest a major revision. Some smaller remarks are given below

PM25 accuracy?

106 explain SINDVI

125-127 irrelevant, this is a standard MODIS product and it doesn’t mater how the data are saved, rather give a reference on this and delete this paragraph

143 unknown???

155 it is about the method and not the exact tools you use – delete all related to ArcGIS as it is irrelaevant, it could be done by other tools just as good

see my remark to 421 below

386 snow

407-408 it might be correlated but you give no direct prove that LST is influenced by PM2.5

421 SINDVI increase of 58 % - this might somebody interpret as the vegetation is spreading, but it might be that it is just in better shape, so add some short explantion on this before (as you define the SINDVI)

Author Response

Response report to reviewer 3

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

I am pleased  with the revision and I think the paper can be published

Author Response

Thanks for your very positive comments on our manuscript.

Reviewer 2 Report

The authors addressed the comments raised up in the first round efficiently, improving the clarity of the text and adding the required parts that are essential for a better appreciation of the work.

Minor comment:

-Line 180: For a better, clarity, I would replace “n” with “n=16”.

-The information about the number of pixels in each image/band (3600 * 7200) and the pixel size (5.6 km × 5.6 km at the Equator) should be added for completeness.

Author Response

reviewer report (round 2)

Author Response File: Author Response.docx

Reviewer 3 Report

My remarks were addressed, so I am OK to publish the manuscript.

Author Response

Thanks for your very positive comments on our manuscript.