Analyzing the Relationship between Developed Land Area and Nighttime Light Emissions of 36 Chinese Cities
Round 1
Reviewer 1 Report
The current version offers some minor improvements and addresses some, though not all, of my concerns in the revisions to the Implications section. Nonetheless, I think the paper's argument has been clarified enough to warrant publication.
Prior to publication, the paper (especially the Abstract and the revised Implications section) could still benefit from professional editing for English usage.
Minor comments regarding figures:
Table 1 lists the 36 cities in guobiao numeric order (common in PRC statistical publications) from Beijing to Xining. Figures 3, 4, and 8 start with Beijing but don't present the cities in the standard order -- can you explain, or revise for consistency?
In Appendix Figure 1, RuralNDVI should be RuralNBTI.
Author Response
Response to reviewer’s comments
Dear reviewer, thanks again for your kind comments and suggestions for enhancing quality of this manuscript.
The authors
1.The current version offers some minor improvements and addresses some, though not all, of my concerns in the revisions to the Implications section. Nonetheless, I think the paper's argument has been clarified enough to warrant publication.
Prior to publication, the paper (especially the Abstract and the revised Implications section) could still benefit from professional editing for English usage.
Answer: We have had the paper edited by MDPI’s English Editing Service.
2.Minor comments regarding figures:
Table 1 lists the 36 cities in guobiao numeric order (common in PRC statistical publications) from Beijing to Xining. Figures 3, 4, and 8 start with Beijing but don't present the cities in the standard order -- can you explain, or revise for consistency?
Answer: We have changed the order of the cities in Table 1 to keep it consistent with that of Figures 3,4, and 8.
3.In Appendix Figure 1, RuralNDVI should be RuralNBTI.
Answer: Yes, it should be RuralNBTI. We have corrected the mistake.
Author Response File: 
Author Response.docx
Reviewer 2 Report
The manuscript has been improved and now is easier to read. Here are some comments for the authors.
[1] Line 171: It should be made clear here that CUDV is the unit for NTBI. I do not see Appendix Figure 1.
[2] Line 260: "featured with per pixel DN" should be CUDV?
[3] Caption of Figure 5 seems wrong. It looks like Figure 5 is showing the extent of sub-categories in land development, rather than pixel-based unitless DN values.
[4] Caption of Figure 6 also seems wrong. As CUDV is showing aggregated DN values for pixels in an area, it should no longer be representing per pixel values as in a figure. Furthermore, I do not think Figure 6 is showing sub-categories of developed land.
[5] Figure 7: lg_totalNTBI should be ln_totalNTBI?
[6] Line 507 and Line 511: Line numbers from previous manuscript got copied errorneously?
[7] Line 508: Where is this R-square value from? In Figure 7 there is no over all trend showed. The closest R-square number comes from year 2013.
Author Response
Response to reviewer’s comments
Dear reviewer, thanks again for your kind comments and suggestions for enhancing quality of this manuscript.
The authors
The manuscript has been improved and now is easier to read. Here are some comments for the authors.
[1] Line 171: It should be made clear here that CUDV is the unit for NTBI. I do not see Appendix Figure 1.
Answer: Yes, CUDV is the unit for NTBI. The CUDV was embedded in Appendix Figure 1.
[2] Line 260: "featured with per pixel DN" should be CUDV?
Answer: Yes, we have corrected the mistake.
[3] Caption of Figure 5 seems wrong. It looks like Figure 5 is showing the extent of sub-categories in land development, rather than pixel-based unitless DN values.
Answer: Yes, we have corrected the mistake.
[4] Caption of Figure 6 also seems wrong. As CUDV is showing aggregated DN values for pixels in an area, it should no longer be representing per pixel values as in a figure. Furthermore, I do not think Figure 6 is showing sub-categories of developed land.
Answer: Yes, we have corrected the mistakes.
[5] Figure 7: lg_totalNTBI should be ln_totalNTBI?
Answer: As strongly suggested by another reviewer, we used the normal logarithm of totalNTBI (lg_totalNTBI) instead of the natural logarithm of totalNTBI.
[6] Line 507 and Line 511: Line numbers from previous manuscript got copied?
Answer: Yes, we have deleted the line numbers erroneously copied from previous manuscript.
[7] Line 508: Where is this R-square value from? In Figure 7 there is no overall trend showed. The closest R-square number comes from year 2013.
Answer: The R-square value has been changed to the mean adjusted R2 of 0.799 in both the abstract and conclusion.
Author Response File: Author Response.docx
Reviewer 3 Report
Dear authors,
Your paper on “Examining the relationship between developed land 2 and associative nighttime light emissions of China’s 3 36 key cities (2000-2013)”, is an interesting additions to large number of studies to relate NTL emission and relating it with urban growth and socio-economic variable. However, I have several concerns related to the set-up of your research and methodology, which would require clarification:
- I do not fully understand your title – what are “associative nighttime light emissions” – the term associative is not clear? Maybe consider to modify your title.
- Another term, which would require reconsideration is “satellite nighttime light emissions” – the satellite is not having NTL emission but captures these emissions.
- Another term, to reconsider is “associative TotalNTBI”
- Line 37 – the urbanization figures seems wrong, could you please check e.g., https://www.un.org/development/desa/publications/2018-revision-of-world-urbanization-prospects.html
- Please also better clarify what is the innovation of your work (e.g., in Line 44 and70-71) – there have been many studies on NTL and urbanization also with the focus on China. Please clarify what is your specific contribution.
- Line 63-64 need clarification/reformulation “changing NTLE may be attributed to direct/indirect influences of driving factors with high complexity”?
- Table 1 the population data seems wrong e.g. Beijing has 21 Mio inhabitants and not 2.1 Mio.
- Line 100: I would suggest not to use the term primary data – e.g. land use data are secondary data.
- Line 115: which years are selected and why?
- Line 118-119: why do you use UTM – for the entire country of China?
- Line 122: 100 GCP seems very few points – and how have these points been used.
- Line 145 (equation 1): Too me, this seems to be a masking operation and not a resolution sharpening – in principle you ignore all NTL outside your built-up mask.
- Line 165 (the index) - what does it explain? The results and conclusions seems obvious: the larger cities have the most NTL emissions.
- Table 3 and 4 -> many of your factors are not significant and should be not used. Did you analyze before doing the regression the collinearity of your factors?
- Overall the paper would require a revision in terms of content, structure and language. Please also try to shorten and get more concise.
Author Response
Response to reviewer’s comments
Dear reviewer, thanks again for your kind comments and suggestions for enhancing quality of this manuscript.
The authors
Dear authors,
Your paper on “Examining the relationship between developed land 2 and associative nighttime light emissions of China’s 3 36 key cities (2000-2013)”, is an interesting additions to large number of studies to relate NTL emission and relating it with urban growth and socio-economic variable. However, I have several concerns related to the set-up of your research and methodology, which would require clarification:
- I do not fully understand your title – what are “associative nighttime light emissions” – the term associative is not clear? Maybe consider to modify your title.
Answer: We have deleted the term associative in the title of the revised version.
- Another term, which would require reconsideration is “satellite nighttime light emissions” – the satellite is not having NTL emission but captures these emissions.
Answer: The term was changed to ‘satellite observed nighttime light emissions’ was used.
- Another term, to reconsider is “associative TotalNTBI”
Answer: The term ‘associative TotalNTBI’ was changed to ‘associated TotalNTBI’.
- Line 37 – the urbanization figures seems wrong, could you please check e.g., https://www.un.org/development/desa/publications/2018-revision-of-world-urbanization-prospects.html
Answer: We used the 2017 data reported by UN. Since annual global urban population data will change, we have updated the population data according to the latest version of UN report.
- Please also better clarify what is the innovation of your work (e.g., in Line 44 and70-71) – there have been many studies on NTL and urbanization also with the focus on China. Please clarify what is your specific contribution.
Answer: We have clarified our contributions in the discussion section.
- Line 63-64 need clarification/reformulation “changing NTLE may be attributed to direct/indirect influences of driving factors with high complexity”?
Answer: What we meant is that the changing NTLE cannot not be simply attributed to some socioeconomic driving factors. It is not enough to address the relationship between human settlement and associated NTLE, given there are direct/indirect influences of socioeconomic driving factors. We also emphasized such shortcomings in Lines 63-71 and the discussion section.
- Table 1 the population data seems wrong e.g. Beijing has 21 Mio inhabitants and not 2.1 Mio.
Answer: Many thanks. We have corrected the table.
- Line 100: I would suggest not to use the term primary data – e.g. land use data are secondary data.
Answer: By primary data, we meant major data rather than first-handed data. But, to avoid potential confusion, we deleted the word ‘primary’ in the revised version.
- Line 115: which years are selected and why?
Answer: To be consistent with national LULC data, DMSP/OLS data were in 2000, 2005, 2008, 2010, and 2013.
- Line 118-119: why do you use UTM – for the entire country of China?
Answer: Given the popularity of UTM system and possibility of international comparison., we decided to use UTM projection in our study. In one of our parallel studies, the recently released 30-m global land cover products (http://www.globeland30.cn/GLC30Download/index.aspx), which were projected in WGS_1984_UTM system, will be used to produce better DMSP/OLS data for international comparative study.
- Line 122: 100 GCP seems very few points – and how have these points been used.
Answer: As shown in Figure 2, because the raw DMSP/OLS data is rather coarse, it is very difficult to find enough GCPs for it. For land cover products and raw DMSP/OLS data acquired in each year, 100 GCPs with distinct geographical features were carefully selected by overlapping the digitalized maps, land cover maps and other auxiliary datasets. In view of the file size and limited computer processing capacity, these points were used to wrap the maps with the second order polynomial method, and the resulting RMSE was within one pixel.
- Line 145 (equation 1): Too me, this seems to be a masking operation and not a resolution sharpening – in principle you ignore all NTL outside your built-up mask.
Answer: Many thanks. We have replaced the term resolution sharpening with edge sharpening.
- Line 165 (the index) - what does it explain? The results and conclusions seems obvious: the larger cities have the most NTL emissions.
Answer: The indices in line 164 to 167 reflected the total NTL emissions associated with the total developed land and the three sub-categories of developed lands (urban, rural, and the other developed lands).
Although larger cities tend to have most NTL emissions, our analysis results have found the relationship between developed land and NTL emissions are more complex. First, there is much variation is the NTL emissions associated with different sub-categories of land. For instance, Zhengzhou’s RuralNTBI was higher than that of Guangzhou and Shenzhen, while its UrbanNTBI and OtherNTBI were lower than that of Guangzhou and Shenzhen. The resulting TotalNTBI of Zhengzhou was lower than that of Guangzhou but higher than Shenzhen. Therefore, TotalNTBI and its three sub-categories of NTBIs can help explain the spatio-temporal patterns of NTL emissions influenced by developmental inequality between different regions. Secondly, although both the urban developed land and the other developed land were in good agreement with their associated NTBIs, the relationship between rural developed land and RuralNTBI was relatively weak. This has led to the mismatched relationship between the total developed land and TotalNTBI in some cities, whose rural developed land was larger than their urban developed land.
- Table 3 and 4 -> many of your factors are not significant and should be not used. Did you analyze before doing the regression the collinearity of your factors?
Answer: Since panel liner regression uses time series data, the collinearity of factors may exist. In this study, there is only moderate collinearity between the factors (GDP, POP, and Secondary). However, remarkable variation in city-specific developmental level will reduce the interpreting power of factors. Tables 3 and 4 contain two or three independent variables, whose coefficients are insignificant or marginally significant (Tertiary, p =0.061). We kept these seemingly insignificant or marginally significant variables because linear regression cannot reveal the direct/indirect effects of these independent variables in depth. As shown in Table 5, the SEM analysis results suggest that these seemingly insignificant or marginally significant variables do have significant direct/indirect effects on Total developed land and associated totalNTBI. Therefore, we decided to keep these insignificant or marginally significant variables in Tables 3 and 4 for further SEM analysis.
- Overall the paper would require a revision in terms of content, structure and language. Please also try to shorten and get more concise.
Answer: We have revised the manuscript to make it more logical and clear to read. Meanwhile, this latest version of manuscript was edited by MDPI’s English Editing Service.
ssociative in the title of the revised version.
- Another term, which would require reconsideration is “satellite nighttime light emissions” – the satellite is not having NTL emission but captures these emissions.
Answer: The term was changed to ‘satellite observed nighttime light emissions’ was used.
- Another term, to reconsider is “associative TotalNTBI”
Answer: The term ‘associative TotalNTBI’ was changed to ‘associated TotalNTBI’.
- Line 37 – the urbanization figures seems wrong, could you please check e.g., https://www.un.org/development/desa/publications/2018-revision-of-world-urbanization-prospects.html
Answer: We used the 2017 data reported by UN. Since annual global urban population data will change, we have updated the population data according to the latest version of UN report.
- Please also better clarify what is the innovation of your work (e.g., in Line 44 and70-71) – there have been many studies on NTL and urbanization also with the focus on China. Please clarify what is your specific contribution.
Answer: We have clarified our contributions in the discussion section.
- Line 63-64 need clarification/reformulation “changing NTLE may be attributed to direct/indirect influences of driving factors with high complexity”?
Answer: What we meant is that the changing NTLE cannot not be simply attributed to some socioeconomic driving factors. It is not enough to address the relationship between human settlement and associated NTLE, given there are direct/indirect influences of socioeconomic driving factors. We also emphasized such shortcomings in Lines 63-71 and the discussion section.
- Table 1 the population data seems wrong e.g. Beijing has 21 Mio inhabitants and not 2.1 Mio.
Answer: Many thanks. We have corrected the table.
- Line 100: I would suggest not to use the term primary data – e.g. land use data are secondary data.
Answer: By primary data, we meant major data rather than first-handed data. But, to avoid potential confusion, we deleted the word ‘primary’ in the revised version.
- Line 115: which years are selected and why?
Answer: To be consistent with national LULC data, DMSP/OLS data were in 2000, 2005, 2008, 2010, and 2013.
- Line 118-119: why do you use UTM – for the entire country of China?
Answer: Given the popularity of UTM system and possibility of international comparison., we decided to use UTM projection in our study. In one of our parallel studies, the recently released 30-m global land cover products (http://www.globeland30.cn/GLC30Download/index.aspx), which were projected in WGS_1984_UTM system, will be used to produce better DMSP/OLS data for international comparative study.
- Line 122: 100 GCP seems very few points – and how have these points been used.
Answer: As shown in Figure 2, because the raw DMSP/OLS data is rather coarse, it is very difficult to find enough GCPs for it. For land cover products and raw DMSP/OLS data acquired in each year, 100 GCPs with distinct geographical features were carefully selected by overlapping the digitalized maps, land cover maps and other auxiliary datasets. In view of the file size and limited computer processing capacity, these points were used to wrap the maps with the second order polynomial method, and the resulting RMSE was within one pixel.
- Line 145 (equation 1): Too me, this seems to be a masking operation and not a resolution sharpening – in principle you ignore all NTL outside your built-up mask.
Answer: Many thanks. We have replaced the term resolution sharpening with edge sharpening.
- Line 165 (the index) - what does it explain? The results and conclusions seems obvious: the larger cities have the most NTL emissions.
Answer: The indices in line 164 to 167 reflected the total NTL emissions associated with the total developed land and the three sub-categories of developed lands (urban, rural, and the other developed lands).
Although larger cities tend to have most NTL emissions, our analysis results have found the relationship between developed land and NTL emissions are more complex. First, there is much variation is the NTL emissions associated with different sub-categories of land. For instance, Zhengzhou’s RuralNTBI was higher than that of Guangzhou and Shenzhen, while its UrbanNTBI and OtherNTBI were lower than that of Guangzhou and Shenzhen. The resulting TotalNTBI of Zhengzhou was lower than that of Guangzhou but higher than Shenzhen. Therefore, TotalNTBI and its three sub-categories of NTBIs can help explain the spatio-temporal patterns of NTL emissions influenced by developmental inequality between different regions. Secondly, although both the urban developed land and the other developed land were in good agreement with their associated NTBIs, the relationship between rural developed land and RuralNTBI was relatively weak. This has led to the mismatched relationship between the total developed land and TotalNTBI in some cities, whose rural developed land was larger than their urban developed land.
- Table 3 and 4 -> many of your factors are not significant and should be not used. Did you analyze before doing the regression the collinearity of your factors?
Answer: Since panel liner regression uses time series data, the collinearity of factors may exist. In this study, there is only moderate collinearity between the factors (GDP, POP, and Secondary). However, remarkable variation in city-specific developmental level will reduce the interpreting power of factors. Tables 3 and 4 contain two or three independent variables, whose coefficients are insignificant or marginally significant (Tertiary, p =0.061). We kept these seemingly insignificant or marginally significant variables because linear regression cannot reveal the direct/indirect effects of these independent variables in depth. As shown in Table 5, the SEM analysis results suggest that these seemingly insignificant or marginally significant variables do have significant direct/indirect effects on Total developed land and associated totalNTBI. Therefore, we decided to keep these insignificant or marginally significant variables in Tables 3 and 4 for further SEM analysis.
- Overall the paper would require a revision in terms of content, structure and language. Please also try to shorten and get more concise.
Answer: We have revised the manuscript to make it more logical and clear to read. Meanwhile, this latest version of manuscript was edited by MDPI’s English Editing Service.
Author Response File: Author Response.docx
Round 2
Reviewer 3 Report
Dear authors,
Thanks for the revised version of your map on “Relationship Between Developed Land and Satellite Observed Nighttime Light Emissions of China’s 36 Key Cities (2000–2013)”. Many aspects have been improved and questions have been clarified. I still have some issues/questions and suggestions before publishing the paper – most of my comments aim at to improve the reporting on your methods and results:
Abstract Line 28etc. last sentence: Could you be more specific?
Line 50-51: “assessment of developmental inequality” -> could be more specific or add an example
Line 65: too coarse NTL data - > Here ISS night-time images can be a alternative see e.g.:
A. Z. Kotarba and S. Aleksandrowicz, “Impervious surface detection with
nighttime photography from the international space station,” Remote Sens.
M. Kuffer, K. Pfeffer, R. Sliuzas, H. Taubenböck, I. Baud, and M. v. Maarseveen, "Capturing the Urban Divide in Nighttime Light Images From the International Space Station," IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, pp. 1-9, 2018.
Line 81: “In terms of geographic zoning” -> Not clear - do you mean you selected the 36 largest cities of all provinces?
Figure 1: Legend misses the administrative boundaries - not clear what they are?
Table 1: What is the reference year of these statistics? If available it might be interesting to add the annual population growth rate?
Line 111: China - doing a very quick check has 11 different UTM zones - using a different zone for one place will lead to large distortions – this is very problematic. UTM should be used only within one zone! For such an analysis a different projection systems should have been used. How did you deal with this - this is still not clear to me?
Line 187 ect – this section is not very clear to me -> You need to be more specific
Line 199etc and Figure 3: Besides absolute growth - it would be also interesting to add relative growth - large cities will have very commonly a large absolute growth
Figure 5 and 6: Boundaries are missing in the legend
Figure 6: Caption text: why dynamics? the values seems to be DN values?
Table 3: I would suggest to add also the standardized coefficients?
Line 299-301 this sentence is not clear
Figures 9: Please add the meaning of the figures - are they coefficients?
Line 444: I would remove the decimals and add 1000 separators to make the stats more readable.
Author Response
Response to Reviewer3’s Comments
Dear reviewer,
From the bottom of the heart, we sincerely thank for your kind comments and suggestion on solidly improving the quality of this manuscript. As you will find, what we corrected based on your suggestion were highlighted in red across the manuscript.
The authors
####################################################################
Comments and Suggestions for Authors
Dear authors,
Thanks for the revised version of your map on “Relationship Between Developed Land and Satellite Observed Nighttime Light Emissions of China’s 36 Key Cities (2000–2013)”. Many aspects have been improved and questions have been clarified. I still have some issues/questions and suggestions before publishing the paper – most of my comments aim at to improve the reporting on your methods and results:
Point 1: Abstract Line 28etc. last sentence: Could you be more specific?
Response 1: We rephrased this sentence.
Point 2: Line 50-51: “assessment of developmental inequality” -> could be more specific or add an example!
Response 2: We added an example.
Point 3:Line 65: too coarse NTL data - > Here ISS night-time images can be a alternative see e.g.:
A. Z. Kotarba and S. Aleksandrowicz, “Impervious surface detection with nighttime photography from the international space station,” Remote Sens.
M. Kuffer, K. Pfeffer, R. Sliuzas, H. Taubenböck, I. Baud, and M. v. Maarseveen, "Capturing the Urban Divide in Nighttime Light Images From the International Space Station," IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing, pp. 1-9, 2018.
Response 3: Thanks for your provided references. We used the ISS night-time images as the alternative.
Point 4:Line 81: “In terms of geographic zoning” -> Not clear - do you mean you selected the 36 largest cities of all provinces?
Response 4: To a large degree, 36 largest cities of all provinces across geographic zoning (such as northern China, central China, and et al shown in Table 1) were selected. Besides, China’s geographic zoning major considers the remarkable variation in natural conditions, socioeconomic complexity, and developmental inequality.
Point 5: Figure 1: Legend misses the administrative boundaries - not clear what they are?
Response 5: The administrative boundary was added in the legend.
Point 6: Table 1: What is the reference year of these statistics? If available it might be interesting to add the annual population growth rate?
Response 6: The latest data in 2013 was shown in Table 1. Herein, to provide more detailed information, both annual growth rates for population and GDP in 2000-2013 were listed int this table.
Point 7:Line 111: China - doing a very quick check has 11 different UTM zones - using a different zone for one place will lead to large distortions – this is very problematic. UTM should be used only within one zone! For such an analysis a different projection systems should have been used. How did you deal with this - this is still not clear to me?
Response 7: We appologize for unclear explain in earlier version of response. Yes, it is impossible to simply perform UTM conversion for the whole China’s territory just use one zone code. Because we only study 36 largest cities, so we just clipped each city cluster (including the target cities and surrounding cities). And for each city cluster, UTM conversion was performed according to their specific UTM zone coding (43-53N). We clarify our processing method in section 3.2.1.
Point 8:Line 187 ect – this section is not very clear to me -> You need to be more specific
Response 8: We rephrased this section to make it more specific and readable.
Point 9: Line 199etc and Figure 3: Besides absolute growth - it would be also interesting to add relative growth - large cities will have very commonly a large absolute growth
Response 9: It is quite right. Then we added relative growth rates in this line.
Point 10:Figure 5 and 6: Boundaries are missing in the legend
Response 10: The boundaries were added in the legend.
Point 11:Figure 6: Caption text: why dynamics? the values seems to be DN values?
Response 11: We changed this caption.
Point 12:Table 3: I would suggest to add also the standardized coefficients?
Response 12: Unlike the commercial software such as SPSS Amos for panel data analysis, the free R software and the plm library cannot produce the standardized coefficients.
Point 13:Line 299-301 this sentence is not clear?
Response 13: We rephrased this sentence to make it more readable.
Point 14:Figures 9: Please add the meaning of the figures - are they coefficients?
Response 14: The meaning of the figures (direct coefficients) was added in Figure 9.
Point 15: Line 444: I would remove the decimals and add 1000 separators to make the stats more readable.
Response 15: We removed the decimals and add 1000 separators in this line.
Author Response File: Author Response.docx
