Correlation Analysis between Urban Green Space and Land Surface Temperature from the Perspective of Spatial Heterogeneity: A Case Study within the Sixth Ring Road of Beijing

Round 1

Reviewer 1 Report

The introduction is appropriate and provides the context, but it should stress the implication of using different analysis methods. Moreover, the introduction must explain why spatial heterogeneity matters. For example, in other studies, spatial heterogeneity has been linked to spatial resilience:

Allen, C. R., Angeler, D. G., Cumming, G. S., Folke, C., Twidwell, D., & Uden, D. R. (2016). Quantifying spatial resilience. Journal of Applied Ecology, 53(3), 625-635.

Garcia, E., & Vale, B. (2017). Unravelling sustainability and resilience in the built environment. Routledge.

Please include a research question (only one) and explain the hypothesis that sustains the research question and the knowledge gap it tries to fill.

Since the focus of the paper, and its contribution, relies on the methodology used and the adjustments made to fit the case study, it could be beneficial to include a literature review section of processes and methods used in other studies, strengths and weaknesses and highlight the knowledge gaps that your methodology will try to cover. Part of the content in the discussion section would fit best in a literature review section.

To reach a wider audience, it could be beneficial for the article to describe in detail, one by one, all the methods used in the methodology section, from data collection to analysis.

In the data sources and data collection, please explain the "corrections" done and provide examples. Include the formulas used for each method (PCA and all the variables included in Table 1) and give a numeric example so that readers can understand and follow all the calculations and the results obtained step by step. Authors should allow scholars to replicate the test and get the same results.

The discussion section is too long and distracts the reader from the link between the analysis and the conclusion.  

In the conclusion section, please keep the terminology coherent, namely, do not name the same thing differently; otherwise, the reader will assume they are different, for example, inhomogeneous and heterogeneous. 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 2 Report

The article deals with the correlation of land-surface temperature in urban areas and green spaces, with the focus on urban heterogeneity. As study area, Beijing is considered. Although some of the findings are trivial (as bigger the area of UGS as higher the cooling effect, forest have the highest effect,…) the results contain also some interesting findings.

 Remarks:

- as only LST derived from remote sensed data is used in the study, the title is somehow misleading. The article is not considering “urban thermal environment” in general, but only LST is considered. Thus the title should be changed.

- as the title correctly reflects, the study is on the CORRELATION of urban green spaces and land-surface temperature. It is scientifically not correct to conclude from correlation to cause –effect as done in particular in the result discussion. Other factors affecting the micro climate are not considered, e.g. cold wind corridors, so the authors should be more careful talking about impact.

 

- line 60: the former refers to the abundance…including … abundance. Abundance including abundance makes no sense, change the sentence.

- line 100 following. The role of the two datasets used for extracting UGS is somehow unclear. If Sentinel data was used as primary source for extracting UGS, it should be mentioned first. It should also clearly be addressed: as for the first dataset the term Landsat-data is used, for the second Sentinel-data should be used. ESA is just the download provider of the Sentinel data.

- line 110. it is unclear what is meant by “a slight deviation”. Is it a geometric deviation or what is that? What is a “simple” correction? What is a “sufficient” accuracy? The whole paragraph is somehow unclear and not scientific sound and thus needs a complete revision.

- line 120: UGS from Landsat? in the paragraph before it was mentioned, that it was extracted from Sentinel data! This a contradiction and a big confusion here!

- line 132: .. by Chavez in the US … Improper/missing reference

- once again: the whole description of the data preparation for UGS is based on Landsat data, which was – according to chapter 2.2 – not used! And if it was used, here the question is why if Sentinel data has a better quality/resolution

- line 174 and fig 3: according fig. 3 1800x1800 window has a slightly higher r2. So why not this window size has been used?

- line 188: x bar is missing: … x bar is the average…

- equation 1 and 2 are both according the implementation in ArcGIS. For the calculation results the spatial weight matrix is crucial. Here some explanations should be given.

- table 2: the formula for SHAPE_MIN seems not to be correct. Is there a summation symbol missing?

 

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Reviewer 3 Report

1.       We proposed a new approach to optimize the selection of landscape metrics, identify the least and most effective metrics to establish geographically weighted regression (GWR) model, and plot the distribution of local regression coefficients to investigate the spatially heterogeneous cooling effects of greenspaces.

2.       In the abstract section, the authors mentioned that they proposed a new approach to optimize the selection of landscape metrics. However, the information on the new approach in the introduction section is not enough. Even, they failed to introduce the research status on the selection of landscape metrics.

3.       The innovation of this study should be strengthened in the introduction. Because the application of GWR model can be found everywhere. It is not a very new topic.

4.       The legend should be added in the Fig.1 for the maps before/after processing.

5.       Why use the correlations between the PLANDs and the LSTs to determine the optimized window size, while not other landscape metrics. Also how many landscape metrics were calculated here should be supplied here.

6.       What method was used to selected the best subsets of the metrics?

7.       The more detail information on the process of the GWR model should be added.

8.       Why the landscape metrics used in the section 3.3.2 were different those in the section 3.4.

9.       In the conclusion section, author claimed that they used new approach to…., I am curious what is the new approach they meant.

Author Response

Please see the attachment.

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

The latest version acknowledges most of the comments and suggestions previously made. 

The letter from the authors was very helpful in understanding what they were trying to do. I commend the authors' openness and flexibility to accommodate all the comments suggested.

Reviewer 3 Report

I am satisfied with the revisions done by the authors.