Quantification of Urban Heat Island-Induced Contribution to Advance in Spring Phenology: A Case Study in Hangzhou, China
Round 1
Reviewer 1 Report
The revision has improved some formal parts of the manuscript, but the technical issues remain and the quantitative significance of the results is still not adequate.
-Abstract: again, the sensors/products used and the study period are missing.
-Figure 10: the very low correlation (R2 is not reported in the figure) suggests that the outcomes are not so evident and the variables are not related. The authors highlighted only the P value, but it is not enough. The R2 is more significant, and it is low. Also, in figure 8b, the slope is around 1 day, that is well below the accuracy of the SOS data.
-Line 408: “The LST in spring was selected in this study, because of its stronger correlation with SOS”: what is the value of this strong correlation? A strong correlation, for instance, can be expected to be greater of 0.75 in terms of R2.
-Line 432: “extremely significant correlation (P<0.01) with the LST”. The P value does not explain the correlation, but the significant level. The correlation must be quantified by the R2.
-In the legend of figure 11 b “UHI” is still present.
-Any classification of the different vegetated cover types in the study area is presented.
-The LST variation in the rural area across the years is mainly driven by the different atmospheric and climatological factors (solar irradiation, soil humidity, precipitation variability), and responds in a different way with respect to the LST in the urban area. How are these factors considered in the study, and how are separated by the urbanization effects? How are these atmospheric factors across the years?
-Overall, the outcomes show a poor relationship between LST and SOS, focus of the proposed investigation: the P value alone cannot be considered a metric validating the response, but the R2 is the main parameter to evaluate (i.e. how the variation in the dependent variable that is predictable, and quantified, by the input variable)
Author Response
Please see the attachment.
Author Response File: 
Author Response.pdf
Reviewer 2 Report
The study aims to present the research on “Quantification of urban heat island-induced contribution to advance in spring phenology: a case study in Hangzhou, China.” The manuscript is presented clearly and nicely. The Paper is interesting but needs some modification before publish. Thus, I would like to suggest minor revisions.
- Need to add aims of the paper in to abstract
- The objectives are not clear in the introduction
- Figure 4: Legends in Temperature need to shows as a range such as < 17.5, 17.5-19.5 etc
- Figure 7 is not clear
- How your methodology can be used for other areas
- Are there any limitations? Could you explain it?
- Discussions and conclusions are fine
Author Response
Please see the attachment.
Author Response File: Author Response.pdf
Reviewer 3 Report
The paper underwent thorough review and met the suggestions given in the previous review.
Author Response
General comment:
The paper underwent thorough review and met the suggestions given in the previous review.
General Response: We greatly appreciate your general and specific comments last round, which are very important and helpful to improve our manuscript.
Round 2
Reviewer 1 Report
The new version presents the same results of the previous round revision, that do not reveal a relationship between SOS and LST.
Figure 10: the results confirm the low correlation (R2 are 0.02, 0.05, 0.14 and 0.2) suggesting that the outcomes are not so evident and the variables are not related. Highlighting only the P value as result is not enough. The R2 is the more significant, suggesting that no significant relation was found between input/output.
The authors said: “The result with low values of P and R2 in this study indicates that the SOS and LST were statistically related but discrete distributed because of some uncertainty such as coarse resolution data. We think it makes sense overall. In future, data with higher resolutions should be used to reduce the uncertainty caused by the data and make results more reliable”
Firstly, if the R2 are the values above, I cannot see the statistically relationship between SOS and LST. It means that the relationship (y/x) found will provide great RMSE errors (it can be computed), leading to a useless relation. Data with higher resolution can cause lower R2: what is expected by the authors with higher resolutions is not proved.
Author Response
General comment:
The new version presents the same results of the previous round revision, that do not reveal a relationship between SOS and LST.
【Answer】We greatly appreciate your general and specific comments, which are very important and helpful to improve our manuscript. In this study, although the R2 of the correlation between LST and SOS was relatively low, it still showed a statistically significant relationship, which has met our goals. The specific points addressed are given in the list below.
1.Figure 10: the results confirm the low correlation (R2 are 0.02, 0.05, 0.14 and 0.2) suggesting that the outcomes are not so evident and the variables are not related. Highlighting only the P value as result is not enough. The R2 is the more significant, suggesting that no significant relation was found between input/output.
【Answer】Thanks for your suggestion. Although the R2 was relatively low (R2=0.02, 0.05, 0.14, 0.2), statistically significant correlations between SOS and average daytime LST were observed in both the urban (P<0.05) and rural (P<0.01), and generally it made sense.
2.The authors said: “The result with low values of P and R2 in this study indicates that the SOS and LST were statistically related but discrete distributed because of some uncertainty such as coarse resolution data. We think it makes sense overall. In future, data with higher resolutions should be used to reduce the uncertainty caused by the data and make results more reliable”.
Firstly, if the R2 are the values above, I cannot see the statistically relationship between SOS and LST. It means that the relationship (y/x) found will provide great RMSE errors (it can be computed), leading to a useless relation. Data with higher resolution can cause lower R2: what is expected by the authors with higher resolutions is not proved.
【Answer】Thanks for your suggestion. First, although the R2 was relatively low (R2=0.2), the P value was significant (P<0.05), which to a certain extent reflected the statistical relationship between LST and SOS. Regrettably, there was not corresponding high-resolution data to conduct this research and calculate the R2 of the correlation between LST and SOS in this study. Previous studies found that the coarse spatial resolution of satellite data could make the retrieval and interpretation of phenological dates particularly challenging in mixed canopies [1]. Therefore, higher resolution data could improve the accuracy of phenological extraction, thereby making the relationship more significant. In future research, we will try to use data fusion methods to obtain higher resolution data for research.
[1] Piao, S.; Liu, Q.; Chen, A.; Janssens, I.A.; Fu, Y.; Dai, J.; Liu, L.; Lian, X., Shen, M.; Zhu, X. Plant phenology and global climate change: Current progresses and challenges. Global Change Biology 2019, 25, 1922-1940. DOI: 10.1111/gcb.14619
This manuscript is a resubmission of an earlier submission. The following is a list of the peer review reports and author responses from that submission.
Round 1
Reviewer 1 Report
General comment:
The authors present a study evaluating the impact of urbanization on plant phenology at the China city of Hangzhou. Vegetation index-based phenology data (start of growing season (SOS)) and land surface temperature (LST) data from MODIS were used to analyze the SOS characteristics and the potential contributions of the LST to the variations in plant phenology.
This work is only a simple way to analyze this topic, with also several flaws (e.g. the accuracy of SOS evaluation and the SOS meaning related to the different vegetation types aver a large area).
The outcomes reveal the poor statistical significance of the proposed investigation.
-Any classification of the different vegetated cover types in the study area is presented.
-The UHI is not computed, but simply the LST is used. The effect of the urban area (UHI) on rural zones very distant is questionable
Specific comments:
-Abstract: the first general part is too long, and some items are not directly addressed in the text. The abstract must highlight what was done, the sensors and data set used, the methodology, the study period, and the novelty of the proposed work.
- End of introduction: what is the novelty of the proposed work with respect to the available literature? The study area cannot be considered a novelty
-Figure 1: a bar of km-scale is missing
-Figure 2 and related text. The gaussian fitting function is a simple method to infer the SOS. More sophisticated approaches were used in the literature (e.g TIMESAT, BFAST models and so on)
-Lines 136-137: “the plant phenology data with a resolution of 500 meters cannot satisfy the demand, so the EVI data with a resolution of 250 meters was used to explore at a more sophisticated scale”. To move from 500 m to 250 m is not a step reaching a “sophisticated” scale, the latter being of the order of a few tens of meters.
-Figure 2a: the R2=0.54 cannot be considered a satisfactory correlation, as highlighted by the point dispersion. What is the error induced by the linear regression? What is the accuracy of the estimated SOS?
-UHI: from figure 1, how can the warming (UHI) of the urban area affect the sites S1-S5 placed several kilometers far from the urban core?
-Section 2.3: the methodology to address the LST-phenology issue is trivial
-The UHI is not computed, but the analysis simply deals with the LST
-Figures 5 and 6: The different SOS should be associated with the different types of vegetations present in the study area, otherwise the geophysical meaning and impact of SOS is not clear.
-Figure 8 caption: the temporal reference is missing
-Figure 10: the very low correlation (R2 is not reported) suggests that the outcomes are not so evident and the variables are not related
Reviewer 2 Report
The study aims to present the research on “Quantification of urban heat island-induced contribution to advance in spring phenology: a case study in Hangzhou, China.” The manuscript is presented clearly and nicely. The Paper is interesting but needs some modification before publish. Thus, I would like to suggest minor revisions.
- Introduction is fine
- Better to incorporate China maps in Figure 1. It will help readers to capture the study area.
- There are several citations are missing in the methodology section.
- Why did you select 2006,2010,2014, and 2018?
- Result Discussion and conclusions are OK.
- Did you have any limitations?
- How proposed methodology can be used in other study areas.
Reviewer 3 Report
This ms quantified UHI contribution to advanced SOS in Hangzhou, China.
Major comments:
Lines 99-101: please explain why the authors selected the test locations, especially those in rural area. For example:
- Why rural locations are all on one side of the city? I understand that those locations are within the Hangzhou boundary, but do they need to be within the boundary? I would select rural locations surrounding the city rather than on one side. It would bring a more complete picture of UHI as heat is not transferred in a single direction.
- Do rural and urban locations have the same vegetation / plant type to allow phenology comparison?
- Considering the very coarse resolution of RS data, how did the author consider mixed-pixel problem at test locations?
Figure 6 and section 3.2 does not make sense for me. I believe we can only compare (in a meaningful way) phenology of the same plant / vegetation. What is the point of comparing SOS of a forest patch and a block of building? I note that although the global phenology product provides phenometrics for every terrestrial pixel, it does not mean that they are all meaningful (or same quality).
Minor comments:
Line 109: replace “time” by “temporal”
Line 110-111: give a brief explanation because this point is important
Line 128: “matched the LST” in what sense? Note that the two datasets (LST and EVI) have different spatiotemporal resolution.
Lines 133-144: Given this information, I think the authors should drop MCD12Q2 to avoid unnecessary confusion for readers.
Section 2.3: I would present Figure 3 first (then equations) because Figure 3 is a lot easier to understand.
Figure 4: need to differentiate 3 lines (buffer_10, buffer_20 and urban) better. Change the color ramp as well because middle color is white, it is the same with background color. I would try to use “heat” (yellow to red) on this figure.
Reviewer 4 Report
It is an interesting study. The manuscript can be improved by considering the following.
It is suggested that the objectives and procedures are better presented in the abstract.
In the introduction, it is suggested that the original character of the article be presented.
The authors need to cite studies that show results from similar studies and also the differences between this study and others already carried out.
In lines 55 and 56 it is stated that “At present, many studies have paid attention to impacts of urbanization on the change of plant phenology”. It is suggested that studies that dealt with this topic be referenced.
In item 2.1 “Study area”, the city of Hangzhou must be better characterized, inserting, for example, the total urban and rural population.
In lines 93 and 94, it is stated that “The annual average temperature and annual precipitation is 17.8°C and 1454 mm, respectively”. It is suggested that the study area be better characterized and that maximum and minimum values be inserted in the extreme seasons and that the variability of precipitation be presented in the different seasons of the year.
Basic cartographic conventions must be included in all maps. For example, all maps must have the graphic scale (Figures: 1, 4, 5, 6, 7). All information entered in Figure 1 must be in English.
In item "4. Conclusions”, the conclusions should be explored, addressing the main problems proposed in the investigation.
