Do Emotional Perceptions of Visible Greeneries Rely on the Largeness of Green Space? A Verification in Nanchang, China

Round 1

Reviewer 1 Report

Dear authors,

I found your research and article interesting. I have two remarks that should be clarified in the content of your publication.

1. why the Sentinel-2A satellite data was used only from the holiday period, and why pictures of faces were collected all year round. Was it not possible to extract the photos from the summer months?

2. the text should contain some characteristics of these parks, I mean the average area, the smallest, the largest park, their character, vegetation, and the occurrence of water.

Author Response

Point 1: Why the Sentinel-2A satellite data was used only from the holiday period, and why pictures of faces were collected all year round. Was it not possible to extract the photos from the summer months?

 

Response 1: We thank the reviewer for raising this question. Because of the subtropical monsoon climate in Nanchang, the vegetation type is mainly evergreen broad-leaf forest. We accepted your professional advice that curate data of landscape metrics in a period across the whole year. I collected the whole year's data and then re-analyzed the full-text data. Please refer to manuscript for details of the modification. “Because of the subtropical monsoon climate in Nanchang, the vegetation type is mainly evergreen broad-leaf forest [51]. The satellite images were used for the whole year 2020 as a data source with less than 5% cloudiness.”

 

Point 2: The text should contain some characteristics of these parks, I mean the average area, the smallest, the largest park, their character, vegetation, and the occurrence of water.

 

Response 2: Thank you very much for your suggestion. I have added the introduction of the landscape characteristics such as water ratio, park area in Table 1.

Author Response File: Author Response.docx

Reviewer 2 Report

I really enjoyed reading this fresh and attractive manuscript. Nevertheless, I have a few comments that concern mainly data analysis.

L26-L28: „When NDVI exceeded 38.2%, posted sentiment was no longer responsive to the largeness of green space. Increasing the amount of green space promoted positive emotions when the NDVI of green space was lower than 38.2%“. I think these two sentences would make better sense if their order were reversed.

L104-L108: „And the research subjects are not limited by devices, except for young and middle-aged people who use social software. The photos of SNS can include people of many ages, such as infants, children, and the elderly“. I guess what the authors wanted to say, but it's not described very clearly. Is the meaning following? „Despite the use of electronic devices is limited mainly to young and middle-aged people who use social software, their photos on SNS include often people of all ages, i.e. infants, children, and old people, making them also research subjects.“ My English is not perfect either, I suggest to the authors let the entire text be edited by a native speaker.

L124-126: In the first aim, you hypothesized that PGVI, NDVI, green ratio, and elevation should correlate. In the second aim, you hypothesized that they have a different effect on emotional expression. But if there was a very strong correlation between quantitative indicators, then their different effect on emotions would be counterintuitive. If you assume that the correlation is there but is not that strong, then it might be best to reformulate the second goal by looking for which of the quantitative indicators best explains the emotional expression.

L137: „The study area of this research was covered a large part of green spaces in Nanchang“ – delete „was“.

L146-L147: „Sentinel-2A is widely used [31]. For this study, Sentinel-2A satellite data was used…“ Why not e.g.: „For this study, widely used Sentinel-2A [31] was utilized“ or „we utilized Sentinel-2A, which is widely used for this purposes [31]“. Your formulation is awkward.

L148: „summer moth with dense vegetation were chosen to be June to September 2020“. I'm from the other side of the world, so I just assume that the vegetation was studied over the summer because it is an area of deciduous temperate forests. Could the authors better specify the type of vegetation they studied.

L174-L176: „we downloaded photos in each study area from January 2020 to December 2021“. Given that the authors themselves state that the vegetation was examined over the summer to analyze dense vegetation, shouldn't the photos of the faces also be just from the period of dense vegetation? During the winter, the vegetation may be sparse, and perhaps the trees may be even bare. I can imagine that such vegetation cannot have a positive effect on emotions or can even have a depressing effect. I would suggest either limiting the dataset to the summer only or even better, adding information whether faces were scanned in the winter or in the summer as another explanatory variable. Could COVID-19 and COVID restrictions affect your research (e.g., fewer possibilities to enjoy a presence in the city center without vegetation = less opportunity to take a happy photo there)?

L201-L202: „As a result, we evaluated the mean values of NDVI, green area ratio and elevation in the study areas. Finally, we obtained several landscape metrics such as NDVI, elevation and green ratio“. Why not cut this information down to just one sentence that does not repeat words NDVI, elevation, green ratio…

L256-L257: „GVI, green ratio, and NDVI, as essential quantitative elements of greenness, can represent the amount of greenness in three dimensions and are important dependent variables in this study“. Not dependent but explanatory (or independent) variable.

L259 and further: Analysis of variance (ANOVA) was applied to detect the effects of PGVI, green ratio and elevation (independent variables) on happy, sad, neutral and PRI scores (dependent variables) in different categories of NDVI“… „Finally, Pearson's correlation analysis examined the effects of varying land- 266 scape factors on emotional expression factors.“ I don't understand why the authors analyze the data so awkwardly. Why do they divide NDVI into categories that are necessarily arbitrary, compare them using ANOVA, when there are still large differences between the individual values and thus large SEs for the given category, and then correlate something separately for each group? A regression or even better general linear model would cost them fewer degrees of freedom and enable the work with the original quantitative data on NDVI (as opposed to ANOVA), and they could work with all the data at once and declare a clearer relationship between emotions (dependent variable) and NDVI (explanatory variable) (as opposed to correlation). Further, building of individual models for different characteristics of greenery and their comparison using AIC would allow the authors to solve which parameter of greenery had the strongest effect on emotions.

L271: We collected the main landscape factors such as PGVI, NDVI green ratio and elevation in the study area“. This sentence seems unnecessary to me at the beginning of the results.

L281-L282„Based on the analysis results, it is known that the NDVI value of the research location is 0.089 to 0.723 shown in Figure 3(b). The values of the green ratio ranged from 0.021 to 0.864, with an average value of 40.3%“. Why repeat this information when it is already in the table?

L301-L302: The results showed significant differences(P<0.05) in happy, neutral and PRI emotions between the different NDVI groups“. Ok, but regression would make it possible to describe this relationship by an equation. In addition, you could consider second- and higher-order polynomials to describe this relationship, which probably won't be simply linear.

L316: Figure 4a. How do you explain the relatively high Happy Score at the lowest level of NDVI? Is it possible to generalize and possibly discuss what people did in the photos (sitting at cafes, buying goods, or doing other activities that are not so common in the middle of greenery but made them happy)?

L319 and further, mainly Table 5. It seems inappropriate to me to divide the data into four subgroups and test them separately. I think it would be much more efficient to use a general linear model that examines the regression between emotions (dependent variable) and PGVI, the green ratio and elevation (explanatory variables), and with NDVI taken into account as a covariate, either in its original quantitative form (preferred variant) or in the form of categories. Such an analysis is called ANCOVA.

L342 and Table 6. There are correlations between the dependent variables. It is therefore very possible that they are also collinear. This could be measured using the variation inflation factor, and thus it could be clearly described which part of explained variability is shared between them.

L384-L386: „Since people prefer to share optimistic images, users tend to upload their happy messages on SNS [61]. This problem can be addressed by expanding the sample size [16]“. I do not know how the extension of the sample could solve this bias. If the choice is still limited to what individual people want to upload to social networks, this bias will be independent of the amount of data (amount of studied people). We have to force people to increase the number of photos they produce and post so that they would capture also negative emotions. For example, the annoyance of having to take too many photos :)

Author Response

Point 1: L26-L28: „When NDVI exceeded 38.2%, posted sentiment was no longer responsive to the largeness of green space. Increasing the amount of green space promoted positive emotions when the NDVI of green space was lower than 38.2%. “ I think these two sentences would make better sense if their order were reversed.

 

Response 1: Thanks for your correction for the sequence of sentences. I modified the data analysis method and changed the correlation analysis to multiple linear regression as you suggest. Please refer to L26-29 of the original text for details of the modification. “Multivariate linear regression indicated that PGVI was estimated to have a significant contribution to facial expression. Increasing the amount of PGVI promoted happy and PRI scores, while at the same time, neutral sentiments decreased with increasing PGVI.”

 

Point 2: L104-L108: „And the research subjects are not limited by devices, except for young and middle-aged people who use social software. The photos of SNS can include people of many ages, such as infants, children, and the elderly“. I guess what the authors wanted to say, but it's not described very clearly. Is the meaning following? „Despite the use of electronic devices is limited mainly to young and middle-aged people who use social software, their photos on SNS include often people of all ages, i.e. infants, children, and old people, making them also research subjects.“ My English is not perfect either, I suggest to the authors let the entire text be edited by a native speaker.

 

Response 2: This expression has been modified. The controversial remarks had been changed. Please refer to manuscript for details of the modification. “Facial expressions reflect the perception of human emotions, which can be analyzed by modern facial recognition technology [15]. There are several advantages of obtaining facial expression data from SNS photos, including exposing human emotions more realistically.”

 

Point 3: L124-126: In the first aim, you hypothesized that PGVI, NDVI, green ratio, and elevation should correlate. In the second aim, you hypothesized that they have a different effect on emotional expression. But if there was a very strong correlation between quantitative indicators, then their different effect on emotions would be counterintuitive. If you assume that the correlation is there but is not that strong, then it might be best to reformulate the second goal by looking for which of the quantitative indicators best explains the emotional expression.

 

Response 3: I modified assumptions which are more accurate. We focus on the impact of visual greeneries (PGVI) and green space areas (NDVI) on emotional expression. Please refer to manuscript “Increasing the number of NDVI doesn’t always promote positive effects. Emotional expressions of visitors in public green spaces are related to PGVI.”

Point 4: L137: „The study area of this research was covered a large part of green spaces in Nanchang“ – delete „was“.

 

Response 4: I deleted “was”. Please refer to manuscript “The study area of this research covered a large part of green spaces in Nanchang”.

 

Point 5: L146-L147: „Sentinel-2A is widely used [31]. For this study, Sentinel-2A satellite data was used…“ Why not e.g.: „For this study, widely used Sentinel-2A [31] was utilized“ or „we utilized Sentinel-2A, which is widely used for this purposes [31]“. Your formulation is awkward.

 

Response 5: Please refer to manuscript “For this study, we utilized Sentinel-2A, which is widely used for this purpose”.

 

Point 6: L148: „summer months with dense vegetation were chosen to be June to September 2020“. I'm from the other side of the world, so I just assume that the vegetation was studied over the summer because it is an area of deciduous temperate forests. Could the authors better specify the type of vegetation they studied.

 

Response 6: I have added the description of the local climate and vegetation type. Please refer to manuscript for details of the modification. “Because of the subtropical monsoon climate in Nanchang, the vegetation type is mainly evergreen broad-leaf forest [51].”

 

Point 7: L174-L176: „we downloaded photos in each study area from January 2020 to December 2021“. Given that the authors themselves state that the vegetation was examined over the summer to analyze dense vegetation, shouldn't the photos of the faces also be just from the period of dense vegetation? During the winter, the vegetation may be sparse, and perhaps the trees may be even bare. I can imagine that such vegetation cannot have a positive effect on emotions or can even have a depressing effect. I would suggest either limiting the dataset to the summer only or even better, adding information whether faces were scanned in the winter or in the summer as another explanatory variable. Could COVID-19 and COVID restrictions affect your research (e.g., fewer possibilities to enjoy a presence in the city center without vegetation = less opportunity to take a happy photo there)?

 

Response 7: We thank the reviewer for raising this question. Because of the subtropical monsoon climate in Nanchang, the vegetation type is mainly evergreen broad-leaf forest. We accepted your professional advice on the data source for the whole year. I collected the whole year's data and then re-analyzed the full-text data. Please refer to manuscript for details of the modification. “Because of the subtropical monsoon climate in Nanchang, the vegetation type is mainly evergreen broad-leaf forest [51]. The satellite images were used for the whole year 2020 as a data source with less than 5% cloudiness.” We explained the impact of the COVID-19 epidemic, please refer to L879-882 “Due to the impact of the COVID-19 epidemic in 2020, many people were wearing face masks [33]. The chances of taking pictures of facial expressions are reduced in green spaces. To avoid data collection restrictions, we chose to analyze facial expressions from 2020 to 2021.” The data in this study are less affected by the COVID epidemic. This article does not focus on COVID-19 cases.

Point 8: L201-L202: „As a result, we evaluated the mean values of NDVI, green area ratio and elevation in the study areas. Finally, we obtained several landscape metrics such as NDVI, elevation and green ratio“. Why not cut this information down to just one sentence that does not repeat words NDVI, elevation, green ratio…

 

Response 8: Duplicates have been removed. Please refer to manuscript “As a result, we evaluated the mean values of landscape metrics that correlated with emotional perception, such as NDVI, green area ratio and elevation in the study areas.”

 

Point 9: L256-L257: „GVI, green ratio, and NDVI, as essential quantitative elements of greenness, can represent the amount of greenness in three dimensions and are important dependent variables in this study“. Not dependent but explanatory (or independent) variable.

 

Response 9: I have changed word “elements” to “independent variables”. Please refer to manuscript “NDVI and PGVI, as essential quantitative independent variables of greenness, can represent the amount of greenness in different dimensions in this study.”

 

Point 10: L259 and further: Analysis of variance (ANOVA) was applied to detect the effects of PGVI, green ratio and elevation (independent variables) on happy, sad, neutral and PRI scores (dependent variables) in different categories of NDVI“… „Finally, Pearson's correlation analysis examined the effects of varying land- 266 scape factors on emotional expression factors.“ I don't understand why the authors analyze the data so awkwardly. Why do they divide NDVI into categories that are necessarily arbitrary, compare them using ANOVA, when there are still large differences between the individual values and thus large SEs for the given category, and then correlate something separately for each group? A regression or even better general linear model would cost them fewer degrees of freedom and enable the work with the original quantitative data on NDVI (as opposed to ANOVA), and they could work with all the data at once and declare a clearer relationship between emotions (dependent variable) and NDVI (explanatory variable) (as opposed to correlation). Further, building of individual models for different characteristics of greenery and their comparison using AIC would allow the authors to solve which parameter of greenery had the strongest effect on emotions.

 

Response 10: We appreciate the reviewer for this kind recommendation. We have adapted the method. We place emphasis on NDVI and PGVI as independent variables. NDVI represents the areas of green spaces, and PGVI represents the frequencies to experience actual visual greenery. Please refer to manuscript. “First, we classified the data according to the calculated value range by Jenks natural breaks classification of NDVI and PGVI into four categories to compare the differences between sites for each type of green space. Second, we obtained the face expression data from Sina microblog using FireFACE software such as, happy, sad, neutral and PRI scores. Third, Split-plot ANOVA (SPANOVA) was used to detect the effect of frequency to experience greenery (PGVI) on facial expressions in regions with different green space areas (NDVI). Since elevation is an intrinsic element of public green space, it was used as a random factor. SPANOVA was applied to detect the combined effects of NDVI and PGVI on perceived emotions (happy, sad, neutral and PRI scores). When the results showed significant differences, we used the Duncan test at the 0.05 significance level for post hoc comparisons to address the uneven number of replicates be-tween the data groups. In addition, the mean values of NDVI, PGVI, green ratio and elevation for each location were tested for correlation to compare whether there was an association between the amount of greenness in the different dimensions. Finally, we screened landscape indicators without multiple co-linearity for multivariate linear regression (MLR) to examine the effects of landscape metrics on emotional expression factors.”

Point 11: L271: We collected the main landscape factors such as PGVI, NDVI green ratio and elevation in the study area“. This sentence seems unnecessary to me at the beginning of the results.

 

Response 11: Unnecessary sentences have been removed.

 

Point 12: L281-L282„Based on the analysis results, it is known that the NDVI value of the research location is 0.089 to 0.723 shown in Figure 3(b). The values of the green ratio ranged from 0.021 to 0.864, with an average value of 40.3%“. Why repeat this information when it is already in the table?

 

Response 12: Data already shown in figure3 has been deleted.

 

Point 13: L301-L302: The results showed significant differences(P<0.05) in happy, neutral and PRI emotions between the different NDVI groups“. Ok, but regression would make it possible to describe this relationship by an equation. In addition, you could consider second- and higher-order polynomials to describe this relationship, which probably won't be simply linear.

 

Response 13: We are so grateful for your kind recommendation. We have adapted the method and used multivariate linear regression (MLR) to describe the effects of NDVI and PGVI on emotional expression factors. Please refer to manuscript Chapter3.4.

 

Point 14: L316: Figure 4a. How do you explain the relatively high Happy Score at the lowest level of NDVI? Is it possible to generalize and possibly discuss what people did in the photos (sitting at cafes, buying goods, or doing other activities that are not so common in the middle of greenery but made them happy)?

 

Response 14: Detailed explanation has been added, Please refer to manuscript. “The lowest level of NDVI had high happy and PRI scores. It was due to the fact that a low level of NDVI can also provide a high PGVI, which can promote positive emotions.”

 

Point 15: L319 and further, mainly Table 5. It seems inappropriate to me to divide the data into four subgroups and test them separately. I think it would be much more efficient to use a general linear model that examines the regression between emotions (dependent variable) and PGVI, the green ratio and elevation (explanatory variables), and with NDVI taken into account as a covariate, either in its original quantitative form (preferred variant) or in the form of categories. Such an analysis is called ANCOVA.

 

Response 15: Thank you very much for your suggestion. We focus on NDVI and PGVI as independent variables. NDVI represents the proportion of green area, and PGVI represents the proportion of actual visual greenery. So, we used split-plot design of ANOVA and MLR to examine the effects of these two landscape elements on emotional expression factors. Please refer to manuscript, multivariate linear regression of green space metrics to facial expression scores were shown in table 6 and figure 5.

Point 16: L342 and Table 6. There are correlations between the dependent variables. It is therefore very possible that they are also collinear. This could be measured using the variation inflation factor, and thus it could be clearly described which part of explained variability is shared between them.

 

Response 16: Multivariate linear regression can accurately reflect the relationship between independent variables and dependent variables. We changed data analysis methods to explore the relationship between landscape elements and sentiment scores. After disregarding the effect of multicollinearity, we used MLR to examine the effects of landscape elements on emotional expression. Please refer to manuscript Table 6 and Figure 5.

 

Point 17: L384-L386: „Since people prefer to share optimistic images, users tend to upload their happy messages on SNS [61]. This problem can be addressed by expanding the sample size [16]“. I do not know how the extension of the sample could solve this bias. If the choice is still limited to what individual people want to upload to social networks, this bias will be independent of the amount of data (amount of studied people). We have to force people to increase the number of photos they produce and post so that they would capture also negative emotions. For example, the annoyance of having to take too many photos :)

 

Response 17: I have made modifications to the statement. Please refer to manuscript “Since people prefer to share optimistic images, users tend to upload their happy messages on SNS. We should increase the number of photos collected, so that negative emotions can also be captured.”

Author Response File: Author Response.docx

Round 2

Reviewer 1 Report

Congratulations. This is a good article

Reviewer 2 Report

I appreciate the effort the authors put into correcting the manuscript and I have no further comments.