1. Introduction
Urban heat island (UHI) is the phenomenon that temperature is higher in urban areas than in their surrounding rural areas. It is widely observed all around the world and generates a lot of ecological and environmental problems, for example, increasing air pollution [1], causing biodiversity loss [2], increasing energy consumption [3], and increasing health risks [4,5]. It is urgent to find an effective measure to mitigate the UHI.
Vegetation could provide shade and transpiration, and thus decrease environmental temperature and mitigate the UHI effect [6,7]. Greenspace amount (e.g., green cover radio or percent of greenspace) showed a significant negative relationship with temperature or UHI intensity, and increasing greenspace amount has been widely accepted and adopted to mitigate UHI [8,9]. However, urban greenspace cannot expand without limitation in most cities. It is very difficult if not impossible to further increase greenspace amount. Increasing the cooling efficiency of urban greenspace with a given amount of greenspace receives increased concern from both researchers and decision-makers.
Based on the theory of landscape ecology, urban greenspace mosaics with impervious surface area display a variety of spatial forms, for example in size, shape, connectivity, fragmentation, etc. Besides the amount of urban greenspace, its spatial configuration also significantly impact urban thermal conditions, and optimizing the spatial configuration of urban greenspace is another effective approach to increase the cooling efficiency of urban greenspace and improve the urban thermal environment [10,11]. Current knowledge was mainly obtained based on the remotely sensed land surface temperature (LST) because of its large spatial coverage and easy data accessibility.
Compared to LST, air temperature (AT) had more direct connections to human comfort and health [12,13,14]. There were significant differences between AT and LST in magnitudes as well as their spatial and temporal pattern [15]. Therefore, the relationship between the spatial pattern of greenspace and LST may be quite different from that between the spatial pattern of greenspace and AT, and findings concluded based on LST may not be effective in managing urban greenspace to decrease AT.
Some studies have investigated the relationship between AT and the spatial configuration of urban greenspace. A field experiment showed that trees planted in groups had lower AT than linearly planted trees and individually planted trees [16]. A simulation reported that clustered trees produced more cooling effects and displayed lower AT than scattered trees [17]. The cooling effect of urban parks was positively related to park size, patch shape of greenspace, and woodland fragmentation [18]. A study demonstrated that greenspace patch size significantly impacts AT through two opposite directions (i.e., increasing patch size can increase transpiration rate but decrease the shading cooling effect) and inferred an optimal configuration (i.e., patch size) of greenspace to cool the environment [19]. As AT has strong diurnal variations, the impacts of urban greenspace on the microclimate are different between day and night [20], but the role of the spatial configuration of urban greenspace is far from thoroughly understood. In addition, the cooling or heating effect of greenspace spatial configuration on AT may be scale-dependent. For example, Qian et al. [21] found that in addition to increasing the proportion of surrounding greenspace within a radius of 100 m, increasing the ED within a radius of 500–1000 m could reduce the temperature. Therefore, the scale effect of the relationship between AT and the spatial configuration of greenspace should also not be ignored.
Taking Changsha, China as an example, this study aims to investigate the scale dependence of the relationship between AT and spatial pattern of greenspace in the hot summer. Specifically, we attempt to answer four questions: (1) What is the relationship between AT and spatial configuration of urban greenspace? (2) What is the relative importance of spatial composition and spatial configuration in decreasing urban air temperature? (3) How do these answers change with the alteration of the spatial scale? and (4) What are the diurnal variations of these impacts? The findings of this study can provide suggestions for guiding the planning and urban management of the greenspace system in Changsha and other rapidly developing subtropical cities.
2. Materials and Methods
2.1. Study Area
Changsha (28°12′ N, 112°59′ E), the capital city of Hunan province, is a leading mega city in central South China (Figure 1). It has a subtropical monsoon climate with an annual temperature of 17.2 °C, an annual accumulated temperature of 5457 °C, and annual precipitation of 1361.6 mm. Changsha has long and hot summers with ~85 days warmer than 30 °C and ~30 days warmer than 35 °C in a year. The vegetation is dominated by subtropical evergreen broad-leaved forests. Xiangjiang River flows across the city. In 2020, the permanent resident population was 10.05 million and the urbanization rate was 82.6 percent. This city experienced rapid urban expansion, with the urban built-up area increasing from 53 km2 in 1978 to 435 km2 in 2017 [22] and the urban thermal environment problem becoming more and more prominent. The ecological environment problem caused by the heat island effect has become an obstacle to the economic development of Changsha. We focused on the highly developed areas (i.e., areas within the Third Ring Road region) (Figure 1). This area is hilly with flat areas surrounded by low mountains with the elevation ranging from −6 to 349 m [22].
2.2. Dataset and Methods
2.2.1. Temperature Data
AT observed at 36 weather stations on 16 sunny days in August 2020 was applied in this study considering data availability of AT and high spatial resolution satellite remote sensing images. August was selected as it is the hottest month. Maintained by the Changsha Meteorological Bureau, these weather stations are distributed within the third ring road and are mainly located in the highly developed built-up area (Figure 1). The climate variables include AT and wetness with a one-hour temporal resolution. We estimated the daytime temperature by averaging the temperature from 6:00 to 20:00 and the nighttime temperature by averaging the temperature from 20:00 to 6:00.
2.2.2. Mapping Land Cover
The Gaofen 2 remote sensing images acquired on 3 November 2019 were used to map urban greenspace assuming there are no significant changes in urban greenspace in one year. The image of November can effectively map urban greenspace as the vegetation is mainly evergreen. We performed geographical calibration and coordinate system transformation on remote sensing images and used panchromatic bands to sharpen multispectral bands. We obtained four bands (red, green, blue, and near-infrared) of multispectral images with a 1-m resolution after image fusion. To improve the accuracy of object classification in the shadow region, the data of the shadow region was extracted separately for object segmentation. Samples of all the classification elements were evenly selected in the study area, and 20% of the selected sample set was reserved as the verification samples, while the other samples were used to train the support vector machine (SVM) classifier. The trained SVM classifier was used to classify the multispectral images, and the land use classification map of the study area was obtained by combining visual interpretation and correction.
The four land cover types were included in the classification map: vegetation cover, impervious surface, water surfaces, and bare surfaces. The main component of vegetation cover was trees, shrubs, and grasses. The main component of impervious surfaces was road and building. After classification, we selected 150 random samples for each class to conduct an accuracy assessment. The final overall accuracy of the land cover map was 91.5% [22].
2.2.3. Measuring Spatial Pattern of Urban Greenspace
Percent of greenspace (PLAND) and spatial configuration of greenspace were calculated based on the 1-m resolution land cover map. We included five configuration metrics: (1) mean patch size (AREA_MN), (2) edge density (ED), (3) mean patch shape index (SHAPE_MN), (4) patch density (PD), and (5) landscape shape index (LSI), based on the following principals: (1) easy to calculate and interpret, (2) important in theory and practice, (3) frequently used [23,24,25]. Table 1 shows the descriptions and equations of these landscape metrics. The configuration metrics were quantified with Fragstats 4.2.
We calculated landscape metrics at 15 different scales (i.e., areas with different buffer radii of 30 m, 50 m, 100 m, 150 m, 200 m, 250 m, 300 m, 350 m, 400 m, 500 m, 700 m, 900 m, 1500 m, and 2000 m surrounding the weather station, hereafter called R30, R50, R100, R150, R200, R250, R300, R350, R400, R500, R600, R700, R900, R1500, R2000, respectively) to answer the question of how the relationship between AT and spatial pattern of urban greenspace responds to the change of scale (ref).
2.2.4. Statistical Analyses
We first conducted Pearson correlation analysis to explore the bivariate relationship between AT and each of the landscape metrics (e.g., percent of greenspace, edge density). We conducted Pearson correlation analysis instead of Spearman correlation analysis for the following reasons: (1) Both AT and landscape metrics are continuous values. (2) AT and landscape metrics show clear linear relationships. (3) AT, percent of greenspace, and most configuration metrics are normally distributed. (4) Pearson correlation has been widely used to investigate the relationship between temperature and spatial pattern of urban greenspace [21,23]. The partial Pearson correlation analysis was then performed to analyze the effect of greenspace spatial configuration on AT while controlling for the impacts of landscape composition because the spatial configuration metrics are usually highly correlated with percent of greenspace [23,26]. We then performed multiple linear regression to investigate the effects of urban greenspace spatial patterns on AT using the stepwise algorithm to select explanatory variables. Stepwise multiple linear regression has been widely applied to explore the effects of landscape patterns on urban thermal environment [26,27]. Finally, we compared the relative contribution of spatial composition and configuration using variance partitioning [3,23]. The variation of AT was decomposed into three fractions: (1) unique effects of PLAND, (2) unique effects of configuration, and (3) joint effects.
All the statistical analyses were performed using R statistical software version 4.0.5 for different scales (buffers), respectively to test the scale effects. Daytime AT and nighttime AT were independently investigated to understand the diurnal difference in the relationship between urban greenspace spatial pattern and AT. Figure 2 shows the flowchart of this study.
3. Results
3.1. Descriptive Statistics
AT showed strong spatial heterogeneity and great diurnal variation. The daytime temperature ranged from 31.8 °C to 33.6 °C with an average of 32.8 °C and a standard deviation (SD) of 0.39 °C. The night temperature ranged from 28 °C to 31 °C with an average of 29.72 °C and an SD of 0.624 °C (Figure 3).
The landscape metrics also displayed strong spatial variation and changed with the change of scale (i.e., buffer radius). The average PLAND was 52.55% for R30 and decreased slightly with the increase of the buffer radius, to 37.43% for R2000 (Figure 4a). SD of PLAND showed a similar decreasing pattern (from 29.54% to 8.89%) with the increase of buffer radius. The average and SD of AREA_MN showed a pattern of increase first and then decrease, with the highest value for R100 (i.e., 0.19 ± 0.4 patches per ha) and the lowest value for R2000 (i.e., 0.11 ± 0.05 per ha) (Figure 4b). The highest average and SD of PD, ED, and SHAPE_MN were 1039 ± 932 patches per km2, 738.2 ± 450.9 m per ha, and 1.62 ± 0.38, respectively, appeared at R30, and then decreased with the increase of buffer radius (Figure 4c–e). The average and SD of LSI increased with the increase of buffer radius, reaching the highest value of 82.73 ± 15.36 (Figure 4f).
3.2. Results of Correlation Analysis
Both daytime and nighttime AT showed a significant negative correlation with PLAND (Figure 5). The correlation between daytime AT and PLAND showed a “U”-shaped pattern (i.e., increasing when the buffer radius is smaller than 100 m and decreasing when the buffer radius is larger than 700 m). The correlation between nighttime AT and PLAND showed an increasing trend. The daytime AT had a tighter correlation with PLAND than the nighttime AT when the buffer radius was smaller than 900 m, while nighttime AT had a tighter correlation with PLAND for R1500 and R2000.
The Pearson correlation showed that daytime AT significantly and positively correlated with PD, ED, and LSI when the buffer radius was smaller than 200 m, and significantly and negatively correlated with AREA_MN for all scales except R1500 and R2000 (Figure 6). The highest correlation was for R150, R50, R50, and R100, respectively for PD, ED, LSI, and AREA_MN. The nighttime AT showed a significant and negative correlation with AREA_MN and a significant positive correlation with the other four configuration metrics (e.g., PD for R150, ED for R100–R500, LSI for R50–R2000, SHAPE_MN for R400–R2000).
The partial correlations between AT and configuration metrics were less significant than the Pearson correlation. Only ED (R30, R50, R100) and LSI (R30, R50, R150) showed a significant correlation with daytime AT after controlling the effect of PLAND. Nighttime AT showed a tighter partial correlation with landscape metrics with PLAND controlled (i.e., R200–R2000 for ED, R100–R2000 for LSI, R150, and R250 for AREA_MN).
3.3. Results of Multiple Linear Regression
Landscape metrics of urban greenspace explained 14% (R2000) to 54% (R50) of the spatial variation of the daytime AT. PLAND was included in the stepwise model and showed negative regression coefficients at all scales (Table 2). ED positively impacted daytime AT for R30, R50, and R100, and the impact was not significant for R1500. LSI positively impacted daytime AT for R150 (significant), R200, and negatively impacted daytime AT for R1500. AREA_MN positively and significantly impacted daytime AT for R50. PD and SHAPE_MN were not included in the stepwise model for all scales (Table 2).
Landscape metrics of urban greenspace explained 14% (R30) to 55% (R400) of the spatial variation of the nighttime AT. PLAND was included in the stepwise model for all scales except for R100–R200 showing negative regression coefficients, and significantly impacted nighttime AT for R30, R50, R700, and R1500 (Table 3). PD significantly negatively impacted nighttime AT for R100, R250, R300, and R400–R600, and the impact was not significant for R150 and R350. ED significantly positively impacted nighttime AT for R400–R2000, and negatively impacted nighttime for R100–R150. LSI significantly positively impacted nighttime AT from R100–R300, and negatively impacted nighttime for R350. AREA_MN negatively impacted nighttime AT for R100 (significant), R150–R300, and positively impacted nighttime AT for R350, R700–R2000. SHAPE_MN positively significantly impacted nighttime AT for R50, R200 (significant), and the negative impact was not significant for R50 (Table 3).
3.4. Results of Variation Partitioning
The relative importance of spatial composition and spatial configuration of urban greenspace in explaining the variation of AT varied among scales and showed diurnal variations. In the daytime, both spatial composition and spatial configuration of urban greenspace impacted AT at R30–R200 and R1500 (Figure 7a). For R 250–R900 and R2000, only the spatial composition of urban greenspace impacted daytime AT. Generally, spatial composition had stronger independent impacts on daytime AT than spatial configuration. For R150 and R200, the joint effects of spatial composition and spatial configuration of urban greenspace were higher than those of the independent effects.
At nighttime, the relative importance of spatial configuration and spatial configuration on nighttime AT was much more complex. The spatial configuration of urban greenspace had stronger impacts on nighttime AT than spatial composition for R100–R350, while at other scales, spatial composition of urban greenspace was more important in explaining the variation of nighttime AT (Figure 7b). At all scales, the joint effects of spatial composition and spatial configuration of urban greenspace were lower than the independent effects. Interestingly, the joint effects of spatial composition and spatial configuration of urban greenspace were negative for R400, R500, R600, and R1500 (Figure 7b).
4. Discussion
4.1. Percent of Greenspace Impacts Air Temperature
We found a significant negative relationship between percent of greenspace and AT and the relationship was consistent between daytime and nighttime. This confirmed previous findings that urban greenspace could cool down the city by evapotranspiration and providing shade [25]. Greenspace cover showed a stronger correlation with daytime AT than nighttime AT when the buffer radius was smaller than 900 m. This was because the overall cooling effect of the canopy on surface temperature was the result of shading and evapotranspiration in the daytime, which could convert solar radiation into latent heat fluxes. At night, canopy evapotranspiration tended to weaken as photosynthesis stops, in part due to the tree canopy blocking of longwave radiation and the loss of canopy turbulence [28]. The quick decrease in the correlation between daytime AT and greenspace cover with the increased buffer radius was possible because of water. With the increase of the scale, the proportion of water increased, which led to the weakening of the relationship between greenspace and temperature.
In general, the correlation between vegetation and temperature increased volatility with the increase of the scale. During the day, the correlation increased first and then decreased, and the first turning point appeared at 150 m buffer, the daytime correlation decreased sharply at 900 m buffer, and the nighttime correlation increased with the increase of the scale, and the first turning point appeared at 100 m (Figure 5). This was consistent with existing research [29]. This result quantitatively confirmed previous studies that claim that the cooling effect of green areas could extend for hundreds of meters [9,21,30]. With the increase of scale, the cooling effect of greenspace decreased due to the interference of other external factors [6].
4.2. Greenspace Configuration Impacts Air Temperature
Patch density of greenspace did not significantly impact daytime AT but significantly and negatively impacted nighttime AT (Table 2 and Table 3), indicating that the fragmented greenspace could decrease AT. This finding was not consistent with previous findings based on LST, for example in Beijing [21,31], but was consistent with the findings in Shanghai [24]. The possible negative relationship between PD and nighttime AT was that the fragmented greenspace could help heat release at night.
Edge density of greenspace showed a significant positive relationship with both daytime AT (at a smaller scale) and nighttime AT (at a larger scale) (Table 2 and Table 3). This indicated that with the same amount of greenspace coverage, greenspace with more edges has a higher AT, consistent with previous studies [23,28]. However, Qian et al. obtained a significant negative relationship between AT and edge density of greenspace in Beijing, possibly because the increase of edge density may lead to the increase of shade provided by trees [21]. However, higher ED also indicated a more fragmented greenspace, which may reduce the effectiveness of evapotranspiration. The increase of edge density led to a decrease in evapotranspiration, which likely outweighed the increased cooling effect by shade and thus resulted in a positive correlation between edge density and AT in the low latitude regions [28,32].
The mean patch size of greenspace showed a significant negative relationship with AT by Pearson correlation, but after controlling for the effect of the percent of greenspace, the correlation weakened significantly (Figure 6). The negative impacts of mean patch size on AT were consistent with the findings by Qian et al. [21]. This was possible because of the high correlation between the percent of greenspace and the mean patch size of greenspace [23].
The land shape index of greenspace showed a positive relationship with both daytime and nighttime AT for both Pearson correlation and partial Pearson correlation. The mean shape index of greenspace showed a significant positive relationship with the daytime temperature at R50 buffer and showed a significant positive relationship with nighttime temperature for R400–R2000, but after controlling for the effect of percent of greenspace, the correlation was not significant (Figure 6). Consistent with the results of existing studies, the shape of greenspace was negatively correlated with the park cold island effect [33]. However, Jaganmohan et al. suggested that complex greenspaces provided less cooling in smaller areas and none in larger areas. For smaller greenspaces, this may be because the increased irregular shape provided longer greenspace to communicate with the surroundings and more cool air could be transported from the greenspace and vice versa [34]. In this study, the correlation between the proportion of greenspace area and temperature decreased with increasing scale, and the shape index correlation increased with increasing scale, making the shape index positively correlated with temperature at all scales. Shashua-Bar and Hoffman’s [35] study showed that the greenspace shape index had little effect on the cooling effect, while other studies concluded that the location and shape of the blue-greenspace had a significant effect on the cooling effect [36]. The existing shape index is controversial in terms of its effect on the thermal environment and may be related to the existing spatial pattern of greenspaces.
4.3. Diurnal Variations of the Greenspace Spatial Configuration Impact on AT
Spatial pattern of urban greenspace explained more variation of nighttime AT than daytime AT (Table 2 and Table 3). A study in Seoul, Korea showed that percent of greenspace in addition to other landscape variables explained higher AT variation in nighttime (with R2 between 0.6–0.7) than in daytime (with R2 between 0.3–0.5) [37]. Percent of tree cover and percent of impervious surface area effectively impact nighttime AT but not daytime AT in Knoxville, USA [6]. Our study also showed that few spatial configuration metrics had a significant impact on daytime AT, indicating greenspace spatial configuration impacts nighttime AT more than daytime AT. The study of Howe et al. identified percent of tree cover as the driving factor of nighttime AT, but no driving factors of daytime AT were identified [6]. The diurnal variations of the greenspace spatial configuration impact on AT demonstrated a complex energy process between the earth surface, atmosphere, and human disturbance.
4.4. Limitations and Future Research Recommendations
This study has some limitations and further investigations are highly recommended. Firstly, the greenspace in this study included both trees and grassland. Though grasslands only cover a small fraction of the study area, separating them may better reveal the relationship between the spatial configuration of greenspace and AT [38]. Secondly, we investigated the impact of horizontal greenspace configuration on AT but did not consider the vertical tree canopy structure that has significant impacts on LST [27]. Thirdly, we only investigated the impacts of greenspace spatial configuration on AT in hot summer when the city experiences heat stress. Previous studies have demonstrated that greenspace spatial configuration also significantly impacts the urban thermal environment in winter and the effects may be different to those in summer [39]. Therefore, similar studies can be conducted to explore the greenspace spatial configuration impacts on AT in the colder winter in Changsha, China [40]. Finally, this study only studied one of the driving factors of human thermal comfort. In future studies, the spatial configuration of urban greenspace should be linked to human thermal comfort considering more variables such as humidity and solar radiation to design more cost-effective spatial configuration optimization strategies [4,5,41,42,43].
5. Conclusions
This study investigated the impact of greenspace spatial patterns on daytime AT and nighttime AT in the hot summer of Changsha, China. The results showed that both spatial composition and spatial configuration of urban greenspace significantly impact both daytime and nighttime AT. The spatial pattern of urban greenspace explained more variations of nighttime AT than daytime AT. The spatial pattern of urban greenspace within the 50 m buffer best explained the variation of daytime AT and that within the 400 m buffer explained the most variation of nighttime AT. The effect of greenspace patterns on AT was scale-dependent and varied diurnally. The spatial pattern of greenspace had a greater effect on daytime AT at a small scale and a greater effect on nighttime AT at a large scale. The relative importance of landscape composition and spatial configuration on AT was scale-dependent and varied diurnally. Based on the findings, we suggest the following strategies to optimize the spatial pattern of urban greenspace to decrease the summer AT: (1) Increasing percent of greenspace by planting more trees where possible. For example, roof greening can be an effective choice considering the wet weather in Changsha, China [22]. (2) Adding more small-sized greenspace patches can effectively decrease nighttime AT. Considering the scarcity of land for developing big greenspace patches, developing scattered small parks in the residential areas can be more effective. (3) Reducing patch complexity of greenspace should also be carried out in urban greenspace planning and management.
Conceptualization, X.L. and Q.T.; methodology, X.L., Q.T. and H.L.; formal analysis, Q.T., Q.Q., Z.W., Z.C., L.H., H.L. and Y.F.; investigation, Q.Q. and Z.W.; data curation, Q.Q., Z.W. and Y.F.; writing—original draft preparation, X.L. and Q.T.; writing—review and editing, X.L., Q.T. and H.L.; supervision, X.L.; project administration, X.L.; funding acquisition, X.L. All authors have read and agreed to the published version of the manuscript.
Data available on request from the authors.
The authors declare no conflict of interest.
Footnotes
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.
Enlarge this image.Figure 4. Violin plot of landscape metrics for different scales. PLAND represents Percent of greenspace, PD represents Patch density, ED represents Edge density, LSI represents Landscape shape index, AREA_MN represents Mean patch size, SHAPE_MN represents Mean patch shape index.
Enlarge this image.Figure 5. Pearson correlation between PLAND of greenspace and AT (daytime and nighttime).
Enlarge this image.Figure 6. Pearson correlation and partial Pearson correlation between AT and configuration metric. Cor-day: correlation between average daytime temperature and configuration metrics; Pcor-day: Partial correlation coefficients between average daytime temperature and configuration metrics; Cor-night: correlation between average nighttime temperature and configuration metrics; Pcor-night: Partial correlation coefficients between average nighttime temperature and configuration metrics. PLAND represents Percent of greenspace, (a) PD represents Patch density, (b) ED represents Edge density, (c) LSI represents Landscape shape index, (d) AREA_MN represents Mean patch size, (e) SHAPE_MN represents Mean patch shape index.
Summary of the used landscape metrics.
| Landscape Metric | Equation | Description (Unit) |
|---|---|---|
| Percent of greenspace (PLAND) |
|
The proportion of greenspace area (%). |
| Patch density (PD) |
|
Number of greenspace patches divided by the total landscape area (n/km2) |
| Edge density (ED) |
|
The total perimeter of vegetation canopy patches per km2 within an analysis unit (m/ha). |
| Landscape shape index (LSI) |
|
The total length of greenspace (or perimeter) divided by the minimum length of greenspace edge (or perimeter) possible for a maximally aggregated class. |
| Mean patch size (AREA_MN) |
|
The average area of vegetation canopy patches within an analysis unit (ha). |
| Mean patch shape index (SHAPE_MN) |
|
The average shape index of vegetation canopy patches within an analysis unit. |
Summary of the stepwise regression analysis for summer daytime air temperature (*** p < 0.001, ** p < 0.01, * p < 0.05).
| Scales | PLAND | PD | ED | LSI | AREA_MN | SHAPE_MN | R2 |
|---|---|---|---|---|---|---|---|
| 30 | −0.005 * | 0.0003 * | 0.32 | ||||
| 50 | −0.012 *** | 0.0009 *** | 1.252 ** | 0.54 | |||
| 100 | −0.009 ** | 0.0005 * | 0.40 | ||||
| 150 | −0.008 * | 0.068 * | 0.42 | ||||
| 200 | −0.010 * | 0.041 | 0.34 | ||||
| 250 | −0.015 *** | 0.30 | |||||
| 300 | −0.016 *** | 0.31 | |||||
| 350 | −0.016 *** | 0.32 | |||||
| 400 | −0.016 *** | 0.32 | |||||
| 500 | −0.017 *** | 0.32 | |||||
| 600 | −0.018 *** | 0.34 | |||||
| 700 | −0.019 *** | 0.35 | |||||
| 900 | −0.018 *** | 0.30 | |||||
| 1500 | −0.05 * | 0.004 | −0.044 | 0.21 | |||
| 2000 | −0.018 * | 0.14 |
Summary of the stepwise regression analysis for summer nighttime air temperature (*** p < 0.001, ** p < 0.01, * p < 0.05).
| Scales | PLAND | PD | ED | LSI | AREA_MN | SHAPE_MN | R2 |
|---|---|---|---|---|---|---|---|
| 30 | −0.018 * | 0.14 | |||||
| 50 | −0.012 ** | 0.515 | 0.26 | ||||
| 100 | −0.0009 * | −0.001 | 0.363 ** | −0.642 * | 0.36 | ||
| 150 | −0.001 | −0.002 | 0.342 ** | −0.57 | 0.40 | ||
| 200 | 0.092 * | −1.013 | 1.598 ** | 0.49 | |||
| 250 | −0.009 | −0.002 ** | 0.162 *** | −1.169 | 0.50 | ||
| 300 | −0.010 | −0.002 ** | 0.139 *** | −1.138 | 0.50 | ||
| 350 | −0.047 | −0.002 | 0.003 | −0.007 | 1.097 | −0.771 | 0.48 |
| 400 | −0.036 | −0.002 * | 0.003 *** | 0.55 | |||
| 500 | −0.035 | −0.002 * | 0.003 *** | 0.52 | |||
| 600 | −0.035 | −0.001 * | 0.003 *** | 0.49 | |||
| 700 | −0.051 *** | 0.002 *** | 3.872 | 0.48 | |||
| 900 | −0.054 | 0.003 ** | 4.27 | 0.48 | |||
| 1500 | −0.064 *** | 0.003 *** | 4.605 | 0.52 | |||
| 2000 | −0.066 | 0.003 *** | 4.610 | 0.51 |
References
1. Lei, Y.; Davies, G.M.; Jin, H.; Tian, G.; Kim, G. Scale-dependent effects of urban greenspace on particulate matter air pollution. Urban For. Urban Green.; 2021; 61, 127089. [DOI: https://dx.doi.org/10.1016/j.ufug.2021.127089]
2. Youngsteadt, E.; Ernst, A.F.; Dunn, R.R.; Frank, S.D. Responses of arthropod populations to warming depend on latitude: Evidence from urban heat islands. Glob. Chang. Biol.; 2017; 23, pp. 1436-1447. [DOI: https://dx.doi.org/10.1111/gcb.13550] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/27809387]
3. Li, X.; Zhou, Y.; Yu, S.; Jia, G.; Li, H.; Li, W. Urban heat island impacts on building energy consumption: A review of approaches and findings. Energy; 2019; 174, pp. 407-419. [DOI: https://dx.doi.org/10.1016/j.energy.2019.02.183]
4. Li, X.; Li, X.; Tang, N.; Chen, S.; Deng, Y.; Gan, D. Summer outdoor thermal perception for the elderly in a comprehensive park of Changsha, China. Atmosphere; 2022; 13, 1853. [DOI: https://dx.doi.org/10.3390/atmos13111853]
5. Deng, Y.; Gan, D.; Tang, N.; Cai, Z.; Li, X.; Chen, S.; Li, X. Research on outdoor thermal comfort and activities in residential areas in subtropical China. Atmosphere; 2022; 13, 1357. [DOI: https://dx.doi.org/10.3390/atmos13091357]
6. Howe, D.A.; Hathaway, J.M.; Ellis, K.N.; Mason, L.R. Spatial and temporal variability of air temperature across urban neighborhoods with varying amounts of tree canopy. Urban For. Urban Green.; 2017; 27, pp. 109-116. [DOI: https://dx.doi.org/10.1016/j.ufug.2017.07.001]
7. Zhen, M.; Hong, F.; Zhou, D. The relationship between spatial arrangement and environmental temperature of residential areas in Xi’an. Indoor Built Environ.; 2019; 28, pp. 1288-1300. [DOI: https://dx.doi.org/10.1177/1420326X19860872]
8. Jung, M.C.; Dyson, K.; Alberti, M. Urban landscape heterogeneity influences the relationship between tree canopy and land surface temperature. Urban For. Urban Green.; 2021; 57, 126930. [DOI: https://dx.doi.org/10.1016/j.ufug.2020.126930]
9. Ziter, C.D.; Pedersen, E.J.; Kucharik, C.J.; Turner, M.G. Scale-dependent interactions between tree canopy cover and impervious surfaces reduce daytime urban heat during summer. Proc. Natl. Acad. Sci. USA; 2019; 116, pp. 7575-7580. [DOI: https://dx.doi.org/10.1073/pnas.1817561116]
10. Masoudi, M.; Tan, P.Y. Multi-year comparison of the effects of spatial pattern of urban green spaces on urban land surface temperature. Landsc. Urban Plan.; 2019; 184, pp. 44-58. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2018.10.023]
11. Liao, W.; Guldmann, J.-M.; Hu, L.; Cao, Q.; Gan, D.; Li, X. Linking urban park cool island effects to the landscape patterns inside and outside the park: A simultaneous equation modeling approach. Landsc. Urban Plan.; 2023; 232, 104681. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2022.104681]
12. Patz, J.A.; Campbell-Lendrum, D.; Holloway, T.; Foley, J.A. Impact of regional climate change on human health. Nature; 2005; 438, pp. 310-317. [DOI: https://dx.doi.org/10.1038/nature04188]
13. Tan, J.; Zheng, Y.; Tang, X.; Guo, C.; Li, L.; Song, G.; Zhen, X.; Yuan, D.; Kalkstein, A.J.; Li, F. The urban heat island and its impact on heat waves and human health in Shanghai. Int. J. Biometeorol.; 2010; 54, pp. 75-84. [DOI: https://dx.doi.org/10.1007/s00484-009-0256-x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/19727842]
14. Arbuthnott, K.G.; Hajat, S. The health effects of hotter summers and heat waves in the population of the United Kingdom: A review of the evidence. Environ. Health; 2017; 16, 119. [DOI: https://dx.doi.org/10.1186/s12940-017-0322-5] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29219088]
15. Cao, J.; Zhou, W.; Zheng, Z.; Ren, T.; Wang, W. Within-city spatial and temporal heterogeneity of air temperature and its relationship with land surface temperature. Landsc. Urban Plan.; 2021; 206, 103979. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2020.103979]
16. Zhao, D.; Lei, Q.; Shi, Y.; Wang, M.; Chen, S.; Shah, K.; Ji, W. Role of species and planting configuration on transpiration and microclimate for urban trees. Forests; 2020; 11, 825. [DOI: https://dx.doi.org/10.3390/f11080825]
17. Lai, D.; Liu, Y.; Liao, M.; Yu, B. Effects of different tree layouts on outdoor thermal comfort of green space in summer Shanghai. Urban Clim.; 2023; 47, 101398. [DOI: https://dx.doi.org/10.1016/j.uclim.2022.101398]
18. Xiao, X.; Zhang, L.; Xiong, Y.; Jiang, J.; Xu, A. Influence of spatial characteristics of green spaces on microclimate in Suzhou industrial park of China. Sci. Rep.; 2022; 12, 9121. [DOI: https://dx.doi.org/10.1038/s41598-022-13108-1]
19. Jiao, M.; Zhou, W.; Zheng, Z.; Wang, J.; Qian, Y. Patch size of trees affects its cooling effectiveness: A perspective from shading and transpiration processes. Agric. For. Meteorol.; 2017; 247, pp. 293-299. [DOI: https://dx.doi.org/10.1016/j.agrformet.2017.08.013]
20. Alonzo, M.; Baker, M.E.; Gao, Y.; Shandas, V. Spatial configuration and time of day impact the magnitude of urban tree canopy cooling. Environ. Res. Lett.; 2021; 16, 084028. [DOI: https://dx.doi.org/10.1088/1748-9326/ac12f2]
21. Qian, Y.; Zhou, W.; Hu, X.; Fu, F. The heterogeneity of air temperature in urban residential neighborhoods and its relationship with the surrounding greenspace. Remote Sens.; 2018; 10, 965. [DOI: https://dx.doi.org/10.3390/rs10060965]
22. Hu, L.; Fan, C.; Cai, Z.; Liao, W.; Li, X. Greening residential quarters in China: What are the roles of urban form, socioeconomic factors, and biophysical context?. Urban. For. Urban. Green.; 2023; 86, 128020. [DOI: https://dx.doi.org/10.1016/j.ufug.2023.128020]
23. Zhou, W.; Wang, J.; Cadenasso, M.L. Effects of the spatial configuration of trees on urban heat mitigation: A comparative study. Remote Sens. Environ.; 2017; 195, pp. 1-12. [DOI: https://dx.doi.org/10.1016/j.rse.2017.03.043]
24. Zhou, W.; Cao, F. Effects of changing spatial extent on the relationship between urban forest patterns and land surface temperature. Ecol. Indic.; 2020; 109, 105778. [DOI: https://dx.doi.org/10.1016/j.ecolind.2019.105778]
25. Guo, G.; Wu, Z.; Chen, Y. Complex mechanisms linking land surface temperature to greenspace spatial patterns: Evidence from four southeastern Chinese cities. Sci. Total Environ.; 2019; 674, pp. 77-87. [DOI: https://dx.doi.org/10.1016/j.scitotenv.2019.03.402]
26. Li, X.; Zhou, W.; Ouyang, Z. Relationship between land surface temperature and spatial pattern of greenspace: What are the effects of spatial resolution?. Landsc. Urban Plan.; 2013; 114, pp. 1-8. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2013.02.005]
27. Chen, J.; Zhan, W.; Jin, S.; Han, W.; Du, P.; Xia, J.; Lai, J.; Li, J.; Liu, Z.; Li, L. et al. Separate and combined impacts of building and tree on urban thermal environment from two- and three-dimensional perspectives. Build. Environ.; 2021; 194, 107650. [DOI: https://dx.doi.org/10.1016/j.buildenv.2021.107650]
28. Chen, J.; Jin, S.; Du, P. Roles of horizontal and vertical tree canopy structure in mitigating daytime and nighttime urban heat island effects. Int. J. Appl. Earth Obs. Geoinf.; 2020; 89, 102060. [DOI: https://dx.doi.org/10.1016/j.jag.2020.102060]
29. Yan, H.; Fan, S.; Guo, C.; Hu, J.; Dong, L. Quantifying the impact of land cover composition on intra-urban air temperature variations at a mid-latitude city. PLoS ONE; 2014; 9, e102124. [DOI: https://dx.doi.org/10.1371/journal.pone.0102124]
30. Zhang, J.; Cui, P.; Song, H. Impact of urban morphology on outdoor air temperature and microclimate optimization strategy base on pareto optimality in northeast China. Build. Environ.; 2020; 180, 107035. [DOI: https://dx.doi.org/10.1016/j.buildenv.2020.107035]
31. Li, X.; Zhou, W.; Ouyang, Z.; Xu, W.; Zheng, H. Spatial pattern of greenspace affects land surface temperature: Evidence from the heavily urbanized beijing metropolitan area, China. Landsc. Ecol.; 2012; 27, pp. 887-898. [DOI: https://dx.doi.org/10.1007/s10980-012-9731-6]
32. Keno, T.B.; Nengcheng, C.; Xiang, Z.; Dev, N. Spatial configuration and extent explains the urban heat mitigation potential due to green spaces: Analysis over Addis Ababa, Ethiopia. Remote Sens.; 2020; 12, 2876.
33. Yang, G.; Yu, Z.; Jørgensen, G.; Vejre, H. How can urban blue-green space be planned for climate adaption in high-latitude cities? A seasonal perspective. Sustain. Cities Soc.; 2020; 53, 101932. [DOI: https://dx.doi.org/10.1016/j.scs.2019.101932]
34. Jaganmohan, M.; Knapp, S.; Buchmann, C.M.; Schwarz, N. The bigger, the better? The influence of urban green space design on cooling effects for residential areas. J. Environ. Qual.; 2016; 45, pp. 134-145. [DOI: https://dx.doi.org/10.2134/jeq2015.01.0062] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26828169]
35. Shashua-Bar, L.; Hoffman, M.E. Vegetation as a climatic component in the design of an urban street: An empirical model for predicting the cooling effect of urban green areas with trees. Energy Build; 2000; 31, pp. 221-235. [DOI: https://dx.doi.org/10.1016/S0378-7788(99)00018-3]
36. Bowler, D.E.; Buyung-Ali, L.; Knight, T.M.; Pullin, A.S. Urban greening to cool towns and cities: A systematic review of the empirical evidence. Landsc. Urban Plan.; 2010; 97, pp. 147-155. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2010.05.006]
37. Ha, J.; Choi, Y.; Lee, S.; Oh, K. Diurnal and seasonal variations in the effect of urban environmental factors on air temperature: A consecutive regression analysis approach. Int. J. Environ. Res. Public Health; 2020; 17, 421. [DOI: https://dx.doi.org/10.3390/ijerph17020421]
38. Zhou, W.; Huang, G.; Cadenasso, M.L. Does spatial configuration matter? Understanding the effects of land cover pattern on land surface temperature in urban landscapes. Landsc. Urban Plan.; 2011; 102, pp. 54-63. [DOI: https://dx.doi.org/10.1016/j.landurbplan.2011.03.009]
39. Zhou, W.; Qian, Y.; Li, X.; Li, W.; Han, L. Relationships between land cover and the surface urban heat island: Seasonal variability and effects of spatial and thematic resolution of land cover data on predicting land surface temperatures. Landsc. Ecol.; 2014; 29, pp. 153-167. [DOI: https://dx.doi.org/10.1007/s10980-013-9950-5]
40. Peng, Y.; Gan, D.; Cai, Z.; Xiao, M.; Shu, D.; Wu, C.; Yu, X.; Li, X. Landscape features impact the spatial heterogeneity of visitation density within a comprehensive park: What are the seasonal and diurnal variations?. Forests; 2023; 14, 1627. [DOI: https://dx.doi.org/10.3390/f14081627]
41. Lin, T.-P.; Matzarakis, A. Tourism climate and thermal comfort in sun moon lake, Taiwan. Int. J. Biometeorol.; 2008; 52, pp. 281-290. [DOI: https://dx.doi.org/10.1007/s00484-007-0122-7]
42. Lin, T.-P.; Matzarakis, A.; Hwang, R.-L. Shading effect on long-term outdoor thermal comfort. Build. Environ.; 2010; 45, pp. 213-221. [DOI: https://dx.doi.org/10.1016/j.buildenv.2009.06.002]
43. Lin, T.-P.; Matzarakis, A. Tourism climate information based on human thermal perception in Taiwan and eastern China. Tour. Manag.; 2011; 32, pp. 492-500. [DOI: https://dx.doi.org/10.1016/j.tourman.2010.03.017]
© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/). Notwithstanding the ProQuest Terms and Conditions, you may use this content in accordance with the terms of the License.