2.1 Study area sample choice and measurement point setting
Based on UAV (Unmanned Aerial Vehicle; DJI MAVIC) aerial photos with a perspective elevation of 150 m, and previous research on parks and green spaces in SIP, we selected 15 green spaces in parks as the research objects. The selection of our research objects was mainly based on the following aspects. First, considering the size range of the 15 green park spaces, and by reference to previous research reports, we adopted 4 ha and 10 ha as the critical point in defining small, medium, and large size green spaces (Gao et al. 2012; Wu et al. 2007) to guarantee that the spaces were comparable. The 15 green spaces were divided into small-size green spaces (area < 4 ha), medium-size green spaces (area = 4–10 ha) and large-size green spaces (area > 10 ha). Secondly, we recorded the spatial characteristics of the 15 green spaces. The green coverage rate ranged from 28.73–94.35%, the canopy density ranged from 0.21 to 0.85, and the selected green spaces had water bodies of different sizes. Within each green space, we distinguished between green spaces a short, medium, or long distance from water (lawn), sparse forest (canopy density range of 0.4 to 0.6), dense forest (canopy density range of 0.7 to 1.0) and other patches. Green spaces a short/medium/long distance from water were classified based on the distance between the boundary of the green space and that of the water body inside the green space. Since the distance between the green space boundary and that of the water bodies differed, green spaces a short/medium/long distance from water were determined based on the distance ratio. For example, if the distance between the boundary of the water body and that of the green space within the researched sample is 200 m (i.e., the ratio is 1), then a green space which is 50 m (ratio of 1/4) away from the boundary of the water body is defined as a green space a short distance from water. A green space which is 100 m (ratio of 1/2) away from the water body boundary is defined as a green space a medium distance from water. A green space which is 150 m (ratio of 3/4) away from the boundary of the water body is defined as a green space a long distance from water. Each research sample was defined as a circular area with a radius of 10 m. In each research sample, three observation points were distributed evenly. Figure 1 and Table 1 show the detailed characteristics of the 15 green spaces. In the numerical simulation study on the effect of the shape and area of the water body, the shapes of water bodies in the 15 green spaces were classified into representative and typical banded water, massive water, and annular water (see Fig. 2).
Table 1
Green space outline and structural geometry
Green space types
|
Name of green space
|
Green space outline
|
Area of green space (m2)
|
Perimeter of green space (m)
|
Perimeter-area ratio
|
Small-size green space
(Area < 4ha)
|
Wenxing Square
|
|
26649
|
800
|
0.03
|
North of Shin Kong Place
|
|
7711.39
|
354
|
0.04
|
Lakeside garden complex
|
|
9532.20
|
780
|
0.08
|
Xinghai Park
|
|
32419.30
|
1010
|
0.03
|
Pingtan School
|
|
14560.89
|
768
|
0.05
|
Medium-size green space
(Area = 4-10ha)
|
Lotus Park
|
|
66644.81
|
1540
|
0.02
|
Dushu Lake Tunnel
|
|
43020.41
|
1850
|
0.04
|
Hongfenglin Park on Guanfeng Street
|
|
71697.57
|
1050
|
0.01
|
Jinjidun
Park
|
|
43609.21
|
855
|
0.02
|
Water Paradise green space
|
|
72596.15
|
1260
|
0.02
|
Zhongtang Park
|
|
84586.27
|
2050
|
0.02
|
Large-size green space
(Area > 10ha)
|
Baitang Botanical Garden
|
|
230520.95
|
2640
|
0.01
|
Fangzhou
Park
|
|
102740.25
|
1260
|
0.01
|
Greenbelt of Nano Technology park
|
|
122666.00
|
1890
|
0.02
|
Central
Park
|
|
142379.46
|
1460
|
0.01
|
Note: Perimeter-area ratio = Perimeter / area of green space. The ratio characterizes the shape of the green space. |
2.3 Statistical analyses
Previous research suggested that urban cold islands, formed by parks consisting mainly of green spaces and water bodies, effectively mitigated the urban heat island effect and improved the urban thermal environment. With a greater area of green space and water bodies, it is easier to form strong local circulation, producing a greater effect on the surrounding thermal effect. Therefore, there is a significant positive correlation between the cooling effect, the area of woodland, and water bodies, which are the key factors that influence the internal and external thermal environment of a park. In this research, we first calculated the average value of temperatures and humidity at each hour between 10:00 and 14:00 at the three observation points. This procedure was carried out for all small-size, medium-size, and large-size green spaces. Then, by processing the temperature and humidity data, we obtained the cooling and humidifying intensity at each site. The three repeated contrast points were located at an empty area about 100 m away from the sampling points. We used the SPSS22.0 software to analyze correlations between the perimeter, area, perimeter-area ratio, leaf area index, canopy density, and other structural characteristics of the green spaces and their cooling and humidifying effects to construct a prediction model. The formula for leaf area index (LAI) is given as Eq. 1
$$LAI=In{\tau }_{\phi i}/-{K}_{\phi i}$$1
In the formula, LAI represents the leaf area index, \({\tau }_{\phi }\) represents the direct radiation transmission (or visible sky ratio) coefficient in each zenith angle area, \({\tau }_{\phi i}\) represents the i-th zenith angle division, and K represents the extinction coefficient of the canopy. The calculation formula for canopy density is expressed as Eq. 2:
where CD represents the canopy density, β represents the sky pixel value of the plant canopy, and α represents the pixel value of the plant canopy. We then used Duncan's method of multiple comparisons to examine the differences in the cooling and humidifying intensity of the green spaces (a short/medium/long distance from water), thus analyzing the influence of water bodies on the cooling and humidifying effect of green spaces. We also adopted numerical modeling to research further the cooling and humidifying effect of banded water, massive water, and annular water. We defined the cooling intensity (cooling effect) as the difference between the air temperature at the contrast point and the average air temperature at each research point. The calculation is shown in Eq. 3:
where t1 represents the air temperature at the contrast point, and t2 represents the average air temperature at each measurement point in the sample site. ∆T, the difference between the above temperatures, is defined as the cooling effect. We defined the humidifying intensity (humidifying effect) as the difference between the relative humidity of the contrast point and the average air relative humidity of each measurement point. The calculation is shown in Eq. 4:
$$\varDelta H={h}_{1}-{h}_{2}$$4
where h1 represents the average air humidity of each measurement point in the green space, and h2 represents the air humidity of the contrast point. ∆H, the difference between humidities, is defined as the humidifying effect (Zhang et al. 2010).
We used a numerical simulation method to study the effects of increasing the area of water bodies by 5% and 10%, and the effect of water shape (banded water, massive water, and annular water areas) on cooling and humidification. Figure 3 shows the simulation images in a 150×150 m area. Considering the characteristics of the simulation software and of the measurement site, the model had a total of 50×50×40 grids, and the grid resolution was 3×3 m. The parameter settings of this simulation input are presented in Table 2. In this study, the Root Mean Square Error (RMSE) and Mean Absolute Percentage Error (MAPE) of air temperature and relative humidity were 0.14℃ and 0.95% and 1.56% and 2.69%, respectively.
Table 2
Input parameter setting in simulations
Parameter
|
Setting
|
Start time of simulation
|
7:00 on 18 July, 2019
|
Simulation time
|
12h
|
Simulated initial temperature/relative humidity
|
31.13℃/78.75%(The measured value of the day)
|
Wind direction/speed
|
northeastern/5.5m/s
|
Roughness length
|
0.01
|
The multiple regression and numerical simulation models of the cooling and humidifying effects presented above produced statistically significant results. They constitute a useful method to evaluate and predict the cooling and humidifying effects of different types of green spaces. For data processing and statistical analysis in this experiment, we used SPSS22.0 (SPSS Inc.) software. For the numerical simulation, we used the ENVI-met software, while we used Microsoft Excel2019 and Origin8.0 software to produce the charts.