Effects of Tree Arrangement and Leaf Area Index on the Thermal Comfort of Outdoor Children’s Activity Space in Hot-Humid Areas

Abstract

Hot-humid areas have long, hot summers and poor outdoor thermal comfort (OTC). The urban heat island (UHI) effect exacerbates the deterioration of OTC in hot-humid areas, seriously affecting the thermal safety of children’s outdoor activities. In this study, 60 scenes were simulated using ENVI-met based on different leaf area index (LAI) and planting arrangements to explore how tree LAI and planting arrangements affect the small-scale thermal environment during hot summer months and to assess OTC using the Universal Thermal Climate Index (UTCI). The research shows that (1) high LAI trees optimize OTC more than low LAI trees, but low LAI trees can be planted multiple times to achieve the level of optimization of high LAI trees; (2) increasing the number of trees optimizes the OTC of the study area, reducing the UTCI by up to 3.7 °C with increased planting compared to unshaded areas; (3) thickening the shade in the east–west direction optimizes the OTC of the study area more than thickening the shade in the north–south direction, with too much north–south shade optimizing the OTC of the study area by only 0.01%. This study provides practical advice for the design of planting in outdoor CAS in hot-humid areas.

1. Introduction

In the context of increased urbanization, the UHI effect has increased significantly [1,2,3]. Under this influence, the incidence of heat-related morbidity in children has increased substantially [4]. Available research shows that children are more sensitive to heat than adults [5,6]. Under the same conditions, children showed a higher thermal sensation vote than other age groups [7]. This is due to differences in children’s different body conditions, clothing insulation and metabolic rates [8,9]. For urban children, urban parks provide a place for them to get in touch with nature and outdoor activities [10]. Playing in the natural environment is not only good for children’s physical and mental health but also helps to improve their cognitive skills [11,12]. However, many modern outdoor CAS designs lack specific thermal comfort considerations, and although they comply with current design codes, they may still be in unsafe thermal environments [13]. This phenomenon is due to a combination of (1) high temperature and humidity climates and the UHI effect [14]; (2) the play equipment being in an unshaded open space [15]; and (3) rides that promote high temperature transfer [16]. It is therefore very important to improve the OTC of the children’s activity area to provide a comfortable place for children to move around.

Qi et al. [17] administered a questionnaire to 303 children in Wuhan during the summer months, and the results showed that both children and guardians wanted to improve outdoor comfort in the summer by lowering the temperature. Additionally, Backlin et al. [18] found that shading with trees was the most suitable means of shading for young children in a simulation of overheating risk in 440 kindergarten courtyards in Sweden. In addition to this, many studies related to children’s OTC have identified trees as an essential part of OTC [19], but few have proposed thermal comfort-based planting strategies for outdoor CAS.

Trees are an important landscape element in the city [20]. Proper planting design not only brings beautiful natural landscapes but also mitigates UHIs through transpiration and shading [21,22]. The transpiration of trees absorbs some of the heat from the surrounding environment, and shade is an important influence in mitigating UHIs [23]. In the same shade, the air temperature (Ta) is lower in the shade of a tree than in the shade of a building, due to the absorption of solar radiation by the leaves and branches of the tree and the effect of transpiration [24]. The canopy consists of branches and leaves, the leaf area index (LAI) is often used to describe canopy cover [25], and the ability of trees to improve the OTC is correlated with LAI [26]. Shahidan et al. [27] investigated the ability of conspecific trees with different LAIs to improve the OTC and found that a Tsuga chinensis with an LAI of 6.1 was significantly better than a Tsuga chinensis with an LAI of 1.5. They also found differences in the ability of different tree species to improve the OTC. Both Mesua ferrea and Hura crepitans were able to reduce Ta below the canopy, but Mesua ferrea had a better cooling effect. This was also found by Ruzana et al. [28] In the same direction, in streets with Ulmus pumila and Firmiana simplex, the OTC of the Ulmus pumila streets is obviously better than that of streets with Firmiana simplex. The reasons may be as follows: (1) Different tree species can form shadows in different areas due to the LAI and crown width [29]. (2) Different xylem structures of different tree species lead to different water use efficiencies [30]. (3) The process of stoma opening and closing differs among tree species [31].

Under the same tree coverage, different tree planting forms will produce different areas of shade and different forms of space under trees, which will have different effects on the OTC [32,33]. Planting forms such as “solitary, rows and groups” are widely used in today’s landscape design [34]. Zhao et al. [35] studied the ability of different tree layouts to regulate the local microclimate, and the research results show that the regulation effect of multiple impacts is far greater than that of single impacts. This result reinforces the stereotype that “more cultivation is better for OTC”. However, this conventional view was broken by a study by Ouyang et al. [36], who explored the effect of tree cover on OTC at different urban densities and showed that the optimum efficiency of OTC was achieved with 20–30% tree cover regardless of increasing building density, suggesting that there is a threshold for trees to optimize OTC in local spaces and that too many trees cannot further optimize the local OTC. Liu et al. [37] further found that the OTC of the environment can vary significantly between trees with different canopy forms that form different percentages of cover under the same planting density. This shows that tree cover is also a major factor in environmental OTC and that a better OTC can be achieved with fewer trees by choosing reasonable tree planting. These studies validate the view of Tan et al. [38] that more trees are not better in special scenes and that the positive impact of trees on the thermal environment will be wasted if they are not planted appropriately. Therefore, landscape architects need to select appropriate tree species according to the characteristics of different scenes and choose rational and efficient planting forms in the urban environment to maximize the positive effects of trees on OTC [39].

In summary, trees have an undeniable role in mitigating the urban heat island effect and creating a good natural landscape. However, little previous research has explored how the placement of trees in outdoor children’s areas combined with LAI affects microclimate and OTC. This is especially true in hot-humid areas, where humid and hot climatic conditions expose OTC to additional risks. As children are a sensitive group to thermal environments, the impact of trees on OTCs in CAS and planting strategies deserves more research and attention. The significance of this study was to highlight the multiple effects of tree LAI and planting form on the OTC of children’s outdoor activity areas in the hot-humid areas, and identify planting designs that effectively improve OTCs in CAS and guide tree planting designs for outdoor CAS in hot-humid areas. Therefore, we chose a real outdoor CAS and an idealized site in the typical hot and humid climate city of Guangzhou (China) to explore the following:

(1)

How does the combination of different tree species and planting number affect OTC?

(2)

How can OTC be improved through greater efficiency planting forms in CAS?

(3)

How can outdoor children’s activity equipment be arranged based on thermal comfort principles?

2. Methodology

The research framework is shown in Figure 1. The study was conducted in two stages (Section 2.5) and a total of 60 scenarios were simulated using ENVI-met. The accuracy of the ENVI-met model was verified by taking real measurements within the original site. The first stage was conducted in a real CAS in Guangzhou city, where the original tree species within the site were replaced with six tree species commonly found in Guangzhou. In the second stage, the trees around the site were removed and an ideal site was created with typical Guangzhou City trees with LAIs of 1–6. In addition, to describe the thermal comfort of humans, the Universal Thermal Climate Index (UTCI) values output by ENVI-met were used as a judgement criterion. The study concludes with a summary and recommendations for the design and adaptation of the CAS.

2.1. Study Sites

Guangzhou (112.8° E–114.2° E, 22.3° N–24.1° N) is partially located in the south of the Tropic of Cancer. The daily solar trajectory of Guangzhou during the summer months (June–September) was calculated according to Ecotect Analysis 2011 software (Figure 2), where the yellow dot indicates the position of the sun, the line connecting the yellow dot to the square box indicates the solar rays, and the middle rectangle indicates the study area. Overall, the sun rises to the northeast and sets to the northwest, with no significant changes in the shadow orientation of the study area during this period. Therefore, modelled and measured values for a typical summer day in July were used to assess the OTCs of outdoor CAS throughout the summer.

With an average annual Ta of 22 °C and an average annual RH of 77%, [40,41] Guangzhou is a hot-humid area. According to statistical data, the average summer Ta in Guangzhou is 27.5–29 °C, with a maximum Ta of 35.9 °C in mid to late July, a relative RH of approximately 80%, an average solar radiation of 137 MJ/m² and a prevailing wind direction of southeast (135°) [42,43]. In this study, on-site measurements of meteorological data in Guangzhou city were conducted on 13 July, and the actual instruments and measured data are shown in

Table 1. Instruments and technical parameters.

	Measured Instrument	Measured Parameters	Instrument Range	Instrument Accuracy	Record Time Interval
	Kestrel5500 (USA)	Wind speed	0~5 m/s	±0.05 m/s	Machine records 5 min/time
	HOBO Pro (USA)	Air temperature	−40~70 °C	±0.5 °C	Machine records 5 min/time
	HOBO Pro (USA)	Relative humidity	0~100%	±2.5%	Machine records 5 min/time
	TBQ-2 (China)	Solar radiation	280~3000 nm	10.436 µV/Wm2	Note 2 min/times

and Figure 3. The dominant wind direction was 135° during the measurement period, the wind speed (WS) ranged from 0.5 m/s to 4.3 m/s, the average WS was 1.74 m/s, the average Ta was 35.2 °C, and the average RH was 47.9%. The measured results are in line with the typical summer daily climate characteristics of Guangzhou.

Trials were conducted in Ersha Island Sports Park (113.3° E, 23.1° N) in the Guangzhou urban area. Its activity area is approximately 0.56 hectares in size and is located in the southwest corner of the park (Figure 4). The landscape elements are mainly trees and children’s play equipment, with trees including Terminalia neotaliala, Cinnamomum camphora, Ficus religiosa, Pterocarpus indicus, Antidesma bunius, and Alstonia scholaris. Play equipment includes slides, swings, seesaws and suspension bridges. Four monitoring points were selected to measure Ta and RH, with monitoring point 1 (P1) under a single tree outside the study area; monitoring point 2 (P2) under multiple trees in the study area; monitoring point 3 (P3) in a full sun unshaded clearing in the study area; and monitoring point 4 (P4) under a single tree in the study area.

2.2. Establishment of the ENVI-met Model

ENVI-met is a three-dimensional microclimate model based on computational fluid dynamics and thermodynamics [44], which is widely used in the field of assessing OTCs [45,46]. Its vegetation system is three-dimensional, so different tree models can be built with a variety of crown shapes and leaf spatial distributions [47]. The original tree model in ENVI-met may not be suitable for this study due to latitude, longitude, and climate zone [48], so a three-dimensional tree model of the current state of the study site was created in the Albero component of the ENVI-met 5.0.3 software (

Table A1. Current 3D plant model.

Model and Related Properties of Current Trees in ENVI-met
	Name	Cinnamomum camphora		Name	Pterocarpus indicus
	Tree height	11.2 m		Tree height	12.2 m
	Canopy	9.7 m		Canopy	9.4 m
	Under branch height	2.1 m		Under branch height	2.6 m
	Foliage shortwave Albedo	16%		Foliage shortwave Albedo	18%
	LAI	3.5 m2/m2		LAI	2.7 m2/m2
	Name	Ficus religiosa		Name	Terminalia neotalial (Underage plant)
	Tree height	10.6 m		Tree height	9.7 m
	Canopy	8.3 m		Canopy	4.2 m
	Under branch height	2.4 m		Under branch height	4.1 m
	Foliage shortwave Albedo	15%		Foliage shortwave Albedo	13%
	LAI	1.2 m2/m2		LAI	1.4 m2/m2
	Name	Alstonia scholaris		Name	Antidesma bunius
	Tree height	9.2 m		Tree height	7.4 m
	Canopy	8.4 m		Canopy	7.9 m
	Under branch height	2.2 m		Under branch height	2.3 m
	Foliage shortwave Albedo	16%		Foliage shortwave Albedo	17%
	LAI	1.9 m2/m2		LAI	2.4 m2/m2

). Six medium-to-large trees commonly found in Guangzhou (Lagerstroemia speciosa, Mangifera indica, Terminalia neotaliala, Bombax ceiba, Ficus virens and Bauhinia blakeana) were also selected to build three−dimensional tree models of the experimental group (

Table A2. Experimental Group 3D model.

3D Model of Plants in ENVI-met Group and Its Related Properties
	Name	Lagerstroemia speciosa		Name	Mangifera indica
	Tree height	9.1 m		Tree height	9.7 m
	Canopy	7.6 m		Canopy	7.4 m
	Under branch height	2.1 m		Under branch height	2.7 m
	Foliage shortwave Albedo	19%		Foliage shortwave Albedo	27%
	LAI	4.4 m2/m2		LAI	2.5 m2/m2
	Name	Bauhinia blakeana		Name	Terminalia neotaliala
	Tree height	13.8 m		Tree height	13.3 m
	Canopy	9.3 m		Canopy	6.9 m
	Under branch height	2.8 m		Under branch height	4.2 m
	Foliage shortwave Albedo	31%		Foliage shortwave Albedo	13%
	LAI	6.1 m2/m2		LAI	6.4 m2/m2
	Name	Bombax ceiba		Name	Ficus virens
	Tree height	9.6 m		Tree height	8.1 m
	Canopy	8.4 m		Canopy	9.0 m
	Under branch height	3.9 m		Under branch height	2.5 m
	Foliage shortwave Albedo	17%		Foliage shortwave Albedo	23%
	LAI	2.2 m2/m2		LAI	3.1 m2/m2

).

A complete ENVI-met Albero model requires 9 parameters: tree height, crown width, under branch height, leaf area density, leaf reflectance, root depth, root width, root morphology and root area density. Liu et al. [49] showed that root depth, root width, root morphology and root area density had no significant effect on the simulation results of microclimate indicators in hot-humid areas, while leaf reflectance and leaf area index had a significant effect on the simulation results. Root depth, root width, root form and root area are therefore the software defaults, and all other parameters were measured by the authors (

Table A3. The first step is to study the morphological characteristics of trees.

Type	Project	Lagerstroemia speciosa	Mangifera indica	Bauhinia blakeana	Terminalia neotaliala	Bombax ceiba	Ficus virens
Shape of tree−crown	Tree height (m)	9.1	9.7	11.8	13.3	9.6	8.1
	Crown amplitude (m)	7.6	7.4	9.3	6.9	8.4	9.0
	Under branch height (m)	2.1	2.7	2.8	4.2	3.9	2.5
Blade properties	Foliage short albedo	0.19	0.27	0.31	0.13	0.17	0.23
	LAI (m2/m2)	4.4	2.5	6.1	6.4	2.2	3.1
	LAD (m2/m3)
	1 m	−	−	−	−	−	−
	2 m	−	−	−	−	−	−
	3 m	0.61	0.35	0.72	−	−	0.52
	4 m	0.65	0.35	0.73	−	0.32	0.57
	5 m	0.68	0.36	0.76	0.66	0.35	0.68
	6 m	0.76	0.39	0.83	0.69	0.41	0.96
	7 m	0.94	0.46	0.96	0.74	0.54	1.20
	8 m	0.78	0.60	1.20	0.83	0.87	0.69
	9 m	0.63	0.44	1.70	0.98	0.46	−
	10 m	−	0.34	1.10	1.25	−	−
	11 m	−	−	0.84	0.97	−	−
	12 m	−	−	−	0.74	−	−
	13 m	−	−	−	0.21	−	−

). Tree height, crown width and height under branches were measured using a rangefinder, leaf surface reflectance was measured using a spectrophotometer (Lambda 950 UV-Vis-NIR spectrophotometer, PerkinElmer, Waltham, MA, USA) and leaf area density (LAD) was calculated using the empirical algorithm proposed by Lalic et al. [50] using tree height and LAI (Figure 5), as shown in Equation (1).

$$LA{D}_X={{\displaystyle \int }}_0^X{L}_m{\left(\frac{h-Z_m}{h-Z}\right)}^{n}exp\left[n\left(1-\frac{h-Z_m}{h-Z}\right)\right]d$$

(1)

where 0$\le Z\le Z_m$, n = 6; $Z_m\le Z\le h$, n = 0.5

In the equation, LAI is the leaf area index (m²/m²), h is the tree height (m), Lm is the leaf area density obtained at height m (m), Z_m is the height (m) corresponding to the maximum leaf area density, and LAI was measured using a plant canopy analyser (Top-1300, China). To reduce the complexity of the model, the grass was set at a uniform 25 mm.

In this study, meteorological conditions on a typical summer day were used as background meteorological data inputted to ENVI-met, and the measured day meteorological data are shown in Figure 3. The specific ENVI-met model parameters are shown in

Table 2. ENVI-met model parameter settings.

Boundary Conditions of the Simulation Process by ENVI-met Model
Guangzhou	23°13′ N; 113°20′ E
Simulation date	On 13 July 2022
Simulation time	From 8:00:00 to 19:00:00
Model dimensions	X−Grids:86 Y-Grids:83 Z-Grids:35
Grid cell	dx = 1 dy = 1 dz = 1
Wind speed (10 m)	0–1.7 m/s
Wind direction (N:0, S:180)	135°
Cloud cover	0
Air temperature	26.1–37.9 °C
Relative humidity	38.76–74.6%
UTCI index calculation	Bio-met process
Results visualization	Leonardo visualization tool

.

2.3. ENVI-met Validation

Although the accuracy of the ENVI-met software has been proven many times [51,52,53], the simulation results may still be inaccurate due to factors such as the object of the simulation, the geographical environment, and the size of the site. In this study, root mean squared error (RMSE), mean absolute error (MAE) and R-squared (R²) were used as evaluation indicators of the numerical accuracy of the simulations. The calculation formula is shown in Equations (2) and (3).

$$RMSE=\sqrt{\frac{1}{n}{{\displaystyle \sum }}_{i=1}^n{\left(y_i^{*}-y_i\right)}^2}$$

(2)

$$MAE=\frac{1}{n}{{\displaystyle \sum }}_{i=1}^n\left|y_i^{*}-y_i\right|$$

(3)

where “$y_i^{*}$” is the simulated value, “$y_i$” is the measured value and “n” is the number of measurements.

2.4. The OTC Evaluation Index

UTCI is defined as the equivalent ambient temperature (°C) of the reference environment and is a common outdoor thermal comfort evaluation indicator [54]. The UTCI is performed by combining physiological models of respondents, clothing models and meteorological factors (Ta, RH, WS and mean radiation temperature (Tmrt)) to represent the level of thermal effects of the environment [55]. It is appropriate for assessments in all climatic zones and seasons [56]. This study outputs the ENVI-met coupled thermal comfort evaluation index UTCI as a reference for the OTC. The amount of thermal sensation in the UTCI is shown in

Table 3. UTCI thermal stress classification (adapted from Peter Bröde [57]).

Thermal Sensitivity	UTCI (°C):	Thermal Sensitivity	UTCI (°C):
Neutral	26	Slight cold stress	0–9
Moderate heat stress	26–32	Moderate cold stress	−13–0
Strong heat stress	32–38	Strong cold stress	−27–(−13)
Very strong heat stress	38–46	Very strong cold stress	−27–(−40)
Extreme heat stress	>46	Extreme cold stress	<−40

.

2.5. Case Studies Description

This study was divided into two stages (

Table 4. Case studies description.

Site Situation		Describe
Serial number	Image	“A–B–C”: where “A” denotes tree species, “B” denotes site boundary conditions and “C” denotes the number of trees on the site.
Sp−E−7		“Sp” denotes existing trees on the site, “E” denotes the presence of arborvitae at the site boundary. In this scene, the accuracy of the ENVI-met model is verified.
The first stage
Serial number	Image	Serial number	Image	Serial number	Image
Ls−E−7		Bb−E−7		Fv−E−7
Mi−E−7		Bc−E−7		Tn−E−7
“Ls” denotes Lagerstroemia speciosa, “Mi” denotes Mangifera indica, “Bb” denotes Bauhinia blakeana, “Tn” denotes Terminalia neotaliala, “Bc” denotes Bombax ceiba, and “Fv” denotes Ficus virens.
The second stage
Serial number	Image	The second stage of the study unfolded based on the results of R2 between Bb−D−7 and Bb−E−7, and Mi−D−7 and Mi−E−7. The specific planting forms for the second stage are shown below. Where “D” denotes no trees on the site boundary, L1 indicates a tree with an LAI of 1; L1−6 indicates trees with LAI of 1 to 6, respectively; L1−6−D−1 indicates scenes with LAI of 1 to 6 and planting quantities 1; L1−6−D−(1−9) indicates scenes with LAI of 1 to 6 and planting quantities 1 to 9.
Bb−D−7
Mi−D−7
Serial number	Image	Serial number	Image	Serial number	Image
L1−6−D−1		L1−6−D−2		L1−6−D−3
L1−6−D−4		L1−6−D−5		L1−6−D−6
L1−6−D−7		L1−6−D−8		L1−6−D−9

). Stage 1: six different scenes were created, using common Guangzhou trees to replace the current trees on the site to compare the ability of different trees to regulate the OTC. The first stage recorded UTCI values at 1.0 m pedestrian height at four monitoring sites. To consider more uncertain boundary conditions and to further explore the ability of different tree LAI and different planting arrangements to regulate OTC, we conducted a second stage of research. Stage 2: select the scenes Bb−E−7 and Mi−E−7 and remove the trees around the field to obtain the new scenes Bb−D−7 and Mi−D−7. Analysis of the correlation coefficients showed that the removal of trees around the site did not change the trend of UTCI within the activity area. On this basis, 54 scenes of LAI combined with planting arrangements were created based on the physical characteristics of typical trees in Guangzhou [58] (10 m tall, 8 m crown width and 2 m height under branches) with a foliar shortwave albedo of 0.18 (the software default). The microclimate parameters at 1.0 m in the red area, including Ta, Tmrt, RH, WS and UTCI, were recorded, and the average value was taken. The unshaded scene was used as the control group to compare the impact of tree LAI and planting arrangement on OTC in the study area.

3. Results

3.1. ENVI-met Accuracy Analysis

In a large number of existing studies, the reliability of the ENVI-met model output is usually verified by comparing it with the measured air temperature and relative humidity [45,59]. As shown in

Table 5. P1–P4 simulation values RMSE and MAE analysis table.

Meteorological Elements	Indicators	P1	P2	P3	P4
Air temperature	RMSE/°C	0.69 °C	1.22 °C	0.61 °C	0.58 °C
	MAE/°C	0.56 °C	1.07 °C	0.57 °C	0.49 °C
	R2	0.94	0.85	0.98	0.90
Relative humidity	RMSE/%	2.16%	1.29%	2.54%	1.74%
	MAE/%	1.79%	1.03%	2.09%	1.49%
	R2	0.93	0.96	0.92	0.95

, we calculated and analysed the errors in the simulated values relative to the measured values. The measured and simulated values of Ta are within 1.9 °C, and the measured and simulated values of RH are within 5.3%. The RMSE and MAE for each monitoring point Ta ranged from 0.58 to 1.22 °C and 0.49 to 1.07 °C, respectively, and the RMSE and MAE values for RH ranged from 1.29 to 2.53 % and 1.03 to 2.07 %, respectively. The R² of Ta at each monitoring point was between 0.85 and 0.94. The R² of the RH at each monitoring point was between 0.92 and 0.96. Figure 6 shows the simulated and measured fitting results for Ta and RH at each monitoring site. The results show that the developed ENVI-met model is reliable and can be used in this study.

3.2. Effects of Different Tree Species on OTC

The UTCI for each scene over time is shown in Figure 7. The UTCI varies between 33.1 and 46.2 °C. It peaks at 15:00 and then gradually decreases. At 15:00, the UTCI varied between 43.9 and 44.1 °C for P1, between 42.1 and 43.1 °C for P2, between 46.1 and 46.2 °C for P3, and between 42.1 and 46.0 °C for P4. Replacing the original tree species on the site has little effect on the UTCI of P1 and P3 but has a significant effect on P2 and P4. At P2, the capacity gradient of the tree-optimized OTC was Bb−E−7 > Ls−E−7 > Tn−E−7 > Fv−E−7 > Bc−E−7 > Mi−E−7, whereas at P4 it was Bb−E−7 > Fv−E−7 > Ls−E−7 > Bc−E−7 > Mi−E−7 > Tn−E−7.

For trees with smaller canopies and lower LAI, the difference in UTCI between P2 and P4 was significant, with Mi−E−7 having a temperature difference between P2 and P4 of 2.59 °C at 10:00 and 0.78 °C at 15:00. Bc−E−7 had a temperature difference between P2 and P4 of 2.79 °C at 10:00 and 3.03 °C at 15:00. In contrast, trees with large canopies and high LAI had a smaller temperature difference between P2 and P4 during the day, with a maximum temperature difference of only 0.46 °C at 10:00 and 0.22 °C at 15:00. Overall, trees with high LAI are more capable of optimizing OTC than trees with low LAI, and planting multiple trees is more capable of optimizing OTC than single trees. Increasing the quantities of plantings significantly enhances the ability of low LAI trees to optimize OTC, but not significantly for high LAI trees. This is because trees with low LAI, which have a low LAD, have a significant increase in LAD within the study area due to the increased number of plantings resulting in the branches of the trees crossing, but trees with high LAI, which have a high LAD, and the branches of other trees are unable to significantly increase the LAD within the study area. Therefore, we can determine that planting multiple trees with low LAI and small canopies can achieve the cooling effect of a single tree with high LAI and a large canopy.

3.3. Effect of Number Planting and LAI on OTC

To further investigate the effects of tree planting arrangements and LAI on the OTCs of the CAS, two scenes, Bb−E−7 and Mi−E−7, were chosen to remove the trees around the site to obtain two new scenes, Bb−D−7 and Mi−D−7. Correlation coefficients were analysed for the UTCI of P2 and P4 for periods for the four scenes. Analysis of the correlation coefficients between the UTCI of P2 and P4 was carried out for the four scenes. The results are shown in Figure 8. The correlation between Bb−D−7 and Bb−E−7 was extremely high, as was the correlation between Mi−D−7 and Mi−E−7. The removal of the trees around the site will not change the results of the Stage 1 study. The second phase of the study will be conducted based on this result. The two stages of the study were based on two time periods, 10:00 and 15:00, for analysis.

3.3.1. Air Temperature (Ta)

Figure 9 shows the effect of tree planting arrangements and LAI on Ta when multiple trees are planted. The Ta trends were essentially the same for all scenes. Ta was lower in all scenes than in the control group without trees. The two peak periods for children’s outdoor activities are 10:00 and 15:00. The mean Ta of each scene decreased as the quantity of planting increased. The mean Ta of L₁₋₆−D−9 at 10:00 was 32.74 °C, while the mean Ta of L₁₋₆−D−9 was 32.21 °C. The mean Ta of L₁₋₆−D−9 at 15:00 was 38.6 °C, while the mean Ta of L₁₋₆−D−9 was 38.07 °C. The mean Ta of each scene decreases as LAI increases. At 10:00, the mean Ta of L₁−D−(1−9) is 32.70 °C, and the mean Ta of L₆−D−(1−9) is 32.38 °C. At 15:00, the mean Ta of L₁−D−(1−9) was 38.61 °C, and L₆−D−(1−9) had a mean Ta of 38.32 °C.

3.3.2. Mean Radiant Temperature (Tmrt)

Figure 10 shows the effect of tree planting arrangements and LAI on Tmrt when multiple trees are planted. The trend in Tmrt was generally consistent across scenes, and Tmrt was significantly lower in all scenes in the study area than in the control group. Tmrt differences within the study area varied with time, with small differences between 11:00–14:00 and 18:00–19:00 but significant differences between 8:00–11:00 and 14:00–17:00. At 8:00, there was a slight decrease in Tmrt for L₁₋₆−D−3, but L₁₋₆−D−4 began to fall sharply by 11.3, 13.9, 15.3, 16.2, 16.3 and 16.4 °C. At 15:00, there was a slight decrease in Tmrt for L₁₋₆−D−5, but L₁₋₆−D−6 began to drop significantly, by 7.9, 9.7, 10.5, 11.1, 11.5 and 11.7 °C, respectively. Overall, trees with high LAI reduced Tmrt more, with four trees planted at 10:00 significantly reducing Tmrt and six trees planted at 15:00 significantly reducing Tmrt (Figure 11).

3.3.3. Relative Humidity (RH)

Figure 12 shows the effect of tree planting arrangements and LAI on RH when multiple trees are planted. The highest RH occurred at 8:00, and the lowest RH occurred at 16:00. RH was significantly and negatively correlated with Ta and Tmrt. The RH of the scenes with more planting was slightly higher than that of the scenes with less planting. At 10:00, the mean RH of L₁₋₆−D−1 was 58.59%, and the mean RH of L₁₋₆−D−9 was 60.13%. At 15:00, the mean RH of L₁₋₆−D−1 was 34.77%, and the mean RH of L1−6−D−9 was 36.57%. There was slightly higher RH for scenes with high LAI compared to scenes with lower LAI. At 10:00, the mean RH of L₁−D−(1−9) was 59.17%, and that of L₆−D−(1−9) was 59.52%. At 15:00, the mean RH of L₁−D−(1−9) was 35.44%, and that of L₆−D−(1−9) was 36.01%.

3.3.4. Wind Speed (WS)

In this study, a simple forcing simulation with a 135° wind direction was used. The results are shown in Figure 13. The trend of WS variation was consistent for different scenes, but the effect was not significant. Wind speed is higher in all scenes than in the control group without trees. WS increases with increasing LAI, and the WS was greatest for L₆−D−(1−9), with 1.02 m/s at 10:00 and 1.12 m/s at 15:00. The arrangement of the trees also slightly affects the WS, and the WS of L₁₋₆−D−8 was the largest. It was 1.03 m/s at 10:00 and 1.13 m/s at 15:00.

3.3.5. Universal Thermal Climate Index (UTCI)

Figure 14 shows the effect of tree planting arrangements and LAI on UTCI when multiple trees are planted. The trend in UTCI is generally consistent across scenes, with the highest UTCI occurring at 15:00 h in all scenes, but the lowest UTCI occurring at different times.

Situations where the lowest UTCI occurs at 8:00 were L₃₋₆−D−1, L₂₋₆−D−2 scenes, and L₁₋₆−D−(3−6) scenes. Scenes where the lowest UTCI occurs at 19:00 were L₁−D−2 and L₁₋₂−D−1 scenes. At 19:00, UTCI increased continuously with the number of plantings, with L₁₋₆−D−9 having 0.42, 0.47, 0.42, 0.34, 0.27 and 0.22 °C higher UTCI compared to L₁₋₆−D−1, respectively. This suggests that the high planting density of trees can optimize the daytime OTC in the study area. Dense trees worsen the OTC of the study area at night. This is because the study area with high LAD absorbs more solar radiation during the day and then heats up and radiates outwards at night [60].

A significant correlation was also found between the quantity of plantings and UTCI (

Table A4. Correlation between the quantity planted and UTCI at 8:00 and 15:00.

8:00				15:00
Scene	Scenes	Mean Difference (°C)	Significance	Scene	Scenes	Mean Difference (°C)	Significance
L1−6−D−1	L1−6−D−2	−0.01161	p = 0.974	L1−6−D−1	L1−6−D−2	−0.04742	p = 0.862
	L1−6−D−3	0.23735	p = 0.501		L1−6−D−3	−0.04702	p = 0.863
	L1−6−D−4	1.43709 **	p = 0.000		L1−6−D−4	−0.03371	p = 0.901
	L1−6−D−5	1.41117 **	p = 0.000		L1−6−D−5	0.26466	p = 0.333
	L1−6−D−6	1.38812 **	p = 0.000		L1−6−D−6	1.52980 **	p = 0.000
	L1−6−D−7	1.47803 **	p = 0.000		L1−6−D−7	1.69460 **	p = 0.000
	L1−6−D−8	1.49673 **	p = 0.000		L1−6−D−8	1.72052 **	p = 0.000
	L1−6−D−9	1.56238 **	p = 0.000		L1−6−D−9	1.74250 **	p = 0.000
L1−6−D−2	L1−6−D−3	0.24896	p = 0.480	L1−6−D−2	L1−6−D−3	0.00040	p = 0.999
	L1−6−D−4	1.44871 **	p = 0.000		L1−6−D−4	0.01371	p = 0.960
	L1−6−D−5	1.42278 **	p = 0.000		L1−6−D−5	0.31208	p = 0.255
	L1−6−D−6	1.39973 **	p = 0.000		L1−6−D−6	1.57722 **	p = 0.000
	L1−6−D−7	1.48965 **	p = 0.000		L1−6−D−7	1.74202 **	p = 0.000
	L1−6−D−8	1.50834 **	p = 0.000		L1−6−D−8	1.76794 **	p = 0.000
	L1−6−D−9	1.57399 **	p = 0.000		L1−6−D−9	1.78993 **	p = 0.000
L1−6−D−3	L1−6−D−4	1.19975 **	p = 0.001	L1−6−D−3	L1−6−D−4	0.01331	p = 0.961
	L1−6−D−5	1.17382 **	p = 0.002		L1−6−D−5	0.31167	p = 0.256
	L1−6−D−6	1.15077 **	p = 0.002		L1−6−D−6	1.57682 **	p = 0.000
	L1−6−D−7	1.24069 **	p = 0.001		L1−6−D−7	1.74162 **	p = 0.000
	L1−6−D−8	1.25938 **	p = 0.001		L1−6−D−8	1.76754 **	p = 0.000
	L1−6−D−9	1.32503 **	p = 0.000		L1−6−D−9	1.78952 **	p = 0.000
L1−6−D−4	L1−6−D−5	−0.02592	p = 0.941	L1−6−D−4	L1−6−D−5	0.29837	p = 0.276
	L1−6−D−6	−0.4897	p = 0.889		L1−6−D−6	1.56352 **	p = 0.000
	L1−6−D−7	0.04094	p = 0.907		L1−6−D−7	1.72831 **	p = 0.000
	L1−6−D−8	0.05964	p = 0.865		L1−6−D−8	1.75423 **	p = 0.000
	L1−6−D−9	0.12529	p = 0.722		L1−6−D−9	1.77622 **	p = 0.000
L1−6−D−5	L1−6−D−6	−0.02305	p = 0.876	L1−6−D−5	L1−6−D−6	1.26515 **	p = 0.000
	L1−6−D−7	0.06686	p = 0.849		L1−6−D−7	1.42994 **	p = 0.000
	L1−6−D−8	0.08556	p = 0.808		L1−6−D−8	1.45587 **	p = 0.000
	L1−6−D−9	0.15121	p = 0.668		L1−6−D−9	1.47785 **	p = 0.000
L1−6−D−6	L1−6−D−7	0.08991	p = 0.798	L1−6−D−6	L1−6−D−7	0.16479	p = 0.546
	L1−6−D−8	0.10861	p = 0.758		L1−6−D−8	0.19072	p = 0.485
	L1−6−D−9	0.17426	p = 0.621		L1−6−D−9	0.213	p = 0.436
L1−6−D−7	L1−6−D−8	0.01870	p = 0.958	L1−6−D−7	L1−6−D−8	0.02592	p = 0.924
	L1−6−D−9	0.08435	p = 0.811		L1−6−D−9	0.04791	p = 0.860
L1−6−D−8	L1−6−D−9	0.06565	p = 0.852	L1−6−D−8	L1−6−D−9	0.02198	p = 0.936

“**” p < 0.01.

), with L₁₋₆−D−4 significantly improving OTC in the morning. In the afternoon L1−6−D−6 significantly improved OTC. Of these, at 8:00, the mean difference between L₁₋₆−D−1 and L₁₋₆−D−4 was 1.44 °C. However, the mean difference was only 0.24 °C compared to L₁₋₆−D−3. At 15:00, the average difference between L₁₋₆−D−1 and L₁₋₆−D−6 was 1.5 °C. However, it was only 0.26 °C compared to L₁₋₆−D−5. Compared to the control, the UTCI of L₁₋₆−D−6 was reduced by 5.3, 6.2, 6.7, 7.2, 7.4 and 7.6%. However, the UTCIs of L₁₋₆−D−7 to L₁₋₆−D−7 were highly overlapping, and their average difference was only 0.16 °C. Compared with L₁₋₆−D−6, L₁₋₆−D−7 is reduced by only 0.12, 0.14, 0.16, 0.18, 0.19, and 0.2 °C. The UTCI is reduced by only 0.02%, 0.03%, 0.04%, 0.04%, and 0.05%, while the maximum optimization of UTC for L₁₋₆−D−9 compared with L₁₋₆−D−7 is only 0.01%.

LAI was also significantly correlated with UTCI (

Table A5. Correlation between the LAI and UTCI at 8:00 and 15:00.

8:00				15:00
Scene	Scenes	Mean Difference (°C)	Significance	Scene	Scenes	Mean Difference (°C)	Significance
L1−D−(1−9)	L2−D−(1−9)	0.56419	p = 0.095	L1−D−(1−9)	L2−D−(1−9)	0.43027	p = 0.303
	L3−D−(1−9)	0.96461 **	p = 0.005		L3−D−(1−9)	0.70856	p = 0.093
	L4−D−(1−9)	1.24602 **	p = 0.000		L4−D−(1−9)	0.93895 *	p = 0.028
	L5−D−(1−9)	1.46004 **	p = 0.000		L5−D−(1−9)	1.13070 **	p = 0.009
	L6−D−(1−9)	1.59267 **	p = 0.000		L6−D−(1−9)	1.23156 **	p = 0.004
L2−D−(1−9)	L3−D−(1−9)	0.40042	p = 0.233	L2−D−(1−9)	L3−D−(1−9)	0.27829	p = 0.504
	L4−D−(1−9)	0.68183 *	p = 0.045		L4−D−(1−9)	0.50868	p = 0.224
	L5−D−(1−9)	0.89585 **	p = 0.009		L5−D−(1−9)	0.70043	p = 0.096
	L6−D−(1−9)	1.02848 **	p = 0.003		L6−D−(1−9)	0.80129	p = 0.058
L3−D−(1−9)	L4−D−(1−9)	0.28140	p = 0.400	L3−D−(1−9)	L4−D−(1−9)	0.23039	p = 0.580
	L5−D−(1−9)	0.49543	p = 0.141		L5−D−(1−9)	0.42214	p = 0.312
	L6−D−(1−9)	0.62806	p = 0.064		L6−D−(1−9)	0.52300	p = 0.212
L4−D−(1−9)	L5−D−(1−9)	0.21403	p = 0.521	L4−D−(1−9)	L5−D−(1−9)	0.19175	p = 0.645
	L6−D−(1−9)	0.34665	p = 0.300		L6−D−(1−9)	0.29261	p = 0.482
L5−D−(1−9)	L6−D−(1−9)	0.13263	p = 0.691	L5−D−(1−9)	L6−D−(1−9)	0.10086	p = 0.808

“*” 0.01 < p < 0.05, “**” p < 0.01.

). The study showed that trees with high LAI reduced UTCI more. The difference between trees with high LAI was significant compared to trees with low LAI, but not significant compared to trees with slightly lower LAI. At 8:00, the mean difference between L₁−D−(1−9) and L₆−D−(1−9) was 1.59 °C. However, at 15:00, the mean difference between L₅−D−(1−9) and L₆−D−(1−9) was only 0.13 °C, and the mean difference between L₁−D−(1−9) and L₆−D−(1−9) was 1.23 °C, but the mean difference between L₅−D−(1−9) and L₆−D−(1−9) was only 0.10 °C. The specific 10:00 and 15:00 ENVI-met simulation results are shown in Figure 15.

Figure 16 shows the thermal stresses for each scene. At 8:00, the L₁−₆−(1−3) scene had a UTCI of 32.1–35.0 °C in the study area, and they were in a “strong heat stress” state. The L₁₋₆−(1−3) scene had a UTCI of 30.1–31.9 °C in the study area, and they were in a state of “moderate heat stress”. At 9:00, the “strong heat stress” state was reached in all scenes with a UTCI of 32.6−34.5 °C. At 10:00, the lower LAI trees and less planted scenes (L₁₋₂−D−(1−3) and L₁−D−(4−9)) were in a “very strong heat stress” state with a UTCI of 38.4–39.7 °C, and the rest of the scenes showed a significant decrease in UTCI, with the thermal comfort advantage of high LAI trees beginning to show, and with increasing LAI, the UTCI decreased in all scenes, with a UTCI of 36.7–39.3 °C. Between 11:00 and 14:00, it was under a “very strong heat stress” state due to enhanced solar radiation, with a UTCI of 39.1–45.8 °C for all scenes, further reducing the OTC of the study area. At 15:00–16:00, scenes with low LAI and low quantities (L₁₋₂−D−(1−4)) were in extreme heat stress, with a UTCI of 46.1–46.4 °C. Planting high LAI and high numbers of trees in hot-humid areas can be effective in avoiding the emergence of extreme heat stress conditions in summer. The UTCI dropped slightly at 17:00 with a UTCI of 39.7–43.7 °C. It was still all in a state of “very strong thermal stress”. At 18:00, the UTCI was 37.2–38.4 °C, and some scenes (L₁₋₆−D−(1−4) and L₁₋₂−D−5) were still under “very strong heat”. At 19:00, all scenes drop to a “strong heat stress” state with a UTCI of 34.3–34.8 °C.

4. Discussion

For large scale spaces such as cities or streets, increasing the number of trees planted will further improve the OTC [61,62]. However, this strategy has its limitations and may not be applicable to small-scale outdoor CAS in hot-humid areas. Our research shows that haphazard tree planting leads to a wasted cooling effect of trees. Excessive and irrational planting cannot produce an effective improvement in the thermal environment of the study area.

The simulation results for Ta and Tmrt showed that Ta and Tmrt were lower than the control group in all scenes, with the difference in Tmrt being more pronounced. Tmrt is the sum of all short-wave and long-wave radiation fluxes received by the body, and Tmrt decreases with increasing LAI and the number of implants. However, the efficiency of its improvement is limited, and the arrangement of trees on the east and west sides of the study area effectively blocks the morning and afternoon solar radiation, making its Tmrt significantly lower than other scenes. Additionally, it is worth mentioning that, since the radiation in the study area mainly originates from the sun in the morning, while in the afternoon the surrounding environment absorbs a large amount of heat and radiates it outward, this results in better Tmrt reduction by trees in the morning than in the afternoon. Overall, the timing of solar movement and the spatial arrangement of plants together influenced the UTCI of the study area, so placing trees in the south–north direction of the study area did not further improve the OTC of the study area. Proper planting arrangements can maximize solar radiation blockage and help further improve the OTC of CAS.

The results of the RH simulations showed that the number of plantings and LAI did not have a significant effect on the RH of the study area. This is because the RH in hot-humid areas in summer is mainly influenced by the humid tropical marine air and partly by the evaporation of water from its substrate [63], making it difficult to effectively regulate the RH of urban outdoor CAS with fewer trees planted in hot-humid areas. However, in tropical arid regions, trees can significantly increase the RH [64]. Thus, the role of trees in regulating local RH depends mainly on their geographical location and climatic conditions.

Large tree canopies, high LAI and dense stands result in lower wind speeds [65]. However, this study found that the WS under the high LAI canopy was slightly greater than the WS under the low LAI canopy, which may be because the WS and turbulence intensity under the canopy are influenced by the aerodynamic properties of the trees [66], and as the airflow approaches the trees, wind fields of different natures are formed above and below the trees due to the canopy [67]. In this area, a turbulent mixing zone is formed upwind and downwind, which merges to form an equilibrium zone when the airflow blocked by the trees returns to close to the original airflow velocity, and a zone of increased velocity is formed on the side of the trees [68], resulting in a slight increase in WS. The higher density of leaf area and smaller interleaf pores of high LAI plants reduces wind diversion to some extent, so WS will be slightly greater under high LAI trees.

Furthermore, comparisons between different tree species reveal the importance of selecting appropriate tree species in outdoor CAS. There is no doubt that trees with high LAI are more beneficial for OTC [69,70]; we found that planting multiple trees with low LAI can achieve the same cooling effect as planting high LAI trees. This means that the choice of tree species becomes flexible when using trees to improve OTC. If there are some low LAI trees with summer flowering and fruiting, we can choose to plant them in multiples to obtain good landscape benefits and OTC. Additionally, to maximize the positive effect of trees in improving OTC, we believe that trees with low LAI should be planted in multiples to allow canopy overlap, while trees with high LAI should avoid canopy overlap.

In addition, this is not only a case study to investigate the thermal performance of different trees and planting arrangements, but also an evidence-based landscape architecture (EBLA) application to show the general method of design decision-making [71]. Based on the microclimate characteristics of the different scenes, the landscape architect chose the most appropriate one in the arboriculture design. With the continuous replenishment of planting scenes in further research, comprehensive tree planting and selection manuals can be created.

There are still some limitations in this research; for children, their perception system of the environment is not fully developed and their metabolic level is different from that of adults, and this study did not conduct a children’s thermal sensory poll to revise the evaluation index of thermal comfort for children, we only judged its effect of optimizing OTC by the cooling trend and magnitude of trees in different scenes. Additionally, we did not consider the influence of paving and children’s play equipment materials in the site. In future studies, the thermal comfort evaluation index will be revised for children in hot and humid areas and more different complex situations will be simulated.

5. Conclusions

This study presents a detailed analysis of a real children’s activity area in Guangzhou, China, using the validated ENVI-met model simulation to analyse the regional OTC. Based on the ability to optimize the OTC of six typical trees in Guangzhou, further exploring the effect of nine planting arrangements in combination with six LAIs on OTCs, the UTCI was used as an evaluation indicator for OTCs, and the following conclusions were drawn.

(1) The UTCI of the open unshaded space of the outdoor CAS is hardly affected by the surrounding trees. (2) Placing trees in the east and west had a significant effect on the OTC of the study area, while placing trees in other directions had no significant effect on the OTC of the study area. (3) Trees with high LAI have stronger OTC optimization ability than trees with low LAI. Trees with low LAI can achieve the OTC optimization effect of trees with high LAI by appropriately increasing the planting number.

Trees are undoubtedly the most environmentally friendly landscape element for landscape architects, and several planting recommendations have emerged from this study. (1) Outdoor CAS should be shaded by trees. (2) It is possible to create a good OTC through fewer high LAI trees, reducing the heat threat to children when they are outdoors. Smaller play equipment in the outdoor activities of children (seesaws, swings, etc.) can be placed under a large canopy of high LAI trees, while larger play equipment (slides, trampolines, suspension bridges, etc.) should have trees planted on their east and west sides to increase coverage. (3) The selection of trees is flexible, and low LAI trees with good landscape benefits can be planted appropriately.