How Compositions of Landscape Elements Affect Outdoor Thermal Environments: Quantitative Study Along the Urban Riverside

Abstract

Riverside landscape belts are crucial for mitigating urban heat islands and enhancing urban esthetics. This study investigates the thermal environment effects of X21 landscape configurations in riverside belts using field measurements and numerical simulations. The physiologically equivalent temperature (PET) assesses human thermal comfort variations. Main findings demonstrate that the “enclosed tree–shrub–grass” configuration is the most effective plant arrangement for enhancing the riverside landscape belts thermal environment, with grassland identified as the optimal underlying surface configuration. Moreover, PET reveals that “enclosed tree–shrub–grass” spaces provide greater comfort during morning and midday periods, while “shrub–grass” areas are more suitable for the evening. This research provides a theoretical framework and empirical data for urban riverside landscape planning and design, significantly advancing urban thermal environment optimization and esthetic quality.

1. Introduction

The urban heat island (UHI) effect, intensified by population growth and urban expansion, threatens ecological and human health [1,2,3]. This phenomenon leads to public health issues, higher energy consumption, and increased air pollution risks [4,5,6]. However, riverfront landscape belts, combining water and greenery, offer recreational spaces and improve the urban thermal environment.

Recent research highlights the role of riverfront landscapes in regulating air temperature and enhancing thermal comfort. Scholars have explored how plant configurations affect the thermal environment [7,8,9,10]. For example, Srivani et al. found that combining trees and meadows creates a more significant cooling effect [11]. Similarly, GAL revealed that trees most effectively improve thermal conditions [12]. Joseph L. Moss demonstrated that while individual trees have minimal cooling variation, their collective impact is substantial [13].

Water bodies also play a crucial role in mitigating UHI due to their evaporative cooling effect [14,15]. Studies show that rivers significantly reduce surrounding air temperatures, especially when integrated with green spaces. Dong, J’s analysis of the summer thermal environment along 20 rivers in Washington revealed that rivers significantly reduce surrounding air temperatures, with the cooling effect being more pronounced when riverfront green spaces are integrated with the water bodies [16]. Hathway concluded that green spaces influence riverine air temperatures more than the rivers themselves [17]. Coates found that shoreline vegetation affects water body temperatures, with wind, solar radiation, and plant types also impacting water velocity [18]. Additionally, Yang and Jun investigated the impact of different water bodies within urban squares on the surrounding thermal environment and human thermal comfort through questionnaire surveys and field observations [19]. Their findings suggest that combining water bodies with green spaces in riverfront spatial landscapes enhances outdoor thermal environments, with human thermal comfort being optimal when sky visibility in green spaces is reduced.

Field measurements and numerical simulations are the primary research methods employed, which have been confirmed by references [1,20,21,22]. However, field measurements lack temporal continuity and are limited in capturing large-scale spatial and temporal variations [23,24,25]. Additionally, numerical simulation methods have been relatively underutilized in studying the thermal environment of water-green space combinations [26,27]. The literature on enhancing the thermal environment through planning and design offers valuable insights for urban planning [28,29,30]. While extensive studies have examined the thermal environment of riverfront landscape zones, the majority focus on macro-level analyses. Few studies have explored the influence of landscape configurations in riverfront zones with quantitative methods on the local geothermal environment, particularly in relation to varying river widths.

This study investigates how different landscape configurations within riparian greenways of varying river widths affect the local thermal environment and identifies the optimal configuration for riverside park land use. The study first validates the effectiveness of thermal environment simulation tools using measured and simulated data. Then, it establishes 21 typical landscape configuration schemes and evaluates the optimal configuration using physiologically equivalent temperature (PET). Finally, it proposes guidelines for riparian greenway planning and design. This research is significant for mitigating urban heat islands and provides data support for riparian landscape planning and design.

2. Methodology

2.1. Study Area

This study focuses on Xi’an (34°10′–34°27′ N, 108°59′–109°16′ E), a city recognized as the cultural, educational, and economic hub of northwest China. Renowned for its abundant river resources, Xi’an features diverse and picturesque riverside landscapes, making it an ideal subject for this research [31,32,33].

Xi’an, a representative city in northern China, is renowned for its “eight rivers around Chang’an” and abundant river resources, including the Wei River, Ba River, and Feng River [34]. In recent years, with the implementation of the Chinese government’s “Building Greater Xi’an” policy, the city’s urban scale and population have continued to expand, making the riverfront landscape belt a focal point of development. Given the critical role of riverfront landscape zones in mitigating the urban heat island effect and shaping urban character, it is essential to study the thermal environment of Xi’an’s riverfront landscape zones.

The Ba River was chosen as the study area due to its significance as one of Xi’an’s primary rivers, spanning two major urban districts—Baqiao District and the International Trade & Logistics Park (see Figure 1). The riverside landscape along the Ba River is characterized by its diversity and vibrancy, making it a prominent recreational and exercise hub for local residents. As one of the more comprehensively developed urban riverside landscape belts, the Ba River serves as an exemplary case study, offering valuable insights and guidance for the development of similar riverside landscapes.

2.2. Field Measurement

To comprehensively assess the influence of the thermal environment within the riverside landscape belt, field measurements were conducted on a representative summer day. Specifically, 7 July 2021, characterized by clear skies and high temperatures, was chosen as the measurement day. During the study period, the recorded air temperatures ranged from a minimum of 27.1 °C to a maximum of 37.5 °C.

The distribution of measurement locations and the measurement process are depicted in Figure 2. The study comprises three measurement sites: SITE1, located along the riverside landscape belt of a 350 m river segment; SITE2, situated along the riverside landscape belt of a 40 m river segment; and SITE3, positioned along the riverside landscape belt of an 8 m river segment. The instruments utilized for the measurements are detailed in

Table 1. Test instruments and performance parameters.

Name	Photograph	Type	Measurement Element	Resolution	Range	Precision
Multifunction tester		JT2020	Air temperature	0.1 °C	−2–120 °C	±0.2 °C
			Air humidity	0.1%	0–100%	±2%
			Relative humidity	0.1%	0–100%	±2%
			Black ball temperature	0.1 °C	0–50 °C	±0.2 °C

.

2.3. Numerical Simulation

2.3.1. Model Setup

The ENVI-met model (v4.4.5 summer20) was selected in this paper, which is a widely validated software package designed to simulate outdoor thermal environment [21,35,36]. It is an integrated three-dimensional non-hydrostatic model for simulating air-plant interactions on the ground. In addition, ENVI-met is designed for typical microscale analyses, with a horizontal resolution of 0.5–10 m, a typical analysis time of 24–48 h, and a time step of 1–5 s.

To ensure the reliability and accuracy of the ENVI-met simulations, the results were validated through field measurements. To minimize external disturbances, a region with limited recreational activity was selected for the field measurements. The initialization parameters for the simulation are detailed in

Table 2. Model initialization parameter setting.

Type	Parameter	Settings
Location	City	Xi’an
	Coordinate	34°32′ N 108°55′ E
Boundary conditions	Turbulence model	Standard TKE model [1]
	Total simulation time	24 h
Simulation time	Output interval	60 min
	Simulation start time	5:00

.

Ten layers with varying leaf area densities (LAD) were used as input parameters to characterize different plant morphologies in the ENVI-met model. LAD is defined as the total leaf area per unit volume (m²/m³). The models were divided into 10 equal vertical sections, each with a specific LAD value. Foliage shortwave albedo and transmittance values were sourced from the ENVI-met database. While representing vegetation with the dominant species may not fully reflect real-world diversity, this setup provides the best approximation of actual conditions given the available data.

The primary underlying surface types consisted of grassland, impervious pavements, trees, and water features. In addition, the ENVI-met database contains parameters for both building materials and underlying surfaces related to all structural elements.

2.3.2. Ideal Scenarios

To comprehensively analyze the variations in cooling effects across different landscape configurations, 21 idealized models were developed within riverside landscape belts at varying river levels. These models encompass 12 distinct plant configuration scenarios and nine types of underlying surfaces, as detailed in

Table 3. Summary of model setup and different types of landscape configuration information.

Model Parameter Setting
River width	350 m	40 m	8 m
Number of grids	150 × 150	100 × 100	100 × 100
Horizontal resolution	2 m × 2 m	2 m × 2 m	2 m × 2 m
Vertical resolution	1 m × 30 m	1 m × 30 m	1 m × 30 m
Simulation region	300 m × 300 m × 5 m	200 m × 200 m × 5 m	200 m × 200 m × 5 m
Plant configuration ENVI-met data parameter table
Plant name		Database ID
Grass		[LG] Luzerne 18 cm
Shrub		[S1] Scrub 1 m
Tree		[L1] Privet
Water		[WW] Deep water
Underlying surface ENVI-met data parameter table
Name		Database ID
Grass		[G1] Grass 10 cm
Asphalt		[ST] Asphalt
Concrete		[PP] Pavement (concrete)
Water		[WW] Deep water
Different plant configurations
river width of 350 m		river width of 40 m	river width of 8 m
CASE 1-1 Shrub–grass (1:1)		CASE 2-1 Shrub–grass (1:1)	CASE 3-1 Shrub–grass (1:1)
CASE 1-2 Tree–grass (1:1)		CASE 2-2 Tree–grass (1:1)	CASE 3-2 Tree–grass (1:1)
CASE 1-3 Tree–shrub–grass (1:1:1)		CASE 2-3 Tree–shrub–grass (1:1:1)	CASE 3-3 Tree–shrub–grass (1:1:1)
CASE 1-4 Enclosed tree–shrub–grass (1:1:1)		CASE 2-4 Enclosed tree–shrub–grass (1:1:1)	CASE 3-4 Enclosed tree-–shrub–grass (1:1:1)
Different underlying surfaces
river width of 350 m		river width of 40 m	river width of 8 m
CASE 1-5 Grass		CASE 2-5 Grass	CASE 3-5 Grass
CASE 1-6 Asphalt		CASE 2-6 Asphalt	CASE 3-6 Asphalt
CASE 1-7 Concrete		CASE 2-7 Concrete	CASE 3-7 Concrete

.

Specifically, based on field survey and the characteristics of the riverside landscape belt, 21 typical landscape configurations have been developed, which align with the landscape pattern of the Riverside Landscape Belt in Xi’an. All calculated percentages correspond to the areal ratio of individual landscape components within the total spatial arrangement; uniform initialization parameters were systematically applied across the 21 idealized model implementations, maintaining experimental control conditions. The Urban Water System Planning Guidelines categorize rivers into three levels: primary rivers with a width of 100 m or more, secondary rivers ranging from 10 m to 100 m, and tertiary rivers with a width less than 10 m. Based on these standards and the results of river surveys in Xi’an, three river widths—8 m, 40 m, and 350 m—were selected as the focus of this study.

2.3.3. Outdoor Thermal Comfort

There are currently various thermal comfort indices to evaluate the outdoor environment based on the human energy balance and the ambient environment [37], such as PMV (predicted mean vote) [38], SET* (standard effective temperature) [39], and PET (physiologically equivalent temperature) [40]. PET was adopted in this study, as it has been primarily used to evaluate outdoor thermal comfort, which derives from the Munich Energy Balance Model for Individuals [41].

Table 4. PET calibrations of TSV for Xi’an [44,45].

Thermal Sensation	Thermal Stress	PET Range (C)
Very cold	Extreme cold stress	<13.1
Cold	Strong cold stress	−13.1 to 7.6
Cool	Moderate cold stress	−7.6 to 0.8
Slightly cool	Slight cold stress	−0.8 to 9.8
Neutral	No thermal stress	9.8 to 30.7
Slightly warm	Slight heat stress	30.7 to 41.3
Warm	Moderate heat stress	41.3 to 48.2
Hot	Strong heat stress	48.2 to 53.6
Very hot	Extreme heat stress	>53.6

shows the two PET classifications for subtropical and temperate regions. The thermal ranges in the temperate region are lower than those in the subtropical region due to the lower humidity and higher tolerance of cold temperatures in that area [42,43]. Furthermore, the grade of thermal stress for temperate regions was applied to investigate the outdoor thermal comfort in Xi’an.

2.4. Verification Analysis

Root mean square error (RMSE) and d, effective for verifying the deviation between measured and simulation values, were used for the verification and evaluation of the measured and simulated data. RMSE represents the difference between the predicted values and the measured values, and is used to assess the data deviation between them. Additionally, d indicates the degree of closeness between the prediction model and the simulation model, where d = 1 indicates perfect agreement between the measured and simulated data, reflecting the actual error between the two [21,35].

The formulas for RMSE and d are as follows:

$$\mathrm{R}\mathrm{M}\mathrm{S}\mathrm{E}=\sqrt{{\displaystyle \frac{1}{\mathrm{n}}}{{\sum }_{i=0}^n\left(\widehat{{\mathrm{y}}_{\mathrm{i}}}-{\mathrm{y}}_{\mathrm{i}}\right)}^2},$$

(1)

$$\mathrm{d}=1-{\displaystyle \frac{\sum _{i=0}^n{\left({\widehat{\mathrm{y}}}_{\mathrm{i}}-{\mathrm{y}}_{\mathrm{i}}\right)}^2}{\sum _{i=0}^n\left(\left|\left(\widehat{{\mathrm{y}}_{\mathrm{i}}}-\overline{{\mathrm{y}}_{\mathrm{i}}}\right)\right|+\left|{\mathrm{y}}_{\mathrm{i}}-\overline{{\mathrm{y}}_{\mathrm{i}}}\right|\right)}}.$$

(2)

$\widehat{y_i}$: simulated value;

$y_i$: measured value;

$\overline{y_i}$: measured mean value;

n: sample size.

Relevant studies suggest that the time periods of 8:00, 14:00, and 20:00 are representative for mitigating urban heat island effects. Consequently, these three time points were selected for simulation data analysis. These specific time points have also been utilized as samples in prior research.

Finally, the SPSS Statistics v. 24 (IBM Corp., Armonk, NY, USA) was used for statistical analyses.

3. Results

3.1. ENVI-Met Verification

The ENVI-met model validation is based on data collected over a 12 h period during the measurement phase.

Figure 3 shows air temperature trends of the three measurement sites. It is evident that the trends of the simulated and measured values at the three measurement points are largely consistent, though minor discrepancies exist. These differences may result from factors not incorporated into the numerical simulation, including variations in water flow velocity, changes in water surface area, and disturbances from recreational activities.

We can see from Figure 3 that the air temperature initially increases and then decreases. The air temperature during the testing period ranged from 26.7 to 37.5 °C. Through comparison, we found that the temperature of SITE3 is higher than that of SITE1 and SITE2 most of the time. Given that SITE3 is located at the widest section of the river, this suggests that in wider river areas, wind speeds may be higher, which speeds up evaporation of water bodies but also blows cold air away, resulting in less cooling in riverside areas.

To assess the accuracy of the simulated results, the root mean square error (RMSE) and the index of agreement (d) were employed to compare the simulated and measured values. As shown in

Table 5. RMSE and D values of measured and simulated values.

Time	Parameter	SITE 1	SITE 2	SITE 3
8:00	RMSE	0.22 °C	0.25 °C	0.31 °C
	d	0.98	0.91	0.92
14:00	RMSE	0.81 °C	0.79 °C	0.82 °C
	d	0.90	0.92	0.96
20:00	RMSE	1.10 °C	1.09 °C	1.02 °C
	d	0.86	0.87	0.82

, the RMSE values range from 0.22 °C to 1.10 °C, indicating minimal discrepancies between the simulated and measured data. Furthermore, the index of agreement (d) ranges from 0.82 to 0.98, with all values approaching 1, demonstrating strong agreement between the simulated and measured air temperatures. These findings confirm the suitability of ENVI-met for this study.

3.2. Impact of Landscape Configuration on the Thermal Environment

To determine the optimal plant configuration scenario, this study utilizes the temperature difference between idealized plant configuration scenarios and the existing scenario as an evaluation metric. The objective is to identify more effective and rational plant configuration schemes.

3.2.1. Relationship Between Landscape and the Thermal Environment on River Width of 350 m

Figure 4a displays a bar chart comparing air temperature differences between various plant configuration scenarios and the original scenario in the river width of 350 m. At 5:00, the temperature differences for the four scenarios (CASE 1-1, CASE 1-2, CASE 1-3, and CASE 1-4) relative to the original scenario were 0.31 °C, 1.42 °C, 1.8 °C, and 2.33 °C, respectively. CASE 1-4 exhibited the highest temperature difference, while CASE 1-1 showed the lowest. Similarly, at 14:00, the differences were 0.21 °C, 1.31 °C, 1.75 °C, and 2.35 °C, with CASE 1-4 again having the highest difference and CASE 1-1 the lowest. At 22:00, the differences were most pronounced, with CASE 1-4 reaching 3.74 °C and CASE 1-1 recording the lowest difference at 1.92 °C.

A comparison of temperature differences at 5:00 and 14:00 reveals that the temperature differentials in CASE 1-4 increased by 1.44 °C and 1.39 °C, respectively. Furthermore, the temperature differences between the other three scenarios and the baseline scenario exhibited substantial increases.

Figure 4b presents a bar chart illustrating the temperature differences between the river-width of 350 m under three underlying surface configurations (CASE 1-5, CASE 1-6, and CASE 1-7) and the baseline scenario. CASE 1-5 exhibits significant temperature differences at 5:00, 14:00, and 22:00, with the largest difference occurring at 22:00, reaching 1.95 °C, indicating a notable cooling effect of CASE 1-5 during nighttime. At 5:00 and 14:00, the temperature differences between CASE 1-5 and the baseline scenario are relatively close, at 0.37 °C and 0.31 °C, respectively, suggesting a stable cooling capacity of CASE 1-5 in the early morning. In contrast, CASE 1-6 and CASE 1-7 show significant temperature differences at 5:00 and 22:00, reaching 1 °C; however, at 14:00, the temperature difference is negative, which may be related to strong solar radiation absorption and heat storage characteristics, highlighting the differences in daytime temperature regulation among various underlying surfaces.

In summary, among the plant configurations evaluated, CASE 1-4 exhibits the most significant cooling effect, suggesting that the enclosed tree–shrub–grass configuration represents a reasonable landscape design for the river width of 350 m. With respect to varying underlying surfaces, CASE 1-5 demonstrates the most substantial enhancement in the thermal environment, especially during nighttime.

3.2.2. Relationship Between Landscape and the Thermal Environment on River Width of 40 m

Figure 5a displays a bar chart depicting the air temperature differences between various plant configuration scenarios and the baseline scenario in the river width of 40 m. At 5:00, CASE 2-4 records the highest temperature difference of 1.79 °C, whereas CASE 2-1 exhibits the lowest at 0.31 °C. The temperature differences for CASE 2-2 and CASE 2-4 are 1.2 °C and 1.49 °C, respectively. By 14:00, the temperature differences for CASE 2-1, CASE 2-2, CASE 2-3, and CASE 2-4 are 0.08 °C, 0.63 °C, 1.01 °C, and 1.39 °C, respectively, with CASE 2-4 consistently showing the highest value. At 22:00, the temperature differences across all four scenarios rise significantly, with CASE 2-4 reaching the maximum difference of 3.24 °C, underscoring its prominent role in nighttime heat dissipation. Although CASE 2-1 remains the lowest, its temperature difference increases to 1.76 °C.

Compared to the temperatures recorded at 5:00 and 14:00, the temperature difference at 22:00 increased by 1.68 °C and 1.63 °C, respectively, demonstrating the overall effectiveness of plants in enhancing the thermal environment during nighttime. Similarly, the temperature difference for CASE2-4 at 22:00 increased by 1.79 °C compared to 5:00 and 1.85 °C compared to 14:00, further highlighting its superior performance in improving nighttime thermal conditions.

As illustrated in Figure 5b, the temperature differences between the three underlying surfaces (meadow, asphalt, and concrete) of the river width of 40 m and the original scene are presented in a histogram. At 5:00, CASE 2-5 exhibits the largest temperature difference, reaching 0.36 °C, followed by CASE 2-7 at 0.15 °C, and CASE 2-6 with the smallest difference of only 0.06 °C. This suggests that CASE 2-5 demonstrates significant advantages in enhancing the thermal environment within the river landscape zone. At 14:00, the temperature differences for CASE 2-5, CASE 2-6, and CASE 2-7 are all negative, with values of −0.05 °C, −2.53 °C, and −0.95 °C, respectively. This indicates that the cooling effects of these three models are inferior to those of the original scene, likely due to strong solar radiation and surface heat storage characteristics. At 22:00, CASE 2-5 shows the most pronounced temperature difference, reaching 1.78 °C, followed by CASE 2-7 at 0.91 °C, while CASE 2-6 has the smallest difference of 0.78 °C.

Generally, the largest temperature difference between the ideal and original scenes of the river width of 40 m occurs at 22:00, primarily observed in the enclosed tree–shrub–grass and grass underlying surfaces. However, the overall temperature difference is slightly lower than that of the river width of 350 m.

3.2.3. Relationship Between Landscape and the Thermal Environment on River Width of 8 m

As illustrated in Figure 6a, the column chart depicts the air temperature differences between various plant configuration scenes and the original scene in the river with a width of 8 m. At 5:00, the temperature differences are as follows: CASE3-1 is 0.63 °C, CASE3-2 is 1.15 °C, CASE3-3 is 1.48 °C, and CASE3-4 is 2.24 °C. Among these, CASE3-4 exhibits the largest temperature difference, reaching 2.24 °C, while CASE3-1 shows the lowest cooling effect at 0.63 °C, resulting in a difference of 1.61 °C between the two. At 14:00, the temperature differences are 0.09 °C for CASE3-1, 0.21 °C for CASE3-2, and 0.57 °C for CASE3-3. Although CASE3-4 has the highest temperature difference at 1.05 °C, its advantage diminishes compared to the early morning. At 22:00, the temperature differences between the four scenes and the original scene reach their peak for the day. CASE3-4 continues to demonstrate the best performance, with its cooling effect increasing to 3.18 °C. While CASE3-1 remains at the lowest level, its cooling effect also rises significantly to 1.76 °C.

The temperature difference for CASE 3-4 at 22:00 is 0.94 °C higher than at 5:00 and 2.13 °C higher than at 14:00, demonstrating a more pronounced variation. Similarly, the temperature difference for CASE 3-1 at 22:00 increases by 1.4 °C compared to 5:00 and by 2.31 °C compared to 14:00. This significant nighttime cooling effect highlights that the enclosed tree–shrub–grass configuration in the river with a width of 8 m has the most substantial impact on improving the thermal environment.

Figure 6b illustrates the temperature differences between the three underlying surfaces of the river with a width of 8 m (CASE3-5, CASE3-6, and CASE3-7) and the original scene. At 5:00, CASE3-5 exhibits the highest temperature difference at 0.35 °C, followed by CASE3-7 at 0.21 °C and CASE3-6 at 0.11 °C. At 14:00, the temperature differences for all three cases are negative: CASE3-5 is −0.25 °C, CASE3-6 is −2.81 °C, and CASE3-7 is −1.23 °C. This indicates that the cooling effects of these models are inferior to those of the original scene, likely due to strong solar radiation and surface heat storage characteristics. At 22:00, the maximum temperature difference is observed in CASE3-6, reaching 2.25 °C, which may be attributed to the nighttime heat dissipation properties of asphalt. CASE3-5 follows with a temperature difference of 1.75 °C, while CASE3-7 has the smallest difference at 0.85 °C.

In summary, CASE3-5 demonstrates a relatively stable cooling effect, particularly during the early morning and nighttime. The cooling effects of CASE3-6 and CASE3-7 are strongly time-dependent, exhibiting significant cooling during specific periods (e.g., nighttime) but potentially increasing under conditions of intense solar radiation (e.g., at 14:00). Additionally, the cooling effect of river channels is closely associated with factors such as width, water area, and the type of surrounding underlying surfaces. Wider river channels are more effective in enhancing overall cooling capacity.

This paper shows that the temperature difference is most pronounced in a river width of 350 m, least significant in a width of 8 m, and moderate in a width of 40 m. This indicates that wider river channels provide more substantial cooling effects. However, the enclosed tree–shrub–grass configuration and meadow-based underlying surfaces are found to be the most effective landscape solutions for improving the thermal environment of riverside zones, with minimal dependence on river width.

3.3. Human Thermal Comfort

3.3.1. Vegetation Configuration

As illustrated in Figure 7a, the physiologically equivalent temperature (PET) at 14:00 is markedly higher than those recorded at 8:00 and 20:00, with the latter two time points demonstrating comparable PET levels. Notably, the lowest PET values at 8:00 and 14:00 are observed in Case 1-4, registering 20.4 °C and 36.3 °C, respectively, which signifies significant thermal stress. Conversely, the lowest PET value at 20:00 is found in Case 1-1, at 22.6 °C, indicating no thermal stress.

As shown in Figure 7b, the PET at 14:00 is significantly higher than at 8:00 and 20:00. Specifically, the lowest PET values at 8:00 and 14:00 are both recorded in CASE 1-4, measuring 21.3 °C and 38.2 °C, which correspond to no thermal stress and slight heat stress, respectively. At 20:00, the lowest PET is recorded in CASE 1-1, at 22.6 °C, indicating no thermal stress.

As depicted in Figure 7c, the PET at 14:00 is significantly higher than at 8:00 and 20:00, with the latter two times exhibiting relatively similar PET values. Specifically, the lowest PET values at 8:00 and 14:00 are recorded in CASE 1-4, measuring 20.6 °C and 40.2 °C, which correspond to no thermal stress and slight heat stress, respectively. At 20:00, the lowest PET is observed in CASE 1-1, at 22.8 °C, indicating no thermal stress.

Overall, the common characteristics across the three river widths are as follows: the PET peaks at 14:00, while the values at 8:00 and 20:00 are relatively similar and lower. Notably, the lowest PET values at 8:00 and 14:00 occur in CASE 1-4, suggesting that the enclosed tree–shrub–grass configuration offers optimal thermal comfort during midday, largely due to tree shading and the thermal properties of grass. At 20:00, the lowest PET is observed in CASE 1-1, as the difference in specific heat capacity between the water surface and the riverbank generates a pressure gradient and wind after sunset. Unlike shrub–grass, the enclosed tree–shrub–grass configuration impedes wind flow, reducing thermal comfort. Consequently, shrub–grass areas provide superior comfort in the evening. These findings indicate that enclosed tree–shrub–grass spaces are the most comfortable in riverside zones during the morning and midday, while grassy areas are more suitable in the evening.

3.3.2. Underlying Surface

Figure 8 presents the PET trends for the underlying surface scenarios. As depicted in Figure 8a, the PET at 14:00 is significantly higher than at 8:00 and 20:00, with the latter two times showing relatively similar PET values. Specifically, the lowest PET values at 8:00, 14:00, and 20:00 are all recorded in CASE 1-5, measuring 21.4 °C, 42.1 °C, and 22.5 °C, respectively. However, only the PET at 14:00 corresponds to moderate heat stress, whereas the PET values at 8:00 and 20:00 indicate no thermal stress.

Figure 8b demonstrates that the lowest PET at 14:00 is recorded in CASE 2-6, measuring 42.5 °C, corresponding to moderate heat stress. In contrast, the lowest PET values at 8:00 and 20:00 are observed in CASE 2-7, at 22.4 °C and 22.5 °C, respectively, indicating no thermal stress.

Similarly, Figure 8c shows that the lowest PET at 8:00 occurs in CASE 3-5, at 22.4 °C, while the lowest PET at 14:00 is recorded in CASE 3-7, at 43.8 °C. At 20:00, the lowest PET is observed in CASE 3-6, measuring 22.1 °C. These findings indicate that only the PET at 14:00 corresponds to moderate heat stress, whereas the PET values at 8:00 and 20:00 reflect no thermal stress.

Analysis of the PET results for various underlying surfaces reveals that thermal comfort is superior in the morning and evening compared to midday. However, no distinct patterns are evident for the river width of 40 m and the river width of 8 m. Notably, the grassy underlying surfaces of the river with a width of 350 m exhibit significant improvements in thermal comfort throughout the day, indicating that wider river channels more effectively enhance the thermal comfort of riverside grassy areas. Consequently, planners should account for the relationship between meadow underlying surfaces and river width when designing riverside landscape zones.

4. Discussion

4.1. Thermal Environment and River-Width

This study investigates the thermal environment of the Ba River landscape belt in Xi’an, revealing a positive correlation between channel width and its ability to mitigate thermal effects. Specifically, wider channels exhibit a more pronounced cooling effect.

Meanwhile, studies about the influence of rivers on the thermal environment of urban built-up areas have been conducted in cities with different climate zones, also revealing that the river has a significant mitigation of thermal effect, with river width being proportional to the cooling effect, such as in Pohang City in the Republic of Korea [46], Nanjing [47], Chengdu [48], Guangzhou [49], Changsha [50], Shanghai [51,52], Wuhan [53], and other cities in China. This is mainly because wider water bodies have a greater heat capacity, which contributes to absorbing and storing more heat, thereby alleviating the thermal environment significantly surrounding the built environment. This indicates that the effect of rivers in mitigating the thermal environment of surrounding built-up areas is recognized in cities across different climate zones, also confirming the feasibility of the research method proposed in this article.

It is worth noting that Feng [54] explored the impact of the Ba River in Xi’an on the surrounding thermal environment, also revealing that the LST (land surface temperature) of the river was significantly reduced within the distance range of 300 to 500 m and 400 to 600 m. Obviously, this belongs to the large-scale perspective of the Ba River’s mitigating effect on the thermal environment of the surrounding built-up areas. Differently, the width of the Ba River in this study ranges from 8 to 350 m, which exactly complements the small-scale thermal environment research in the Ba River Basin. Based on these findings, during the design period of the Ba River Riverside Landscape in Xi’an, we suggest that methods such as segmental widening of the river channel and the addition of artificial wetlands could be used to increase the width of the natural river channel. Meanwhile, open spaces should be reserved in areas with large water areas where rivers converge as residential activity areas to provide a more comfortable riverside space.

4.2. Thermal Environment and Plant Configurations

We also found that the cooling effects of tree–shrub–grass and shrub–grass configurations were the most significant within the riverside landscape belt compared to the cooling influence of different plant configurations.

This is consistent with the research results on the cooling effect of different plant configurations in urban green space [55], railway stations [56], city squares [57], urban residential areas [58], and other areas. It shows that the space with lawn grass, high canopy closure, and thick canopy could alleviate the local thermal environment effectively, and this holds true for any urban space. The main reason for this result is the multi-tiered shading effect of the tree–shrub configurations, which substantially mitigate direct solar radiation reaching the ground while simultaneously absorbing latent heat through water release during transpiration processes. Concurrently, herbaceous coverage reduces surface albedo and enhances soil moisture retention. Furthermore, plant communities generate a self-reinforcing microclimate through wind speed reduction and humidity elevation, collectively contributing to ambient temperature regulation. These findings demonstrate that tree–shrub configurations exert the most pronounced geothermal moderation effects within the central zones of urban riverfront landscapes.

In addition, Sodoudi [59] demonstrated that both spatial configuration and vegetation types within green spaces jointly influence their cooling efficiency, with peak cooling effects observed at 14:00. Extending this finding, our study further investigated diurnal variations in the cooling performance of vegetation configurations. Specifically, we revealed that tree–shrub–grass configurations exhibit stronger cooling effects during daytime, whereas shrub–grass configurations appear at night. Meanwhile, Chen [60] indicated that tree canopy showed a stronger influence on LST during the day than at night, followed by vertical structure and configuration, also revealing that variation in both daytime LST and nighttime LST combines composition, configuration, and vertical structure of tree canopy. These findings elucidate the pronounced cooling effect exerted by the canopy while also corroborating the underlying mechanisms driving the notably enhanced daytime thermal regulation observed in shrubs and grasses within this paper.

On the whole, in the absence of solar radiation, surface heat dissipation occurs primarily via long-wave radiation to the atmosphere. However, tree canopies impede this heat emission by forming a “thermal insulation layer” effect. In contrast, shrub-covered areas, devoid of overhead tree shelter, allow direct radiative cooling from the surface to the atmosphere. This unimpeded heat loss results in accelerated cooling rates. Furthermore, cold air masses from adjacent riverine zones enhance convective heat removal by penetrating vegetated spaces through airflow. These combined microclimate mechanisms explain the distinct thermal performance of tree–shrub–grass and shrub–grass configurations during daytime and nighttime.

In our study, the effects of different landscape configurations on the local thermal environment during the day and night were considered, which complements the basic theory that landscape configurations optimize the local thermal environment. Therefore, based on our research results, we propose to arrange low shrub grass in the waterfront area of the riverfront landscape to provide a comfortable environment at night, transitioning outwards to shrub and grass structures to enhance daytime shade, thus forming a spatio-temporal balanced composite vegetation configuration.

4.3. Thermal Environment and Underlying Surface

We also found that grass as the underlying surface has the best cooling effect, followed by concrete and asphalt.

Hathway [17], Manavvi [61], and Qi [48] have shown that the cooling effect of rivers would be enhanced while reducing impervious surfaces, also revealing that underlying surfaces such as plants could improve the local thermal environment, while roads aggravate the surrounding thermal environment. These studies suggest that the permeable underlying surface near the river could enhance the cooling effect of the river, which is also confirmed by our research results, which reveal that grass has better water permeability than underlying surfaces such as concrete and asphalt, and its cooling effect is more significant. As well as that, we further analyze the cooling effect of impermeable cushions of different materials and point out that the cooling effect is as follows: grassland > concrete > asphalt; This complements the basic data on the influence of artificial underlayment on the local thermal environment of the riverfront landscape belt.

Totally, grasslands generally have a lower heat capacity and a higher evaporative heat dissipation capacity, which could reduce surface temperature through evaporation. Concrete surfaces are usually smooth, with low thermal reflectivity and relatively high heat absorption, resulting in a high surface temperature. This allows concrete to increase the “heat island effect” during periods of high temperature, especially in cities. The thermal conductivity and heat absorption capacity of asphalt are usually stronger than that of concrete because it has a dark surface and low thermal reflectivity, which easily absorbs the heat of solar radiation.

Based on these research results, we suggest that grass and grass-planting brick pavement should be used as much as possible in the design of the riverfront landscape belt, and the design of “ecological slope protection + grass ditch” should be adopted to replace the traditional hard revetment, which not only improves the local microclimate but also provides more comfortable and leisure space for citizens.

4.4. Limitations and Future Directions

While this study has achieved certain outcomes, several limitations exist. Firstly, the ENVI-met model does not account for water depth’s potential effects on the thermal environment. Actual channel water depth changes could significantly alter evaporation and thermal inertia dynamics. Secondly, the model simplifies vegetation types using a single species, whereas real vegetation is diverse and complex. Future research should explore mixed vegetation allocations’ synergistic effects. Thirdly, this study focuses solely on summer scenarios, with transitional seasons and winter thermal responses needing further investigation. Additionally, practical constraints like economic costs and maintenance requirements were not considered, potentially limiting the findings’ direct applicability.

Future work will involve multi-season simulations and integrating social behavior data (e.g., resident activity patterns) to balance thermal comfort with functional needs. Developing a dynamic vegetation growth model to assess vegetation maturity’s long-term cumulative effects on thermal regulation is also recommended, providing scientific support for waterfront landscape lifecycle management.

5. Conclusions

This study focuses on the Ba River’s riverside landscape zone in Xi’an, a representative northern Chinese city. Twenty-one ideal landscape configuration scenarios were developed and evaluated based on temperature differences from the original scenario, with the physiologically equivalent temperature (PET) used to assess human thermal comfort. Key findings include the following:

(1)

ENVI-met reliability in predicting the local thermal environment of riverside landscape zones is confirmed, with maximum RMSE and d values of 1.10 °C and 0.98, respectively, within an acceptable error range.

(2)

Wider rivers show more significant cooling effects. The enclosed tree–shrub–grass configuration is optimal for enhancing riverside thermal environments, with grass being the most effective underlying surface, especially for nighttime cooling, providing a theoretical basis for riverside landscape planning.

(3)

PET results indicate that enclosed tree–shrub–grass spaces offer the highest comfort in riverside zones during morning and midday, while shrub–grass areas are more suitable at night, which offers spatial guidance for riverside residents.