Assessing the Effects of Exposure to Green Rooftop Spaces on Perceived Restorativeness: A Field Study in Xiamen, China

Abstract

Rapid urbanization is forcing the need to build urban environments that promote healthy and sustainable human settlements with a view to reducing the risks of various diseases. Previous studies have demonstrated that green rooftop spaces may stimulate restorative benefits and enhance their livability. However, this potential remains without empirical research support. To investigate the restoration of green rooftop spaces on human well-being, we conducted a field experiment using psychophysiology measurements. With the help of physiological indicators (SBP, DBP, HR, HRV) and psychological indicators (PRS, BPOMS), the changes of indicators in the experiment were analyzed. We recruited a total of 35 college students who were randomly assigned to one of four rooftop spaces for a field experiment. Physiological and psychological indicators were measured before and after recovery to analyze the effects of rooftop spaces on restorative benefits. Paired sample t-test results revealed that although there were differences in the physiological and psychological measurements related to environmental restorative benefits, both confirmed that green rooftop spaces exhibited stress-reduction effects on participants. Notably, all four rooftop spaces had psychophysiological restoration to varying degrees, with spaces featuring a higher degree of naturalness demonstrating better restoration. Furthermore, a grassy environment had effects on positive psychophysiological responses. Correlation analysis indicated that subjective perception evaluations of green rooftop spaces had positive effects on psychophysiological restoration. The results of the study are conducive to increasing awareness of the restorative value and potential of green rooftop spaces, which can be constructed by designers, thereby contributing to sustainable urban development.

1. Introduction

Due to the steady rise in urbanization, the global urban population is projected to intensify, with an estimated 68% of the population residing in urban areas by 2050 [1,2]. While continuous urbanization has significantly enhanced the convenience of human life, it has concurrently been accompanied by a rise in psychological disorders and health-related concerns among urban populations [3,4,5]. Urgently needed are ways to restore the physical and mental health of modern populations to mitigate the prevalence of mental illness. Recently, a substantial body of studies have demonstrated that connectedness to the natural environment can serve as a compensatory mechanism for the deterioration of individuals’ health, as exposure to natural settings has been shown to promote health recovery and alleviate stress [6,7,8,9,10]. The natural environment offers residents a restorative experience by mitigating stimuli associated with daily stressors and providing an adaptive refuge [11]. Both the Stress Reduction Theory (SRT) and the Attention Restoration Theory (ART), respectively proposed by Ulrich and Kaplan, have been widely recognized as foundational frameworks in research pertaining to restorative landscape design [12,13]. According to SRT, exposure to natural environments was effective in alleviating mental stress and promoting a healthy emotional state. Specific environmental elements were capable of inhibiting the initiation of stress response, with interactions with natural settings eliciting rapid, subconscious emotional reactions that mitigate negative emotions and attenuate physiological disturbances. ART posited that exposure to natural environments facilitated the recovery of directed attention from mental fatigue caused by daily chores. Whereas information processing about directed attention requires more cognitive effort and energy expenditure, the properties of natural landscapes encourage a shift from directed to spontaneous attention, thereby supporting the restoration of well-being compromised by attentional fatigue through interaction with nature. Moreover, Hartig et al. posited that naturalized spaces have psychological benefits of reducing stress and negative emotions, while improving attention and regaining spiritual vitality [14]. Natural environments, when compared to urban settings, tended to provide greater restorative benefits, resulting in significantly positive changes in blood pressure, mood, and concentration [15].

Numerous studies have indicated that urban green spaces, such as urban parks, forest parks, and waterfront parks, possess a high degree of perceived restorative benefits [16,17,18,19]. It is noteworthy that recent studies have shown that pocket parks have the restorative potential to alleviate mental stress [20,21,22]. Due to their strong accessibility, pocket parks can fulfill the psychological needs of individuals residing in high-density urban areas, providing greater opportunities for interaction with nature and contributing to enhanced mental well-being [23]. Despite their limited spatial scale, the natural landscape elements within pocket parks have been shown to alleviate psychological stress among urban populations and support in acquiring restorative experiences [24,25]. The different green space types elaborated above all possess characteristics similar to natural environments and are rich in recreational and experiential qualities, as well as stress-reducing effects. In a modern fast-paced urban life, social competition, interpersonal interaction, and other issues are usually the stressful sources for urban residents. Green spaces with naturalized features contribute to stress reduction, which is related to the psychoevolutionary theory that humans possess an inherent affinity for nature rooted in evolutionary history. Prior studies have used methodologies such as virtual reality and image/video simulation to assess the effects of greening on human stress recovery and confirmed this point.

Additionally, scholars have observed that a green rooftop space, considered a unique type of pocket green space, also has the ability to mitigate stress and enhance both physical and mental health [26]. Recognizing the limited availability of land due to urbanization, the construction of green spaces on building rooftops will furnish additional vacant space for the diverse existence of urban landscapes. In general, individual green rooftop space is usually small in size but has convenient accessibility to the neighboring population. Having lush vegetation is not always possible due to technical (e.g., load capacity), economical, architectural, and other constraints. As a novel type of urban green space, green rooftops offer highly convenient outdoor recreational opportunities for nearby residents. Green rooftops with aesthetically pleasing spaces for recreation and socialization can enhance the quality of life for urban residents and increase community cohesion, while promoting environmental education and sustainability awareness. Simultaneously, green rooftops contribute diverse ecological benefits to cities, such as mitigating the urban heat island effect, improving air quality, reducing carbon footprints, regulating stormwater management, and enhancing urban biodiversity, that are argued to have positive effects on the health and well-being of people [27,28]. In addition, green rooftops help to improve the thermal insulation performance of buildings, thereby reducing building maintenance costs and extending rooftop life [29,30,31,32,33]. Green rooftops can also increase the value of real estate and lead to the development of green industries. It is evident that green rooftop spaces, although limited in extent, play a vital role in creating a healthy and livable environment. In densely populated urban areas, individuals may regain the ability to meet their daily lives through the immersive experience in green rooftop spaces, thereby promoting physical and mental health [34]. Previous research has demonstrated that green rooftop spaces significantly enhanced psychological well-being, contributing to stress alleviation and the facilitation of restorative experiences [34,35]. Currently, some scholars have conducted research on the restorative benefits of vegetation in green rooftop spaces and found that the positive impacts of rooftop spaces with vegetation on environmental restoration were more significant than those without vegetation [36]. Lee et al. investigated the influence of vegetation structure on psychological recovery in rooftop spaces and found that observing natural spaces with structured vegetation was associated with more positive emotions and physiological well-being, while enclosed vegetated spaces with limited openness may not promote physical and mental health [37]. Short-term contact with nature environments has also been shown to facilitate individual well-being [9], and compared to concrete rooftops, taking a short time to appreciate lush green rooftop grass would continue to increase the attention among college students [26]. Furthermore, the perceived experience of green rooftop spaces is more intuitive and contributes to a higher level of restorativeness. Mesimäki et al. provided multi-sensory exposure to green rooftop space for tourists [34], revealing the existence of perceived restorative potential of green rooftop space with a relatively high degree of restorativeness. Nevertheless, existing investigations on the health benefits of green rooftop spaces are usually based on visual stimuli, such as images and virtual environments, to evaluate individuals’ physiological and psychological responses to natural settings [1,26,38,39]. Although such methods can simulate realistic natural environments, they inevitably produce perceptual differences when compared to actual natural environments [40]. In contrast, on-site studies enable individuals to engage directly with the physical environment of green rooftop spaces, providing a more authentic experiential context. In addition, researchers have recognized the restorative benefits of green rooftop spaces and their contribution to enhancing the livability of urban environments. However, empirical evidence remains limited, particularly concerning how specific design interventions can be strategically employed to optimize the restorative potential of green rooftop spaces. Most relevant studies focused on a single space or vegetative features as the research context with relatively single evaluation methods, and comparative analyses examining the potential health benefits across different green rooftop spaces remain absent. Therefore, to comprehensively assess the health benefits provided by green rooftops, this study explored the effects of green rooftop spaces on restorative benefits in terms of both physiological measures and psychological measures through field research. Green rooftops, as a unique type of green space, can further complement the multifaceted content on the health benefits of different types of green spaces by exploring the restorative benefits of green rooftop spaces with differing environmental characteristics.

With the gradual intensification of social competition, college students are prone to psychological and physiological health issues when facing pressures such as coursework and scientific research [41,42]. Green rooftop spaces function as secondary natural environments with which nearby college students regularly engage in their daily lives. And regular exposure to such natural spaces has been shown to reduce stress, promote positive emotional states, and improve concentration [43,44]. Therefore, this study utilizes biofeedback measurement methods, various psychological state scales, and other technical means to investigate the impacts of the perceptual experiences in green rooftop spaces on stress reduction among college students.

The primary objectives of this study are as follows: (1) To investigate the effect of green rooftop spaces on the psychological well-being of college students; (2) To examine the impact of green rooftop spaces on the physiological restoration of college students; (3) To analyze the relationship between environmental perception and stress restoration among college students. Our study aims to elucidate mechanisms through which green rooftop spaces facilitate stress restoration, and the results will be conducive to providing a scientific theoretical basis for designing green rooftop spaces with a focus on promoting health. The structure of this paper is as follows. Section 2 describes the research area and the experimental design of this study. Section 3 describes the results. Section 4 discusses the results, limitations, and implications. The paper ends with a conclusion, summarizing our major findings, in Section 5.

2. Materials and Methods

2.1. Study Sites

In our study, the green rooftop spaces located at the Architecture Experimental Building of a university in Xiamen, China (24°43′16.23″ N, 118°10′7.05″ E) were selected as the experimental site. According to the Chinese standard of climatic regionalization for architecture, Xiamen is situated within a region characterized by a hot summer and warm winter climate, classified as a subtropical marine monsoon climate. Climate data show that the highest average annual temperature occurs in July, reaching approximately 28.5 °C, while the lowest average temperature occurs in January with an average value of 13.3 °C [45]. This study lasted from May to June 2023, encompassing the relatively comfortable climate period in Xiamen. The experimental measurements were performed on the green rooftop spaces of Huaqiao university’s Architecture Experimental Building.

The green rooftop covers an area of 240 m² and incorporates various landscape features such as a pond, gallery, sunlight room, trees, shrubs, and ground cover plants. Additionally, the green rooftop is divided into four recreational spaces: gallery space (GS), wooden square (WS), sunlight room (SR) and planting space (PS), which are typically accessible to the public as a courtyard, providing recreational spaces for nearby students (Figure 1). Therefore, four types of green rooftop spaces were selected for trials to assess their potential effects on the psychological and physiological restoration of college students (

Table 1. Descriptions of the four green rooftop spaces.

Sites	Characteristics
GS	The gallery space is situated adjacent to a polygonal pond and encompasses an area of 69 m2. The key landscape elements within this space include vegetation, water features, a gallery structure, seating, and wooden paving. Notably, the pond occupies approximately half of the total area, contributing to the space’s irregular spatial configuration. The overall spatial arrangement is well ordered, with a relatively structured plant community and a visible green index of approximately 23%. The vegetation is predominantly composed of Punica granatum and Ficus microcarpa, arranged symmetrically to enhance spatial harmony. Furthermore, the space is characterized by clear sightlines, promoting openness and visual connectivity.
WS	An open square encompasses an area of 47 m2 and is primarily covered with a wooden platform, accompanied by a minimal presence of vegetation. The space is equipped with treadmills and seating, providing opportunities for physical exercise and rest. The plant community within this area is relatively simplistic, predominantly composed of Ophiopogon japonicus, resulting in a visible green index of approximately 9%. Additionally, the space is characterized by its openness and the absence of visual obstructions, facilitating unobstructed sightlines.
SR	The sunlight room serves as a designated space for teachers and students to engage in gardening activities. Encompassing an area of 66 m2, its landscape elements include vegetation, wooden paving, as well as tables and chairs. The space is enclosed by glass walls, creating a semi-enclosed yet permeable environment. It features a defined area of shrubbery with a visible green index of approximately 16%. A square planting pond is incorporated within the space, where Pteris species are cultivated in a systematically organized planting pattern, enhancing both the aesthetic and functional qualities of the environment.
PS	The planting space consists of two rectangular planting pools, covering an area of 43 m2. The primary landscape elements include vegetation and tiled paving. The space is characterized by a high proportion of hard paving and an overall open configuration. The plant community exhibits a degree of vertical stratification, although the visible green index remains relatively low. The vegetation is predominantly composed of Morus and Rhododendron. The Mulberry trees are planted individually to emphasize their distinct aesthetic value, while the Azaleas are arranged in clusters within the planting pools, enhancing the visual appeal of the space.

).

2.2. Participants

A total of 35 participants (17 males and 18 females) were recruited online to take part in the experiment.

Table 2. Participants’ attributes.

Gender	Number	Age (Years)	Height (cm)	Weight (kg)	BMI (kg/m2)
Male	17	25.58 ± 2.61	177 ± 4.88	72.30 ± 7.31	23.02 ± 1.48
Female	18	23.33 ± 2.73	165.06 ± 4.83	55.21 ± 8.27	20.22 ± 2.76
Overall	35	24.43 ± 2.94	170.86 ± 7.81	63.51 ± 11.75	21.58 ± 2.61

displays the basic demographic information of participants. All participants were students from the School of Architecture, who regularly visited the experimental site daily and were familiar with the green rooftop spaces.

Participants were in good physical health, with no history of hypertension, cardiovascular disease, or significant physical or psychological trauma. They were instructed to maintain a balanced diet, ensure sufficient sleep, and refrain from engaging in strenuous physical activity prior to the trial. In addition, 24 h before the experiment, they were advised to abstain from consuming tea, coffee, and alcohol. To ensure the experiment’s smooth progress, all potential participants were informed about the experimental procedure and signed an informed consent form prior to the experiment. After the experiment, participants were compensated with a stipend as a token of appreciation for their involvement and support.

2.3. Measurements

2.3.1. Physiological Measures

Physiological indicators measured in this experiment include blood pressure (BP), heart rate (HR), and heart rate variability (HRV). Previous studies have demonstrated that BP, HR, and HRV served as effective indicators of physiological stress and are frequently utilized in studies related to stress restoration [46,47]. Among them, BP consists of diastolic blood pressure (DBP) and systolic blood pressure (SBP). When an individual experiences stress, the sympathetic nerve–adrenal medulla system will produce epinephrine and norepinephrine, leading to elevated blood pressure and constriction of peripheral blood vessels. A portable electronic sphygmomanometer (Omron Healthcare Co., Ltd., Kyoto, Japan) was used in our study to measure BP. HR serves as an objective physiological indicator of arousal modulated by sympathetic and parasympathetic nerves. During mental states such as tension and stress, activation of the sympathetic nervous system leads to an increase in HR. Once these mental states dissipate, the parasympathetic nerves inhibit the heartbeat from returning to a normal frequency in a short time [48]. In this study, a heart rate belt (Polar Electronics Co., Ltd., Oulu, Finland) was selected to continuously record participants’ HR values (Figure 2).

HRV refers to the variation in the time intervals between consecutive heartbeats and reflects the physiological marker of an individual’s capacity to regulate the stress response, which can be used as an objective indicator to evaluate autonomic nervous activity [49]. Analysis methods for HRV mainly include time domain, frequency domain, and nonlinear analysis, with time domain and frequency domain being the most commonly employed across various research disciplines [50]. The specific HRV indicators utilized in this study are summarized in

Table 3. Summary of HRV parameters.

Parameter	Unit	Description
Mean RR	ms	The mean of RR intervals.
SDNN	ms	Standard deviation of RR intervals.
RMSSD NN50 pNN50	ms beats %	Square root of the mean squared differences between successive RR intervals. Number of successive RR interval pairs that differ more than 50 ms. NN50 divided by the total number of RR intervals
Normalized power	n.u.	Powers of LF and HF bands in normalized units.
LF/HF	—	Ratio between LF and HF band powers.

.

The time domain includes parameters such as Mean RR, SDNN, RMSSD, NN50, and pNN50. A systematic review of SDNN values in healthy individuals reported that the average SDNN was 50 ms and the median SDNN was 51 ms [51]. RMSSD is used to assess parasympathetic activity and serves as an indicator of physiological restoration [52], with higher RMSSD values suggesting greater stress relief. In the frequency domain, the fast Fourier transform is applied to convert time domain signals into frequency domain signals, followed by power spectral density analysis. The frequency domain includes four parameters: very low frequency (VLF, 0–0.04 Hz), low frequency (LF, 0.04–0.15 Hz), high frequency (HF, 0.15–0.4 Hz), and LF/HF ratio. LF is primarily associated with sympathetic activity, while HF serves as an indicator of parasympathetic regulation [53], and LF/HF represents the balance between sympathetic and vagal nerves. The VLF signification, however, is not yet clear. Higher LF/HF values indicate increased sympathetic activity or decreased parasympathetic activity, which is associated with elevated acute stress and anxiety. Conversely, lower LH/HF values indicate increased vagal activity, reflecting a greater state of relaxation and autonomic balance [54]. Sympathetic nerves (SNS) and parasympathetic nerves (PNS) are not the only factors influencing an individual’s emotional state, but they offer critical insights into its assessment. For example, during emotional states such as anger or anxiety, sympathetic nerves (SNS) innervate the parasympathetic nerves (PNS). A heart rate monitor (Polar Electronics Co., Ltd., Oulu, Finland) and heart rate belt (Polar Electronics Co., Ltd, Oulu, Finland) were utilized to continuously record participants’ HRV parameters (Figure 2). All instruments were calibrated before the measurements, and specific parameters of the instruments are detailed in

Table 4. Specifications of measurement instruments.

Instrument	Model	Parameter Range	Accuracy
Omron electronic sphygmomanometer	HEM-7202	0–299 mmHg	±3–4 mmHg
Polar heart rate belt	H10	30–240 bpm	±0.3%
Polar heart rate monitor	Vintage V2	0–1000 ms	±0.3%

.

2.3.2. Questionnaire

We developed a questionnaire comprising three parts. The first part collected fundamental demographic information, including gender, age, height, and weight. The second part aimed to assess the extent of psychological restoration by measuring the participants’ psychological indicators. To achieve this objective, a perceived restorative scale (PRS) and brief profile of mood state scale (BPOMS) were used to assess the effect of green rooftop spaces on psychological restoration. The PRS, established by Hartig et al. based on Kaplan’s research [55], is widely recognized for its effectiveness in evaluating the restorative potential of environments [56]. A modified version of PRS proposed by Hartig was adopted in our study, which comprises four dimensional characteristics: ‘being away’, ‘fascination’, ‘extent’, and ‘compatibility’ [12,57]. During the questionnaire survey, research assistants provided explanations of each PRS dimension to ensure participants had a clearer and more intuitive understanding of the scale (

Table 5. The dimensional characteristics of perceived restorative scale.

Dimension	Interpretation
Being away	It can break away from the routine and help ease the tension.
Fascination	The environment is charismatic and has attractive qualities.
Extent	The elements of the environment and landscape are compatible and can allow people to extend the good association.
Compatibility	You can quickly adapt to situations like this and want to do things that match the needs provided by the environment.

). The PRS is used the Likert-7 scale to assess participants’ psychological restoration.

The BPOMS is a widely utilized instrument for assessing emotional states and has been used in prior studies to measure changes in emotional states. It comprises six emotional factors: tension (T), anger (A), fatigue (F), confusion (C), depression (D), and vigor (V). The BPOMS utilizes a 5-point scale, with higher scores for each emotional factor indicating greater emotional intensity. Total mood disturbance (TMD) is calculated based on these six emotional factors using the following equation.

TMD = T + A + F + C + D − V

A lower TMD score signifies a more stable emotional state [58]. The integrated application of PRS and BPOMS aims to assess individual psychological change more broadly, thereby facilitating an in-depth exploration of the potential effects of green rooftop spaces on psychological recovery. Given that individual sensory differences may affect stress restoration, part III of the questionnaire introduced an evaluation of the participants’ perception of the current environment. This evaluation included scenic beauty evaluation (SBE), total satisfaction evaluation (TSE), and landscape preference evaluation (LPE) to explore the relationship between subjective perceived evaluation and stress restoration [59]. Five points were used for SBE, TSE, and LPE, with scores ranging from 1 to 5.

2.4. Experimental Procedure

Our study utilized a within-subjects design in which each participant had to undertake all environmental conditions and receive measurements of the dependent variables. The outdoor experiment was conducted during sunny and windless daytime (8:00–9:30 and 16:00–17:30), coinciding with times when college students engaged in recreational activities on the green rooftop. Given the close proximity of the experimental sites, meteorological measurements revealed that the physical environments remained essentially consistent, with an average air temperature of 25.56 °C, relative humidity of 34.05%, and wind speed of 0.4 m/s. In addition, conducting experiments at this location minimized the impact of acoustic environmental variations on the experimental out-comes. Due to the limited availability of instruments, two participants commenced the experiment simultaneously, with each participant spending approximately 25 min to complete the four experimental stages. Notably, participants were asked to avoid interacting with others or using electronic devices during the experiment. We continuously monitored and measured HR and HRV throughout the experiment.

First, participants were provided with a detailed explanation of the tasks prior to the experiment. Upon their arrival at the experimental site, two participants were randomly assigned to the center of an experimental space to eliminate order effect. Research assistants helped participants wear physiological monitoring devices (heart rate monitors and heart rate belts) and collected their basic demographic information. Baseline measurements of HRV and HR were performed after participants were required to remain calm for 2 min (Stage 1).

Next, a 5-min Trier Social Stress Test (TSST) was administered to induce participants’ psychological stress levels [60]. The stress task has two parts, including a 3-min impromptu speech and a 2-min oral number calculation of moderate difficulty. This stress test ensures that participants accumulate a specific level of stress within a limited time, thereby maintaining consistency in stress induction across all participants. Immediately after completing the stress test, research assistants measured participants’ BP and then instructed them to fill out the first questionnaire (Stage 2).

Prior studies have demonstrated that reliable findings can be obtained through exposure to the natural environment for 3–5 min [61]. According to SRT, significant restorative effects are expected to occur after short-term contact with nature during the recovery phase of perceiving the landscape. Overall, stress levels demonstrated a rapid reduction during the initial 3 min of the recovery phase, followed by a relatively stable restoration effect in the subsequent 2 min. After 5 min, the recovery effects generally plateaued and could potentially decline thereafter. Therefore, the duration of the recovery phase was set at 5 min in this experiment, with the mean HR and HRV recorded during this period serving as the restorative data. During the recovery phase, participants were required to remain seated to perceive the green space, minimizing potential interference from physical activities on the experimental outcomes. BP records were conducted again after 5 min and then participants were instructed to complete the second questionnaire (Stage 3).

Finally, participants had a 2-min break to relax during Stage 4. After Stage 4, participants were guided to the next experimental site where they repeated the above experimental process. The detailed experimental procedure is shown in Figure 3.

2.5. Data Analysis

To ensure the reliability of the experimental results, the internal reliability of the PRS and BPOMS scales was assessed using Cronbach’s alpha. The measure had good reliability, with Cronbach’s α values for PRS and BPOMS scales of 0.91 and 0.83, respectively (p < 0.05). Then, HR and HRV data were analyzed using Kubios HRV Standard software (Version 3.5), a user-friendly tool for HRV analysis developed by the University of Eastern Finland. To mitigate potential distortions in the analysis results caused by human factors, RR interval data were calibrated using Kubios HRV Standard software. First, the data were corrected for artefacts and ectopic beats by comparing each RR interval value against the local average interval. If the RR interval value exceeded the specified threshold, the interval was identified as an artefact and marked for correction. Second, to exclude typical artifacts, ectopic beats, missed beats, and multiple beats, we replaced heart rate intervals of missed beats with a cubic spline interpolation method [62]. Finally, the data were detrended using the smoothness priors method to eliminate nonstationarities in the time series. The cutoff frequency was set below the low frequency band (0.04 Hz), ensuring the preservation of normal short-term HRV components. With reference to previous studies, this study adopted the HRV parameters of Mean RR, SDNN, RMSSD, HF_norm, and LF/HF to investigate physiological responses to natural exposure [63].

We did not expect any significant differences in psychological and physiological levels after the stress test. A one-way analysis of variance (ANOVA) was employed to examine variations in physiological and psychological indicators before perceiving greenness. Next, the Shapiro–Wilk test was performed to assess the normal distribution of all data. It was found that all data satisfied normal distribution, therefore the paired sample t-test was used to analyze the differences in psychological and physiological changes before and after recovery phase. The paired difference was calculated by subtracting the difference before and after the restoration. Then, we used Spearman correlation analysis to explore the relationship between participants’ subjective perceptions and stress restoration benefits. The data summary and analysis were conducted using IBM SPSS Statistics version 26.0.

3. Results

3.1. Data Detection Before Restoration

Following the successful homogeneity of variance test, this study performed an ANOVA to examine the differences in psychophysiological indicators among four rooftop spaces before the restoration phase. As expected, the results showed that the majority of psychophysiological indicators did not show significance, indicating there were generally no differences in the physical and mental state of college students (p > 0.05) after exposure to the stressors (

Table 6. ANOVA results and mean values in psychophysiological indicators among the four green spaces before recovery.

Indicator	GS	WS	SR	PS	df	MS	F	p
Being away	3.23	2.91	2.74	2.91	3	1.44	1.74	0.16
Fascination	2.71	2.71	2.49	2.60	3	0.42	0.33	0.80
Extent	2.57	2.63	2.43	2.49	3	0.28	0.25	0.86
Compatibility	2.54	2.51	2.54	2.37	3	0.24	0.24	0.87
T	2.97	2.43	2.71	2.46	3	2.26	1.66	0.18
A	1.49	1.26	1.20	1.14	3	0.79	3.23	0.03
F	1.86	1.69	2.09	1.86	3	0.94	1.51	0.21
C	1.91	1.66	1.71	1.57	3	0.74	0.97	0.41
D	1.20	1.11	1.11	1.06	3	0.12	1.00	0.40
V	2.14	2.14	1.66	1.77	3	2.2	3.84	0.11
TMD	7.29	6.00	7.17	6.31	3	14.05	1.54	0.21
SBP	117.46	115.34	110.69	114.17	3	280.96	1.95	0.12
DBP	80.09	79.49	78.49	79.00	3	229.00	6.48	0.56
HR	78.74	77.97	78.34	78.69	3	4.44	0.03	0.99
Mean RR	772.31	783.09	781.09	786.31	3	1256.28	0.09	0.97
SDNN	44.08	44.01	43.14	42.63	3	17.21	0.06	0.98
RMSSD	35.52	36.00	33.44	34.98	3	43.04	0.13	0.94
HFnorm	27.54	28.82	25.52	26.56	3	69.24	0.52	0.67
LF/HF	3.64	3.17	4.08	3.36	3	5.51	0.76	0.52

). Thus, the observed differences in the recovery of physiological and psychological states in the later stage can be principally attributed to the varying landscape environments of the green rooftop spaces.

3.2. Effects of Green Rooftop Spaces on Psychological Well-Being

3.2.1. Perceived Restoration Evaluation

Among the four rooftop spaces, PRS levels increased significantly after the recovery phase (Figure 4). The paired sample t-test results showed that the differences in PRS values before and after recovery for GS, WS, SR, and PS were 1.69, 0.84, 1.60, and 0.58, respectively (

Table 7. Evaluation of PRS in different rooftop spaces.

Rooftop Space	Pre/Points	Post/Points	Pre-Post Difference	Cohen’s D	t	p
GS	2.76 ± 1.04	4.46 ± 0.59	−1.69 ± 1.26	1.34	−15.93	0.000 ***
WS	2.69 ± 1.00	3.53 ± 0.78	−0.84 ± 0.96	0.86	−10.26	0.000 ***
SR	2.55 ± 0.92	4.15 ± 0.67	−1.60 ± 1.06	1.51	−17.89	0.000 ***
PS	2.59 ± 1.13	3.17 ± 1.01	−0.58 ± 0.83	0.70	−8.23	0.000 ***

Note: ***: p < 0.001.

). The effect size analysis indicated a substantial difference between pre-recovery and post-recovery, as the Cohen’s d values were all close to 0.80 [64]. We were therefore able to conclude that four rooftop spaces have the potential to facilitate psychological restoration, with their restorative effects ranked as follows: GS > SR > WS > PS. Furthermore, GS and SR exhibited similar restorative benefits. Moreover, the differences in PRS scores between GS and SR were higher than that between WS and PS. The findings indicated that restorative benefits in GS and SR were superior to those of WS and PS.

To further explore the restorative characteristics, we separately analyzed the scores and differences of the PRS dimensions across the different rooftop spaces. The paired sample t-test results indicated that all restorative dimension scores showed significant changes across the four rooftop spaces, among which the changes of all dimensional characteristics values in PS were lower than compared to the other three rooftop spaces. In addition, the fascination dimension value in PS (p < 0.01) was significantly lower than in the other three rooftop spaces (p < 0.001) (Figure 5). Concerning the dimensions of fascination, extent, and compatibility, the scores exhibited a consistent trend with the overall PRS, ranking in descending order as GS, SR, WS, and PS. However, the change of being away in SR (−1.49 ± 1.04) was considerably larger than in GS (−1.34 ± 1.21), while WS (−0.83 ± 0.79) and PS (−0.57 ± 0.74) demonstrated a smaller difference in this dimension relative to GS and PS. Furthermore, the change in value before and after recovery in WS was slightly greater than in PS. The significant increases in PRS dimensions scores illustrated that the green rooftop spaces had generated restorative benefits on participants’ mental exhaustion.

3.2.2. Change of Emotional Factors in BPOMS

Across the four rooftop spaces, emotional factors such as tension, anger, fatigue, confusion, depression, and TMD decreased significantly, while vigor exhibited a substantial increase after perceiving landscape (p < 0.001). The paired sample t-test results indicated that the average score for tension, anger, fatigue, confusion, depression, and TMD decreased by 56.44%, 20.47%, 27.27%, 35.09%, 8.93%, and 63.38%, respectively, while that of vigor increased by 66.32%. Among the emotional factors of the BPOMS, the score for tension and vigor varied considerably before and after recovery, analogous to the TMD assessment. However, the pre-post recovery difference of depression was minimal (Figure 6). The significant changes in the BPOMS confirmed that green rooftop spaces can promote the psychological well-being of college students.

A comprehensive analysis of the restorative effects of various rooftop spaces revealed significant reductions in tension, anger, fatigue, confusion, and depression, along with an increase in vigor across all four rooftop spaces (p < 0.05) (Figure 7). Concerning depression, no significance was found before and after recovery in PS (p > 0.05), and the observed changes were consistently less significant compared to other emotional factors in all rooftop spaces. The differences in TMD values before and after the recovery were highly significant (p < 0.001). The pre-post recovery differences in WS (3.20 ± 2.94) and PS (3.20 ± 2.77) were identical, with both values falling below the average TMD score (4.24 ± 3.45). Moreover, the changes in TMD scores were comparable between GS (5.23 ± 4.07) and SR (5.34 ± 3.35). Overall, the significant reduction in TMD scores suggested that the artificial landscape of green rooftop spaces provided pleasant sensory stimulation for students, contributing to enhanced emotional stability. Among these, GS and SR had the most significant impact on the restoration of psychological stress.

3.3. Effects of Green Rooftop Spaces on Physiological Indicators

3.3.1. BP Change Between Pre and Post of Recovery

As shown in

Table 8. Change in BP before and after recovery.

Indicator	Pre-Recovery /mmHg	Post-Recovery /mmHg	Pre-Post Difference	Cohen’s D	t	p
SBP	114.41 ± 12.16	109.69 ± 10.64	4.73 ± 6.91	0.68	8.10	0.000 ***
DBP	78.26 ± 6.29	77.06 ± 6.73	1.20 ± 5.99	0.20	2.37	0.019 *

Note: *: p < 0.05, ***: p < 0.001.

, both SBP and DBP exhibited a significant decrease after recovery (p < 0.05). SBP decreased by 4.73 mmHg with a moderate effect size (Cohen’s d = 0.68, p < 0.001), while DBP decreased by 1.20 mmHg with a smaller effect size (Cohen’s d = 0.20, p < 0.05), indicating that rooftop spaces positively contribute to stress restoration. The paired sample t-test results indicated a significant decrease in SBP after recovery in GS and SR (p < 0.001), while no significance in SBP was observed before and after recovery in WS and PS (p > 0.05). The average SBP decreased by 9.69 mmHg in GS and 5.06 mmHg in SR. Moreover, DBP displayed the most pronounced decrease in GS, with a reduction of 3.94 mmHg, whereas there was no significant change in DBP before and after recovery in WS, SR, and PS (p > 0.05) (Figure 8).

3.3.2. Change of HR and HRV Before and After Recovery

The results showed that participants’ HR exhibited varying degrees of decline after observing the rooftop landscape, suggesting that the types of green rooftop space were associated with distinct levels of restorative effects (Figure 9). We found that significant declines in HR occurred in GS (p < 0.01) and SR (p < 0.05). The most pronounced decline of HR was observed in GS (2.37 ± 3.81 bpm), followed by SR (1.60 ± 3.85 bpm). These findings suggest that site GS is associated with a higher degree of restoration compared to the other three rooftop spaces.

For HRV parameters, increases in Mean RR, SDNN, RMSSD, and HF_norm, along with decreases in LF/HF, indicate improved recovery. In the time domain analysis, we found that Mean RR, SDNN, and RMSSD increased after recovery (Figure 9). The increases in Mean RR after viewing the landscapes of the four rooftop spaces were as follows: GS (23.89 ± 40.55 ms), WS (4.54 ± 62.57 ms), SR (17.69 ± 42.31 ms) and PS (10.97 ± 68.49 ms). The paired sample t-test results aligned with the HR changes, showing significant increases in Mean RR increases for GS (p < 0.01) and SR (p < 0.05). Additionally, while SDNN exhibited slight increases in GS, SR, and PS and decreases in WS, no significant differences were observed among the four rooftop spaces. Similarly, exposure to the landscapes of the rooftop spaces under pressure led to increased RMSSD values, as follows: GS (5.48 ± 13.55 ms), WS (2.76 ± 10.40 ms), SR (5.54 ± 10.52 ms), and PS (4.40 ± 11.49 ms). Furthermore, the differences in RMSSD before and after recovery were significant for all rooftop spaces except WS (p > 0.05). In the frequency domain analysis, the increases in HF_norm after recovery for the four rooftop spaces were as follows: GS (5.17 ± 9.58 n.u.), WS (3.87 ± 11.65 n.u.), SR (5.59 ± 16.60 n.u.), and PS (4.63 ± 11.94 n.u.). The paired sample t-test results revealed that the increases in HF_norm were significant in GS (p < 0.01) and PS (p < 0.05). After recovery, LF/HF values in GS and SR showed a downward trend, only with GS exhibiting a significant difference before and after recovery (p < 0.01). On the contrary, LF/HF values in WS and PS increased slightly, but no significant differences were observed before and after recovery.

3.4. The Relationship Between Subjective Perception Evaluation and Stress Restoration

3.4.1. Subjective Perception Evaluation and Psychological Well-Being

According to the Spearman correlation coefficient and significance levels among the parameters, subjective perceptions exhibited a significant positive correlation with the majority of psychological indicators (Figure 10). Compared to the change in BPOMS, the correlation between subjective perceptions and the PRS increases was notably higher, with moderate correlations ranging from 0.3 to 0.6.

Among the emotional factors, the increase in V inhibited the strongest correlation with SBE, followed by the decreased value in TMD, A, T, and D. Both the decrease in TMD and the increase in V were significantly correlated with TSE. Additionally, the increase in V showed the highest correlation with LPE, followed by the decrease in TMD. Simultaneously, subjective perceptions were found to have a significant positive effect on PRS increases, with LPE exerting the greatest influence, followed by TSE and SBE. These findings indicate that perceived evaluation had a positive effect on the psychological restoration of college students.

3.4.2. Subjective Perception Evaluation and the Physiological Restoration

The correlation analysis revealed that subjective perceptions were positively correlated with changes in most physiological parameters. SBE had a significantly positive effect on SBP reduction, followed by a decrease in LF/HF. Similar trends occurred in the relationship between TSE and physiological changes. Nevertheless, LPE was significantly and positively correlated only with SBP reduction (Figure 11). Overall, subjective perceptions exhibited significant positive association with SBP decreases compared with LF/HF decreases.

4. Discussion

This study aimed to determine the effect of green rooftop spaces on physical and mental health to enhance understanding of their restorative benefits. Consistent with previous findings [34,35,65], our results confirmed that green rooftop spaces contribute to both physiological and psychological well-being. Despite differences in physiological and psychological outcomes, both sets of data confirmed that rooftop spaces effectively reduced stress among college students. Additionally, the study compared the restorative effects of different green rooftop spaces, demonstrating that the four spaces, each with distinct spatial characteristics, provided varying degrees of restorative benefits.

4.1. Effects of Green Rooftop Spaces on Psychological Restoration

The results indicated a significant increase in PRS scores, a substantial reduction in negative emotions, and a notable increase in positive emotions after recovery. As expected, the psychological measures demonstrated the positive restorative benefits of green rooftop spaces on college students. The significant increase in PRS scores across all four rooftop spaces suggested that brief immersion in these environments can facilitate a restorative experience. Additionally, the changes in BPOMS revealed that participants’ positive emotions increased and negative emotions decreased after recovery. Specifically, vigor increased significantly, while tension, anger, fatigue, confusion, and depression showed significant reductions. These findings support the notion that short-term exposure to natural environments can positively influence emotional well-being [9].

Judging from the PRS evaluation results, the four rooftop spaces exhibited varying restorative effects due to the difference in landscape characteristics. Our study revealed that PRS increases were greater in GS and SR compared to the other two rooftop spaces. This result can be attributed to the presence of natural landscape elements in sites GS and SR, in accord with previous results [36,37,64], indicating that green spaces with a higher degree of naturalness would facilitate better restoration. Additionally, the PRS increase in site WS was higher than in PS, likely due to the lack of proper maintenance and the greater prevalence of hard pavement in PS. This finding further supports the affordance theory, which emphasizes the role of ground texture in shaping environmental perception [66].

In addition, the trends of TMD changes across different rooftop spaces were similar to the PRS results. Among them, GS and SR exhibited the most pronounced effects in alleviating psychological stress among college students, further confirming that spaces with a higher degree of naturalness exhibit stronger stress-reduction effects. However, the decrease in TMD was slightly greater in SR than in GS. This divergence may be attributed to the fact that SR is surrounded by glass and has a semi-enclosed and permeable environment, providing participants with buffer spaces against the negative health effects of stressful events. This is consistent with early studies by Kaplan, which found that well-isolated and escapable environments are more effective in restoring positive emotions [67]. It also actually supports the prospect-refuge theory that environments with clear sightlines and concealed spaces possess the capacity to enhance well-being restoration [68]. Prior studies have also demonstrated that landscape spaces with a certain level of enclosure were likely to provide restorative benefits [69]. Moreover, the variations in TMD across different rooftop spaces revealed that different landscape spaces held different capacities to influence participants’ emotions. It is in line with findings by Huang et al. [64] and Wang et al. [70], which highlight the differential impact of landscape features on emotional recovery.

To sum up, the changes in PRS and BPOMS before and after recovery corroborated previous findings that rooftop spaces with naturalized features provided restorative effects and helped reduce negative emotions. This study assessed the restorative benefits of the green rooftop spaces based on visual stimuli. Whereas, in the early stages of visual processing, individuals primarily focused on the information of the matching scene, which aligned with the biophilia hypothesis that humans possessed an intrinsic tendency to seek connections with natural lifeforms. During perceiving landscape, attention often occurred unconsciously, and undirected attention did not require significant cognitive efforts at this time. Therefore, observing the landscape environment of green rooftop spaces is conducive to reducing distractions and facilitating restorative experiences.

4.2. Effects of Green Rooftop Spaces on Physiological Restoration

As can be seen from Section 3.2, most physiological indicators underwent significant positive changes. Significant positive changes in physiological indicators (namely BP, HR, and HRV parameters) were observed in the more naturalized sites such as GS and SR, indicating that highly naturalized spaces offer greater restoration potential compared to those with monotonous landscape elements, a conclusion supported by the observed psychological changes. This is in accord with earlier findings that environments enriched with natural elements promote greater physiological restoration [13,14,15]. The superior restorative potential of GS can be attributed to its structured vegetation and expansive water features. Prior studies have consistently demonstrated that environments incorporating structured vegetation designs and water bodies are more effective in reducing stress [37,70,71,72], consistent with our results. Moreover, the significance of water for stress reduction has been posited by prior research, reporting its crucial role in environmental restoration [72]. This is similar to the findings of Yin et al. [73], which observed that the combination of blue and green spaces was more conducive to promoting psychophysiological restoration. Thus, integrating water features into natural environments can enhance the perception of nature and foster a greater sense of relief. In contrast, the vegetation community in SR lacks well-structured designs but contains a substantial grassland area. Such landscape characteristics have been associated with better restorative benefits, aligning with previous reports suggesting that grassy environments had a higher likelihood of restoration [64]. This finding can be interpreted by the fact that participants had more opportunities to engage with vegetation easily in closer proximity to grassy environments, potentially enhancing higher restorative effects. The explanation could align with the perspective of the biophilia theory, which posits that humans have an innate tendency to connect with other forms of life [74]. In contrast, WS demonstrated limited restorative benefits and, in some cases, may have even elicited negative emotions. This outcome could be due to WS being predominantly paved with wood and featuring less vegetation, which tends to create a sense of monotony and rigidity in terms of visual perception. Conversely, PS exhibited a more favorable restorative effect than WS, which differed from the psychological results. This discrepancy may be due to the longer time required to complete psychological measurements compared to physiological assessments, leading to differences in the observed outcomes [75]. Previous study has indicated that processing complex scenes demands greater cognitive effort [76]. The positive changes in physiological arousal can be construed as the inhibition of the sympathetic nervous system and the enhancement of parasympathetic activity. When individuals engage with natural environments, the cerebral cortex can be activated, triggering involuntary attention and increasing receptivity to external information. Therefore, short-term exposure to natural environments has the potential to reduce acute stress levels, providing empirical support for the stress reduction theory [13].

Overall, the restorative effects of the four rooftop spaces followed the order: GS > SR > WS > PS, with GS possessing the highest restorative benefit. The superior restorativeness of GS can be attributed to its richer vegetation community, higher green visibility, and the integration of a large water body, which collectively create an artificial natural environment with a high degree of naturalization. Empirical studies have indicated that vegetation constitutes the environmental element to which individuals direct their gaze for the longest duration in natural settings, suggesting that the environmental components most likely to attract visual attention may also play a primary role in facilitating recovery [77]. Additionally, previous research has confirmed that environments combining blue and green spaces are more effective in improving health benefits. It can be seen that integrating water bodies into environments contributes to higher levels of well-being and is of great significance in creating healthy landscape environments. Therefore, environments incorporating natural elements exert greater physiological restorative effects compared to built environments that lack such features [64]. Site SR demonstrated greater effectiveness in promoting physical and mental health, which may be explained by its semi-enclosed space and the presence of a certain area of grass. This finding was similar to the previous studies which suggest that enclosed landscape spaces may induce higher restorative benefits. However, other research has reported better restorative effects in more open natural spaces, which was not consistent with the results of this study. It is hypothesized that the cause of the different findings may be mainly due to specific characteristics of the space. As a semi-enclosed space surrounded by glass, SR maintains visual permeability, allowing for a connection with the external environment and fostering positive associations with natural surroundings. In addition, it has been reported that spaces offering a degree of enclosure while maintaining openness and allowing the observation of others’ activities can provide stronger restorative benefits [69]. The limited impact of site WS on psychological and physiological restoration may be attributed to its structurally homogeneous design, minimal vegetation, and lack of spatial variation, which can induce feelings of confusion and disengagement. Similar results also appeared in PS. Although PS contained more natural vegetation than site WS, its poor maintenance and extensive hard surfaces resulted in less restorative benefits. In summary, GS and SR, which feature higher levels of naturalness, provided better restoration compared to WS and PS, which were characterized by more monotonous landscape elements. In a word, the presence of vegetation and water in green rooftop spaces can relieve stress and promote positive emotions, thus improving physical and mental health and facilitating recovery. Consequently, urban planners should emphasize the construction of green rooftop spaces with a view to creating natural shelters for promoting health. Along with social competition, development of interpersonal relationships, and other stressors, health concerns have become increasingly prevalent. The artificial natural environment of green rooftop spaces will be able to provide non-medical means of healing, suggesting that their inclusion in urban public health policies could help mitigate healthcare burdens to some extent.

Both the physiological and psychological effects observed in this study align with Ulrich’s model of affective response to natural environments, which posits that natural settings facilitate psychological and physiological stress reduction [78]. Notably, while differences existed between physiological and psychological outcomes, the overall trends were highly consistent and posited that green rooftop spaces held restorative potential, especially in environments with a higher degree of naturalness. Following exposure to stressors, participants who engaged with green rooftop spaces experienced gradual emotional improvement. Simultaneously, their bodies ceased activating the ‘fight-or-flight’ responses, leading to inhibited sympathetic nervous activity and the promotion of positive physiological changes. This is in line with previous findings reporting the close relationship between psychological and physiological changes during the recovery phase [79]. Meanwhile, we demonstrated that there was a significant correlation between subjective perception evaluations and psychophysiological changes. Although the specific results were not entirely consistent, they contributed to the growing body of evidence highlighting the importance of healthy landscapes. Compared with physiological indicators, correlations between subjective perception evaluations and changes in psychological indicators were more significant. This suggests that individuals who perceive an environment more positively may experience greater psychological recovery. Therefore, prioritizing the sensory experience of visitors in the design of green rooftop spaces is crucial for maximizing restorative benefits. Furthermore, our study synthetically applied multiple psychophysiological indicators for analyzing the restorative benefits of rooftop spaces compared with previous studies, which more robustly explained the effects of green rooftop space landscapes on stress reduction and physical well-being. Although the findings of this study reveal that various green rooftop spaces can alleviate physiological stress to some extent, individual physiological responses are subject to influences such as genetic characteristics and living environments. As a result, these variations may not fully capture the overall effectiveness of stress recovery. Furthermore, the underlying mechanisms driving these physiological responses warrant further empirical investigation and validation.

4.3. Limitations and Implications

This study revealed the health effects of green rooftop spaces through a field experiment. Nevertheless, several limitations should be taken into consideration when applying our findings. First, as the study sample consisted exclusively of college students, further research involving more diverse demographic groups is necessary to enhance the generalizability of the results. Second, our study was limited to a qualitative analysis of the effects of green rooftop spaces on psychological and physiological responses. Further research should aim to quantitatively assess the specific environmental characteristics and their influence on the quality of environmental restoration in rooftop spaces, thereby contributing to a more comprehensive understanding of environmental psychophysiological restoration. Third, this study was conducted under comfortable climatic conditions and focused solely on the influence of visual perception, whereas real rooftop environments involve more complex factors. For example, environmental elements such as air quality may be closely linked to restorative benefits. In addition, research on the restorative benefits of green spaces was based on experiments conducted at a specific time, lacking consideration of dynamic environmental changes. Subsequent studies could address this limitation by examining variations across different times of day and seasonal changes to explore the health effects of green rooftop spaces from both cross-sectional and longitudinal perspectives. Thus, future study is warranted involving a more diverse participant pool and incorporating other environmental factors (i.e., olfactory and acoustic perception) to conduct a more in-depth investigation, allowing for a comprehensive understanding of the effect of rooftop spaces on human well-being.

5. Conclusions

In this study, we conducted a field experiment to assess the effects of green rooftop spaces on the physiological and psychological well-being of college students. Our main conclusions were drawn.

First, while differences were observed between physiological and psychological measurements regarding the restorative benefits of the environment, both sets of data confirmed the stress-reduction effects of green rooftop spaces, which provided valuable insight into the restorative potential of artificial natural environments. Compared to physiological indicators, changes in psychological indicators were more pronounced in assessing the restorative benefits of green rooftop spaces. This may be attributed the participants’ long-term residence in the area, which fostered a stronger sense of place attachment and emotional preferences toward the green rooftop environment. In contrast, variations in physiological parameters offered an objective measure of the restorative effects of green rooftop spaces on both physical and mental health. Second, the four rooftop spaces exhibited varying degrees of restorative benefits, with those featuring a higher degree of naturalness demonstrating superior restorative effects compared to spaces dominated by extensive wooden pavement but minimal vegetation, and poorly maintained landscapes. These findings underscore the importance of incorporating natural elements and enhancing spatial experiences when designing restorative rooftop spaces. Third, subjective evaluations of green rooftop spaces positively influenced psychophysiological restoration. Notably, the correlation between subjective perceptions and PRS increases was stronger than that with changes in emotional factors. Among the physiological parameters, subjective perceptions showed the stronger positive correlation with the SBP decreases, followed by LF/HF decreases. These findings highlight the importance of prioritizing sensory experiences in the initial stages of designing green rooftop spaces to maximize their restorative benefits.

Our findings support a preliminary conclusion that green rooftop spaces have the potential to enhance both physiological and psychological well-being, with those characterized by a high degree of naturalness providing better restorative benefits. The combination of physiological and psychological measurements contributes to a more comprehensive understanding of the effects of green rooftop spaces on psychophysiological restoration and has reference value for the construction of restorative green rooftop spaces. The research findings offer a strategic direction for the sustainable development of building rooftops, contributing to enhancement of urban livability. However, this study primarily focused on visual perception, overlooking the potential influence of other sensory factors on environmental restoration. Future research should incorporate a multisensory approach to examine how sensory interactions influence restorative benefits under varying conditions.