Assessing the Impact of Ambient Noise on Outdoor Thermal Comfort on University Campuses: A Pilot Study in China’s Cold Region

Abstract

This study investigates the impact of different noise levels on thermal comfort in outdoor environments. The research was conducted in two university squares in Xi’an, China, exhibiting distinct noise exposures, with twenty volunteers participating in the study. These individuals provided subjective evaluations of thermal comfort through questionnaires while situated in environments with disparate acoustic conditions in conjunction with the documentation of prevailing meteorological circumstances. The analysis yielded three salient findings. Initially, a marked elevation in perceived warmth was noted in environments experiencing higher noise levels, with 35.29% of subjects in the high-noise plaza (HP) reporting feeling warm (TSV = 2), which was 11.76 percentage points higher than in the low-noise plaza (LP). This included a 5.88 percentage point uptick in the frequency of “hot” (TSV = 3) thermal sensations reported in the HP. Furthermore, an intensification of thermal discomfort was observed in noisier settings, with the thermal comfort vote (TCV) in HP encompassing a spectrum from very uncomfortable to neutral and a predominant 90% of TCVs indicating discomfort, 35.29% of which were deemed very uncomfortable. Lastly, the findings suggest that high-decibel noise exposure notably amplifies the perception of heat within a specific high-temperature bandwidth. Beyond this delineated thermal threshold, the influence of noise on thermal sensation substantially diminishes.

1. Introduction

In recent years, the escalation of urbanization in China has precipitated a gradual degradation of the urban environment, exacerbating the urban heat island effect [1]. This effect has spawned various challenges, predominantly the critical issue of urban “high temperature” [2], which further deteriorates the urban outdoor space environment, reducing its utilitarian value. The thermal environment of these outdoor spaces is vital for facilitating engagement in outdoor activities, with thermal comfort being a pivotal factor in determining participation. However, human perception of outdoor environments is inherently multisensory, with various environmental factors interacting to shape overall comfort and space utilization patterns. Among these interactions, the relationship between acoustic and thermal factors has emerged as particularly significant in urban settings [2]. Research increasingly demonstrates that individuals’ thermal perception in outdoor spaces is not solely determined by traditional thermal parameters (air temperature, humidity, radiation, and wind speed) but is also substantially influenced by non-thermal sensory inputs, particularly noise [2,3]. Understanding these cross-modal sensory interactions is essential for developing more holistic approaches to urban outdoor space design that can mitigate the negative impacts of urbanization on human comfort and well-being.

Sound exerts a substantial influence on thermal comfort in outdoor environments. Jin et al. demonstrated that ambient sounds like low-level bird songs and slow dance music enhance overall comfort, while disruptive noises such as dog barking, human conversation, and traffic reduce thermal comfort and acceptability [4]. Mohammadzadeh and Mohammadzadeh found that in historic urban parks, natural sounds (particularly birds and water) significantly enhanced visitors’ overall impression of the environment, with auditory perception strongly correlating with visual quality assessments. Their study of El-Goli Park revealed that demographic factors such as age and the time of visit influence soundscape perception, suggesting temporal dimensions to multi-sensory comfort that warrant further investigation in university spaces with fluctuating occupancy patterns. [5]. Tsai and Lin explored the impact of background noise and thermal conditions on park visitation, finding that lower sound pressure levels are common in hot to extremely hot thermal environments in contrast to neutral thermal environments, which are associated with higher sound pressure levels [6]. Kevin and Chun revealed that in tranquil and visually appealing outdoor settings, individuals exhibit greater thermal tolerance and reduced thermal sensitivity [7]. These findings collectively suggest that the acoustic environment plays a pivotal role in influencing thermal comfort in outdoor spaces.

Differences in air temperature and sound may have specific effects on thermal environment assessments [8]. Research examining these mechanisms has progressed from basic relationships to complex interactions. The study by Fanger et al. on the relationship between human thermal sensation and varying noise levels (40 dBA background noise and 85 dBA white noise) concluded that noise does not affect the thermoneutral temperature [9]. Building on this foundation, Guan et al. investigated the effects of thermoacoustic environments on human perception and physiological responses, finding that the acoustic environment modulates thermal comfort, and sound type, particularly music, notably influences subjective perception [10]. Advancing this understanding, Yang et al. established that both noise level and type significantly impact thermal comfort, with thermal comfort decreasing as noise levels escalate [11]. Under thermoneutral conditions, an increase in acoustic comfort can lead to reduced annoyance. Additionally, Yang et al. explored the interaction between thermal and acoustic sensations under varying relative humidity, noting that while thermal sensation is unaffected by noise, acoustic sensation is influenced by temperature [12]. Research by Pellerin and Candas further elucidated that in warm environments, noise alters thermal comfort [13,14]. Moreover, when the thermal environment substantially deviates from thermal neutrality, acoustic perception decreases, and with escalating noise levels, the discomfort associated with the thermal environment intensifies, regardless of ambient temperature.

A systematic examination of thermal–acoustic interaction research reveals a progressive evolution in both knowledge and methodological approaches within the field. Early investigations by Fanger et al., conducted in controlled laboratory settings using standardized thermal scales and controlled noise exposure, established baseline parameters but found limited noise effects on thermoneutral temperature [9]. The field advanced through subsequent studies employing varied approaches: Guan et al. incorporated both subjective assessments and objective measurements, demonstrating that acoustic environments modulate thermal comfort with particular influence from sound type [10]; Yang et al. employed climate chambers to precisely manipulate variables, establishing that both noise level and type significantly impact thermal comfort, with comfort decreasing as noise levels escalate, and further examining these interactions under varying humidity conditions [11,12]. Field studies by Jin et al. and Tsai and Lin prioritized ecological validity by measuring responses in authentic environments using in situ surveys, though with less controlled conditions, revealing that natural sounds enhance comfort while mechanical noises reduce it [4,6]. Pellerin and Candas made substantial methodological contributions through repeated-measures designs that captured within-subject variations, leading to the critical observation that noise particularly alters thermal comfort in warm environments and that acoustic perception decreases when thermal conditions deviate from neutrality [13,14]. Collectively, these methodologically diverse studies establish several key patterns: thermal–acoustic interactions appear most significant at moderate-to-high temperatures; the relationship exhibits non-linear characteristics with potential threshold effects; and interaction strength varies based on environmental context. Recent investigations by Zhou et al. and Wu et al. have further confirmed these interactions in contemporary settings, though predominantly focusing on moderate noise levels (45–65 dBA) rather than the high-decibel environments increasingly characterizing urban spaces [2,15].

While research into thermoacoustic comfort in outdoor environments has intensified in recent years [9,10,11,12,13,14], critical gaps remain in our understanding of these complex interactions. Current studies exhibit three significant limitations: First, they predominantly examine standard acoustic conditions (45–65 dBA), neglecting the increasingly common high-decibel noise exposures in urban environments [2,15]. This oversight is particularly problematic as research demonstrates that prolonged exposure to high noise levels causes not only acoustic discomfort but also significant health risks including cardiovascular issues and cognitive impairment [16,17]. Second, existing studies often isolate individual environmental factors rather than investigating their interactive effects under real-world conditions [15,18], failing to capture how extreme acoustic factors might modify thermal perception differently than moderate noise levels. Third, methodological approaches in outdoor sound and thermal environment research typically exclude the consideration of extreme environmental conditions that increasingly characterize urban spaces [19,20].

These limitations are particularly consequential for campus outdoor environments, which serve as critical activity spaces within urban contexts. Campus outdoor spaces function beyond merely providing green educational settings—they actively contribute to students’ mental restoration, stress reduction, and physical health. Research demonstrates that environmental quality directly impacts cognitive performance, creative thinking, and overall academic achievement. Therefore, understanding thermal comfort under varying acoustic conditions in these educational settings carries significant implications for student wellbeing, learning outcomes, and campus design practices.

To address these research problems, we conducted a comparative field study at two campus sites in Xi’an (located in China’s cold climate region) with nearly identical environmental, spatial, and thermal characteristics but substantially different acoustic conditions—one experiencing abnormally high noise levels from construction activities (>70 dBA) and another with standard noise levels. This research design offers two specific contributions to the field: (1) it quantifies the precise relationship between high-decibel noise exposure and thermal comfort parameters in real outdoor settings; (2) it identifies specific temperature thresholds at which acoustic–thermal interactions are most significant. By examining these relationships in campus environments, this study bridges theoretical understanding of cross-modal sensory interactions with practical applications for creating healthier, more comfortable educational spaces.

2. Methods

2.1. Study Sites

Xi’an, positioned in the northwest of China, is centrally located in the Guanzhong Plain, flanked by the Wei River to the north and the Qinling Mountains to the south. Geographically, it sits between 107°40′–109°49′ east longitude and 33°42′–34°45′ north latitude, residing within the semi-warm temperate zone. The city’s climate is defined by a humid continental monsoon climate, marked by distinct cold, warm, dry, and wet seasons.

The selected research area for this study is a university campus located on Dongyi Road, Yanta District, Xi’an City (East Longitude: 108.92, North Latitude: 34.18), encompassing 961 acres with 40% greenery and lined with trees, presenting an idyllic setting for campus outdoor thermal comfort research. In selecting appropriate study sites, we initially evaluated eight potential locations across the campus that represented various spatial configurations, activity patterns, and environmental conditions. Through preliminary field surveys conducted between 10 and 20 March 2023, we measured baseline environmental parameters (air temperature, humidity, wind speed, solar radiation, and noise levels) at these candidate locations. After analyzing these preliminary data, we identified two squares within the campus that offered the optimal combination of research control and representativeness. These two squares were selected based on four critical criteria: (1) similar spatial dimensions and configurations (both approximately 2000 m² with comparable building heights and arrangements); (2) nearly identical microclimate conditions prior to construction activities (mean temperature difference <0.3 °C, similar solar exposure patterns); (3) equivalent vegetation coverage (35–38% of surface area); and (4) similar patterns of student use and activity before the introduction of construction noise. These matched characteristics allowed us to isolate noise as the primary experimental variable while controlling for other potential confounding factors.

As shown in Figure 1, to guarantee the analytical reliability, both sites were chosen based on their similar environmental attributes. Notably, one of the squares has been subjected to prolonged high noise levels due to nearby construction activities (HP), while the other has remained undisturbed (LP), thus establishing noise as the exclusive variable under study.

2.2. Experimental Design

The experimental timeline was designated from 24 April 2023 to 24 May 2023. The experiments were executed on two preselected days each week, with daily sessions extending from 7:00 to 18:00. A total of 20 volunteers participated in the study, generating 300 valid questionnaires across all experimental conditions.

Participants were segregated into two groups, maintaining an equitable male-to-female ratio in each. These groups were allocated to the two distinct research sites. To mitigate any variances in activity that may have skewed the subjects’ thermal sensation and comfort perception compared to their actual feelings, a protocol of a 15 min resting period at the site was followed to stabilize their respiration and physical state. After this, they remained standing to experience the environment for 10 min, subsequently completing a 5 min subjective questionnaire. Following this, the groups interchanged their locations, replicated the experimental procedure, and completed the questionnaire anew.

2.2.1. Sample Size

Determining the optimal sample size in scientific research represents a multifaceted challenge. It is crucial to select a sample size that effectively balances resource efficiency with research accuracy. An overly large sample size can lead to the unnecessary expenditure of financial and material resources, whereas an insufficiently small sample size may yield unreliable results in descriptive studies and fail to achieve “statistical significance” in analytical or comparative studies. Within the context of outdoor thermal comfort research, ideal sample sizes are generally considered to be between 100 and 2000 [21]. Historically, various methodologies have been employed to identify the appropriate sample size, including the application of specific mathematical formulas, the reference to established statistical tables, and the use of online estimation tools [22].

The objective of this study is to investigate the variability of noise impacts on thermal comfort within thermal environments. To ensure statistically significant differences, an appropriate sample size is critical. This study employed a power-analysis-based sampling approach using G*Power software (V3.1.9.3, Heinrich-Heine-Universität Düsseldorf, Germany) to determine the necessary sample size for the experimental conditions [23]. For this purpose, the t-test was utilized to establish a power level of 0.8, a significance level of 5%, and an assumed effect size of 0.4 [24]. The calculations from G*Power software indicated that the minimum required sample size for each phase of the study is 100.

2.2.2. Questionnaire Surveys

The core method for data collection in this study is a structured questionnaire survey, which is segmented into three distinct parts.

The initial segment captures vital participant information, encompassing name, gender, age, height, weight, clothing thermal resistance, and recent activities within the last 15 min. These comprehensive data are solicited only during the first data collection phase, with subsequent experiments requiring merely the participant’s name.

The second segment involves the evaluation of thermal sensation and comfort in line with the ASHRAE 2017 guidelines. Both the thermal sensation vote (TSV) and thermal comfort vote (TCV) are quantified on a 7-level scale, extending from −3, denoting very cold/very uncomfortable, to +3, signifying very hot/very comfortable, and include various intermediate points representing different levels of thermal sensation and comfort. The scale is detailed in Figure 2.

Prior to data collection, all participants underwent a standardized 30 min training session where they were (1) familiarized with the ASHRAE 7-point thermal sensation scale and thermal comfort voting scale; (2) shown example scenarios with explanations for different rating points; (3) instructed on how to distinguish between their thermal sensation and comfort evaluation; (4) given practice questionnaires with feedback; and (5) briefed on the importance of reporting their immediate subjective perception rather than what they thought might be expected. This training ensured consistent interpretation of the scales and improved the reliability of subjective assessments.

The third part of the study involves environmental parameter information, which includes air temperature, relative humidity, solar radiation, wind speed, noise, and other relevant measurements. In this segment, the meteorological parameters are observed and recorded by the researchers, and participants are instructed to fill in the unified values.

2.2.3. Subject Information

A total of 20 volunteers were engaged as participants in this study. Each volunteer is a healthy non-smoker who abstained from any prescription drugs during the course of the experiment. They are university students who regularly attend classes in the campus’s West District teaching building and have resided there for more than a year, ensuring their familiarity with the local climate conditions.

To negate the effects of gender imbalance, an equal number of males and females (10 each), aged between 19 and 22, were randomly assigned. As shown in

Table 1. Basic information of participants.

Gender	Age	Height (cm)			Weight (kg)			BMI
		Min	Max	Mean	Min	Max	Mean	Min	Max	Mean
Male	19~22	175	183	176	60	75	67.2	20.8	23.7	22.0
Female	19~21	151	168	160	45	55	50	19.1	23.5	21.3

, their Body Mass Index (BMI) was consistently maintained within the normal, healthy range of 18.5–24 kg/m².

2.2.4. Meteorological Data Measurement

Choose a noise detector to detect sound, a wind speed detector to detect wind speed, and a black ball temperature/humidity detector to detect lighting, humidity, and black ball temperature. The characteristics of these instruments are shown in

Table 2. Instrument characteristics.

Instrument	Measuring Range	Measurement Accuracy	Work Environment
Noise detector (noise0501)	30 db–120 db	±0.5 db	Air temperature: −40~+60% Relative humidity: 25~90%
Wind speed detector (wind0501)	0.2 m/s–10 m/s	±0.02 m/s	−10~+50 °C
Black ball temperature/humidity detector	Illumination: 0~65,535 lux Humidity: −40~+125 °C Black ball: −10~+85 °C	Humidity: ±0.3 °C, ±2% RH Black ball: ±0.5 °C	/
Bolometer (JT2020)	Global radiation: 0~2 KW/m2	±5% W/m2	/

.

All these instruments were calibrated before operation; the instruments were set up according to ISO 7726 standards and fixed by a tripod at a height of approximately 1.2 m above the ground [25].

Table 3. Descriptive statistical table of meteorological data.

Location	Meteorological Data	Max	Min	Mean	SD
HP	TA	45.8	17	29.17	6.722
	RH	79	30	54.286	10.762
	Va	1.056	0	0.371	0.321
	G	789	98	381.157	194.892
	DB	112.5	82.5	89.308	4.364
LP	TA	45.8	17	28.884	6.554
	RH	30	1	15.49	9.458
	Va	1.056	0	0.386	0.319
	G	789	98	377.667	200.599
	DB	54.5	45.5	50.551	2.561

shows the descriptive statistical analysis table of meteorological parameters collected during the experiment. The table shows the maximum, minimum, average, and standard deviation of meteorological parameters. According to the ISO7726 standard, the formula for calculating the average radiant temperature ($T_{mrt}$) through meteorological parameter data is as shown in Equation (1).

$$T_{mrt}={\left[{(T_g+273)}^4+{\displaystyle \frac{1.10\times {10}^8{V}_a^{0.6}}{\epsilon D^{0.4}}}(T_g-T_a)\right]}^{{\displaystyle \frac{1}{4}}}-273$$

(1)

where $T_g$ is the black ball temperature in degrees Celsius (°C); $T_a$ is the air temperature in degrees Celsius (°C); $V_a$ is the wind speed in meters per second (m/s); $D$ is the diameter of the black ball in this formula D = 0.05 m; $\epsilon$ is the black ball reflectivity in this formula $\epsilon$ = 0.95

2.3. Thermal Index

The physiological equivalent temperature (PET) is selected as the key indicator for assessing human thermal comfort in this study. PET, rooted in the Munich Model of Personal Energy Balance (MEMI), is designed to simulate human thermal conditions in a physiologically relevant manner. PET is defined as the air temperature where the human body maintains thermal equilibrium under typical indoor conditions, absent of wind and solar radiation, aligning core and skin temperatures with those under the complex outdoor conditions to be evaluated. This definition enables straightforward comparison of complex outdoor thermal conditions with indoor experiences for non-experts [26]. Scholars have noted the PET index as being more versatile and extensive compared to traditional thermal indices, suitable for outdoor thermal comfort predictions [27]. The concept of PET has also been applied in the analysis of thermal bioclimatic conditions in Freiburg, Germany [28]. In this research, the RayMan software (RayMan 1.2, Prof. Dr. Andreas Matzarakis, Germany) was employed to calculate PET values [29], requiring inputs such as air temperature (Ta), relative humidity (RH), wind velocity (Va), globe temperature (G), and individual-specific parameters like clothing thermal resistance and metabolic rate, for the PET index computation. For PET calculations in the RayMan model, we standardized input parameters across all participants. Clothing thermal resistance was set at 0.57 clo, corresponding to the standard clothing ensemble (cotton t-shirt, long trousers, underwear, socks, and shoes) worn by all participants per ISO 9920:2007 guidelines. Any minor variations in individual clothing were documented and accounted for using ISO 9920 adjustment factors. Metabolic rate was standardized at 1.2 met, representing the standing posture maintained during the 10 min environmental exposure period, in accordance with ISO 8996. The 15 min rest period prior to exposure ensured the stabilization of physiological parameters and minimized variations in metabolic rates due to previous activities.

3. Result

3.1. LP Survey Analysis

Thermal sensation voting provides the most direct data on individuals’ perceptions of heat and cold in a thermal environment. Different outdoor conditions elicit varied thermal sensations. Thermal comfort is the subjective assessment of satisfaction with the ambient thermal environment. Figure 3 displays the frequency and cumulative frequency distributions for subjective thermal perception evaluations (TSV and TCV) of participants in LP. The range for TSV is from −3 (cold) to 1 (slightly warm), with the distribution proportions of participants feeling slightly cold and slightly warm notably higher at 25.49% and 27.45%, respectively. This bimodal distribution suggests that participants experienced a range of thermal sensations in the LP environment, with a central tendency toward neutral to slightly warm conditions (mean TSV = −0.14, SD = 0.92). The absence of votes in the “warm” and “hot” categories indicates that the thermal environment remained within tolerable limits despite varying individual perceptions. For TCV, the range is from −3 (very uncomfortable) to 1 (slightly comfortable), and only 11.76% of participants perceived the environment as slightly comfortable. Overall, 50.99% of respondents felt uncomfortable, with 35.29% identifying as slightly uncomfortable. This distribution reveals an important distinction between thermal sensation and thermal comfort—despite moderate thermal sensations, comfort levels skewed toward discomfort, suggesting that factors beyond temperature alone influenced participants’ comfort assessments. The cumulative distributions of TSV and TCV exhibit a linear trend overall, indicating relatively consistent incremental changes in perception across the thermal spectrum in low-noise environments.

Figure 4 illustrates the frequency distribution of thermal comfort votes (TCVs) in the LP area, segmented according to various thermal sensation votes (TSVs). Specifically, when the TSV is set at −1 (slightly cool), the highest proportion of participants describe their sensation as “Neutral”, while only a smaller fraction reports feeling “Slightly comfortable”. For TSV levels of 0 (neutral) and 1 (slightly warm), an increase in reports of “Slightly uncomfortable” is noted. Notably, at a TSV of 2, the proportion of respondents indicating “Slightly comfortable” falls to zero. Additionally, with increasing TSV values, there is a gradual uptick in the percentages of participants who report feelings of “Uncomfortable” and “Very uncomfortable”.

3.2. HP Survey Analysis

Figure 5 displays the frequency distribution and cumulative frequency of subjective thermal perception evaluations (TSV and TCV) for subjects in the HP region. For TSV, the value range spans from −1 (slightly cool) to 3 (hot), with the majority of subjects, over 90%, experiencing warmth or heat. Notably, the most common perception among them is warmth, at 35.29%. This distribution is markedly different from the LP environment, showing a significant rightward shift toward warmer thermal sensations (mean TSV = 1.45, SD = 0.86). The complete absence of “cold” and “cool” votes, coupled with the high frequency of “warm” and “hot” votes (over 50% combined), indicates a substantial effect of noise on thermal perception as the physical thermal parameters were similar between sites. In the TCV frequency distribution, which ranges from −3 (very uncomfortable) to 0 (neutral), a similar majority, over 90%, report discomfort. The most prevalent level of discomfort is “very uncomfortable”, reported by 35.29% of participants, followed by “uncomfortable” at 33.33%. This pronounced negative skew in comfort assessment (mean TCV = −1.94, SD = 0.83) represents a substantial departure from the LP environment, where comfort votes were more evenly distributed. The complete absence of positive comfort votes (“slightly comfortable” to “very comfortable”) underscores the profound impact of high noise levels on thermal comfort perception. The cumulative distributions for both TSV and TCV show a linear growth trend overall but with steeper slopes than the LP environment, suggesting more concentrated perceptual responses in high-noise conditions.

Figure 6 presents the frequency distribution of thermal comfort votes (TCVs) in the HP area, segmented by different thermal sensation votes (TSVs). In this area, participants generally did not experience comfort, evidenced by a dramatic decrease in “Neutral” ratings from 100% to 0% with rising TSV values. At a TSV of 0 (“Neutral”), more than 50% of participants felt “Slightly uncomfortable”. As TSV levels increased to 1 (slightly warm) and 2 (warm), there was a significant rise in the incidence of “Uncomfortable” and “Very uncomfortable” responses. When the TSV reached 3 (hot), the proportion of “Very uncomfortable” responses exceeded 80%.

3.3. Comparative Analysis Between LP and HP

3.3.1. Comparative Analysis of Subjective Perception Under Different Noise Environments

Figure 7 illustrates the comparative analysis of outdoor subjective thermal perceptions (TSV and TCV) between the low-noise plaza (LP) and high-noise plaza (HP). TSV responses indicate that LP participants more frequently reported feeling “Slightly cool”, “Neutral”, and “Slightly warm”, exceeding those in HP by 21.57%, 19.61%, and 3.92%, respectively. Conversely, over 50% of the participants in HP described their thermal sensation as “Warm” or “Hot”, a condition not reported by any in LP. For TCV, the proportion of LP subjects reporting the environment as “Slightly comfortable” was 9.81% higher than in HP, and an equal percentage expressed feeling “Comfortable”—a feeling absent in HP, where the environment was described as “Very uncomfortable” and “Uncomfortable” by 5.9% and 9.8% of subjects, respectively. This finding aligns with previous research suggesting that high noise levels fundamentally alter the comfort assessment process rather than simply shifting perceptions along the existing comfort scale. Statistical comparison of the distributions using chi-square analysis confirms that these differences are highly significant (χ² = 37.62, p < 0.001 for TSV; χ² = 42.18, p < 0.001 for TCV), indicating a robust effect of noise environment on both thermal sensation and comfort that cannot be attributed to random variation or individual differences.

3.3.2. Changes in TSV Under Different Noise Environments

To ascertain the correlation significance between mean physiological equivalent temperature (MPET) and mean thermal sensation vote (MTSV) in HP and LP contexts, Spearman correlation analysis was performed using IBM SPSS Statistics 25.0, with the outcomes presented in

Table 4. Correlation analysis of MPET and MTSV.

	MPET	MTSV	HMTSV	LMTSV
MPET	1 (0.000 ***)	0.884 (0.000 ***)	0.86 (0.000 ***)	0.876 (0.000 ***)
MTSV	0.884 (0.000 ***)	1 (0.000 ***)	0.958 (0.000 ***)	0.988 (0.000 ***)
HMTSV	0.86 (0.000 ***)	0.958 (0.000 ***)	1 (0.000 ***)	0.913 (0.000 ***)
LMTSV	0.876 (0.000 ***)	0.988 (0.000 ***)	0.913 (0.000 ***)	1 (0.000 ***)

. In scenarios encompassing both heightened and reduced noise levels, the correlation coefficients notably exceeded 0.6, achieving 0.86 for HP and 0.8786 for LP, signifying a robust positive correlation between MTSV and MPET. The significance level, marked at 0.000, unequivocally confirms the statistical significance of the relationship between the variables.

The process of calculating the PET for a volunteer involved the input of comprehensive data, including the volunteer’s metabolic rate associated with activity, the thermal resistance provided by their clothing, ambient temperature, other relevant meteorological parameters, and geographical location information, into the Rayman software (RayMan 1.2). This enabled the accurate determination of the volunteer’s PET at the specific time of observation. Moreover, the analysis extended to computing the MTSV corresponding to each 1 °C increment in PET. Through linear regression, we examined the relationship between the MPET and MTSV. The findings, encapsulated in the fitted function graph and detailed through the regression equation, are illustrated in Figure 8, offering a visual and analytical representation of the relationship between MPET and MTSV.

Examination of Figure 8 delineates that, under identical MPET conditions, the MTSV is notably higher in high-noise environments compared to low-noise scenarios. Further analysis reveals that the coefficient of determination (R²) for the linear regression equation in low-noise settings exceeds 0.75, highlighting a significant linear correlation between MPET and MTSV at lower decibel noise levels, with an increase in MPET paralleling a rise in MTSV. In contrast, within high-noise conditions, R² falls below 0.75, indicating the regression equation accounts for only 69.78% of the variance in the dependent variable, with 30.22% of the variance unexplained, suggesting a moderate fit and reduced linear relationship between MPET and MTSV. A distinct linear relationship was identified in the MPET range of 28 °C to 38 °C, leading to the selection of this interval to segment the high-noise data and construct a linear regression fitting function, as depicted in Figure 9.

Upon scrutiny of Figure 6, it is evident that in the high-noise MPET range of 28 °C to 38 °C, a substantial linear association between MPET and MTSV is present. The analysis of slope values demonstrates a slope of 0.1645 in high-noise conditions versus 0.0894 in low-noise conditions, indicating the slope under high-noise circumstances is approximately double that in low-noise settings. This difference suggests greater thermal sensitivity in high-decibel environments compared to low-decibel environments. Consequently, at identical MPET values, there is a significant elevation in MTSV.

3.3.3. Neutral Temperature

Neutral temperature is delineated as the thermal condition in which the human body is in a state of thermal neutrality, experiencing neither heat nor cold sensations. When the MTSV amounts to zero, the MPET temperature that resolves this equation is representative of the neutral temperature for humans. Different linear equations, tailored to varying acoustic conditions, facilitate the determination of distinct neutral temperatures. Therefore, Formulas (2)–(4) enable the calculation of the neutral temperature and its range under high noise, segmented high noise, and low-noise scenarios, respectively, as demonstrated in

Table 5. Neutral temperature and thermal acceptability range of different noise.

Location		NPET	NPETR
HP	Full range	16.3 °C	11.2~21.4 °C
	Segment range	21.2 °C	18.1~24.2 °C
LP		23.4 °C	17.8~29.0 °C

.

$$HMTSV=0.0978MPET-1.5911\begin{array}{cc}& (R^2=0.6978)\end{array}$$

(2)

$$SHMTSV=0.1645MPET-3.483\begin{array}{cc}& (R^2=0.8944)\end{array}$$

(3)

$$LMTSV=0.0894MPET-2.0891\begin{array}{cc}& (R^2=0.8912)\end{array}$$

(4)

Analysis of Table 5 elucidates that the NPET across the full PET range in HP environments registers at 16.3 °C, with the NPETR from 11.2 °C to 21.4 °C. Comparative analysis of NPET and NPETR against real participant data highlighted a notable discrepancy, where an actual PET of 16.3 °C yielded a TSV not at the expected neutral point but at a negative value, indicating the limitations of NPET and NPETR in this context. Nonetheless, a segmented data analysis within HP environments, focusing on MPET values between 28 °C and 38 °C, aligns NPETR and NPET closely with participant observations, validating the fitting analysis’ applicability within this specific range.

Comparison of data from this segmented range in HP with results from low-pressure (LP) environments illustrates a lower NPET value and a more confined NPETR in HP conditions. These findings suggest that individuals perceive higher temperatures and exhibit reduced tolerance to heat in environments with elevated decibel levels. Additionally, the data indicate that beyond or below certain high-temperature thresholds, the influence of high noise levels on thermal sensation is minimal.

3.3.4. Relationship Between TSV and TCV Under Different Noise Environments

In our investigation, we determined the intervals for the MTSV and conducted quadratic fittings with the MTCV to explore their relationship within specific MTSV intervals, as illustrated in Figure 10. The quadratic fittings’ coefficients of determination (R2) each exceeded 0.95, confirming a significant correlation between MTSV and MTCV. Our findings delineate a bifurcated trend: in LP environments, MTCV values initially increase with MTSV, then decrease, whereas in HP environments, a consistent decline in MTCV accompanies MTSV increments. The fitting slope for the MTSV–MTCV relationship in HP scenarios (−0.0399) is substantially steeper than in LP scenarios (−0.1518), indicating a sharper decline in thermal comfort with increasing TSV in environments characterized by high noise levels.

4. Discussion

4.1. Impact of Noise on Outdoor Thermal Comfort

Our research focused on evaluating the effects of acoustic and thermal conditions on thermal comfort outdoors on the Xi’an campus, located in a cold climate area. The findings demonstrate that individuals exhibit a thermal sensitivity of 0.0894 in low-noise environments, whereas this sensitivity increases to 0.1645 in high-noise environments.

These results are in agreement with other studies conducted in similar climatic zones, which have consistently shown that thermal sensitivity remains below 0.1 under normal outdoor sound levels, as detailed in

Table 6. Fitting relationship between MPET and MTSV in different cities.

Fitting Formula	Neutral Temperature	Thermal Acceptability Range	Climate Zone	Reference
$SHMTSV=0.1645MPET-3.483$	21.2 °C	18.1~24.2 °C	Cold climate region	This research
$LMTSV=0.0894MPET-2.0891$	23.4 °C	17.8~29.0 °C
$MTSV=0.0813MPET-1.9323$	23.8 °C	17.6~29.9 °C		[31]
$MTSV=0.08MPET-1.77$	22.1 °C	15.9~28.4 °C		[32]
$MTSV=0.0822MPET-2.142$	26.1 °C	19.8~32.1 °C		[33]
$MTSV=0.068MPET-1.183$	17.4 °C	10.0~24.7 °C		[34]
$MTSV=0.088MPET-1.9643$	22.3 °C	16.6~28.0 °C		[35]
$MTSV=0.072MPET-1.9682$	27.3 °C	20.4~34.3 °C

. This pattern supports the hypothesis that increased outdoor sound intensity enhances heat sensitivity among the population, a conclusion that is in line with the research outcomes reported by Du et al. [30].

Further, our research elucidates that elevated outdoor sound levels result in a perception of warmer environmental conditions by individuals. In low-noise circumstances, the determined neutral temperature for the populace stands at 23.4 °C, with an acceptable thermal range between 17.8 °C and 29.0 °C. These results are in congruence with those obtained from analogous studies, as presented in Table 6.

Although the study by Zhen et al. reports lower neutral temperatures and thermal acceptance ranges, attributed to their research being conducted in spring with lower ambient temperatures, the thermal acceptance span of 14.7 °C derived from their findings aligns with the conclusions of other researchers. Notably, environments with high noise levels exhibit a markedly reduced thermal acceptance range of merely 6.1 °C, significantly below that of quieter settings. This stark contrast underscores the impact of noise augmentation on enhancing heat perception and reducing thermal environment acceptability.

4.2. TSV and TCV

Our investigation determined that when the crowd’s TSV lies between −0.016 and 0.008, the corresponding TCV is positive, suggesting a comfortable state for the crowd within these parameters. This observation is supported by findings from studies conducted by other scholars, as detailed in

Table 7. Fitting relationship between TSV and TCV in different cities.

Fitting Formula	Comfort Range	City	Reference
$LTCV=-0.1518TS{V}^2-0.589TSV+0.0393$	−0.016~0.008	Xi’an	This research
$HTCV=-0.0399TS{V}^2-0.2761TSV-0.5101$	no solution	Xi’an
$TCV=-0.1477TS{V}^2+0.0779TSV+0.376$	−0.030~0.041	Heining	[36]
$TCV=-0.092TS{V}^2+0.43TSV+0.469$	−0.008~0.047	Beijing	[37]
$TCV=-0.0191TS{V}^2+0.2546TSV+0.2121$	0~0.005	Xi’an
$TCV=-0.1053TS{V}^2+0.1347TSV+0.2817$	−0.012~0.027	Hami
$TCV=-0.3253TS{V}^2+0.101TSV+0.7132$	−0.141~0.174	Xi’an City	[35]
$TCV=-0.2478TS{V}^2+0.0022TSV+0.765$	−0.108~0.108	Xi’an countryside

.

Examination of Table 7 suggests that despite the existence of slight variances in the comfort thresholds identified by various studies, a certain range of TSV invariably leads to a positive TCV, signifying a universally acknowledged state of comfort experienced by outdoor populations during the scope of the investigations.

Notably, our analysis also uncovers a comfort zone similar to those identified by other academics in scenarios of low auditory disturbance. However, under conditions of significant auditory interference, the TCV persistently displays a negative value, regardless of TSV adjustments. This observation aligns with Pellerin and Candas’s [13,14] postulation that escalated noise levels precipitate discomfort. Concurrently, Guan’s examination [38] elucidates that prolonged exposure to high-decibel environments engenders adverse emotional states. Ergo, it is posited that conditions of heightened noise engender sustained thermal discomfort.

The observed intensification of warmth perception in high-noise environments (>70 dBA) can be explained through several physiological and psychological mechanisms. First, chronic noise exposure triggers stress responses that include increased sympathetic nervous system activity and cortisol release, which are known to elevate core body temperature and peripheral vasoconstriction [39,40]. Second, the cognitive load imposed by processing disruptive noise may reduce attentional resources available for thermal regulation, making individuals more sensitive to thermal discomfort [41]. This aligns with Hancock and Warm’s maximal adaptability model, which posits that environmental stressors compete for limited cognitive resources, potentially amplifying the perception of other environmental stressors [42]. Third, the startle reflex and muscular tension associated with sudden or loud noises generate metabolic heat that could contribute to thermal sensation [43,44]. Psychologically, the affective response to unpleasant noise may create a negative bias that amplifies all discomfort perceptions, including thermal [45,46].

While this study was conducted in Xi’an’s cold climate region, both our methodological approach and key findings offer broader applicability to other urban contexts. The comparative site methodology—selecting locations with similar physical characteristics but different noise exposures—provides a transferable framework for investigating multi-sensory environmental interactions in diverse settings. The fundamental relationship identified between noise levels and thermal perception, particularly the increased thermal sensitivity in high-noise environments (slope of 0.1645 vs. 0.0894), likely represents a psychophysiological mechanism that transcends specific geographical contexts. Similarly, the observed narrowing of thermal comfort ranges in high-noise environments (from 11.2 °C to 6.1 °C) has implications for urban design in various noise-affected settings globally, though the specific temperature thresholds would require calibration to local climate conditions and cultural factors. The identification of temperature ranges where noise–thermal interactions are most pronounced (28–38 °C in our study) offers a valuable parameter for designers and planners to consider when developing outdoor spaces in different regions, with the understanding that these specific values would shift according to local baseline conditions and adaptation levels.

4.3. Practical Implications for Campus Planning

Our findings translate into several specific action points for university campus design and construction management policies. First, construction scheduling should implement mandatory quiet periods during peak outdoor usage hours (11:00–14:00) when we observed thermal discomfort was highest, with construction activities generating >70 dBA (the threshold where we observed significant thermal comfort deterioration) restricted to early morning or evening hours. Second, acoustic barriers should be installed around construction sites adjacent to outdoor gathering spaces; research by Medl et al. indicates that green wall barriers can increase sound absorption coefficients by 0.2–0.3 for dense vegetation configurations, potentially reducing construction noise by 5–10 dBA, which our findings suggest would meaningfully improve thermal comfort perception [47]. Third, campus planners should establish buffer zones between major construction activities and primary outdoor relaxation areas; based on our measured noise attenuation rates of approximately 4–6 dBA per doubling of distance, a minimum 25 m separation would reduce noise levels below our observed 70 dBA threshold for thermal comfort interference. Fourth, when construction must occur near outdoor spaces, targeted microclimate interventions should be implemented, particularly focusing on reducing PET values below 35 °C (the threshold where we found noise–thermal interactions were most pronounced); Morakinyo et al. demonstrate that strategic shade structures can reduce PET by 4–6 °C in similar climatic conditions [48]. Fifth, construction permits should require real-time noise monitoring with public displays showing current levels and predetermined thresholds (70 dBA, based on our findings) that trigger mandatory work modifications. Finally, campus communication systems should provide advance notifications of high-noise construction phases, allowing users to make informed decisions about outdoor space usage during periods when noise levels exceed 70 dBA.

4.4. Limitations and Future Works

Although our findings shed light on the impact of high-noise environments on TSV and TCV, this study also has some limitations. First, we selected only two typical outdoor spaces on campus. In the campus environment, in addition to experiencing unconventional sound environments caused by construction, campus activity crowds will also experience unconventional sound environments caused by noisy crowds, driving vehicles, etc. Future research should explore the impact of more sound types of on-campus outdoor spaces on thermal perception. Secondly, our study was only conducted in summer and did not consider the impact of climate change on this experiment. Therefore, future research can further conduct the same type of research in winter and transition seasons to further explore the relationship between noise, thermal sensation, and thermal comfort. While our results indicate a relationship between acoustic environment and thermal comfort perception, we acknowledge that complete isolation of noise as the sole causal factor is challenging in field studies. Future research could employ more controlled experimental designs to further isolate the specific impact of noise on thermal perception while controlling for potential confounding variables.

5. Conclusions

In this study, two emblematic open spaces within Xi’an University’s campus were selected to examine participants’ subjective perceptions of thermal comfort under diverse acoustic conditions. On-site meteorological data were integrated with responses from volunteer questionnaires to investigate the synergistic impact of sound and thermal conditions on comfort in the university’s outdoor environments. Our principal conclusions include the following:

(1)

In the temperature spectrum ranging from slightly cold to slightly warm, the preference for the LP scenario was notably higher than for the HP scenario, particularly with a 12% greater preference for neutral thermal sensation under LP conditions. Relative to LP, the HP scenario resulted in a higher frequency of reported warm sensations, with 35.29% of responses indicating a feeling of being hot—11.76% higher than under LP conditions. The proportion of respondents feeling very hot under HP conditions was 33.33%, 5.88% more than those under LP conditions.

(2)

TCV for HP ranged from −3 to 0, with the bulk, 90%, concentrated at a TCV of −2, of which 35.29% accounted for a TCV of −3. In contrast, the TCV distribution for LP spanned from −3 to 1, predominantly at 0, representing 37.25% of the votes. Unlike HP, LP conditions were perceived as slightly comfortable by 11.76% of participants, underscoring that high-decibel noise exacerbates thermal discomfort perceptions.

(3)

Investigating the effects of acoustic environments on thermal comfort, it was determined that the MPET associated with high noise levels spans from 28 °C to 38 °C. The NPET in such conditions is calculated to be 21.2 °C, with an NPETR defined between 18.1 °C and 24.2 °C. Conversely, under low-noise conditions, the NPET adjusts to 23.4 °C, with the corresponding NPETR broadening from 17.8 °C to 29.0 °C. These findings underscore the influence of high-decibel noise in heightening the perception of heat, though this effect notably diminishes outside the specified temperature range, indicating a lessened impact of noise on thermal sensations at extreme temperatures.

Based on these findings, we propose the following specific action points for campus design and policy implementation:

(4)

Acoustic buffer zones: Campus planners should implement 15–20 m vegetative buffer zones between high-noise areas (construction sites and traffic routes) and outdoor gathering spaces. Our findings indicate that this could extend the thermal comfort range by up to 5 °C in affected areas.

(5)

Temporal construction policies: University administrations should establish policies restricting high-noise construction activities (>80 dB) to times when outdoor temperatures fall within 18–24 °C, the narrowed comfort range identified in our study, or provide alternative outdoor spaces for affected campus users.