Outdoor Comfort Optimization in Historic Urban Quarters: From Multisensory Approaches to Operational Strategies Under Resource Constraints

Abstract

During the transition from urban expansion to renewal, optimizing environmental comfort under resource constraints presents critical challenges. While existing research confirms that multisensory interactions critically shape environmental comfort, these insights are rarely operationalized into protocols for resource-constrained contexts. Focusing on historic urban quarters that need to balance modification and preservation, this study quantifies multisensory (acoustic, visual, thermal) interactions and integrations to establish operational resource-optimization strategies. Through laboratory reproduction of 144 field-based experimental conditions (4 sound sources × 3 sound pressure levels × 4 green view indexes × 3 air temperatures), systematic subjective evaluations of acoustic, visual, thermal, and overall comfort were obtained. Key findings demonstrate: (1) Eliminating extreme comfort evaluations (e.g., “very uncomfortable”) within any single sensory domain stabilizes cross-sensory contributions to overall comfort, ensuring predictable cross-domain compensations and safeguarding resource efficacy; (2) Accumulating modest improvements across ≥2 sensory domains reduces per-domain performance threshold for satisfactory overall comfort, enabling constraint resolution (e.g., visual modification limits in historic districts); (3) Cross-domain optimization of environmental factors (e.g., sound source and air temperature) generates mutual enhancement effects, maximizing resource economy, whereas intra-domain optimization (e.g., sound source and sound pressure level) induces competitive inefficiencies. Collectively, these principles establish operational strategies for resource-constrained environmental improvements, advancing sustainable design and governance through evidence-based multisensory approaches.

1. Introduction

Comfortable outdoor environments profoundly affect people’s health and wellbeing [1]. With global urbanization progressing [2], outdoor comfort in cities has been intensively studied, providing key insights for creating high-quality built environments for urban users [3]. As urbanization approaches saturation in many countries, the focus of urban development has shifted from expansion to renewal. This is particularly evident in China, where rapid urbanization has created an urgent need to provide a living environment that meets modern standards of livability for its growing urban population [4]. Therefore, the renewal of old residential communities has emerged as a major national strategy aimed at improving public welfare, guiding urban transformation, and promoting sustainable development [5]. In addition, in line with China’s pursuit of an eco-civilization in the new era, fundamental principles have been established emphasizing resource efficiency and ecological benefits in urban and human settlement development [6]. Consequently, efficiently integrating available resources to enhance environmental quality and comfort has become a critical imperative.

For a long time, many comfort studies focused on isolated sensory domains, evaluating the relationships between comfort and physical environmental factors. Thermal comfort has received the most attention, with meteorological factors such as air temperature, mean radiant temperature, humidity, and wind velocity having been identified as influencing factors [7,8,9]. Studies on visual comfort have mainly focused on aesthetic elements and light conditions. Landscapes such as plants and water features are associated with outdoor visual comfort [10], whereas glare is associated with discomfort [11,12]. Regarding the acoustic domain, comfort evaluation is influenced by both physical and perceptual factors. On the physical side, for example, sound pressure level serves as one of the fundamental metrics, with international guidelines such as the WHO Environmental Noise Guidelines providing health-based recommended exposure levels for different noise sources [13]. On the perceptual side, soundscape research emphasizes the role of subjective experience in shaping acoustic comfort, highlighting the importance of sound source identification and listener context [14]. Empirical studies have further demonstrated how both physical metrics and human perception predict acoustic comfort [15,16,17]. Overall, these studies provide foundational insights for improving environmental quality within each sensory modality, serving as important references for both research and design practice. However, outdoor environments are highly complex. People are exposed to all those environmental factors simultaneously and usually do not distinguish their individual effects. Therefore, adopting multisensory approaches is essential for achieving a comprehensive understanding of outdoor comfort [18]. More importantly, recognizing the interactions and combined effects of multiple sensory domains enable potential sustainable solutions under resource constraints. Such implications and value were previously overlooked.

In the past decade, a few studies on urban outdoor comfort began evaluating multisensory interactions. Preliminary evidence on the combined effects comes mainly from subjective evaluation data. For example, Nitidara et al. [19] used the Structural Equation Model (SEM) and revealed that people’s subjective perceptions of acoustic, thermal, and visual domains interact in influencing overall comfort. Cureau et al. [20] found that qualitative data like people’s sensation, perception, comfort, acceptability, and preference in thermal, visual, and acoustic domains are more accurate than using physical environmental data to predict the overall comfort of urban parks. In terms of environmental factors, Du et al. [21] investigated the influence of visual (illumination intensity, LUX), acoustic (sound types and sound pressure level), and thermal (physiological equivalent temperature, PET) factors in urban residential outdoor spaces by collecting the sensation and comfort of the elderly. They found that acoustic and visual factors significantly affected the thermal comfort of the elderly; acoustic and thermal factors also significantly affected visual comfort or sensation. In addition, based on morphological and meteorological data, Li and Liu [22] proposed that visual and thermal factors not only directly influence thermal, aesthetic, and acoustic comfort but also modulate the contribution of the individual comfort domain to overall comfort. However, their onsite measurement did not account for acoustic factors. Moreover, recent research has incorporated both subjective and objective perspectives. For example, Xiao et al. [23] focused on a summer campus environment and established linear correlations between overall comfort and both subjective sensations (i.e., thermal, light, and acoustic sensations) and environmental quality parameters (i.e., PET, LUX, L_Aeq). Similarly, Nian et al. [24], in their study on autumnal urban riverfront spaces, found that both physical parameters and subjective comfort ratings related to the thermal and visual domains significantly influenced overall comfort. Overall, these studies confirm the interrelated nature of multisensory domains in shaping outdoor comfort, emphasizing the importance of multisensory approaches in outdoor comfort analysis. Nonetheless, they offered limited insight into operational principles for comfort optimization under resource constraints.

Against this gap, this study focuses on historic urban quarters. These areas preserve valuable and unique records of urban origins, including historic buildings, road networks, and local cultural heritage. However, many such neighborhoods face significant developmental challenges. In China, for instance, many urban areas developed during early urbanization were constructed without the guidance of modern planning principles [25], resulting in inadequate infrastructure and unsatisfactory environmental conditions. Furthermore, limited funding for renewal initiatives has delayed necessary improvements, contributing to the decline of these areas into disadvantaged neighborhoods [26]. Similar issues have also been observed in other Asian countries [27], indicating a broader regional pattern. Renewing these sites could serve as a powerful impetus for sustainable social, economic, and cultural development [28]. Notably, their inherent limitations (e.g., preservation laws limiting visual modifications [29]) make them ideal study cases for resource-optimization strategies. Furthermore, distinct morphological characteristics [30] and intangible features (e.g., cultural atmosphere, collective memory) [31] in historic urban quarters basically shape unique multisensory interactions. Consequently, context-specific interaction models become necessary, whereas results from existing studies (mostly on modern open spaces such as pedestrian streets [32], high-rise residential neighborhoods [21], urban parks [19,20], and squares [22]) lack direct applicability.

Several studies have been carried out on the environmental comfort in historic urban centers, with a predominant focus on outdoor thermal conditions. For instance, several studies have evaluated existing outdoor thermal comfort to identify the optimal time to visit such heritages. Examples include investigations in Isfahan, Iran [33] and the ancient plazas of Spain [34], both employing integrated approaches such as simulation, microclimatic monitoring, and questionnaire surveys. Other studies have examined the relationship between traditional urban morphologies and thermal comfort, underscoring the potential of microclimatic rehabilitation in urban heritage conservation. Notable examples include studies in Ghardaïa, Algeria, where street geometry factors such as height-to-width ratio and solar orientation were found to significantly affect thermal comfort [35]. Similarly, Li et al. investigated the impact of street morphology using numerical modeling and on-site measurements in a historic district in Zhenjiang, China [36]. In addition, passive strategies such as green and blue infrastructure configurations have been proposed to mitigate thermal discomfort without compromising heritage conservation [37]. While these studies provide valuable empirical and methodological insights, they predominantly address single-sensory aspects, particularly thermal comfort. Holistic comfort strategies are needed to address broader sensory perceptions and enhance overall environmental quality. This necessity is especially salient in the Chinese context, where historic districts exhibit pronounced diversity in climate, cultural practices, and local heritage characteristics.

Therefore, this study aims to quantify interactions between multisensory (acoustic, visual, thermal) comforts and between environmental factors in historic urban quarters, and critically, establish resource-optimization strategies that leverage cross-domain interaction effects to maximize outdoor comfort under constraints. By transforming observed interaction phenomena into operational protocols, this study seeks to advance multisensory approaches toward sustainable solutions.

2. Method

Outdoor comfort research employs diverse methodologies, including field surveys [19], in situ experiments [21], and case-based experiments [37], each offering distinct advantages depending on research objectives. In the present study, the historic quarter of interest faces lagging environmental governance and infrastructure renewal. Therefore, on-site environmental parameters and subjective evaluations may fall within a narrow range or remain at a low level, which can limit analytical insights. To explore a broader spectrum of sensory conditions and enable investigation of possible comfort improvements in such settings, this study utilized laboratory-based experiments, allowing for the extension of controllable parameters. Given the wide array of potential multisensory combinations, a preliminary exploratory survey was first conducted to identify the more influential perceptual factors within the specific historic district. This step ensured that the subsequent laboratory experiments remained both methodologically feasible and contextually relevant.

2.1. Study Case

The context of this study is a historic quarter in Xiamen, China, originally named Baijiacun and now known as Shentian community. It was built in 1927 to accommodate relocated residents [38]. As Xiamen’s first large-scale residential development, it pioneered early municipal construction and embodies the city’s modern urbanization roots. Over a century, its residential buildings preserved original architectural styles and layouts, and the initial road network remains intact. These spaces, cultural heritage, and daily landscapes retain deep significance for both local community and the rapidly evolving city. Today, this neighborhood is home to over 10,000 residents [39]. The area features a variety of functional spaces, including a market, restaurants, shops, as well as recreational venues such as parks and public squares, which collectively serve the daily needs of the community. Nevertheless, the neighborhood faces a critical renewal challenge: despite being located in downtown, the community’s housing, infrastructure, and outdoor environment fail to meet modern livability expectations. Crucially, unlike commercially driven heritage adaptations, renewals here must achieve living quality and comfort while preserving the authentic historic habitat concept, such as architectural integrity and cultural atmosphere. This purpose established Shentian community as an ideal case for developing environmental enhancement strategies under constraints.

2.2. Exploratory Surveys

To ensure that the laboratory experiments remained both feasible and contextually grounded, exploratory surveys were carried out targeting key perceptual elements across visual, acoustic, and thermal domains. Through sensory walks conducted across different seasons, subjective evaluations were collected from residents in typical outdoor settings. Corresponding audio-visual recordings were obtained to serve as experimental materials. Since the findings from the exploratory survey were used primarily to determine experimental conditions rather than to support core conclusions through systematic and comprehensive analysis, only a brief overview of the surveys and outcomes related to the experimental setup is provided here. Further details are available in Appendix A.

Exploratory surveys were conducted across three seasons (spring, summer, and winter). As the climatic comfort in Xiamen is generally comparable between spring and autumn [40], only one transitional season, spring, was included in the survey. Participants were recruited to take part in guided sensewalks, an experiential data collection method that immerses individuals in the environment and integrates multisensory experiences to form environmental evaluations [41]. Originally developed in soundscape research [42], this approach has now been extended to various perceptual dimensions. In this study, a route connecting ten representative outdoor spaces within the case study area was designed. A satellite map of the area, the sensewalk route, and photographs of the ten selected locations are provided in Appendix A (Figure A1).

At each location, participants evaluated four aspects of the environment: (1) comfort level, including acoustic, visual, thermal, and overall comfort; (2) acoustic perception, focusing on the loudness of different sound sources; (3) visual perception of landscape features and sunlight; and (4) thermal perception, including temperature, humidity, and wind speed. The survey generated 600 (3 seasons × 10 locations × 20 participants) on-site questionnaires.

In addition, environmental audio-visual information was collected as experimental materials. At each location, environmental sounds were recorded for 15 min using an ambisonics microphone (AMBEO VR Mic, Sennheiser, Wedemark, Germany) and a multi-track field recorder (F6 Multi-track Field Recorder, Zoom, Hauppauge, NY, USA) during the participant evaluations. To avoid capturing individuals in the imagery, panoramic photos were taken with a camera (Insta360 X3, Insta360, Shenzhen, China) at a height of 1.5 m only after the participants had departed. Furthermore, objective parameters about the visual, acoustic, and thermal environment were measured. It is important to note that these additional data were used solely for characterizing the environmental context and were not included in the subsequent analysis.

The exploratory surveys yield that the loudness of sound sources, plants, and temperature were the key environmental factors influencing single-domain and overall comfort in the historic quarter. Based on these findings, sound sources, sound pressure levels [SPL], green view indexes [GVI], and air temperatures [T_a] were adopted as independent variables for the experiment. Specific categories for each environmental factor were set according to their typical conditions in context and potential optimization levels. These categories were then combined to create various environmental conditions aimed at eliciting diverse comfort votes from participants during subsequent experiments.

2.3. Experimental Conditions and Study Design

An orthogonal design was employed to form experimental conditions. This resulted in a total of 144 unique conditions, formed by combining all levels of the acoustic (12 conditions: 4 sound sources × 3 SPLs), visual (4 GVIs), and thermal (3 air temperatures) variables.

Four representative sound sources, including natural, human, traffic, and construction sounds, were determined based on field questionnaire results and audio recordings. The three sound pressure levels [SPLs] tested were 40 dBA, 50 dBA, and 60 dBA, which were based on both field measurements and psychophysical principles. The levels of 50 dBA and 60 dBA were chosen as representative of typical acoustic conditions in the study area, as confirmed by exploratory surveys conducted across ten sites, which recorded average SPL ranges of 55.0–67.2 dBA in spring, 49.7–66.1 dBA in summer, and 52.9–61.6 dBA in winter. The inclusion of 40 dBA served as a potential optimizable target, representing a noticeably quieter environment. The 10 dBA interval between successive levels was determined based on established psychoacoustic evidence, wherein a 10 dBA increase is widely perceived as an approximate doubling of loudness [43]. This perceptually salient interval is effective in eliciting significant differences in subjective comfort evaluations in prior multisensory research [21], and 50 dBA as a perceptually neutral benchmark is also supported by an outdoor multisensory study [23]. This resulted in 12 acoustic conditions (4 sound sources × 3 SPLs) for the experiment. Audio clips dominated by each target sound source (natural/human/traffic/construction sounds) were selected from the 15 min field recordings. Through careful auditing, continuous 90 s segments were selected to ensure the dominance of the intended source. This duration was determined through a pre-experiment to be sufficient for participants to perceive the acoustic environment and form a stable comfort evaluation. These segments were edited using Adobe Audition software (version 24.2.0.83) to form the final stimuli.

The green view index [GVI] was used to quantify the proportion of green plants visible within the field of view [44]. Semantic image segmentation was applied to the panoramic images captured at the ten locations to calculate the fraction of green pixels. Based on this process, four representative visual conditions were identified, corresponding to GVI values of 0.15, 0.28, 0.47, and 0.60 (Figure 1).

Figure 1. The panoramic images as visual conditions.

For thermal conditions, air temperature [T_a] in the laboratory was adjusted to provide cool, neutral, and warm sensations. The predicted mean vote [PMV] model [45] was employed to determine the T_a values corresponding to these thermal sensations. As for key input parameters of the model, the metabolic rate, clothing insulation, relative humidity, and air velocity were set 1.0 met, 1.15 clo, 50%, and 0.2 m/s, respectively. Calculations identified 16 °C, 24 °C, and 32 °C as the tested temperatures.

The study adopted a within-subject design, where each participant experienced and evaluated all 144 conditions. To manage this, the conditions were divided into three rounds of experiments, with each round testing a single thermal condition. On each experiment day, participants evaluated 48 randomly presented audiovisual conditions. Each participant completed two additional rounds within half a month to test the remaining thermal conditions.

2.4. Environmental Reproduction

The experiment was conducted in a listening room with a controlled physical environment. To ensure accuracy and minimize interference, only one participant participated at a time. The size of the room was 3 × 5 × 3 m³. The participant sat at the center of the room.

During the experiment, video and audio were presented synchronously. To create an immersive experience of outdoor environments, field recordings were used to reproduce virtual audiovisual environments. Research has confirmed that this technology provides the most authentic environmental experiences among a series of reproduction technologies, and participants’ subjective evaluations collected using this method closely align with on-site ratings in outdoor environments [46,47]. The audiovisual recordings were sourced from exploratory surveys (see Section 2.1). To enhance immersive three-dimensional representations, audios that preserve the spatial acoustic characteristics were reproduced via headphones (ATH-M50x, Audio-Technica, Tokyo, Japan). The headphone volume was calibrated to the three SPLs (i.e., 40 dBA, 50 dBA, and 60 dBA) using a head and torso simulator (Type 4128C, Brüel & Kjær, Nærum, Denmark) and a sound level meter (Type 2270, Brüel & Kjær, Nærum, Denmark). Panoramic photos of the recorded scenarios were displayed using a head-mounted display (PICO 4 Ultra, PICO, Beijing, China) to support head-tracked virtual reality. The participant sat on a swivel chair to allow for 360-degree viewing.

The thermal environment was controlled with air conditioning and a humidifier [48]. A temperature and humidity sensor were placed at a height of 1 m near the participant’s seat to ensure real-time monitoring. Before the participant entered the room, the thermal environment parameters had adjusted to the desired levels and remained stable.

2.5. Experiment Process

After entering the listening room, the participants experienced a 30 min acclimatization period before starting the formal experiment [45].

Each trial began with a 30 s exposure to a specific audiovisual condition, after which the participant made the comfort vote. The acoustic comfort, visual comfort, thermal comfort, and overall comfort of the environment were rated using a 7-point scale (−3 = “very uncomfortable”; −2 = “uncomfortable”; −1 = “slightly uncomfortable”; 0 = “neutral”; 1 = “slightly comfortable”; 2 = “comfortable”; 3 = “very comfortable”). The participant took a break every 12 trials. Completing 48 trials required approximately 1.5 to 2 h. The overall experimental workflow, the process of a single trial, and the participant‘s setup are illustrated in Figure 2.

Figure 2. (a) overall experimental workflow, (b) one trial process, and (c) participant equipped with VR and headphones.

2.6. Participants

Since the PMV model is constructed based on the data from individuals with a normal Body Mass Index [BMI] [45], participants with a normal BMI were recruited for this study. A total of 35 college students (16 males) participated in the experiment (see Table 1), an acceptable sample size according to prior multisensory environmental studies [49,50]. Participants were instructed to wear a stipulated ensemble of clothing, which comprised a long-sleeved T-shirt, a jacket, long pants, socks, and shoes. The insulation value for each garment was assigned based on the ASHRAE Standard 55 [51]: a long-sleeved T-shirt (0.25 clo), a jacket (0.60 clo), long pants (0.24 clo), socks (0.02 clo), and shoes (0.04 clo). The total insulation for this ensemble was calculated to be 1.15 clo, equal to the clothing insulation value adopted in the PMV model. The total insulation provided by this clothing was approximately equal to the clothing insulation value adopted in the PMV model (1.15 clo) for calculating air temperature. In addition, participants were instructed to avoid strenuous physical exercise for at least 3 h prior to the experiment.

Table 1. Characteristics of the participants.

Gender		Age (Years)	Height (cm)	Weight (kg)
Women (n = 19)	Mean (S.D.)	21.8 (3.2)	176.4 (8.1)	74.6 (15.5)
Men (n = 16)	Mean (S.D.)	21.0 (2.6)	164.2 (5.8)	53.1 (8.9)

2.7. Statistical Analysis

Due to the absence of four participants during their second-/third-round experiment, 4407 valid questionnaires were collected in total.

Firstly, the effects of single-domain subjective evaluations (i.e., acoustic comfort, visual comfort, thermal comfort) and their interactions on overall comfort were analyzed using multi-way ANOVA. Follow-up Tukey’s HSD post hoc tests were performed to examine significant main effects. Simple main effects analyses were performed to break down significant interactions and identify conditions that effectively and consistently promote overall comfort.

Secondly, conditional probability analysis was employed to investigate cumulative effects for achieving overall comfort. Huang et al. [48] first introduced this method to evaluate the one-vote veto power of multisensory domains in their indoor comfort research, and Du et al. [21] later applied it to outdoor comfort research. Unlike these prior veto-focused applications, the current study adopted conditional probability analysis to quantify resource allocation efficiencies enabled by multisensory accumulation. Fundamentally grounded in probability theory, this method calculates the probability P(A|S), defined as the probability of event A occurring given the occurrence of premise event S. For this investigation, event A represented achieving comfortable overall comfort, operationally characterized by overall comfort ratings within the [1, 3] range. Premise events (S_t,n) corresponded to specific combinations of improvements across multisensory comfort domains, modeling distinct resource allocation strategies. As detailed in Table 2, comfort rating thresholds spanning from “uncomfortable (−2)” to “very comfortable (3)” were established for the lowest rating within each sensory domain. A binary classification then categorized each domain as meeting the threshold (1) or not meeting it (0) otherwise. Finally, the number of improved sensory domains was systematically expanded from zero to one, two, and ultimately all three domains to evaluate multisensory accumulation.

Table 2. Premise events (S_t,n) identified as combinations of improvements across multisensory comfort domains (acoustic, visual, and thermal comfort) for conditional probabilities analysis.

Thresholds	Uncomfortable (−2)	Slightly Uncomfortable (−1)	Neutral (0)	Slightly Comfortable (1)	Comfortable (2)	Very Comfortable (3)
Improvements (Classified by Thresholds)	0: [−3, −2) /1: [−2, 3]	0: [−3, −1) /1: [−1, 3]	0: [−3, 0) /1: [0, 3]	0: [−3, 1) /1: [1, 3]	0: [−3, 2) /1: [2, 3]	0: [−3, 3) /1: [3]
Premise events (St,n identified as combinations of conditions)	All three comfort domains improved
	S−2,1 = B1, C1, D1	S−1,1 = B1, C1, D1	S0,1 = B1, C1, D1	S1,1 = B1, C1, D1	S2,1 = B1, C1, D1	S3,1 = B1, C1, D1
	Two comfort domains improved
	S−2,2 = B1, C1, D0	S−1,2 = B1, C1, D0	S0,2 = B1, C1, D0	S1,2 = B1, C1, D0	S2,2 = B1, C1, D0	S3,2 = B1, C1, D0
	S−2,3 = B1, C0, D1	S−1,3 = B1, C0, D1	S0,3 = B1, C0, D1	S1,3 = B1, C0, D1	S2,3 = B1, C0, D1	S3,3 = B1, C0, D1
	S−2,4 = B0, C1, D1	S−1,4 = B0, C1, D1	S0,4 = B0, C1, D1	S1,4 = B0, C1, D1	S2,4 = B0, C1, D1	S3,4 = B0, C1, D1
	One comfort domain improved
	S−2,5 = B1, C0, D0	S−1,5 = B1, C0, D0	S0,5 = B1, C0, D0	S1,5 = B1, C0, D0	S2,5 = B1, C0, D0	S3,5 = B1, C0, D0
	S−2,6 = B0, C1, D0	S−1,6 = B0, C1, D0	S0,6 = B0, C1, D0	S1,6 = B0, C1, D0	S2,6 = B0, C1, D0	S3,6 = B0, C1, D0
	S−2,7 = B0, C0, D1	S−1,7 = B0, C0, D1	S0,7 = B0, C0, D1	S1,7 = B0, C0, D1	S2,7 = B0, C0, D1	S3,7 = B0, C0, D1
	None of the comfort domains improved
	S−2,8 = B0, C0, D0	S−1,8 = B0, C0, D0	S0,8 = B0, C0, D0	S1,8 = B0, C0, D0	S2,8 = B0, C0, D0	S3,8 = B0, C0, D0

Note: B, C, D, respectively, represent acoustic, visual, and thermal comfort. Subscript ‘_t’ of S_t,n reflect the threshold value. Subscript ‘_n’ of S_t,n reflect the specific combinations of comfort domain improvements (e.g., n = 1: all domains; n = 2: acoustic and visual; …). S_−1,2 = B₁, C₁, D₀, for instance, denotes a condition where the lowest ratings in both the acoustic (B) and visual (C) domains have been improved to the level of “slightly uncomfortable (−1)”.

Thirdly, multi-way ANOVA was used to analyze the effects of environmental factors (i.e., sound source, sound pressure level [SPL], green view index [GVI], and air temperature [T_a]) and their interactions to identify which specific factors significantly contribute to overall comfort. Follow-up Tukey’s HSD post hoc tests were performed on significant main effects. Simple main effects analyses were performed to break down significant interactions.

3. Result

3.1. Effects of Acoustic, Visual and Thermal Comforts on Overall Comfort

Significant main effects were found for acoustic, visual, and thermal comfort on overall comfort by ANOVA (Table 3). Voting on one comfort domain was positively associated with overall comfort (Figure 3). Tukey’s HSD post hoc tests showed significant differences between any two votes in one comfort domain, except between “very uncomfortable (−3)” and “uncomfortable (−2)” for visual comfort (p = 1.000).

Table 3. Multi-way ANOVA of the impact of multi-domain comforts on overall comfort.

	df	F	Sig.	ηp2
Acoustic comfort	6	221.857	0.000 ***	0.237
Visual comfort	6	121.819	0.000 ***	0.146
Thermal comfort	6	19.116	0.000 ***	0.026
Acoustic comfort × Visual comfort	36	3.406	0.000 ***	0.028
Visual comfort × Thermal comfort	35	2.659	0.000 ***	0.021
Thermal comfort × Acoustic comfort	36	3.016	0.000 ***	0.025

Note: Partial eta-squared (η_p²) values are reported as measures of effect size, representing the proportion of variance in the dependent variable attributable to each factor or interaction, after accounting for other effects in the model. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 3. Mean of overall comfort of the (a) acoustic, (b) visual, and (c) thermal comfort domain at different levels. The error bars represent 95% confidence intervals. n.s. indicates the comparison is not significant.

Significant interactions were also observed between two single-domain comforts, as presented in Table 3. Figure 4 breaks down these interactions. Follow-up analyses further examined the statistical significance of their simple main effects (Table 4).

Figure 4. Interaction effects between (a) acoustic × visual, (b) visual × thermal, and (c) thermal × visual comforts.

Table 4. Results of simple main effects analysis on the interaction effects of multi-domain comforts.

		df	F	Sig.	η2		df	F	Sig.	η2
	Visual comfort					Thermal comfort
Acoustic comfort	3	6	62.023	0.000 ***	0.080	3	6	33.763	0.000 ***	0.045
	2	6	158.652	0.000 ***	0.182	2	6	172.224	0.000 ***	0.195
	1	6	292.862	0.000 ***	0.291	1	6	129.163	0.000 ***	0.153
	0	6	151.490	0.000 ***	0.175	0	6	161.415	0.000 ***	0.185
	−1	6	93.865	0.000 ***	0.116	−1	6	138.268	0.000 ***	0.162
	−2	6	36.660	0.000 ***	0.049	−2	6	55.284	0.000 ***	0.072
	−3	6	6.479	0.000 ***	0.009	−3	6	13.880	0.000 ***	0.019
	Thermal comfort					Acoustic comfort
Visual comfort	3	5	27.221	0.000 ***	0.031	3	6	24.991	0.000 ***	0.034
	2	6	59.262	0.000 ***	0.077	2	6	43.147	0.000 ***	0.057
	1	6	69.885	0.000 ***	0.089	1	6	27.317	0.000 ***	0.037
	0	6	99.551	0.000 ***	0.122	0	6	37.792	0.000 ***	0.050
	−1	6	140.924	0.000 ***	0.165	−1	6	60.445	0.000 ***	0.078
	−2	6	41.785	0.000 ***	0.055	−2	6	50.680	0.000 ***	0.066
	−3	6	8.201	0.000 ***	0.011	−3	6	27.009	0.000 ***	0.036
	Visual comfort					Acoustic comfort
Thermal comfort	3	6	2.279	0.034 *	0.003	3	6	4.575	0.000 ***	0.006
	2	6	16.446	0.000 ***	0.023	2	6	13.890	0.000 ***	0.019
	1	6	16.891	0.000 ***	0.023	1	6	15.585	0.000 ***	0.021
	0	6	12.717	0.000 ***	0.018	0	6	14.879	0.000 ***	0.020
	−1	6	6.057	0.000 ***	0.008	−1	6	8.650	0.000 ***	0.012
	−2	6	4.920	0.000 ***	0.007	−2	6	6.473	0.000 ***	0.009
	−3	5	2.368	0.037 *	0.003	−3	6	3.650	0.001 **	0.005

Note: * p < 0.05, ** p < 0.01, *** p < 0.001.

As illustrated in Figure 4, most simple slopes are aligned in the same direction. However, the positive association between acoustic comfort and overall comfort fluctuated when visual comfort was rated as “very uncomfortable (−3)” (Figure 4a). Similarly, the positive association between visual comfort and overall comfort fluctuated when thermal comfort was rated as “very comfortable (3)” or “very uncomfortable (−3)” (Figure 4b). A comparable pattern was observed for the positive association between thermal comfort and overall comfort, which fluctuated when acoustic comfort was rated as “very comfortable (3)” or “very uncomfortable (−3)” (Figure 4c).

As shown in Table 4, simple effect analysis revealing that all simple main effects of acoustic, visual, and thermal comfort were statistically significant. Notably, the effect size of the simple main effects of acoustic, visual, and thermal comfort depended on the comfort ratings in other domains. Specifically, the smallest effects of any single-domain comfort occurred when another domain was rated as “very comfortable (3)” or “very uncomfortable (−3)”. For example, acoustic comfort exhibited medium (0.140 > η² > 0.060) or large (η² > 0.140) effect sizes for thermal comfort ratings within the range of [−2, 2], but only small effect sizes (0.060 > η² > 0.010) when thermal comfort was rated as “very comfortable (3)” (η² = 0.045) or “very uncomfortable (−3)” (η² = 0.019). This finding suggests that extreme evaluations (“very comfortable (3)”, “very uncomfortable (−3)”) in one domain attenuate the effects of other domains.

In summary, the stable and effective contribution of a single sensory domain to overall comfort requires avoiding extreme evaluations in other domains.

3.2. Accumulative Effects of Multisensory Comforts on Achieving Overall Comfort

Figure 5 illustrates the conditional probability of achieving overall comfort (rating within the range of [1, 3]) based on the occurrence of premise events (see Table 2). The horizontal axis under the heat map reflects progressive thresholds for incremental improvements in individual sensory comfort domains, and the vertical axis reflects the expansion of improved comfort domains. Analysis adopted the >50% probability of achieving overall comfort as a reference target to quantify how cumulative multisensory improvements reduce per-domain threshold requirements. Improvement measures were operationally defined as elevating the lowest evaluation ratings in a single domain, with baseline conditions retaining “very uncomfortable (−3)” lowest ratings as the methodological control. As presented in the bottom row of Figure 5, under such a baseline, the probability of achieving overall comfort remained consistently below 50%. Notably, even when peak ratings attained “very comfortable (3)” across all domains, the probability reached only 34.63%.

Figure 5. Probability of achieving a comfortable overall comfort (the overall comfort falls within the range of [1, 3]) when the premise event S_t,n in Table 2 occurs. A, V, T represent acoustic comfort, visual comfort, and thermal comfort, respectively.

Elevating the lowest evaluation ratings in a single domain showed the capacity to achieve the 50% probability target. Specifically, elevating the acoustic domain’s lowest rating to “comfortable (2)” yielded an 80.19% success probability, while comparable visual domain improvements achieved 68.62%. Expanding improvements to two domains substantially enhanced outcomes. For example, joint elevation of the lowest ratings in both acoustic and visual domains to “neutral (0)” increased the probability to 61.14%. Ultimately, improving all three sensory domains simultaneously exhibited maximal tolerance for the lowest evaluation ratings, maintaining 50.84% probability while requiring only “uncomfortable (−2)” ratings. These findings quantify a threshold-reduction effect whereby incorporating one additional domain enables a two-level relaxation in per-domain requirements while preserving >50% success probability.

In addition, among the three sensory domains, the acoustic domain proved to be the most efficient for improvement, followed by the visual domain, with the thermal domain being the least effective.

3.3. Effects of Multisensory Environmental Factors on Overall Comfort

As shown in Table 5, significant main effects were found for the sound source, sound pressure level (SPL), green view index (GVI), and air temperature (T_a) by multi-way ANOVA.

Table 5. Multi-way ANOVA of the impact of environmental factors on overall comfort.

	df	F	Sig.	ηp2
Sound source	3	669.045	0.000 ***	0.315
SPL	2	394.162	0.000 ***	0.153
GVI	3	116.019	0.000 ***	0.074
Ta	2	9.747	0.000 ***	0.004
Sound source × SPL	6	46.105	0.000 ***	0.060
Sound source × Ta	6	2.828	0.009 **	0.004
Sound source × GVI	9	1.714	0.080	0.004
SPL × GVI	6	1.042	0.396	0.001
SPL × Ta	4	1.899	0.108	0.002
Ta × GVI	6	0.216	0.972	0.000

Note: Partial eta-squared (η_p²) values are reported as measures of effect size, representing the proportion of variance in the dependent variable attributable to each factor or interaction, after accounting for other effects in the model. * p < 0.05, ** p < 0.01, *** p < 0.001.

Figure 6a,b showed that the influences of two acoustic factors were pronounced. Tukey’s HSD post hoc tests revealed that the mean values for overall comfort significantly differed between any two sound sources. Particularly, natural sounds scored the highest overall comfort, followed by human, traffic, and construction sounds. In addition, across all sound sources tested in this study, a lower SPL scored significantly higher overall comfort than a higher SPL. Similarly, a higher GVI always scored significantly higher overall comfort than a lower GVI (Figure 6c). Regarding the thermal factor, neutral temperature (24 °C) scored a significantly higher overall comfort vote than air temperatures that deviate from neutral conditions (i.e., 16 °C, 32 °C). However, there was no significant difference between overall comfort votes at the cool temperature (16 °C) and warm temperature (32 °C) (Figure 6d).

Figure 6. Mean of overall comfort for different categories of (a) sound source, (b) SPL, (c) GVI, and (d) T_a. The error bars represent 95% confidence intervals. n.s. indicates the comparison is not significant.

Table 5 also demonstrated statistically significant interactions between the effects of SPL and sound source, and between the effects of T_a and sound source. Figure 7 breaks down these interactions. Follow-up analyses further examined the statistical significance of their simple main effects.

Figure 7. Interaction effects between environmental factors, presented from complementary perspectives, including (a) SPL × sound source, (b) T_a × sound source, (c) sound source × SPL, and (d) sound source × T_a. Dashed lines indicate that simple main effects are not significant.

As shown in Table 6, SPL significantly affected overall comfort under human, traffic, or construction sounds, but not under natural sounds (p = 0.899). In addition, T_a significantly affected overall comfort only under natural sounds, but not under human (p = 0.314), traffic (p = 0.107), or construction sounds (p = 0.659). These differences can be easily visualized by Figure 7a,b.

Table 6. Results of simple main effects analysis on the interaction effects of environmental factors.

		df	F	Sig.	η2
	Sound source
SPL	Natural	2	0.107	0.899	0.000
	Human	2	170.592	0.000 ***	0.072
	Traffic	2	191.718	0.000 ***	0.081
	Construction	2	173.111	0.000 ***	0.073
	Sound source
Ta	Natural	2	14.446	0.000 ***	0.007
	Human	2	1.160	0.314	0.000
	Traffic	2	2.233	0.107	0.000
	Construction	2	0.421	0.657	0.000
	SPL [dBA]
Sound source	40	3	80.376	0.000 ***	0.052
	50	3	227.774	0.000 ***	0.135
	60	3	455.770	0.000 ***	0.238
	Ta [°C]
Sound source	24	3	295.116	0.000 ***	0.168
	16	3	209.196	0.000 ***	0.126
	32	3	178.569	0.000 ***	0.109

Note: * p < 0.05, ** p < 0.01, *** p < 0.001.

Simple main effects analysis also demonstrated that the sound source had significant effects on overall comfort under each level of SPL or T_a. However, its main effect changed with each level of the other environmental factors. The more the overall comfort benefited from SPL (mean of overall comfort: 40 > 50 > 60 dBA), the less it was affected by the sound source (η²: 0.052 < 0.135 < 0.238) (Figure 7c). Conversely, the more the overall comfort benefitted from T_a (mean of overall comfort: 24 > 16/32 °C), the more it was affected by the sound source (η²: 0.168 > 0.126 > 0.109) (Figure 7d).

4. Discussion

4.1. Eliminating Extreme Discomfort Guaranteeing Cross-Domain Compensation

Multisensory interactions were observed between subjective evaluations of individual domains. A notable finding was that when any single domain was rated as “very comfortable (3)” or “very uncomfortable (−3)”, contributions from other domains became unstable, manifesting as diminished or fluctuating effects. This finding is supported by Qin et al. [52], who investigated the effects of relative humidity and SPL on thermal comfort vote. Their results demonstrated that sound could cross-modally influence thermal perception, and that this effect diminishes under extreme thermal conditions (e.g., beyond a PET of 43 °C). This phenomenon offers both positive insights and warnings. On the one hand, pronounced advantages in one sensory domain may cause people to partially overlook deficiencies in others. Lau and Choi’s [32] study of high-density urban outdoor environments, where a perceived tranquil and attractive outdoor environment led to significantly lower thermal sensitivity during hot summers. On the other hand, pronounced disadvantages in one sensory domain can also inhibit compensatory effects from other domains. Such volatility invalidates resource investments in non-extreme domains, particularly under resource-limited conditions.

Consequently, a fundamental operational rule emerges for multisensory optimization strategies: resources should better first be directed toward domains exhibiting extreme discomfort, since their presence suppresses cross-modal compensation. Only after neutralizing these disruptive influences can investments in other domains yield reliable returns, establishing a cost-efficient safeguard for integrated sensory systems.

4.2. Accumulating Multisensory Enhancements Relaxing Per-Domain Threshold

Unlike veto-focused conditional probability analyses, such as Du et al.’s study [21], this research employed conditional probability analysis to quantify the resource allocation efficiencies enabled by multisensory accumulation. The results demonstrated a threshold-reduction effect: incorporating one additional sensory domain permitted a two-level relaxation in per-domain performance requirements while sustaining >50% probability of achieving overall comfort. In particular, expanding improvements to ≥2 domains substantially increased tolerance for lower individual comfort ratings while maintaining satisfactory overall comfort. For example, achieving >50% overall comfort probability requires only that the lowest ratings reach “neutral (0)” in both acoustic and visual domains.

These insights enable a pragmatic resource-reallocation strategy to bypass site-specific limitations. Specifically, accumulating modest improvements in operationally feasible domains delivers viable outcomes where modifications in constrained domains face prohibitive barriers. For historic districts governed by rigid visual preservation mandates, this approach resolves renewal challenges by accumulating achievable enhancements in acoustic and thermal domains instead of pursuing visual upgrades.

4.3. Optimizing Cross-Domain Factors Generating Mutual Reinforcement

In terms of environmental factors, cross-domain optimization of environmental factors generates mutually reinforcing effects on overall comfort. This phenomenon manifested concretely in the synergistic interaction between sound source types and T_a. Specifically, favorable sound sources amplified thermal comfort contributions to overall comfort, while comfortable thermal conditions similarly intensified acoustic benefits. For instance, only when the overall comfort benefited most from the sound source (i.e., natural sounds) was it significantly affected by T_a. Such thermal-acoustic reciprocities align with previous studies. For example, Li and Liu [22] proposed that the contribution of acoustic comfort to overall comfort strengthened as thermal conditions improved (measured by universal thermal climate index), confirming the same reinforcement pattern. Cureau et al. [20] also found that the effect of acoustic comfort on overall comfort was not significant in one park but showed significance in another with relatively high thermal acceptance. In the same vein, Du et al. [21] reported that acoustic comfort has a profound impact on overall comfort in spring (T_a: 13.88~28.81 °C), but its effects decrease in hot summer (T_a: 23.81~39.80 °C).

Conversely, sound source types and sound pressure levels (SPL) demonstrated mutually suppressive effects on overall comfort, suggesting that factors within a single sensory domain may compete for perceptual resources. For example, the more the overall comfort benefited from SPL (40 > 50 > 60 dBA), the less it was affected by the sound source (η²: 0.052 < 0.135 < 0.238). A plausible explanation is that exposure to environmental factors involving the same sensory domain triggers similar processing (e.g., auditory perception). When a salient favorable environmental factor is perceived, it may depress the sensation resources allocated to process other factors, thereby limiting their collective contribution.

These findings dictate a targeted investment strategy: prioritizing cross-sensory factor pairs (e.g., sound source × air temperature) over isolated intra-domain upgrades can maximize resource efficiency, particularly when budgetary flexibility exists.

4.4. A Multi-Sensory Strategy and Its Illustrative Applications

Collectively, these findings support a three-step approach for improving outdoor comfort under resource constraints. Firstly, addressing domains with extreme discomfort must be prioritized because persistent extreme disadvantages block predictable contributions from other sensory domains. Secondly, accumulating improvements across ≥2 sensory domains ensures viable overall comfort while resolving difficult constraints. Thirdly, when possible, optimizing environmental factors across different senses further boosts efficiency through mutual reinforcement effects.

The core strength of this strategy lies in providing a clear framework for understanding how different sensory domains interact to influence overall comfort. It thereby offers practical guidance for the integrated and optimized use of multi-sensory resources, ensuring that investments can be strategically directed to maximize enhancements in environmental comfort, directly aligning with the imperatives of resource efficiency and sustainable development in urban renewal.

In historical districts, where visual modifications are often restricted, the strategy supports a systematic process. It begins by assessing whether essential interventions are required to address extreme discomfort in any domain, for example, visual nuisances such as poorly managed garbage sites. Priority should be given to resolving these issues while preserving historical visual character. Subsequently, feasible strategies can be proposed for other modifiable sensory domains, such as acoustics and thermal comfort. In the case study presented here, both subjective evaluations and environmental factors confirmed that the acoustic domain had the strongest impact on overall comfort. This suggests acoustic improvement as a promising starting point. Interventions can even leverage district-specific features, such as narrow alleys and enclosed courtyards, to create urban quiet zones, or draw on local cultural context to design distinctive soundscapes. Well-designed soundscapes have also been shown to encourage pedestrian lingering [53,54], therefore support building urban slow spaces, aligning well with the livable and leisurely usage desired in historical settings [54]. Finally, the proposed interventions should be further evaluated to estimate their contribution to overall comfort and their ability to compensate for project constraints, ultimately confirming their effectiveness and economic efficiency. Additionally, although this study focuses on a historic quarter, the approach is also adaptable and could be flexibly adjusted to various common urban renewal constraints.

4.5. Limitations and Future Research

While the study presents valuable insights, several limitations should be acknowledged.

Firstly, the tested environmental conditions were sourced from a specific historic district case study. The relative influence of sensory domains on comfort may vary with cultural, geographic, and urban morphological factors. Therefore, generalizing these results to all urban settings should be done with caution. For instance, although the selected SPLs reflect typical acoustic conditions within this specific context, they do not cover higher noise levels (e.g., 70–80 dBA), which are common in other urban settings. Future research incorporating a broader range of diverse historical contexts is needed to validate and refine these observations and explore improvement strategies that perform under more challenging conditions.

Secondly, although high-fidelity audiovisual environments were reproduced in the lab, ecological validity remains a concern. For example, to maintain a consistent metabolic rate for the application of the PMV model, participants remained seated throughout the experiments. This posture does not fully replicate the dynamic nature of real-world outdoor behavior. Moreover, the T_a were set based on the PMV model, which was originally developed for indoor environments and may not fully reflect the complexity of outdoor thermal experience (e.g., solar radiation). Future research should aim to develop more ecologically valid thermal exposure protocols for lab study and incorporate realistic human movement patterns.

Thirdly, incorporating more environmental factors and individual variables would provide a more comprehensive understanding of outdoor comfort and broaden the scope of potential applications. For example, incorporating quantitative morphological characteristics (e.g., street height-to-width ratios or building heights) could better account for microclimate effects [35,36]. Furthermore, similar to many studies in the multisensory comfort domain [55], our participant sample was relatively homogeneous, consisting primarily of younger adults, which may limit the generalizability of the results to broader populations. Future studies would benefit from including more diverse demographic groups.

5. Conclusions

This study investigated outdoor comfort in historic urban quarters through a multisensory experimental approach, establishing resource-optimization strategies that leverage cross-domain interactions to maximize comfort under constraints. The findings support a practical three-step approach for resource-constrained improvements: addressing extreme discomfort to stabilize systems; then distributing improvements across accessible domains to bypass constraints; finally, targeting cross-factor synergies to maximize economy, with key empirical insights corresponding to each step are summarized below:

(1) Addressing domains with extreme discomfort must be prioritized because extreme evaluations (e.g., “very uncomfortable”) in any single sensory domain block predictable contributions from other sensory domains. Conversely, without extreme ratings, contributions from other sensory domains to overall comfort remained stable.

(2) When domain-specific modifications face prohibitive barriers (e.g., visual preservation limits in historic districts), shifting resources to feasible domains (like acoustics and thermal comfort) enables feasible solutions. Improving two or more sensory domains simultaneously increased the probability of overall comfort and reduced reliance on high thresholds in any single domain.

(3) Cross-domain integration outperforms isolated optimization for resource allocation economy because environmental factors within the same domain often competed in their contribution to overall comfort (e.g., sound source vs. SPL in acoustics). In contrast, cross-domain factors (e.g., sound source × air temperature) exhibited mutually reinforcing effects, supporting integrated over isolated optimization.