Skip to Content
MDPIMDPI

Advanced

SustainabilitySustainability
PDF
  • Article
  • Open Access

23 August 2025

Spatial Variations in Urban Outdoor Heat Stress and Its Influencing Factors During a Typical Summer Sea-Breeze Day in the Coastal City of Sendai, Japan, Based on Thermal Comfort Mapping

and
1
School of Architectural Engineering, Tongling University, 1335 Cuihu 4th Road, Tongling 244061, China
2
Department of Architecture, Faculty of Architecture, Tohoku Institute of Technology, 35-1 Yagiyamakasumicho, Taihaku Word, Sendai 982-8577, Miyagi, Japan
*
Author to whom correspondence should be addressed.
Sustainability2025, 17(17), 7627;https://doi.org/10.3390/su17177627
This article belongs to the Section Air, Climate Change and Sustainability
Version Notes
Order Reprints

Abstract

Sea breezes alleviate coastal heat stress via cooling and humidifying. Sendai, Japan, in 2015 had a population of 1.08 million and an area of 786 km2. Integrating the WRF model with RayMan, this study employs the PET index to assess spatiotemporal distributions of thermal comfort and heat stress, and their influencing factors, on typical summer sea-breeze days in Sendai, Japan. Results indicate that in the coastal zone, PET was primarily regulated by air temperature (Ta) and relative humidity (RH). In contrast, wind speed was the dominant influence on urban/inland zones, with Ta and RH contributing more during the evening. Sea breezes markedly improved the thermal environment in the coastal zone, suppressing PET increases. PET in urban and inland zones exhibited an initial rise followed by a decline, with the inland zone experiencing sustained extreme heat stress for 3 h. Among regions experiencing extreme heat stress, inland zones showed the highest proportion (17.75%), while coastal zones had the lowest (2.14%). Proportions across the three zones were similar under nighttime conditions with no thermal stress, with the urban zone exhibiting a slightly lower proportion. This study provides a theoretical basis for climate-adaptive urban planning leveraging sea breezes as a resource.

1. Introduction

Over recent decades, urban expansion has markedly accelerated, with over half the global population currently residing in urban areas [1]. Human activities contribute significantly to urban overheating, which stems from the combined effects of global warming and the urban heat island (UHI) phenomenon [2]. Based on the urban thermal balance theory, the UHI can be attributed to the combined influence of one or more factors, including diminished land cover, enhanced solar energy absorption by heat-storing materials, increased anthropogenic heat emissions, and constrained urban airflow [3]. Numerous studies have demonstrated that global warming [4], rapid urbanisation [5], and the UHI effect [6] collectively exacerbate the deterioration of the urban thermal environment, significantly impacting thermal comfort perception of residents and amplifying heat-related health risks [7]. The 2022 State of the Global Climate report of the World Meteorological Organization further underscores the imperative for decisive international action to strengthen the response to climate change, especially given the significant detrimental effects of associated phenomena—including extreme weather events, UHI effects, and air pollution—on urban thermal comfort. In response, urban planners and designers are pursuing diverse initiatives to mitigate these impacts [8,9]. This phenomenon has also spurred growing scholarly interest in research on outdoor thermal comfort (OTC) [10].
Outdoor human comfort is a critical metric for assessing urban liveability. Specifically, thermal comfort is defined as a subjective psychological state reflecting an individual’s satisfaction with the surrounding thermal environment. Unlike ambient air or land surface temperature, which quantify only the physical thermal properties of the environment, thermal comfort involves human perception and satisfaction with the thermal environment [11]. This perceptual state is comprehensively influenced by three primary factor categories: physical (e.g., air temperature, solar radiation, relative humidity, and wind speed), physiological (e.g., metabolism, activity level, age, and gender), and psychological (e.g., lifestyle, habits, and comfort expectations). Furthermore, OTC not only directly impacts the physical and mental health of individuals [12] but also influences their outdoor behavioural patterns. It also plays a crucial role in promoting outdoor activities, social interactions [13], and sustainable urban development [14]. Moreover, OTC is widely recognised as one of the most significant environmental factors affecting resident participation in outdoor activities [15].
Research on urban outdoor thermal comfort (OTC) combines field measurements [16] and numerical simulations [17] to improve microclimatic environments. Simulations use tools like ENVI-met5.7, RayMan1.2, and SOLWEIG1.0. Thermal indices like PET (physiological equivalent temperature) [18], UTCI (Universal Thermal Climate Index) [19], SET (Standard Effective Temperature) [20], perceived temperature, predicted mean vote/predicted percentage dissatisfied, effective temperature, and wet-bulb globe temperature [21] assess comfort and heat stress. Despite diversity, these indices primarily use four meteorological inputs: air temperature (Ta), relative humidity (RH), wind speed (v), and mean radiant radiation (G), with PET, UTCI, and SET being most widely applied [22].
Global OTC studies span South America [23], Europe [24], Africa [25], Asia [26], and Australia [27]. Factors influencing individual thermal perception are well studied [28], with microclimate-dependent physical factors being most critical [14]. Ta and v are identified as the most decisive meteorological variables [29], though impacts vary by climate zone [30]. Landscape elements (e.g., water bodies, vegetation type/density) also differentially influence OTC in urban spaces [31].
Owing to the ease of maritime transport and effects of economic agglomeration, a major proportion of global economic activities is concentrated in coastal regions. This concentration has led to over half of the global population (approximately three billion people) living in urban areas within 200 km of the coastline [32]. In these zones, the sea breeze, a characteristic mesoscale circulation system driven by the land–sea thermal contrast, arises from the differential heat capacity between water bodies and terrestrial surfaces. The high heat capacity of the ocean moderates rapid daytime temperature increases, whereas urban surfaces and construction materials efficiently absorb and store substantial amounts of solar radiation [33,34]. This thermal disparity propels the inland transport of cooler marine air, establishing a local wind system that frequently develops during summer. This inherent temperature regulation capability constitutes the underlying physical mechanism by which sea breezes ameliorate the thermal environment in coastal cities.
Recently, coastal cities have increasingly started to acknowledge the pivotal role of sea breezes in lowering ambient temperatures and mitigating the UHI effect [35]. Relevant research has robustly demonstrated this effect. Sasaki et al. [36] combined numerical simulations with field observations from Japan and confirmed the high efficacy of sea breezes in alleviating summer heat. Papanastasiou et al. [37] observed that sea breezes induced a rapid 4.0 °C temperature drop during heatwaves in Athens, Greece. Similarly, Kolokotsa et al. [38] found that temperatures in Chania, Greece, were 4 °C lower on days with westerly sea breezes than on days with northerly winds, as ancient walls obstructed sea-breeze penetration, thus hindering UHI mitigation. Zhou et al. [39] further quantified the cooling intensity of sea breezes in Adelaide, Australia. These insights highlight the versatility of sea breezes as a natural cooling resource, with particularly pronounced cooling effects in wide-open areas such as coastlines [40].
Through its cooling and humidifying effects, sea breezes significantly mitigate human heat stress and create more comfortable thermal environments in coastal cities [41]. Consequently, the cooling effect of sea breezes in coastal areas and their influence on OTC have drawn increasing attention. In recent years, extensive investigations have been performed on OTC in various climatic zones worldwide, particularly in renowned coastal cities and regions such as Shanghai [42], Hong Kong [43], Nagoya [44], Melbourne [45], Singapore [46], and Umeå [47]. Multiple studies [48,49,50] have reported complex interactions between sea breezes and the UHI phenomenon, demonstrating that the sea-breeze cooling effect effectively reduces UHI intensity and enhances urban OTC. For example, Emmanuel and Johansson [51] reported that the absence of a sea breeze during hot afternoons directly caused uncomfortable thermal conditions in Colombo, Sri Lanka. Similarly, Lopes et al. [52] demonstrated that sites influenced by a sea breeze generally experienced higher thermal comfort levels during daytime on typical days in Madeira Island, Portugal. Researchers have used various thermal indices for quantitative assessment: Papanastasiou et al. [37] studied the combined impacts of sea breezes on air temperature, aerosol levels, and resident thermal stress along the east coast of central Greece; subsequently, Anjos et al. [53] explored the relationship between sea breeze front advancement and PET variations in Sergipe, northeastern Brazil. However, the impact of sea breezes on OTC in coastal cities exhibits spatial heterogeneity, which is primarily governed by coastal proximity and urban morphological features. Guo and Zhao [54] revealed a critical threshold of 2.5 km beyond which the cooling effect of the ocean via sea breezes on urban areas may gradually attenuate or dissipate—an effect attributed to modifications in urban morphology and land cover patterns. This highlights the imperative for site-specific OTC studies within the shoreline zone, which is a designated region of intensive and multifaceted human activities [55]. At the macro scale, the effectiveness of wind (including sea breezes) at alleviating UHIs, and thereby improving OTC, is fundamentally controlled by urban morphological attributes. Specifically, morphological parameters including building density, layout, and orientation significantly modulate wind velocity and air temperature distributions, with building density playing a dominant role in localised temperature elevation [33]. He et al. [56] conducted a systematic analysis of the interconnections among urban form, ventilation efficiency, the UHI effect, and outdoor thermal comfort in a dense high-rise district of coastal Sydney, Australia (Cfa, Köppen climate classification). Their research focused on the evaluation of district-level ventilation performance and its associations with urban morphology, the UHI effect, and relative humidity while examining the facilitative and restrictive roles of sea breezes as a natural cooling mechanism in densely built environments.
However, research on the impact of summer sea breezes on outdoor thermal comfort (OTC) in coastal cities remains limited, particularly regarding environmental drivers and their spatiotemporal heterogeneity. This study aims to address the following fundamental questions:
(1) How do thermal comfort indices and heat-stress levels spatiotemporally vary along coastal–urban–inland gradients during typical sea-breeze days in Sendai, Japan?
(2) Under summer sea-breeze conditions, how do environmental factors differentially regulate thermal comfort indices across coastal gradient zones (coastal, urban, and inland areas)?
To resolve these questions, we integrated the WRF and RayMan models to quantify spatiotemporal proportions of heat-stress levels across distinct coastal-distance zones, elucidate zonal regulatory mechanisms of key environmental factors on OTC, and ultimately provide theoretical foundations for climate-resilient urban planning in coastal cities.

2. Data and Methods

2.1. Study Area

Sendai, Japan experienced a total population growth from 0.086 to 1.08 million and an expansion in land area from 17.5 to 786 km2 between its establishment as a city in 1889 and the year 2015 [57] (Figure 1). Notably, statistical analyses indicate that the annual mean temperature of Sendai has risen at a rate of 2.3 °C per century alongside urban expansion and population growth. As a coastal city, judicious utilisation and scientific analysis of its diverse geographical environment and unique topographical conditions can contribute to improving urban thermal comfort [58,59]. Sendai has a humid subtropical climate and is classified as Cfa under the Köppen climate classification system [60]. The terrain is bounded by mountains to the west and the Pacific Ocean to the east, with its elevation gradually increasing from the eastern coast toward the western mountains and with a relative relief of approximately 200 m. This study subdivided the research area into three zones based on distance from the coast (Figure 1): the coastal (0–7 km from the coast), urban (7–15 km from the coast), and inland (more than 15 km from the coast) zones.
Figure 1. Study area and computational domains. The arrow direction indicates the ‘sea breeze’ direction defined in this study, which represents the prevailing wind direction during summer in the research area. (a) Map of Japan; (b) range of each nesting domain (For details, see Table 1); (c) satellite imagery of Domain 3 showing the coastal zone (0–7 km from the coast), urban zone (7–15 km from the coast), inland zone (>15 km from the coast), and sea breeze direction.
Table 1. Computational parameterisation of the WRF model.

2.2. Determination of Study Day

For this study, we defined days with southeasterly winds as “sea-breeze days”. The screening criteria required simultaneous fulfilment of the following meteorological conditions: sunshine duration ≥ 40%; no rainfall; cloud cover < 5%; and during the period from sunrise to sunset, the existence of persistent and stable onshore winds lasting for at least 2 h. Meteorological data were obtained directly from the official website of the Japan Meteorological Agency. The Sendai local meteorological station is located in the urban area of Sendai, approximately 9 km from the coast (38°15′43.60″ N, 140°53′49.52″ E). The representativeness of 5 August 2016, selected as a representative sea-breeze day in this study, has been verified through multi-dimensional meteorological data comparison in previous studies [61]. Based on the sea-breeze arrival time and meteorological conditions, the designated study day was 5 August 2016 (Figure 2).
Figure 2. Meteorological parameters for 5 August 2016, including sunshine duration, temperature, wind direction, and wind speed.

2.3. Thermal Comfort Index

2.3.1. Physiological Equivalent Temperature Calculation

PET is defined as the equivalent air temperature at a specific location (indoors or outdoors) that would elicit the same core and skin temperatures in the human body as those achieved in a standard indoor environment under thermal equilibrium conditions. PET is an index developed based on the Munich Energy Balance Model for Individuals (MEMI). As one of the most commonly used thermal comfort indices, PET serves as a standard tool for evaluating thermal environmental changes predicted in urban and regional planning studies [62,63]. MEMI is a thermo-physiological model grounded in the human thermal equilibrium equation. This model incorporates all fundamental thermoregulatory processes, such as peripheral vasoconstriction and vasodilation as well as physiological sweating rate.
The advantage of PET lies in its unit, degrees Celsius (°C), which enables people from diverse backgrounds to intuitively assess human thermal comfort or heat stress induced by the outdoor thermal environment. The calculation procedure comprises the following steps:
(a) For a given set of meteorological parameters, the MEMI model is employed to compute the human physiological thermoregulatory response (including mean skin temperature and core body temperature).
(b) The calculated mean skin temperature and core body temperature values from step (a) are inputted to the MEMI model under the following standard reference conditions: V = 0.1 m/s, water vapor pressure = 12 hPa, Tmrt equated to Ta, and Ta as determined from the solution of the energy balance equations.
(c) The Ta obtained in step (b) represents the PET value.
The RayMan model, which is widely adopted for PET calculation, is a thermal–environmental assessment tool developed according to the guidelines of the German Association of Engineers. This model accurately quantifies radiative fluxes in urban built environments by processing parameters including TA, RH, cloud cover, temporal data, surface emissivity, and solid angle factors. To derive PET values under specific thermal conditions, the model requires meteorological input parameters (TA, radiation intensity, V, RH, cloud cover, temporal data, and geographical coordinates) and individual physiological attributes (age, sex, body weight, height, activity level, and clothing thermal resistance). In this study, the input parameters comprised TA, RH, radiation intensity, and V. All data were directly obtained from WRF model simulations or calculated from simulation outputs. PET values were computed using standardised physiological parameters (age: 35 years, height: 1.75 m, metabolic rate: 80 W/m2, clothing insulation: 0.9 clo, weight: 75 kg, and sex: male).

2.3.2. Source Data Acquisition

The meteorological input parameters for the RayMan model in this study were obtained from the output of the Advanced Research Weather Research and Forecasting (ARW-WRF) model, which was developed by the National Centre for Atmospheric Research and the National Centres for Environmental Prediction (NCEP) [64]. The ARW-WRF model is widely employed in short-term weather forecasting and in atmospheric process and long-term climate simulations. The simulation period and core physics configurations of the model used in this study are presented in Table 1. The simulation data have been validated [65]. The model was validated through Bias, Root-Mean-Square Error (RMSE), correlation coefficient, and error analysis. The results indicate a significant correlation between simulated and measured values. However, higher urbanization levels and periods closer to midday exhibited relatively greater Bias, RMSE, and errors in the simulation results.
The meteorological data required for the RayMan model include Ta, solar radiation, v, and RH. These parameters can be directly simulated or derived using the WRF model. The WRF output variable T2 represents the thermodynamic temperature (in K) 2 m above ground level, which can be converted to the air temperature data required by the RayMan model. The downward shortwave flux at ground surface (SWDOWN variable) from the WRF output directly serves as the surface solar shortwave radiation input for the RayMan model.
The WRF model provides 10 m eastward wind speed (U10) and 10 m northward wind speed (Q10). The 10 m horizontal wind speed, V, is calculated using the Pythagorean theorem:
V = U 10 2 + Q 10 2
RH is defined as the ratio of the actual vapor pressure in the air to the saturation vapor pressure at a given temperature.
Surface flux data from the NCEP reanalysis dataset does not include the RH parameter directly. Therefore, RH must be calculated based on specific humidity (q; Pa units), air pressure (p; Pa units), and air temperature (T2; K units) using the following formula:
R H = 100 w w s ≈ 0.263 p q e x p 17.67 ( T 2 − T 0 T 2 − 29.65 − 1
where w is vapor pressure (Pa), ws is saturated vapor pressure (Pa), and T0 is the reference temperature (typically 273.16 K).

3. Results and Discussion

3.1. PET Distribution Map

3.1.1. Spatiotemporal Variations in PET

We calculated fifteen sets of PET data at hourly intervals during sea-breeze activity (07:00–21:00). As depicted in Figure 3, illustrating the hourly average PET values across the study area and the three zones during sea-breeze activity, along with corresponding average PET curves, the coastal zone exhibited markedly lower PET values than the other zones from 08:00 onward, particularly during 08:00–12:00. In the coastal zone, PET reached 36.8 °C at 08:00, remained around 37 °C until 12:00, and declined after 13:00. Conversely, the urban zone peaked at 42.5 °C at 10:00, followed by a continuous decline. The inland zone recorded the latest PET peak (44.3 °C) at 12:00, representing the highest value observed throughout the study period.
Figure 3. Curves showing the hourly variation in mean PET for the entire study area and the coastal, urban, and inland zones.
We calculated the PET for all simulated points and visualised spatiotemporal patterns of PET across the study area through spatial mapping (Figure 4). The distribution maps clearly demonstrate that the coastal zone consistently maintained lower PET values during all observation periods, whereas areas with high PET values concentrated in the western and northern regions. The western high-PET zone, located within the dashed circle on the western side of the PET (08:00–14:00) map in Figure 4, migrated westward starting at 08:00, reaching its peak intensity between 10:00 and 12:00 with minimal displacement from 11:00 to 12:00. PET values declined progressively after 12:00, and this thermal core completely dominated the study area by 17:00. The northern high-PET zone advanced from the coastal zone through the urban core starting at 08:00, forming a stable thermal core in the northern region by 10:00. At 11:00, PET values continued rising, with northward migration until the zone exited the study area at 13:00. After 15:00, following the dissipation of the western and northern thermal cores, elevated PET values persistently clustered in urban built-up areas, consistent with the previously documented heat-trapping effect in high-density building zones [57].
Figure 4. Spatial distributions of physiologically equivalent temperature (PET), air temperature (TA), wind speed (V), relative humidity (RH), and solar radiation (G) in the study area.

3.1.2. Relationship Between Thermal Comfort Indices and Environmental Factors

The environmental factors examined in this study include Ta, RH, v, wind direction, and solar radiation. Figure 4 illustrates the spatial distribution patterns of variations in PET, Ta, v, wind direction, RH, and solar radiation across the study area at fifteen time points.
At 07:00, all environmental factors registered low values. However, regions with lower relative humidity showed higher PET values. By 08:00, localised zones near urban coastal areas demonstrated both low relative humidity and elevated PET. Concurrently, solar radiation intensified, with higher levels observed in the western inland region. Here, offshore winds (land-to-sea circulation) contributed to increased PET, a pattern that persisted until 14:00.
At 10:00, onshore winds established a consistent landward transport with gradually increasing velocity. Nevertheless, the northern inland area experienced weak sea breeze penetration and low v, leading to a sustained high Ta and PET until 15:00. PET decreased significantly only after sea breezes pervaded the entire study area. This may be due to the fact that the sea breeze brings cooler and more humid air, effectively reducing the thermal load on the environment, thus lowering the PET.
From 15:00 onward, diminished solar radiation coincided with elevated PET in high-temperature and low-humidity zones. These findings demonstrate substantial correlations between environmental factors and PET, with distinct spatial heterogeneity. Therefore, we conducted zonal correlation analyses on the coastal, urban, and inland zones to specifically explore relationships between different environmental factors and PET in these areas.
Figure 5 shows the correlation heatmaps between the thermal comfort indices and three environmental factors (TA, RH, and V) across the three study regions. The correlations of each respective factor with the others within each region are discussed in detail below.
Figure 5. Correlation heatmaps between thermal comfort indices and the three environmental factors (TA, V, and RH) across the three study zones (coastal, urban, and inland). Significance tiers: **** (p < 0.0001), *** (p < 0.001), ** (p < 0.01), * (p < 0.05), ns (p ≥ 0.05).
(1) Relationship Between Air Temperature and Thermal Comfort Indices
Ta maintained a statistically significant positive correlation with the thermal comfort indices throughout the coastal zone with a monotonically increasing trend (Figure 5). By 13:00, the correlation coefficient (r) exceeded 0.9, and this strong correlation persisted until 21:00, peaking at 19:00 (r > 0.99). In the urban zone, these parameters exhibited a positive correlation from 09:00 onward, showing a general upward trend and reaching its maximum at 18:00 (r > 0.99). In the inland zone, a positive correlation developed starting at 12:00, with a progressively strengthening correlation also peaking at 18:00 (r > 0.99). In summary, the correlation between TA and PET increased across all three zones, with the coastal zone consistently exhibiting the strongest correlation, followed by the urban zone, and being weakest in the inland zone. Notably, all three zones demonstrated exceptionally high correlations (r > 0.98) after 17:00.
(2) Relationship Between Relative Humidity and Thermal Comfort Indices
As illustrated in Figure 5, thermal comfort typically decreases under high-humidity conditions, with a significant correlation observed between them, reaching statistical significance at 13:00 with a correlation coefficient exceeding −0.8. This strong correlation persisted until 19:00 before sharply diminishing. In the urban and inland zones, the correlations remained weak prior to 17:00 but intensified thereafter. At 18:00, peak correlations were observed in all three zones, with correlation coefficients exceeding −0.9.
(3) Relationship Between Wind Speed and Thermal Comfort Indices
As shown in Figure 5, significant correlations (r > −0.8) were observed in the coastal zone from 07:00 to 10:00, with the strongest correlation at 07:00 (r > −0.9). In both the urban and inland zones, high correlations (r > −0.8) were maintained from 07:00 to 16:00, except at 13:00 in the urban zone (r > −0.7). In the urban zone, the peak correlation occurred at 10:00 (r > −0.9), whereas in the inland zone, it was observed at 15:00 (r > −0.9). Overall, the coastal zone exhibited the weakest correlation between v and the thermal comfort indices, whereas the urban and inland zones demonstrated comparable, statistically significant stronger correlations.
Overall, Ta and the thermal comfort indices exhibited statistically significant positive correlations (r > 0.9) in all three zones during the period of 17:00–21:00. In the coastal zone, these parameters demonstrated stronger correlations, particularly in the early morning and afternoon. RH showed significant negative correlations (r < −0.8) with the thermal comfort indices during 17:00–19:00 across all three zones, taking values below −0.9 at 18:00. The coastal zone exhibited notably higher correlations, especially prior to 17:00. Wind speed, v, was negatively correlated with the thermal comfort indices in all zones. Consistently high correlations (mean r ≈ −0.8) were observed in the urban and inland zones, whereas the coastal zone displayed relatively lower correlations, except during the early morning hours.

3.1.3. Quantitative Analysis of Factors Influencing Thermal Comfort Based on XGBoost and SHAP

To explore the impacts of various environmental factors on thermal comfort, this study constructs a thermal comfort prediction model based on extreme gradient boosting (XGBoost) and Shapley Additive exPlanations (SHAP) and realizes the quantitative analysis of influencing factors through interpretable machine learning techniques.
XGBoost, developed by Chen et al. in 2015 [66], is an efficient ensemble learning method. It optimizes classifiers based on the gradient boosting framework and is characterized by high prediction accuracy, high computational efficiency, and low cost [67]. Meanwhile, it possesses certain interpretability, thus gaining extensive application and achieving remarkable results in numerous fields [68,69]. In the field of machine learning, this method is often used to model the relationship between influencing factors and response variables (such as thermal comfort status in this study).
SHAP, developed by Lundberg and Lee (2017) [70], is an interpretable machine learning model. Its core lies in quantifying the contribution of each feature through a feature attribution method, thereby cumulatively deriving the final prediction result. Specifically, SHAP adopts a game-theoretic approach to attribute the prediction output to its input features, providing a cohesive and theoretically grounded method for local explanations [71]. This technique not only clarifies feature contributions but also enhances the transparency and trustworthiness of complex models [72]; in this study, SHAP is used to evaluate the significance of environmental factors to thermal comfort. SHAP values facilitate an in-depth understanding of feature contributions by assigning each feature an importance value for a specific prediction. In terms of interpretation, SHAP analysis has both global and local explainability, helping to understand the contribution of individual input features to specific predictions. In terms of the nature of influence, a negative SHAP value indicates that the feature has a negative impact on thermal comfort, while a positive value indicates a positive impact. In terms of feature value magnitude, in SHAP analysis, variables are ranked by importance; red values represent larger feature values, and blue indicates smaller feature values. The colour bars on the right side of the figures serve as legends, clearly illustrating the correspondence between colour gradients and related feature values.
To evaluate the positive and negative impacts of different environmental factors on the thermal comfort index, this study investigated the SHAP values of the three zones throughout the entire study period (07:00–21:00) and visualized these SHAP values within the study area across the three zones during the aforementioned time periods (for details, see Appendix Figure A1) to assess their contribution to the thermal comfort index. Figure A1 illustrates the marginal impacts of various factors in the model, where each dot represents an independent data point. The SHAP scatter plots in the figure provide a detailed visualization of feature importance and their impact on the predictive model. Meanwhile, the influence weights of various factors on the thermal comfort index are depicted through mean absolute SHAP values, with higher mean absolute SHAP values indicating a greater impact of the feature on the model output. The existence of multiple factors with different influence weights highlights the multifaceted nature of thermal comfort.
From the distribution of all SHAP values (Figure A1, Figure A3 and Figure A4), it can be observed that air temperature (TA) consistently exhibits a positive impact, while wind speed (V) consistently shows a negative impact across all time periods and zones. Specifically, as temperature increases, SHAP values rise, corresponding to higher physiological equivalent temperature (PET) values (lower comfort); conversely, as wind speed increases, SHAP values decrease, corresponding to lower PET values (higher comfort). This is consistent with the physiological responses of the human body to thermal environments. Relative humidity (RH) has an insignificant impact on PET, with only a weak influence during certain time periods in the coastal zone and even lower influence at all times in the urban and inland zones.
In this study, the mean absolute SHAP values of each environmental factor across all study time periods in the three zones were summarized separately to interpret the differences and variations in the influence weights of each environmental factor on the thermal comfort index in each zone (Figure 6). In the coastal zone, wind speed (V) was the key factor affecting the thermal comfort index during 7:00–10:00, with the highest importance score, after which air temperature (TA) became the most important factor. In the urban and inland zones, wind speed (V) acted as a key factor influencing the thermal comfort index for a longer duration, from 7:00 to 15:00. Overall, wind speed (V) had the strongest impact on the thermal comfort index in the inland zone, with the longest duration of strong influence; air temperature (TA) exerted a high and stable impact on the coastal zone throughout the entire study period, while its strong impact on the urban and inland zones was mainly concentrated in the evening. Relative humidity (RH) had a very low impact on the thermal comfort index in the urban and inland zones, and its impact on the coastal zone was also limited, with the highest absolute SHAP value of 0.352 observed at 14:00.
Figure 6. Hourly mean absolute SHAP values across the three study zones (coastal, urban, and inland) throughout the study timeline (07:00–21:00). (1) Coastal zone; (2) urban zone; (3) inland zone.

3.2. Heat Stress Distribution Map

3.2.1. Heat Stress Classification

Different ranges of PET values correspond to distinct levels of human thermal sensation. To explicitly interpret PET results, this study classified PET values according to the criteria adopted for thermal sensation grading shown in Table 2.
Table 2. Human thermal perception classes. Based on PET ranges, different thermal sensation and heat stress levels were defined, which are represented by distinct colours.

3.2.2. Spatiotemporal Characteristics of Heat Stress Classification

PET values across the study area were classified according to Table 2 and supplemented by stacked histograms (Figure 7), with hourly mapping (Figure 8) providing the spatiotemporal analysis. Extreme heat stress (perceived as “very hot”) primarily occurred during 8:00–16:00. Its emergence in the study area commenced at 8:00, showing a continual spread that peaked at 10:00 (45.1%). Severe heat stress (“hot”) persisted from 7:00 to 16:00, with notably high prevalence (> 80% at 13:00 and 14:00) from 8:00 to 15:00. Moderate heat stress (“warm”) spanned 7:00–17:00; higher areal proportions occurred in the early morning (7:00–8:00) and late afternoon (15:00–17:00), reaching ~80% at 7:00 and 16:00, while remaining below 20% from 8:00 to 14:00. Slight heat stress (“slightly warm”) was observed at 7:00 and 16:00–21:00. Its coverage increased steadily from 16:00, reached a maximum (86.9%) at 18:00, and subsequently declined. No thermal stress (indicating full comfort) spread during the period of 18:00–21:00. Its areal proportion exhibited persistent increase, culminating at 81.5% by 21:00.
Figure 7. Proportional distribution of heat stress classifications at hourly intervals (7:00–21:00) throughout the study area.
Figure 8. Spatial distribution of thermal perception (TP) at hourly intervals (7:00–21:00) across the study area.
Figure 8 reveals the spatiotemporal distribution of heat stress. Except for localised extreme heat stress occurring near the coastal and urban zones at 08:00, the main areas of extreme heat stress were concentrated closer to the northern and western parts of the study area, with a higher proportion observed in the north. At 07:00, only a small, sporadically distributed proportion was present within the study area. Between 08:00 and 17:00, severe heat stress dominated most areas, except for sporadic moderate heat stress areas in the coastal zone and extreme heat stress near the northern and western regions. Subsequently, the severe heat stress zone gradually shifted inland. At 07:00, moderate heat stress prevailed across the remaining areas (accounting for 78.5%), except for slight heat stress in the west and otherwise sporadic severe heat stress. By 08:00, the coverage of moderate heat stress decreased and became primarily distributed in the coastal and western areas. From 09:00 to 14:00, moderate heat stress persisted in the coastal zone. Starting at 15:00, moderate heat stress began penetrating inland from the coastal zone, its coverage increasing rapidly until it fully pervaded the inland zone by 17:00. Slight heat stress, distributed across the western part of the study area at 07:00, reappeared in the coastal zone at 17:00 and gradually penetrated inland. By 18:00, its coverage had spread throughout the entire study area. Starting at 19:00, the no thermal stress zone, which had emerged in the coastal zone at 18:00, expanded outward, centring on the urban zone and forming an enfolding pattern around the slight heat stress zone. Concurrently, the extent of slight heat stress within the urban zone progressively diminished and was eventually replaced by no thermal stress.
Overall, during the study day, the western part of the study area generally exhibited higher thermal comfort levels between 07:00 and 09:00. From 09:00 to 12:00, lower thermal comfort was observed in the northern part, while the coastal zone consistently showed higher thermal comfort. The distributional characteristics at 19:00–21:00 indicated that the urban zone consistently displayed lower thermal comfort than other zones. These results suggest that the distance from the coast may significantly influence thermal comfort in coastal cities. The following discussion further elaborates on the distinct spatial zoning characteristics of urban thermal comfort.
The average PET values for all simulated points within the inland area, urban area, and coastal area of the study region were calculated (Table 3, Figure 9). According to the heat stress classification indicated by the color codes in Table 3 (based on Table 2), extreme heat stress primarily occurred in the inland area (10:00–12:00) and the urban area (10:00–11:00). The onset of no thermal stress in the urban area was delayed by approximately one hour compared to the other two areas. Relatively, the duration of strong heat stress in the coastal area was shorter, with no occurence of extreme heat stress. Conversely, the duration of moderate heat stress in this area was relatively longer.
Table 3. Average PET values for sub-regions of the study area (colour coding based on heat stress classification in Table 2).
Figure 9. Thermal perception curves of the coastal, urban, and inland zones during the study period.
Analysis of the average PET curves for the three zones reveals that the curves closely coincided during the hours of 07:00–08:00 and 18:00–21:00. Throughout the period of 09:00–11:00, the heat stress classification in the coastal zone remained relatively stable, whereas it continuously increased in both the urban zone and inland zone. This resulted in substantial differences in heat stress classification levels among the urban, inland, and coastal zones during this period and the subsequent 5 h (approximately 09:00–14:00). Specifically, the temperature difference between the inland and coastal zones was larger during 10:00–14:00, reaching a maximum value of 6.58 °C at 11:00. Similarly, the temperature difference between the urban zone and coastal zone was larger during 09:00–11:00, peaking at 5.18 °C at 10:00.
Thus, it was further observed that during sea-breeze days, the OTC in coastal areas was substantially higher than that in urban and inland areas from 9:00 to 19:00, with the period between 10:00 and 12:00 being particularly notable.

3.2.3. Temporal Distribution of Heat Stress Levels in Different Zones

To further analyse the temporal distribution characteristics of heat stress levels within the three zones shown in Figure 9 and evaluate the conditions in each zone in detail, the time proportions during which different heat stress levels dominated in each zone during the study period (07:00–21:00) were calculated (Figure 10). As shown in Figure 10, on sea-breeze days, statistically significant differences were observed in the time proportions of extreme heat stress and moderate heat stress levels among the three zones. The time proportions of extreme heat stress were 17.75, 10.83, and 2.14% in the inland, urban, and coastal zones, respectively, indicating a considerably lower proportion in the coastal zone. Conversely, the time proportions of moderate heat stress were 18.72, 20.14, and 31.09% in the inland, urban, and coastal zones, respectively, demonstrating a notably higher proportion in the coastal zone. The time proportions of no heat stress showed minimal variation across the three zones, with values particularly close between the inland (14.22%) and the coastal (14.49%) zones. In comparison, the urban zone exhibited a slightly lower proportion at 12.79%. In the Appendix, Figure A2 shows the hourly percentage heatmap of heat stress classification by study zone. This visualisation primarily serves to facilitate the interpretation of the hourly distributions of thermal stress categories presented in Figure 10.
Figure 10. The time proportions dominated by different heat stress levels for each zone (coastal, urban, and inland) during the study period (07:00–21:00). In Figure 10, distinct colours represent different heat stress levels, with percentages indicating the time proportion occupied by each level throughout the 15-h study period.

4. Conclusions

This study investigated the spatiotemporal variation in thermal comfort indices and heat stress classifications in Sendai, Japan during typical sea-breeze days and elucidated the environmental driving factors affecting thermal comfort across coastal–urban–inland gradients. Key findings indicate that coastal areas consistently exhibited superior thermal comfort with the lowest heat stress levels, including minimal prevalence of extreme heat stress, thus demonstrating significant divergence from the urban and inland zones. Whereas v exerts a limited influence on coastal comfort, the advection of cold, moist air via the sea breeze plays a critical regulatory role. Notably, the sea-breeze effect substantially suppresses a rise in thermal comfort indices in coastal areas when extreme heat stress occurs inland. Thermal comfort in urban and inland areas is predominantly governed by v, which is modulated by coastal proximity and urban morphology. The penetration efficiency of sea breezes directly regulates thermal comfort in these regions. During the evening hours, RH influences thermal comfort indices across the entire study area, with its spatial distributions revealing pronounced reduction in urban centres. Consequently, attenuated v due to urban drag during sea-breeze penetration, coupled with lower urban humidity, constitute pivotal factors affecting inland thermal comfort. Urban ventilation-corridor optimisation and humidity-enhancement strategies may extend the spatial coverage of sea-breeze-mediated thermal regulation and improve regional thermal comfort. Our research provides mechanistic insights into sea-breeze-induced thermal environment amelioration in coastal cities, offering a scientific basis for the enhancement of marine airflow in climate-resilient urban planning.
This study focuses on the spatial distribution patterns of sea breeze impacts and directly uses 10 m wind speeds from WRF simulations for PET calculation. Due to the characteristic decrease in wind speed toward the surface in the near-ground layer, this approach may overestimate the wind-induced cooling effect at human height, potentially leading to a moderate overestimation of the PET improvement attributed to sea breezes. It should also be noted that thermal comfort at the actual 1.2 m height is influenced by three-dimensional urban environments, which are not covered in the mesoscale analysis of this study.
The results of this study reflect only the characteristics of typical summer sea-breeze days similar to 5 August 2016. Since sea breezes are significantly affected by factors such as season, pressure, and urban meteorology, the conclusions cannot be directly generalized to other seasons or meteorological conditions, and readers are hereby reminded to interpret them with caution.

Author Contributions

Conceptualization, S.P.; data curation, S.P.; formal analysis, H.W.; funding acquisition, H.W.; methodology, S.P.; project administration, H.W.; resources, H.W.; software, S.P.; supervision, H.W.; validation, S.P.; visualization, S.P.; writing—original draft, S.P.; writing—review and editing, S.P. All authors have read and agreed to the published version of the manuscript.

Funding

This research was funded by the Scientific Research Project of Higher Education in Anhui Province (Grant No. 2024AH051851), the Talent Research Initiation Fund Project of Tongling University (Grant No. 2023tlxyrc27), the Horizontal Research Project of Tongling University (Grant No. 2024tlxyxdz251), and the JSPS KAKENHI grant (No. JP19K04734).

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Acknowledgments

Thanks to Hironori Watanabe of Tohoku Institute of Technology for providing the data.

Conflicts of Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Correction Statement

This article has been republished with a minor correction to the Data Availability Statement. This change does not affect the scientific content of the article.

Abbreviations

The following abbreviations are used in this manuscript:
PETPhysiological equivalent temperature
WRFWeather Research and Forecasting
OTCOutdoor thermal comfort
TaAir temperature
VWind speed
RHRelative humidity
TPThermal perception

Appendix A

Figure A1. Trends in SHAP values with changes in the values of each input variable of the model, along with mean absolute SHAP values, in the coastal zone.
Figure A2. Hourly percentage heatmap of heat stress classification by study zone. In the hourly percentage heatmap of the HSC across study zones, colour gradients visually represent the proportional dominance of each heat stress level. This visualisation primarily serves to facilitate the interpretation of the hourly distributions of thermal stress categories presented in Figure 10.
Figure A3. Trends in SHAP values with changes in the values of each input variable of the model, along with mean absolute SHAP values, in the urban zone.
Figure A4. Trends in SHAP values with changes in the values of each input variable of the model, along with mean absolute SHAP values, in the inland zone.

References

  1. Desa, U.N. The 2014 Revision of theWorld Urbanization Prospects; United Nations Department of Economics and Social Affairs, Population Division: New York, NY, USA, 2015; p. 41. [Google Scholar]
  2. Santamouris, M. Analyzing the heat island magnitude and characteristics in one hundred Asian and Australian cities and regions. Sci. Total Environ. 2015, 512, 582–598. [Google Scholar] [CrossRef] [Scilit]
  3. Stewart, I.D.; Oke, T.R. Local climate zones for urban temperature studies. Bull. Am. Meteorol. Soc. 2012, 93, 1879–1900. [Google Scholar] [CrossRef] [Scilit]
  4. Grimm, N.B.; Faeth, S.H.; Golubiewski, N.E.; Redman, C.L.; Wu, J.; Bai, X.; Briggs, J.M. Global change and the ecology of cities. Science 2008, 319, 756–760. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  5. Zhao, L.; Oppenheimer, M.; Zhu, Q.; Baldwin, J.W.; Ebi, K.L.; Bou-Zeid, E.; Guan, K.; Liu, X. Interactions between urban heat islands and heat waves. Environ. Res. Lett. 2018, 13, 034003. [Google Scholar] [CrossRef] [Scilit]
  6. Chen, L.; Ng, E. Outdoor thermal comfort and outdoor activities: A review of research in the past decade. Cities 2012, 29, 118–125. [Google Scholar] [CrossRef] [Scilit]
  7. Heaviside, C.; Macintyre, H.; Vardoulakis, S. The urban heat island: Implications for health in a changing environment. Curr. Environ. Health Rep. 2017, 4, 296–305. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  8. He, B.J. Towards the next generation of green building for urban heat island mitigation: Zero UHI impact building. Sustain. Cities Soc. 2019, 50, 101647. [Google Scholar] [CrossRef] [Scilit]
  9. Qi, J.D.; He, B.J.; Wang, M.; Zhu, J.; Fu, W.C. Do grey infrastructures always elevate urban temperature? No, utilizing grey infrastructures to mitigate urban heat island effects. Sustain. Cities Soc. 2019, 46, 101392. [Google Scholar] [CrossRef] [Scilit]
  10. Aghamolaei, R.; Azizi, M.M.; Aminzadeh, B.; O’Donnell, J. A comprehensive review of outdoor thermal comfort in urban areas: Effective parameters and approaches. Energy Environ. 2023, 34, 2204–2227. [Google Scholar] [CrossRef] [Scilit]
  11. Nouri, A.S.; Matzarakis, A. The maturing interdisciplinary relationship between human biometeorological aspects and local adaptation processes: An encompassing overview. Climate 2019, 7, 134. [Google Scholar] [CrossRef] [Scilit]
  12. Freitas, C.R.D.; Grigorieva, E.A. A comprehensive catalogue and classification of human thermal climate indices. Int. J. Biometeorol. 2015, 59, 109–120. [Google Scholar] [CrossRef] [Scilit]
  13. Lai, D.; Chen, B.; Liu, K. Quantification of the influence of thermal comfort and life patterns on outdoor space activities. In Building Simulation; Tsinghua University Press: Beijing, China, 2020; Volume 13, pp. 113–125. [Google Scholar] [CrossRef] [Scilit]
  14. Tian, Y.; Hong, B.; Zhang, Z.; Wu, S.; Yuan, T. Factors influencing resident and tourist outdoor thermal comfort: A comparative study in China’s cold region. Sci. Total Environ. 2022, 808, 152079. [Google Scholar] [CrossRef] [Scilit]
  15. Geng, Y.; Hong, B.; Du, M.; Yuan, T.; Wang, Y. Combined effects of visual-acoustic-thermal comfort in campus open spaces: A pilot study in China’s cold region. Build. Environ. 2022, 209, 108658. [Google Scholar] [CrossRef] [Scilit]
  16. Park, M.; Hagishima, A.; Tanimoto, J.; Narita, K.I. Effect of urban vegetation on outdoor thermal environment: Field measurement at a scale model site. Build. Environ. 2012, 56, 38–46. [Google Scholar] [CrossRef] [Scilit]
  17. Morakinvo, T.E.; Lau, K.K.-L.; Ren, C.; Ng, E. Performance of Hong Kong’s common trees species for outdoor temperature regulation, thermal comfort and energy saving. Build. Environ. 2018, 137, 157–170. [Google Scholar] [CrossRef] [Scilit]
  18. Wang, Y.; Ni, Z.; Peng, Y.; Xia, B. Local variation of outdoor thermal comfort in different urban green spaces in Guangzhou, a subtropical city in South China. Urban For. Urban Green. 2018, 32, 99–112. [Google Scholar] [CrossRef] [Scilit]
  19. Cheung, P.K.; Jim, C.Y. Comparing the cooling effects of a tree and a concrete shelter using PET and UTCI. Build. Environ. 2018, 130, 49–61. [Google Scholar] [CrossRef] [Scilit]
  20. Hsieh, C.M.; Jan, F.C.; Zhang, L. A simplified assessment of how tree allocation, wind environment, and shading affect human comfort. Urban For. Urban Green. 2016, 18, 126–137. [Google Scholar] [CrossRef] [Scilit]
  21. Zheng, S.; Zhao, L.; Li, Q. Numerical simulation of the impact of different vegetation species on the outdoor thermal environment. Urban For. Urban Green. 2016, 18, 138–150. [Google Scholar] [CrossRef] [Scilit]
  22. Li, J.; Liu, N. The perception, optimization strategies and prospects of outdoor thermal comfort in China: A review. Build. Environ. 2020, 170, 106614. [Google Scholar] [CrossRef] [Scilit]
  23. Villadiego, K.; Velay-Dabat, M.A. Outdoor thermal comfort in a hot and humid climate of Colombia: A field study in Barranquilla. Build. Environ. 2014, 75, 142–152. [Google Scholar] [CrossRef] [Scilit]
  24. Nikolopoulou, M.; Lykoudis, S. Thermal comfort in outdoor urban spaces: Analysis across different European countries. Build. Environ. 2006, 41, 1455–1470. [Google Scholar] [CrossRef] [Scilit]
  25. Mushore, T.D.; Odindi, J.; Slotow, R.; Mutanga, O. Remote sensing-based outdoor thermal comfort assessment in local climate zones in the rural–urban continuum of eThekwini municipality, South Africa. Remote Sens. 2023, 15, 5461. [Google Scholar] [CrossRef] [Scilit]
  26. He, X.; An, L.; Hong, B.; Huang, B.; Cui, X. Cross-cultural differences in thermal comfort in campus open spaces: A longitudinal field survey in China’s cold region. Build. Environ. 2020, 172, 106739. [Google Scholar] [CrossRef] [Scilit]
  27. Shooshtarian, S.; Lam, C.K.C.; Kenawy, I. Outdoor thermal comfort assessment: A review on thermal comfort research in Australia. Build. Environ. 2020, 177, 106917. [Google Scholar] [CrossRef] [Scilit]
  28. Lai, D.; Lian, Z.; Liu, W.; Guo, C.; Liu, W.; Liu, K.; Chen, Q. A comprehensive review of thermal comfort studies in urban open spaces. Sci. Total Environ. 2020, 742, 140092. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  29. Chen, X.; Xue, P.; Liu, L.; Gao, L.; Liu, J. Outdoor thermal comfort and adaptation in severe cold area: A longitudinal survey in Harbin, China. Build. Environ. 2018, 143, 548–560. [Google Scholar] [CrossRef] [Scilit]
  30. Liu, L.; He, H.; Cai, Y.; Hang, J.; Liu, J.; Liu, L.; Jiang, P.; He, H. Cooling effects of wetland parks in hot and humid areas based on remote sensing images and local climate zone scheme. Build. Environ. 2023, 243, 110660. [Google Scholar] [CrossRef] [Scilit]
  31. Liu, L.; Wu, J.; Liang, Z.; Gao, J.; Hang, J.; Liu, J.; Liu, L. Spatio-temporal analysis of local thermal environment in waterfront blocks along the both sides of pearl river in Guangzhou, China. Case Stud. Therm. Eng. 2024, 53, 103875. [Google Scholar] [CrossRef] [Scilit]
  32. Al-hagla, K.S.; El-sayad, Z.T. Using simulation methods to investigate the impact of urban form on human comfort. Case study: Coast of Baltim, North Coast, Egypt. Alex. Eng. J. 2019, 58, 273–282. [Google Scholar] [CrossRef] [Scilit]
  33. Norouziasas, A.; Ha, P.P.; Ahmadi, M.; Rijal, H.B. Evaluation of urban form influence on pedestrians’ wind comfort. Build. Environ. 2022, 224, 109522. [Google Scholar] [CrossRef] [Scilit]
  34. Wang, Q.; Wang, X.; Meng, Y.; Zhou, Y.; Wang, H. Exploring the impact of urban features on the spatial variation of land surface temperature within the diurnal cycle. Sustain. Cities Soc. 2023, 91, 104432. [Google Scholar] [CrossRef] [Scilit]
  35. Salvati, A.; Kolokotroni, M.; Kotopouleas, A.; Watkins, R.; Giridharan, R.; Nikolopoulou, M. Impact of reflective materials on urban canyon albedo, outdoor and indoor microclimates. Build. Environ. 2022, 207, 108459. [Google Scholar] [CrossRef] [Scilit]
  36. Sasaki, Y.; Matsuo, K.; Yokoyama, M.; Sasaki, M.; Tanaka, T.; Sadohara, S. Sea breeze effect mapping for mitigating summer urban warming: For making urban environmental climate map of Yokohama and its surrounding area. Urban Clim. 2018, 24, 529–550. [Google Scholar] [CrossRef]
  37. Papanastasiou, D.K.; Melas, D.; Bartzanas, T.; Kittas, C. Temperature, comfort and pollution levels during heat waves and the role of sea breeze. Int. J. Biometeorol. 2010, 54, 307–317. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  38. Kolokotsa, D.; Psomas, A.; Karapidakis, E. Urban heat island in southern Europe: The case study of Hania, Crete. Sol. Energy 2009, 83, 1871–1883. [Google Scholar] [CrossRef] [Scilit]
  39. Zhou, Y.; Guan, H.; Huang, C.; Fan, L.; Gharib, S.; Batelaan, O.; Simmons, C. Sea breeze cooling capacity and its influencing factors in a coastal city. Build. Environ. 2019, 166, 106408. [Google Scholar] [CrossRef] [Scilit]
  40. Niu, J.; Liu, J.; Lee, T.C.; Lin, Z.J.; Mak, C.; Tse, K.T.; Tang, B.; Kwok, K.C. A new method to assess spatial variations of outdoor thermal comfort: Onsite monitoring results and implications for precinct planning. Build. Environ. 2015, 91, 263–270. [Google Scholar] [CrossRef] [Scilit]
  41. Ng, E. Policies and technical guidelines for urban planning of high-density cities–air ventilation assessment (AVA) of Hong Kong. Build. Environ. 2009, 44, 1478–1488. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  42. Huang, Z.; Cheng, B.; Gou, Z.; Zhang, F. Outdoor thermal comfort and adaptive behaviors in a university campus in China’s hot summer-cold winter climate region. Build. Environ. 2019, 165, 106414. [Google Scholar] [CrossRef] [Scilit]
  43. Cheung, P.K.; Jim, C.Y. Subjective outdoor thermal comfort and urban green space usage in humid-subtropical Hong Kong. Energy Build. 2018, 173, 150–162. [Google Scholar] [CrossRef] [Scilit]
  44. Watanabe, S.; Nagano, K.; Ishii, J.; Horikoshi, T. Evaluation of outdoor thermal comfort in sunlight, building shade, and pergola shade during summer in a humid subtropical region. Build. Environ. 2014, 82, 556–565. [Google Scholar] [CrossRef] [Scilit]
  45. Nogueira, M.; Lima, D.C.; Soares, P.M. An integrated approach to project the future urban climate response: Changes to Lisbon’s urban heat island and temperature extremes. Urban Clim. 2020, 34, 100683. [Google Scholar] [CrossRef] [Scilit]
  46. Acero, J.A.; Ruefenacht, L.A.; Koh, E.J.; Tan, Y.S.; Norford, L.K. Measuring and comparing thermal comfort in outdoor and semi-outdoor spaces in tropical Singapore. Urban Clim. 2022, 42, 101122. [Google Scholar] [CrossRef] [Scilit]
  47. Yang, B.; Olofsson, T.; Nair, G.; Kabanshi, A. Outdoor thermal comfort under subarctic climate of north Sweden–A pilot study in Umeå. Sustain. Cities Soc. 2017, 28, 387–397. [Google Scholar] [CrossRef] [Scilit]
  48. He, B.J.; Ding, L.; Prasad, D. Outdoor thermal environment of an open space under sea breeze: A mobile experience in a coastal city of Sydney, Australia. Urban Clim. 2020, 31, 100567. [Google Scholar] [CrossRef] [Scilit]
  49. Umezaki, A.S.; Ribeiro, F.N.D.; Oliveira, A.P.D.; Soares, J.; Miranda, R.M.D. Numerical characterization of spatial and temporal evolution of summer urban heat island intensity in São Paulo, Brazil. Urban Clim. 2020, 32, 100615. [Google Scholar] [CrossRef] [Scilit]
  50. Yamamoto, Y.; Ishikawa, H. Influence of urban spatial configuration and sea breeze on land surface temperature on summer clear-sky days. Urban Clim. 2020, 31, 100578. [Google Scholar] [CrossRef] [Scilit]
  51. Emmanuel, R.; Johansson, E. Influence of urban morphology and sea breeze on hot humid microclimate: The case of Colombo, Sri Lanka. Clim. Res. 2006, 30, 189–200. [Google Scholar] [CrossRef] [Scilit]
  52. Lopes, A.; Lopes, S.; Matzarakis, A.; Alcoforado, M.J. The influence of the summer sea breeze on thermal comfort in Funchal (Madeira): A contribution to tourism and urban planning. Meteorol. Z. 2011, 5, 553–564. [Google Scholar] [CrossRef] [Scilit]
  53. Anjos, M.; Lopes, A.; Lucena, A.J.D.; Mendonça, F. Sea breeze front and outdoor thermal comfort during summer in northeastern Brazil. Atmosphere 2020, 11, 1013. [Google Scholar] [CrossRef] [Scilit]
  54. Guo, F.; Zhao, J.; Zhang, H.; Dong, J.; Zhu, P.; Lau, S.S.Y. Effects of urban form on sea cooling capacity under the heatwave. Sustain. Cities Soc. 2023, 88, 104271. [Google Scholar] [CrossRef] [Scilit]
  55. Zhang, H.; Guo, F.; Liu, K.; Wang, J.; Dong, J.; Zhu, P. Spatial differences in thermal comfort in summer in coastal areas: A study on Dalian, China. Front. Public Health 2022, 10, 1024757. [Google Scholar] [CrossRef] [Scilit]
  56. He, B.J.; Ding, L.; Prasad, D. Relationships among local-scale urban morphology, urban ventilation, urban heat island and outdoor thermal comfort under sea breeze influence. Sustain. Cities Soc. 2020, 60, 102289. [Google Scholar] [CrossRef] [Scilit]
  57. Zhou, X.; Okaze, T.; Ren, C.; Cai, M.; Ishida, Y.; Watanabe, H.; Mochida, A. Evaluation of urban heat islands using local climate zones and the influence of sea-land breeze. Sustain. Cities Soc. 2020, 55, 102060. [Google Scholar] [CrossRef] [Scilit]
  58. Zhu, M.; Atkinson, B.W. Observed and Modelled Climatology of the Land-Sea Breeze Circulation over the Persian Gulf. Int. J. Climatol. 2004, 24, 883–905. [Google Scholar] [CrossRef] [Scilit]
  59. Bonnardot, V.; Planchon, O.; Cautenet, S. Sea Breeze Development under an Offshore Synoptic Wind in the South-Western Cape and Implications for the Stellenbosch Wine-Producing Area. Theor. Appl. Climatol. 2005, 81, 203–218. [Google Scholar] [CrossRef] [Scilit]
  60. Peel, M.C.; Finlayson, B.L.; McMahon, T.A. Updated world map of theKöppen-Geiger climate classification. Hydrol. Earth Syst. Sci. Discuss. 2017, 11, 1633–1644. [Google Scholar] [CrossRef] [Scilit]
  61. Peng, S.; Watanabe, H. Analysis and mapping of sea breeze event time in coastal cities: A case study of Sendai. Atmosphere 2022, 13, 1484. [Google Scholar] [CrossRef] [Scilit]
  62. Höppe, P. The physiological equivalent temperature–a universal index for the biometeorological assessment of the thermal environment. Int. J. Biometeorol. 1999, 43, 71–75. [Google Scholar] [CrossRef] [Scilit] [PubMed]
  63. Ohlert, M.; Baruchel, J.; Galez, P. Applications of a Universal Thermal Index: Physiological Equivalent Temperature. J. Phys. D Appl. Phys. 1995, 28, A78–A83. [Google Scholar] [CrossRef] [Scilit]
  64. Powers, J.G.; Klemp, J.B.; Skamarock, W.C.; Davis, C.A.; Dudhia, J.; Gill, D.O.; Coen, J.L.; Gochis, D.J.; Ahmadov, R.; Peckham, S.E.; et al. The Weather Research and Forecasting Model: Overview, System Efforts, and Future Directions. Bull. Am. Meteorol. Soc. 2017, 98, 1717–1737. [Google Scholar] [CrossRef] [Scilit]
  65. Peng, S.; Watanabe, H. Reproducibility Analysis of WRF Model in Urban Climate Research, A Case Study of Sendai. J. Asian Urban Environ. 2022, 91–94. [Google Scholar]
  66. Chen, T.; He, T.; Benesty, M.; Khotilovich, V.; Tang, Y.; Cho, H.; Chen, K. Xgboost: Extreme Gradient Boosting, R Package Version 0.4-2; Xgboost: Seattle, WA, USA, 2015. [Google Scholar]
  67. Fan, J.; Wang, X.; Wu, L.; Zhou, H.; Zhang, F.; Yu, X.; Xiang, Y. Comparison of Support Vector Machine and Extreme Gradient Boosting for predicting daily global solar radiation using temperature and precipitation in humid subtropical climates: A case study in China. Energy Convers. Manag. 2018, 164, 102–111. [Google Scholar] [CrossRef] [Scilit]
  68. Song, L.L.; Wang, S.H.; Yang, C.; Sheng, X. Application Research of Improved XGBoost in Imbalanced Data Processing. Comput. Sci. 2020, 47, 98–103. [Google Scholar]
  69. Li, B.-S.; Li, L.-Z.; Sun, Y.; Zhu, Y.-Q. Intranet Defense Algorithm Based on Pseudo Boosting Decision Tree. Comput. Sci. 2018, 45, 157–162. [Google Scholar]
  70. Lundberg, S.M.; Lee, S.-I. A Unified Approach to Interpreting Model Predictions. In Proceedings of the Neural Information Processing Systems, Long Beach, CA, USA, 4–9 December 2017. [Google Scholar]
  71. Strumbelj, E.; Kononenko, I. Explaining prediction models and individual predictions with feature contributions. Knowl. Inf. Syst. 2014, 41, 647–665. [Google Scholar] [CrossRef] [Scilit]
  72. Ribeiro, M.T.; Singh, S.; Guestrin, C. “Why should I trust you?” Explaining the Predictions of Any Classifier. In Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining, San Francisco, CA, USA, 13–17 August 2016; pp. 1135–1144. [Google Scholar]
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Article Metrics

Citations

Article Access Statistics

Journal Statistics
Multiple requests from the same IP address are counted as one view.
Upcycling Potato Juice Protein for Sustainable Plant-Based Gyros: A Multidimensional Quality Assessment
Previous
Evaluating Corporate Carbon Emissions Reporting: Assessing Transparency and Completeness with the Carbon Integrity Index
Next