1 Introduction

A comfortable urban thermal climate plays an irreplaceable role in assisting people to maintain good health and lower energy use for indoor air conditioning^[1]. However, due to the rapid urbanization in developing countries, the changes in underlying structures and increasing heat discharge from vehicles and buildings have worsened the outdoor thermal climate^[2-3] and greatly affected the thermal comfort in urban space. Consequently, an increasing attention to the outdoor thermal environments and thermal comfort was drawn throughout the world, especially in the hot and humid regions. A variety of methods were adopted for outdoor thermal comfort study. For example, Lin^[4]examined the user thermal comfort in a public square in Taiwan through a field study. Xi et al.^[5] investigated the outdoor thermal comfort of campus clusters in Guangzhou with a questionnaire survey. Ooka et al.^[6] developed a multi-objective optimum design method for a comfortable outdoor thermal urban environment. Ketterer et al.^[7] simulated the thermal human-biometeorological conditions in Stuttgart, Germany by using RayMan and ENVI-met. Then the relative influential factors affecting outdoor thermal comfort mainly considered the urban morphology and thermal properties of urban surfaces, such as the H/W ratio^[8-9], the sky view factor (SVF)^[10-11], the thermal impact of vegetation and water space^[12-13] and the influence caused by the albedo of different ground cover materials^[14-15]. However, most of these current studies on the outdoor thermal comfort mainly focused on the specific test area and lacked comprehensive analysis of both the thermal comfort and the influences of urban spatial morphology and underlying surface characteristics. Quantitative study on the outdoor thermal comfort from the view of urban planning in a highly urbanized city is becoming significant.

The subtropical city of Shenzhen is experiencing dramatic development in mainland China. As an economic special district since 1980, the emergence of Shenzhen led to its rapid growth from a small city into a megalopolis with 14 million inhabitants^[16]. The condition of tremendous growth in population and buildings in Shenzhen is unique and the cultural background and living standard of the local respondents of Shenzhen are also different from the current conditions of other countries^[17]. The long-lasting summertime of Shenzhen characterized by high temperature and high humidity has become a real nuisance to local residents. Nevertheless, few studies have researched on the outdoor thermal comfort by considering different regional spatial layouts with real thermal evaluations of the local residents in the hot and humid summertime of Shenzhen. Therefore, quantitative analysis of the outdoor thermal environments and thermal comfort in the summertime of Shenzhen seems necessary to achieve the climate-conscious urban design.

This paper adopted both questionnaire surveys and simultaneous physical field measurements to conduct the investigation on the outdoor thermal environments and thermal comfort in six different places within the city of Shenzhen during the hottest summertime. Both the thermal preferences on different meteorological parameters and the overall evaluation on the outdoor thermal comfort were illustrated and quantified. Besides, the main factors influencing the outdoor thermal comfort were discussed. Additionally, the impact of both building morphology and underlying surface compositions on outdoor thermal comfort was analyzed.

2 Methods 2.1 Study Areas

This study chose the International Low Carbon City (ILCC) of Shenzhen in the Pearl River Delta region in South China as the study area. The ILCC has an area of 53.42 km² and it has been divided into hundreds of localized urban blocks by planners according to different architectural types and underlying surface compositions to quantitatively express the building parameters and underlying surface conditions of each urban block, as shown in Fig. 1. The field study was conducted at six different outdoor spaces in the ILCC. They belonged to six urban blocks with different neighborhoods and layouts, ranging from dispersed, low-rise building areas to densely populated central areas, as expressed in Fig. 1. Among them, test sites No. 1 and No. 4 were surrounded with intensive high-rise buildings, test sites No. 5 and No.6 exhibited conditions with low-rise buildings, test site No.2 contained a large area of water bodies and vegetation with fewer mid-rise buildings, and test site No. 3 presented an open space with a few scattered trees and low-rise buildings along an asphalt road. With the data collection on the ILCC, the building parameters and underlying surface parameters of the urban blocks corresponding to these six test sites could be identified for further quantitative comparison and analysis.

2.2 Field Survey

The field data collection of each test site included simultaneous questionnaire surveys on human responses and physical measurements of the major meteorological parameters. The outdoor thermal comfort and micro-meteorological parameters in the daytime of the hottest summertime were considered as the main research objects and the whole process was conducted in July, 2013, which was usually the hottest month in Shenzhen. During the field survey, the questionnaires of the six test sites were given out and ended at the same time during the daytime. In order to accurately reflect the thermal environmental evaluations of the respondents, the residents around each test site who had leisurely walked in the outdoor space for about half an hour were mainly selected as the respondents for this questionnaire study. A total of 415 sets of valid records (approximately 70 sets of each test site for consistency of this multi-point survey) were collected in this study. The questionnaire was designed using plain, concise language, and the content mainly consisted of two parts, as shown in Fig. 2. A five-point scale of the thermal environment evaluations was employed for the purpose of simplicity and convenient statistics. Different degrees of evaluation of thermal sensation corresponded to the evaluation scale varying from "-2" to "+2". Additionally, the overall thermal comfort evaluations scales (T_CS) was considered as the main parameter expressing the overall thermal comfort degrees, varying from "very uncomfortable" to "very comfortable".

As the underlying surface surroundings and architectural constructions of these six test sites varied greatly, the micro-meteorological parameters of these six test sites were also different. Therefore, six HOBO micro weather stations were applied to measure the local air temperature (T_a), relative humidity (R_H), wind speed (V), global solar radiation (G) and globe temperature (T_g) of each test site. These micro weather stations had been calibrated before starting the field experiment. They were located in the open space near interviewees while the interview was taking place. All of the micro-meteorological parameters were automatically recorded every 10 s to match the accurate time of each questionnaire. Table 1 displays the specifications of the sensors applied to measure these micro-meteorological parameters. During the process of measurement, the sensors used to measure T_a and R_H were wrapped with protective heat insulation and placedaway from any direct sunlight. All of the sensors were uniformly held at a height of about 1.5 m by considering the usual measuring height of the outdoor local thermal environments and according with the average thermal sensation height of the interviewees.

2.3 Thermal Comfort Indices

Several indices integrating thermal environmental factors and the heat balance of the human body are applied for accessing thermal comfort. Among them, the standard effective temperature (SET^*)^[18] and physiologically equivalent temperature (PET)^[19-20] have been widely used in urban built-up areas for assessing outdoor thermal comfort and obtaining thermally acceptable ranges or the neutral temperature of outdoor spaces^[21-22]. Since occupant thermal sensations and preferences in different regions may contribute to different thermal comfort ranges, the thermal comfort evaluation ranges represented by PET and SET^* of Shenzhen should be assessed by the subjective sensations of the local residents in Shenzhen to reflect the different degrees of thermal comfort.

This investigation considered and compared these two thermal comfort indices of SET^* and PET to generalize the thermal comfort evaluation range benchmarks in Shenzhen. The thermal comfort indices SET^* and PET are expressed in ℃ and taking into account T_a, mean radiant temperature (T_MR), R_H, and V. Among the meteorological parameters, T_MR could be calculated as (ISO 7726 standard):

${T_{{\rm{MR}}}} = {[({T_{\rm{g}}} + 273){\rm{ + }}\frac{{1.10 \times {{10}^8}{V^{0.6}}}}{{\varepsilon {D^{0.4}}}}(Tg - Ta)]^{\frac{1}{4}}} - 273$

(1)

where T_MR is the mean radiant temperature, ℃; T_g is the globe temperature, ℃; V is the wind velocity, m/s; T_a is the air temperature, ℃; D is globe diameter (=0.15 m in this study); and ε is emissivity (= 0.95 for a black globe).

After collecting the questionnaires and measuring the corresponding meteorological parameters from the six test sites, the index PET and SET^* of each questionnaire can be estimated using the free software package RayMan^[23]. With the values of SET^*, PET and the corresponding T_CS of hundreds of questionnaires, the appropriate thermal comfort evaluation range benchmarks represented by SET^* and PET could be obtained with statistical analysis, thus quantitatively describing the different magnitudes of outdoor thermal comfort under the summertime of Shenzhen.

2.4 Characteristic Parameters of Regional Spatial Layout

Considering the surroundings of each test site were different, this paper chose the corresponding urban blocks covering each test site as the basic research areas. In order to comprehensively reflect the characteristics of regional spatial layouts around each test site, the average height of buildings (H), building density (B_D), the coverage ratio of green and water space (S), and sky view factor (S_VF) were adopted as four characteristic parameters. Among them, the parameters of H and B_D mainly reflected the building morphology. The S_VF considered the spatial enclosure degree around the test sites. The S represented the green and ecological effects of surrounding environments.

The parameters of H, B_D, and S were the common indices widely used in the field of urban planning and could be easily calculated with the information of geographical database. The S_VF could be obtained by using the fisheye camera and calculations with RayMan. Table 2 shows the characteristic parameters of six test sites and the pictures of their surroundings. By combining the characteristic parameters of each test site and their corresponding micro-meteorological parameters and T_CS, the relationships between them could be analyzed. The impacts of regional spatial layout on thermal climate and thermal comfort could be revealed.

3 Results and Discussion 3.1 Micro-meteorological Parameters of the Six Test Sites

With the meteorological data from the six micro weather stations of the test sites, the thermal environmental conditions of each test site during the whole survey process could be revealed by presenting the variation curves of air temperature (T_a), relative humidity (R_H), wind speed (V) and global solar radiation (G), as shown in Fig. 3. From an overall perspective, the T_a of these six test sites were mainly between 28-36 ℃, the R_H of them were between 50%-85% and the G of them could reach over 500 W/m² during some investigation periods. The overall weather conditions during the investigation could be expressed with high temperature and high humidity with strong solar radiation.

Then considering the micro-meteorological parameters of T_a, R_H, V and G between the six test sites, they varied greatly during the investigation. The main reason was the diverse underlying surface constituents and regional spatial layouts of these test sites.

Among these six test sites, test site No. 3 and No. 6 had higher values of T_a and G mainly because they lied in sparse low-rise buildings. Sparse low-rise buildings could not effectively block the sunlight from reaching into the pedestrian space and therefore increased the values of T_a and G. The test site No.1 and No.4 had relatively lower values of V and G due to their surrounding densely high-rise buildings. Densely high-rise buildings greatly blocked the solar radiation and also weakened the wind speed. However, the densely spatial arrangement may result in the accumulation of large quantities of building anthropogenic heat exhaust from the building air conditioning systems. Therefore, No.1 and No.4 still had higher values of T_a mainly varying between 29-33 ℃. The test site No.2 had relatively lower values of T_a and G, higher values of R_H and V compared to others due to its surroundings with mid-rise buildings and large areas of vegetation and water bodies. The mid-rise buildings helped block sunlight and decrease the wind speed. The transpiration of vegetation and evaporation from water bodies increased the relative humidity and also caused much cooling effects on lowering the air temperature. To sum up, different regional spatial layouts could contribute to the varied micro-meteorological conditions.

3.2 Evaluation Ranges of Different Meteorological Parameters

By combining the thermal sensation voting results of the meteorological parameters from hundreds of questionnaires and the corresponding real measuring values with micro weather stations, the meteorological parameter ranges corresponding to different thermal sensation votes were obtained.

The value ranges of T_a, R_H, V and G all showed an increase with the augmentation of thermal sensations as a whole, as shown in Fig. 4. The results show that the ranges of T_a expressed as "Moderate", "Warm" and "Hot" by respondents significantly overlapped with each other. The moderate value range of T_a was between approximately 28 to 30 ℃ during the investigation, and respondents would be more inclined to feel "warm" or "hot" when the T_a was higher than 30 ℃. In addition, the value range of feeling a "Moderate" R_H was mainly concentrated in the range of 60%-70%, and the respondents would be highly likely to feel varying degrees of moisture when the R_H reached 75% or more. Additionally, the values of V corresponding with an evaluation of "Moderate" covered a wide range of 0-2.0 m/s, and a relatively small number of the residents chose "Weak" or "Very weak". Few respondents felt "Strong" wind conditions during the investigation. Moreover, different thermal sensations about G by respondents fluctuated greatly from under 10 W/m² to approximately 800 W/m². Because the parameters V and G had instantaneity, the corresponding evaluations by respondents and the simultaneous measured values may contain differences due to time error and instrument error. In conclusion, the results display the thermal preferences of the local residents in Shenzhen on these meteorological parameters under summer climatic conditions.

3.3 Factors Affecting Thermal Comfort

The meteorological parameters and also the human body conditions were usually considered as the influential factors on the outdoor thermal comfort^[24-25]. With the results of T_CS and the human physiological parameters from hundreds of questionnaires together with the simultaneous field measurements, some variables covering the parameters of age (A), gender (G_en), clothing insulation (I_cl), air temperature (T_a), relative humidity (R_H), wind speed (V) and global solar radiation (G) were selected as the factors reflecting the impacts on T_CS. By applying a multiple stepwise regression analysis, the quantitative mathematical equation expressing the relationships between them was then described as follows:

${T_{cs}} = - 0.541{T_a} - 0.022{R_H} + 0.442V - 0.003G - 1.769{I_{cl}} + 15.411$

(2)

where T_a varies from 28.3 to 35.2 ℃; R_H varies between 56.5% to 83.8%; V varies from 0 to 2.35 m/s; G varies within the range of 0.6 to 1 150 W/m², and I_cl varies in the range of 0.36 to 0.67 clo. This equation reveals linearity with the adjusted R square value of 0.765. These five variables of T_a, R_H, V, G and I_cl were reserved as the key factors significantly affecting thermal comfort. Meanwhile, the variables of A and G_en were eliminated due to their less significant effects on the thermal comfort. Therefore, the results demonstrate that the significant correlation factors affecting the thermal comfort evaluations concentrate more on the external environmental conditions rather than the physiological conditions of the human body.

Additionally, the standardized regression coefficients for assessing the effect of these five influential factors were obtained, as shown in Table 3. The results show that the parameters of T_a, R_H, V, G and I_cl have negative relationships with T_CS, while V has a positive relationship with T_CS. The reason may be that more wind helps to increase the skin surface heat dissipation while the increases in T_a, R_H, G impeded the heat dissipation from skin surface in summer. Besides, T_a had the largest influence on the thermal comfort compared to other parameters. Therefore, lower T_a, R_H, G and more wind together with appropriate clothes would contribute to a more comfortable outdoor thermal climate.

3.4 Assessment of the Thermal Comfort Indices

Having calculated the two thermal indices PET and SET^*, the exact range distributions of SET^* and PET corresponding to different values of T_CS could be obtained with the mathematical statistics method, as described in Table 4. The result reveals a clear difference existing between the values of PET and SET^* corresponding to the same T_CS and the range values of SET^* and PET both declined in varying degrees with the increasing T_CS. Furthermore, the thermal comfortable feelings with scales of "+1"and"+2"covered a wide range varying from 28.14-32.83 ℃ expressed by PET and the corresponding range of 24.74-30.45 ℃ represented by SET^*.

The thermal comfortable range of SET^* was about 2-3 ℃ lower than that of PET. The main reason is their different ways of calculating the physiological sweat rate and the heat flows from body surface. The Munich Energy-balance Model for Individual (MEMI) used in PET calculates the heat flows from parts of body surface that are covered and not covered by clothing respectively. Besides, the modeling of the physiological sweat rate and the consideration of standard environment of PET and SET^* are also different. The humidity conditions in standard environment of SET^* is 50 % in relative humidity while PET assumes the vapor pressure of 12 hPa as the standard environmental humidity conditions. A field study by Sasaki^[26] indicated that the SET^* shows the tendency to evaluate a little lower than the actual thermal sensation due to the coverage of the model for calculating the convective heat transfer coefficient used in SET^*. The PET shows the tendency to evaluate a little higher than the actual thermal sensation as the results of PET are greatly influenced from MRT. Therefore, the calculation values of PET and SET^* are also different when given the same thermal environment.

3.5 Impacts of Regional Spatial Layout on Thermal Climate and Thermal Comfort

Considering the different micro-meteorological conditions of each test site caused by their diversified regional spatial layouts, the correlation analysis between the regional spatial layout characteristic parameters of H, B_D, S, S_VF and micro-meteorological parameters of T_a, R_H, V, G, and T_CS were conducted. The pearson correlation coefficients r between them were obtained, as shown in Table 5.

The results show that T_a has negative correlations with the parameters of H, B_D, S and has a positive correlation with S_VF. Among the characteristic parameters, S has a significant negative correlation with T_a, which indicates that increasing the area of water and green space can effectively help lower the air temperature. Conversely, R_H has positive correlations with H, B_D, S and has a negative correlation with S_VF. Besides, the absolute value of r between S and R_H is the biggest. Thus, the water and green space contribute to increasing the relative humidity. Then V has negative correlations with H and B_D, and has positive correlations with S and S_VF. Among these characteristic parameters, the absolute value of r between B_D and V is the biggest, which demonstrates that the wind speed greatly decreases with B_D. Then G has negative correlations with H, B_D, S and a positive correlation with S_VF. Besides, a significant negative correlation exists between H and G and a very significant positive correlation exists between S_VF and D. Therefore, effectively increasing H and decreasing S_VF can help block the strong solar radiation from reaching into the inner pedestrian space. Then from an overall perspective, T_CS has positive correlations with H, B_D, S and has a negative correlation with S_VF. Besides, a significant positive correlation exists between S and T_CS. Thus, appropriately increasing H, B_D, S and decreasing S_VF contribute to providing a thermal comfortable environment.

Therefore, the reasonable configuration of regional spatial layout can greatly improve the summertime conditions with high temperature and high humidity in Shenzhen.

4 Conclusions

1) The overall weather conditions during the investigation could be expressed with high temperature and high humidity with strong solar radiation. The micro-meteorological parameters expressed with the parameters of T_a, R_H, V and G between the six test sites varied greatly during the investigation due to their different regional spatial layouts. During the hot and humid summertime, the moderate range of T_a was between 28 to 30 ℃, and the moderate range of R_H mainly concentrated in the range of 60%-70% by using the results of the thermal sensation votes on these meteorological parameters.

2) With the quantitative regression analysis, T_a, R_H, V, G and I_cl were finally identified as the main influential factors impacting outdoor thermal comfort. Among them, T_a had the greatest effect on thermal comfort. The corresponding value ranges of PET and SET^* with different T_CS were obtained to reflect the different degrees of thermal comfort in Shenzhen. The thermal comfortable evaluations covered a wide range of 28.14-32.83 ℃ by PET and a corresponding range of 24.74-30.45 ℃ by SET^*. The thermal comfortable ranges of SET^* was lower than that of PET due to their different ways of calculating the physiological sweat rate and the heat flows from body surface.

3) By considering the effects of characteristic parameters of regional spatial layout on thermal climate and thermal comfort, correlation analysis between H, B_D, S, S_VF and T_a, R_H, V, G, T_CS were conducted. The results show that S has a significant negative correlation with T_a and a positive correlation with R_H. Then a significant negative correlation exists between H and G and a very significant positive correlation exists between S_VF and G. Overall, appropriately increasing H, B_D, S and decreasing S_VF contribute to increasing T_CS. Reasonable configuration of the regional spatial layout could help provide a thermal comfortable environment.