Comparison of the Thermal Environment by Local Climate Zones in Summer: A Case Study in Suwon, Republic of Korea

Abstract

The thermal environments of five local climate zones (LCZs) in summer were investigated using all measured microclimatic data collected from 9:00 on 3 August until 15:30 on 4 August, 2017. The physiological equivalent temperature (PET) and universal thermal climate index (UTCI) of the human thermal environment were highest in LCZ E_B (paved with scattered trees) and lowest in LCZ 2_B (compact mid-rise with scattered trees) and LCZ 4 (open high-rise) during the daytime, and highest in LCZ 4 and lowest in LCZ D (low plants) during the nighttime. LCZ D and LCZ E_B in the park and LCZ 5 (open mid-rise) revealed similar thermal environments, and LCZ 2_B and LCZ 4 showed higher PET and UTCI values for the thermal environment. The maximum mean difference by location was 6.7 °C for PET and 3.7 °C for UTCI during the daytime, a one-level difference in PET and a two-thirds level in UTCI. During the nighttime, the maximum mean difference was 2.3 °C for PET and 1.9 °C for UTCI, a one-third-level difference. Compared to LCZ 2_B, the PET showed a maximum difference of 11 °C and UTCI of 6 °C, showing a difference of two levels for PET and one level for UTCI.

1. Introduction

Due to rapid urbanization worldwide, 68% of the world’s population is expected to live in cities by 2050 [1]. As a result, global warming and the urban heat island phenomenon frequently occur, which cause the deterioration of the living quality of urban residents [2,3,4,5,6].

Research on the urban climate has mainly focused on the analysis of the thermal environment according to the urban structure type with the theme of the urban heat island phenomenon since the 1970s [2]. Three types of studies have primarily been conducted: (1) investigations of the aspect ratio (H/W) of buildings and street canyons to alleviate the urban thermal environment [7,8,9,10], (2) comparisons of the thermal environment reduction effect according to the arrangement of buildings and trees [11,12,13,14], and (3) assessment of the suitability of different landscape pavement materials [15,16,17].

Concerns about the urban thermal environment have led to research on the urban climate and interest in the function and utilization of spatial resources called green infrastructure. Green infrastructure is a network of natural spaces or spaces close to nature that contribute to the ecological function of the city and the improvement of the quality of life of citizens in urban areas [18,19,20,21]. Urban green infrastructure is known as a very useful tool to increase human thermal comfort in outdoor environments, especially in temperate and Mediterranean climates [22,23]. In addition, studies have shown that urban green infrastructure effectively decreases urban heat and increases thermal comfort in outdoor spaces [24,25]. Therefore, the use of green infrastructure for urban planning has increased [26,27,28].

Among green infrastructure in urban areas, urban neighborhood parks are known to have an air temperature (T_a) reduction effect, contributing towards the urban cool island phenomenon, depending on their size and planting density [25,29,30]. As factors for reducing the temperature of the thermal environment in urban neighborhood parks, the effects of the shading and evapotranspiration of trees and evaporation of water space are felt within parks as well as in their surrounding areas [31,32,33]. There have been many studies on the heat reduction effect of parks, such as those on the cooling effect on T_a due to the layout of a park [34], the cooling effect on T_a due to trees and shading structures in urban parks [35], and the thermal comfort of a park according to changes to the trees and shrubs present [36]. Recently, there have been studies on the effect of heat reduction in green and water spaces using computer simulation programs [12,37]. Currently, the most popular is the ENVI-met program [38,39,40,41,42], which is a three-dimensional microclimate model designed to simulate the interactions of land cover, vegetation, and atmosphere in an urban environment [17]. In particular, studies using computer simulation programs must go through a process of verification using measured climatic data [43]. Most studies have compared the thermal environments within a park, and a few have compared the human thermal environment of urban neighborhood parks and various land use areas surrounding them using measured climatic data [43].

A classification method called a local climate zone (LCZ) was developed for the analysis of the urban heat island phenomenon [44]. There are 17 LCZ types: LCZ 1–10 according to the built type and LCZ A–G according to the land cover type, and an additional four types according to the land cover properties. Each zone is affected by its microclimate according to these types, but there are also zones that are not classified into existing LCZs due to the geographical, cultural, historical, and climatic effects of the region and country [45]. Recently, the world urban database and access portal tools (WUDAPT) project was launched in 2018 [46], and the LCZ generator (https://lcz-generator.rub.de, accessed on 22 December 2022) [47] has been used to attribute LCZs with remote sensing data in many cities worldwide, e.g., [48,49,50,51,52]. Additionally, LCZ mapping using geographic information systems (GIS) is another research stream [53]. Moreover, a global map of LCZs was created “to support earth system modeling and urban-scale environmental science” [54].

LCZs were created for urban heat island research, that is, analysis of T_a changes between urban and rural areas affected by urban morphology and geometry. Based on the human energy balance model, studies are required to include not only T_a but also other climatic factors that affect the human thermal environment, such as humidity, wind speed (u), and solar and terrestrial radiation. In recent years, a few studies that combine LCZs with human thermal sensation/comfort have been conducted worldwide [55,56,57,58,59,60,61,62,63,64,65]. However, most studies have used computer simulation programs, such as RayMan, to estimate the mean radiant temperature (T_mrt), which is the solar and terrestrial radiation absorbed by the human body surface [55,56,57,61]; a multilayer perception network structure algorithm for estimating physiological equivalent temperature (PET) [59]; WRF for u estimation [61]; the MUKLIMO_3 urban climate model for the analysis of HUMIDEX [62]; a mesoscale atmospheric model, Méso-NH, for estimating T_a, T_mrt, and universal thermal climate index (UTCI) [63]; the ENVI-met microclimate simulation program, which tests the correlation between the sky view factor and six microclimate and thermal comfort indexes for the predicted mean vote (PMV) [65,66] and PET [67]; BioKlima 2.6 model (https://www.igipz.pan.pl/Bioklima-agik.html, accessed on 22 December 2022) for UTCI [68]; SOLWEIG model for T_mrt [69]; and, a multi-scale modeling framework, WRF-BEP/BEM and RayMan Pro models, for assessing meteorological distributions and human thermal comfort indices, PET, modified PET (mPET), and UTCI [70]. Only one study [64] measured all required microclimatic data (T_a, relative humidity (RH), u, and T_mrt) for calculating human thermal sensation, PET. Therefore, all required microclimatic data for outdoor human thermal sensation in various LCZs are rarely collected for effective comparisons.

This study investigated the thermal environments of five different LCZs, two locations within an urban neighborhood park and three locations outside of the same urban neighborhood park, by measuring in situ microclimatic data. The study is a part of the 5-year (2016–2021) project ‘Application of spatial environmental information and green infrastructure for integrated management of land and environmental plans’ of the National Institute of Environmental Research in the Republic of Korea. The results for 2016 were published in Urban Forestry & Urban Greening in 2021 [71].

2. Materials and Methods

2.1. Study Area

To compare and analyze the thermal environment of an urban neighborhood park and its surrounding areas, a study site was selected around Hyowon park (37°15′47″ N, 127°02′16″ E; area, 141,565 m²) in the city of Suwon, Gyeonggi-do, Republic of Korea (Figure 1a). Suwon is located near Seoul, the capital city of Korea, and its Köppen climate classification is Dwa (temperate, humid continental, severe dry winter, and hot summer). Hyowon park is a typical urban neighborhood park located in central Suwon, which is mainly comprised of commercial areas, low- and high-rise apartments, and other residential areas.

There were five study locations: the lawn area (LCZ D) and park center (LCZ E_B) inside the park, and the commercial area (LCZ 2_B), low-rise apartment (APT) (LCZ 5), and high-rise APT (LCZ 4) outside the park (Figure 1b and Figure 2;

Table 1. The five study locations.

Location	Aspect Ratio	Building Surface Fraction (%)	Impervious Surface Fraction (%)	Height of Roughness Elements (m)	Local Climate Zone (LCZ)
High-rise apartment (APT)	4	16.5	67.8	76.5	LCZ 4	open high-rise
Low-rise apartment (APT)	0.7	19.3	51.9	15	LCZ 5	open mid-rise
Commercial area	0.8	40.8	100	12.9	LCZ 2B	compact mid-rise with scattered trees
Park center	0	0	100	0	LCZ EB	paved with scattered trees
Lawn area	0.01	2	25.1	0.1	LCZ D	low plants

). The ground surface is a 30 m-diameter grass area in LCZ D, cement bricks in LCZ 5, and clay bricks in the other three locations.

2.2. Measurement Methods

2.2.1. Microclimate

During the measurement period, microclimate data were collected at two locations (LCZ D and LCZ E_B) inside the park and three locations (LCZ 2_B, LCZ 4, and LCZ 5) outside the park in the summer from 09:00 on 3 August to 15:30 on 4 August, 2017.

T_a and RH were measured using HMP155A (Campbell Scientific Inc., Logan, UT, USA; http://campbellsci.com (accessed on 26 December 2022)), and u and direction were measured using Met One 034B-L Windset (Campbell Scientific Inc.) every 1 min (

Table 2. Instruments for microclimatic data.

Instrument
Image	Data	Name	Accuracy	Resolution	Manufacture
	Short- and long-wave Radiation	CNR4 Net Radiometer	≤1% (−40–80 °C)	0.1 Wm−2	Kipp & Zonen lnc.
	Air temperature and Relative humidity	HMP155A	Air temp: ±0.3 °C (−80–60 °C) Relative humidity: 2% (0–90%) 3% (90–100%)	0.01 °C 0.01%	Campbell Scientific lnc.
	Wind speed and direction	Met one 034B-L Windset	Wind speed: ±0.1 ms−1 (≤10.1 ms−1) ±1.1% (≥10.1 ms−1) Wind direction: ±4°	0.001 ms−1
	Datalogger	CR1000	±0.06% (0–40 °C)

).

Shortwave (solar) and longwave (terrestrial) radiation were measured every 5 s using a CNR4 Net-radiometer (Kipp & Zonen Inc., Delft, The Netherlands; http://kippzonen.com (accessed on 26 December 2022)). All measuring equipment was installed at a height of 1.1–1.2 m above the ground, which is the standard height in human biometeorology [72]. All measured data were saved in the CR1000 datalogger (Campbell Scientific Inc.).

Measured microclimate data were analyzed for three time periods: daytime on 3 August (the start of measurement to sunset, 09:00–19:29), nighttime from 3 to 4 August (sunset to sunrise, 19:30–05:40), and daytime on 4 August (sunrise to the end of measurement, 05:41–15:30).

2.2.2. Human Thermal Sensation

The human thermal sensation was based on a human energy balance model to quantify the amount of energy absorbed by the human body from the surrounding environment and the amount of energy emitted from the human body to the surrounding environment [73]. Internationally active models include PMV [74] for indoors, and comfort formula [75,76], PET [77,78,79], and UTCI [80,81] for outdoors. These models were used to analyze the thermal environment felt by humans in indoor and outdoor spaces using T_a, RH, u, and T_mrt. In particular, T_mrt was calculated by using the amount of shortwave and longwave radiation absorbed by the human body, assuming that the environment surrounding the human body is in the form of a sphere and emits the same amount of longwave radiation energy from all surfaces of the sphere to the human body. In this study, shortwave and longwave radiation data measured with a CNR4 Net-radiometer designated as the second class by the International Organization for Standardization were used to calculate T_mrt.

A CNR4 Net-radiometer has two pyranometers, a sensor capable of measuring shortwave radiation energy, and two pyrgeometers, capable of measuring longwave radiation energy, so each sensor has a system that measures the amount of radiation energy from the sky and ground hemispheres (http://www.kippzonen.com/Product/85/CNR4-Net-Radiometer, accessed on 22 December 2022). Equation (1) was used to calculate T_mrt (when the area is shaded, $f_p·K_b{}^{+}=0$) [73,74]:

$$\mathrm{Sunny}\mathrm{area}:T_{mrt}={\left[\frac{f_p·K_b{}^{+}·a_k+0.5\left(K_{d+}{K}_r\right)a_k+0.5·{\epsilon }_p\left(L\downarrow +L\uparrow \right)}{{\epsilon }_p\sigma }\right]}^{0.25}-273.15°\mathrm{C}$$

(1)

where ε_p is the longwave radiation emissivity of the human body (0.97), σ is the Stefan–Boltzmann constant (5.67 × 10⁻⁸ Wm⁻²K⁻⁴), and f_p is the projected area factor of the human body affected by direct beam shortwave radiation, depending on the position of the sun (Equation (2)). Equation (2) combines human standing and walking postures and was calculated using the methods of a previous study [82]. K_b is the direct beam shortwave radiation entering the horizontal plane, and K_b⁺ is the direct beam shortwave radiation entering the vertical plane of the human body surface, with the formula K_b/sin (solar elevation, β). a_k is the absorptivity of shortwave radiation energy (0.7), and the constant 0.5 is the area ratio of the amount of radiation that affects the human body due to the direction of light, assuming that the total surface area of the body is 1.0. K_d is the diffuse beam shortwave radiation from an area where the sky is not obscured by buildings, trees, and so on. To obtain the K_d value at the sunny point, the direct beam shortwave radiation entering the CNR4 pyranometer facing the sky was covered with a human fist for 1 min every 30 min from the start time to the end time of the study [72]. K_r is composed of reflected shortwave radiation energy that is reflected by buildings, trees, and the ground surface. L↓ is the longwave radiation from the sky area, buildings, and trees. L↑ is the longwave radiation energy emitted from the ground surface. The human thermal sensation was classified according to

Table 3. Human thermal sensation levels of physiological equivalent temperature (PET) [83] and universal thermal climate index (UTCI; http://www.utci.org (accessed on 26 December 2022)).

Human Thermal Sensation
Thermal Perception	PET (°C)	UTCI (°C)	Grade of Physiological Stress
very cold	<4	<−40	extreme cold stress
		−27∼−40	very strong cold stress
cold	4∼8	−13∼−27	strong cold stress
cool	8∼13	0∼−13	moderate cold stress
slightly cool	13∼18	9∼0	slight cold stress
neutral	18∼23	9∼26	no thermal stress
slightly warm	23∼29		slight heat stress
warm	29∼35	26∼32	moderate heat stress
hot	35∼41	32∼38	strong heat stress
very hot	>41	38∼46	very strong heat stress
		>46	extreme heat stress

using the measured T_a, RH, u, and T_mrt.

$$f_p=3.34·{10}^{-7}{\beta }^3-6.60·{10}^{-5}{\beta }^2+8.42·{10}^{-4}\beta +0.297$$

(2)

2.3. Statistical Analysis

Microsoft Office Excel 2019 was used to obtain the mean and standard deviation results of T_a, RH, u, T_mrt, PET, and UTCI at all five locations. The significance analysis was conducted through one-way ANOVA using the statistical program IBM SPSS Statistics Version 24 (https://www.ibm.com/spss (accessed on 26 December 2022)).

3. Results

3.1. Microclimate

For measurements taken from 9:00 on 3 August to 15:30 on 4 August in summer, the T_a was 24.8–35.1 °C, RH was 43.6–81.4%, u was 0.0–3.9 ms⁻¹, and T_mrt was 18.1–74.5 °C. These results are typical of the hot summer climate in Suwon (Figure 3).

The T_a during the daytime on 3 August (09:00–19:29) appeared in the order of LCZ 2_B > LCZ 5 > LCZ E_B > LCZ 4 > LCZ D. LCZ 2_B showed a mean 0.2–0.4 °C higher value than the other locations, and park inner locations (LCZ D and E_B) and apartment locations (LCZ 4 and 5) showed a very slight difference (Figure 3 and

Table 4. Mean ± standard deviation values of the five locations during 3–4 August. T_a, RH, u, T_mrt, PET and UTCI mean air temperature (°C), relative humidity (%), wind speed (ms⁻¹), mean radiant temperature (°C), physiological equivalent temperature (°C), and universal thermal climate index (°C), respectively.

Data	Location	Mean ± Standard Deviation
		Ta (°C)	RH (%)	u (ms−1)	Tmrt (°C)	PET (°C)	UTCI (°C)
3–4 August (n = 1831)	Lawn area (LCZ D)	29.6 ± 2.5 a*	60.5 ± 8.7 b	0.5 ± 0.4 a	40.6 ± 18.3 a	35.3 ± 10.2 bc	33.5 ± 5.9 a
	Park center (LCZ EB)	29.6 ± 2.6 a	61.4 ± 9.2 c	0.8 ± 0.5 c	45.3 ± 19.3 c	36.5 ± 10.1 d	34.7 ± 6.2 c
	Commercial area (LCZ 2B)	30.1 ± 2.3 c	59.0 ± 8.5 a	1.3 ± 0.5 e	41.6 ± 16.1 a	34.2 ± 7.8 a	33.7 ± 4.9 a
	Low APT (LCZ 5)	29.5 ± 2.7 a	63.0 ± 10.0 d	0.7 ± 0.3 b	43.6 ± 18.6 b	35.8 ± 10.0 c	34.4 ± 6.1 bc
	High APT (LCZ 4)	29.8 ± 2.3 b	61.0 ± 8.6 bc	1.2 ± 0.5 d	43.3 ± 15.8 b	34.8 ± 8.1 ab	34.1 ± 5.0 b
3 August daytime 09:00–19:29 (n = 630)	Lawn area (LCZ D)	31.0 ± 1.0 a	53.7 ± 2.8 b	0.7 ± 0.5 a	49.8 ± 12.6 a	40.0 ± 6.4 c	36.5 ± 3.4 b
	Park center (LCZ EB)	31.2 ± 0.9 b	54.0 ± 2.9 b	1.0 ± 0.4 c	56.9 ± 10.8 c	42.4 ± 5.2 e	38.4 ± 2.9 e
	Commercial area (LCZ 2B)	31.4 ± 0.9 c	52.4 ± 2.1 a	1.7 ± 0.5 e	49.3 ± 11.4 a	37.9 ± 4.5 a	36.1 ± 2.5 a
	Low APT (LCZ 5)	31.2 ± 1.0 b	55.5 ± 3.4 d	0.9 ± 0.3 b	53.0 ± 11.3 b	40.9 ± 5.4 d	37.7 ± 3.0 d
	High APT (LCZ 4)	31.0 ± 0.8 a	54.4 ± 2.3 c	1.6 ± 0.5 d	52.9 ± 8.9 b	39.2 ± 3.8 b	36.9 ± 2.2 c
3–4 August nighttime 19:30–05:40 (n = 611)	Lawn area (LCZ D)	27.5 ± 1.3 c	67.8 ± 5.2 b	0.3 ± 0.3 a	20.1 ± 0.9 a	24.1 ± 1.2 a	26.9 ± 1.0 a
	Park center (LCZ EB)	27.3 ± 1.3 b	69.5 ± 5.2 c	0.5 ± 0.4 b	23.0 ± 1.1 b	25.1 ± 1.5 b	27.6 ± 1.1 b
	Commercial area (LCZ 2B)	28.4 ± 1.5 e	65.4 ± 5.8 a	1.1 ± 0.4 d	24.4 ± 1.4 d	25.9 ± 1.8 c	28.5 ± 1.3 d
	Low APT (LCZ 5)	27.0 ± 1.2 a	72.1 ± 5.1 d	0.5 ± 0.2 b	23.3 ± 1.3 c	25.0 ± 1.3 b	27.8 ± 1.0 c
	High APT (LCZ 4)	28.0 ± 1.4 d	67.5 ± 5.6 b	0.9 ± 0.3 c	26.1 ± 1.2 e	26.4 ± 1.5 d	28.9 ± 1.1 e
4 August daytime 05:41–15:30 (n = 590)	Lawn area (LCZ D)	30.2 ± 2.9 a	60.3 ± 9.7 ab	0.5 ± 0.4 a	52.1 ± 14.4 a	41.8 ± 9.0 b	37.0 ± 5.3 a
	Park center (LCZ EB)	30.3 ± 3.0 a	60.9 ± 10.0 bc	0.9 ± 0.4 c	55.9 ± 16.0 b	41.9 ± 9.1 b	37.9 ± 5.7 b
	Commercial area (LCZ 2B)	30.5 ± 2.8 a	59.6 ± 9.9 a	1.3 ± 0.5 e	51.2 ± 14.0 a	38.7 ± 7.7 a	36.6 ± 4.9 a
	Low APT (LCZ 5)	30.3 ± 3.2 a	61.8 ± 11.1 c	0.8 ± 0.3 b	54.5 ± 17.3 b	41.5 ± 10.0 b	37.8 ± 6.2 b
	High APT (LCZ 4)	30.3 ± 2.9 a	61.2 ± 9.9 bc	1.1 ± 0.4 d	50.7 ± 15.0 a	38.7 ± 8.7 a	36.5 ± 5.4 a

* a–e are the results of significance analysis of one-way ANOVA (p < 0.05).

). In the nighttime (19:30–05:40), T_a showed greater differences between the locations than during the daytime in the order of LCZ 2_B > LCZ 4 > LCZ D > LCZ E_B > LCZ 5. LCZ 2_B and 4 showed 0.9–1.1 °C and 0.5–0.7 °C higher values than the two locations inside the park, respectively. LCZ 5 showed a slight difference from the two locations inside the park. During the daytime on 4 August (05:41–15:30), the T_a results were similar to those taken during the daytime on 3 August, and LCZ 2_B showed a mean 0.2–0.3 °C higher value than the locations inside the park. T_a showed the highest value in LCZ 2_B in both the daytime and nighttime, and the T_a differences among the locations were greater in the nighttime than in the daytime.

RH was higher in the nighttime than in the daytime. In the daytime on 3 August, RH was in the order of LCZ 5 > LCZ 4 > LCZ E_B > LCZ D > LCZ 2_B, and LCZ 2_B showed mean 1.3–3.1% lower values than the other locations. In the nighttime on 3–4 August, RH was LCZ 5 > LCZ E_B > LCZ D > LCZ 4 > LCZ 2_B, and LCZ 2_B showed a mean 2.1–6.7% lower value than the other locations. RH in the daytime on 4 August showed the same result as the daytime on 3 August, and LCZ 2_B showed a mean 0.7–2.2% lower value than the other locations. Both daytime and nighttime showed the highest value in LCZ 5 and the lowest in LCZ 2_B.

The maximum difference in T_a and RH among the locations occurred at 21:00 between LCZ 2_B and 5: the highest T_a location, LCZ 2_B, was 1.9 °C higher than the lowest T_a location, LCZ 5. For RH, the two locations by contrast showed an 8.6% difference.

The u direction was east-west due to the influence of local wind and showed a similar pattern throughout the measurement period. LCZ 2_B showed mean 0.0–0.9 ms⁻¹ and 0.2–0.8 ms⁻¹ higher values than the other locations in the daytime on 3 August and in the nighttime and daytime on 4 August, respectively. Both daytime and nighttime values were in the order of LCZ 2_B > LCZ 4 > LCZ E_B > LCZ 5 > LCZ D. The maximum difference occurred at 11:30, with the value for LCZ 2_B being 1.5 ms⁻¹ higher than that for LCZ D.

T_mrt appeared in the order of LCZ E_B > LCZ 5 > LCZ 4 > LCZ D > LCZ 2_B during the daytime on 3 August, and LCZ E_B showed a mean 7.6 °C higher value than LCZ 2_B. In the nighttime, LCZ 4 > LCZ 2_B > LCZ 5 > LCZ E_B > LCZ D appeared in the order of T_mrt, and LCZ D showed a mean 6.0 °C lower value than LCZ 4. Additionally, the two locations inside the park showed mean 1.4–4.3 °C lower values than LCZ 2_B. In the daytime on 4 August, LCZ E_B > LCZ 5 > LCZ D > LCZ 2_B > LCZ 4 appeared in this order, and LCZ E_B showed a mean 5.2 °C higher value than LCZ 4. During the daytime, the highest value was shown in LCZ E_B, and during the nighttime, the highest value was shown in LCZ 4. The maximum difference of T_mrt among the locations occurred at 15:00, with the values for LCZ E_B and 5 being 22 °C higher than that for LCZ 2_B.

3.2. Human Thermal Sensation

Comparative analysis using PET and UTCI showed that the human thermal sensation was 21.1–60.2 °C for PET and 24.9–45.2 °C for UTCI for the entire study period from 9:00 on 3 August to 15:30 on 4 August (Figure 4 and Table 4). During the daytime on 3 August (09:00–19:29), only LCZ E_B showed a ‘very hot’ level in PET, a mean of 42.4 °C, and a ‘very strong heat stress’ level in UTCI, a mean of 38.4 °C (Table 3 and Table 4). The other four locations showed one lower level, a ‘hot’ level in PET, a mean of 37.9–40.9 °C, and a ‘strong heat stress’ level in UTCI, a mean of 36.1–37.7 °C. When the period of the high thermal environment during the daytime, 10:00–16:00, is considered, LCZ D, 5, and E_B, but not 4 and 2_B, were at a ‘very hot’ level for PET and a ‘very strong heat stress’ level for UTCI. The maximum difference among the locations was a mean of 6.7 °C in PET and 3.7 °C in UTCI, which meant about a one-level difference in PET and a half-level difference in UTCI (Table 3).

During the nighttime on 3–4 August (19:30–05:40), all locations showed a level of ‘slightly warm’ in PET, a mean of 24.1–26.4 °C, and a level of ‘moderate heat stress’ in UTCI, a mean of 26.9–28.9 °C. LCZ 5 with mature trees, with a mean of 25.0 °C, exhibited a thermal environment similar to that of LCZ E_B, with a mean of 25.1 °C. LCZ 2_B, with a mean of 25.9 °C, and LCZ 4, with a mean of 26.4 °C, with a larger volume of buildings compared to trees, exhibited a higher thermal environment than the park locations, LCZ D (mean 24.1 °C) and E_B because the longwave radiant heat did not decrease at night. The maximum difference among the locations was a mean of 2.3 °C in PET and 1.9 °C in UTCI, with about a one-third-level difference in both PET and UTCI.

During the daytime on 4 August (05:41–15:30), LCZ E_B, D, and 5 showed a level of ‘very hot’ for PET, a mean of 41.5–41.9 °C, and LCZ 2_B and 4 showed one lower level, ‘hot,’ in PET, with both showing a mean of 38.7 °C. In UTCI, all locations showed a level of ‘strong heat stress,’ a mean of 36.5–37.9 °C. However, all locations were in the ‘very hot’ level for PET and ‘very strong heat stress’ for UTCI after 10:00. The thermal environment was higher in the daytime on 4 August than in the daytime on 3 August because during the daytime on 4 August there was a clearer sky with less cloud cover than the previous day.

The maximum difference in PET and UTCI among the locations appeared around 15:00, exactly the same as T_mrt. LCZ E_B and 5 were the locations where T_mrt was the highest during the daytime, showing 11 °C higher for PET and 6 °C higher for UTCI compared to those of LCZ 2_B, where T_mrt was the lowest, which means about a two-level difference in PET and one level in UTCI.

3.3. Statistical Analysis

For significance test T_a results among the five locations during the measurement period, there was no significant difference among LCZ D, E_B, and 5, and the other two locations were found to be significantly different. Interestingly, none of the locations were significantly different in the daytime but were significantly different in the nighttime. This appeared to show the dominant effect on T_a of shortwave radiation in the daytime and longwave radiation in the nighttime.

For RH, there was no significant difference among LCZ 4, D, and E_B, but the other two locations were found to be significantly different for the entire measurement period. In the daytime, the mean differences were very small, and no locations showed significant differences. In the nighttime, LCZ D and 4 had no significant difference, but the other three locations were significantly different.

For u, there were significant differences between all five locations except between LCZ E_B and 5 during the nighttime.

For T_mrt, there were no significant differences between LCZ D and 2_B and between LCZ 4 and 5 for the entire measurement period. In the daytime, the same significant differences were shown on 3 August, but there were no significant differences among LCZ D, 2_B, and 4, and between LCZ E_B and 5 on 4 August. In the nighttime, all five locations were significantly different from each other.

For PET and UTCI, all five locations were significantly different in the daytime on 3 August, but there were no significant differences between LCZ 2_B and 4, and among LCZ D, E_B, and 5 in the daytime on 4 August. In the nighttime, all locations were significantly different except between LCZ E_B and 5 for PET.

4. Discussion

Compared with the results of previous studies, a study in Hungary by Unger et al. [60] showed the highest PET, 35.0 °C, in LCZ 2 (compact mid-rise) during the daytime (13:00–14:00) in summer, which was 3.6 °C higher than the lowest PET in LCZ D. This study showed the opposite results; LCZ 2_B had a mean 2.1–3.1 °C lower than that of LCZ D. This appears to be due to the size of the grass field (30 m diameter) in LCZ D and the higher sky view factor (0.925). The size of the grass field is small and affected by surrounding bricks, thus, the evapotranspiration effect might be reduced. Moreover, LCZ D had an 11.6% higher sky view factor than LCZ 2_B (0.809), which means that more shortwave radiation reached LCZ D and created a higher thermal environment than LCZ 2_B during the daytime. During the nighttime (for 2 h after sunset), Unger et al. [60] found that LCZ 2 and 3 (compact low-rise) showed the highest PETs, 18.7 °C and 18.9 °C, which were a mean 3.6 °C higher than that of LCZ D. In this study, the highest PET values were found for LCZ 4 (open high-rise) rather than LCZ 2_B, and the difference between LCZ 2_B and D was a mean of 1.8 °C, which is half the difference compared to the results of Unger et al. [60]. However, the LCZ D of Unger et al. [51] was located in a rural area rather than inside the city, and the analysis time was set to 2 h after sunset when the difference of longwave radiation by the LCZs is low during the nighttime.

In a study of Serbia by Milošević et al. [61], LCZ 5 (open mid-rise) showed the highest PET value during the daytime (4:00–18:00) in summer, which was 3.9 °C higher than that of the lowest, LCZ A (dense trees). In addition, when LCZ 2 was compared with LCZ A and D, LCZ 2 showed 1.8 °C and 1.2 °C higher results than LCZ A and D, respectively. During the nighttime (19:00–3:00), LCZ 2 showed the highest PET results and showed 6.4 °C and 6.9 °C higher results than LCZ D and A, respectively. In this study, LCZ 2_B showed opposite results in the daytime, with a mean of 2.1–3.1 °C lower than LCZ D, when compared with the results of Milošević et al. [61]. Additionally, LCZ 5 had an up to 0.9 °C difference with LCZ D, showing a very slight reduction compared with the results of Milošević et al. [61] (3.3 °C). For the nighttime results of this study, LCZ 2_B showed a much smaller difference of 1.8 °C with LCZ D.

Both of the above two studies showed very low PET values compared to this study and did not have an LCZ 4 (open high-rise) with high-rise buildings, which showed the highest PET results in the nighttime in this study.

A study in Oberhausen, Germany, by Müller et al. [57] showed mostly high PET values in LCZ 5 and 2 and low PET values in LCZ A because of the mixed effects of shading and evapotranspiration.

A study in Hungary by Kovács and Németh [55] reported that the annual average T_a of LCZ 2 during 2001–2010 was 3 °C higher than that of LCZ 6_A (open low-rise with dense trees). The average T_a difference at 12:00 in the summer of 1981–2010 was 3.8 °C.

Kowk et al. [63] showed that the T_a of LCZ 1/2/3 was higher than that of LCZ A and D by 1.9 °C and 1.92 °C, respectively, in the daytime (13:00–16:00) when the solar radiation was the strongest. T_mrt values among the above LCZs showed differences of 0.8 °C and −3.37 °C, respectively. In the nighttime (3:00–6:00) when the urban heat island effect is most prominent [84], LCZ 1/2/3 temperatures were 2.58 °C and 2.82 °C higher in T_a and 9.1 °C and 9.81 °C higher in T_mrt than LCZ A and D, respectively. LCZ 4/5 (open high/mid-rise) had the highest temperature during the daytime for UTCI, and LCZ A had the lowest. Moreover, T_a and T_mrt values of all LCZs decreased with increasing distance from the city center. In this study, the T_a between LCZ 2_B and D was only 0.3–0.4 °C and 0.9 °C in the daytime and nighttime, respectively. The two LCZs also showed small differences in T_mrt of up to 0.9 °C in the daytime and 4.3 °C in the nighttime. For UTCI, LCZ 5 showed higher values among the five LCZs in the daytime. Interestingly, LCZ 4 showed a lower value in the daytime but had the highest value in the nighttime due to a lower sky view factor that meant that less shortwave radiation reached the human body in the daytime and more longwave radiation was emitted by the surrounding tall buildings in the nighttime.

Lau et al. [64] in Hong Kong, the only study that collected all required microclimatic data for human thermal sensation/comfort of the LCZs, showed that LCZ 2, with a mean value of 55.8 °C, had the highest T_mrt in the summer daytime (10:00–16:00) in June–September, and it was found to be 21.5 °C higher than that of the lowest LCZ D, with a mean of 34.3 °C. Even for PET, LCZ 2 showed a mean value of 44.0 °C, which was 11.0 °C higher than the LCZ D value of 33.0 °C, which was the lowest value. In this study, LCZ E_B had similar T_mrt and PET results as Lau et al. [64], but LCZ D had much higher results due to the small size of the grass field and the high sky view factor.

Therefore, sky view factors and surface materials were the key variables of the thermal environments of LCZs. Vegetation surfaces such as grass fields can reduce thermal environments through evapotranspiration. Higher sky view factor LCZs will have more shortwave radiation in the daytime. By contrast, lower sky view factor LCZs will have more longwave radiation in the nighttime. However, the locations and compositions of the same LCZs can create various thermal environments.

5. Conclusions

In this study, the thermal environments of five LCZs, two in the Hyowon neighborhood park and three with different land uses outside the park, in summer were investigated. The thermal environment was evaluated by measuring T_a, RH, u, and shortwave and longwave radiation from 9:00 on 3 August until 15:30 on 4 August, 2017.

The human thermal environment index was ‘hot’–‘very hot’ for PET and ‘strong heat stress’–‘very strong heat stress’ for UTCI during the daytime, and ‘slightly warm’ for PET and ‘moderate heat stress’ for UTCI during the nighttime. Two locations (LCZ D and E_B) in the park and LCZ 5 showed similar thermal environments, and LCZ 2_B and 4 showed higher thermal environments than locations in the park. The maximum mean difference by location was 6.7 °C for PET during the daytime and 3.7 °C for UTCI, showing a one-level difference in PET and a two-thirds-level difference in UTCI. During the nighttime, the maximum mean difference by location was 2.3 °C for PET and 1.9 °C for UTCI, both of which were lower level differences, a one-third level, than those during the daytime. The maximum difference between locations was the same as T_mrt for both PET and UTCI at around 15:00. Compared to LCZ 2_B, E_B and 5 showed a maximum difference of 11 °C in PET and 6 °C in UTCI, showing a difference of two levels in PET and one level in UTCI.

The limitation of this study is that the measured climatic data were location data only in sunny areas. If a computer simulation was used, whole area data could have been acquired, but location data can only provide the typical climatic data of each micro-scale area. Thus, transitional data among different microclimatic areas could not be obtained. Moreover, in this study, the proportion of shaded areas in the LCZs was not considered. Areas with high-rise buildings, such as LCZ 4, would have different thermal results for the whole area if the shaded portions were included in the analysis. However, this study used all required microclimatic data for human thermal sensation/comfort analysis of the LCZs instead of using computer simulation programs. More in situ microclimatic data (T_a, RH, u, and T_mrt) should be accumulated as a database for the studies of thermal environments of LCZs.

In future studies, computer simulation results with measured microclimatic data should be considered together to investigate the thermal environments of various LCZs.