A Literature Survey of Local Climate Zone Classification: Status, Application, and Prospect

Abstract

Rapid urban expansion and climate change have prompted further investigations into urban thermal climates and the development of local climate zone (LCZ) classification systems. LCZs, proposed 10 years ago, comprise a new and systematic classification of field sites for heat island studies to provide a reference for future LCZ research, so that scholars can understand what research has been done and identify future research trends. We analyzed LCZ studies in a database from 2012 to the present, and identified recurring themes using VOSviewer software, including LCZ mapping, measurement methods, thermal environments, and outdoor thermal comfort, among others. A systematic evaluation was performed using bibliometric analysis in the PRISMA framework—190 relevant studies were selected for subsequent analysis. Descriptive analysis showed that LCZ research has received increasing attention, particularly in China, where more than 60% of the LCZ studies were conducted. The results showed that the maximum number of articles on all themes was 57 articles on LCZ mapping, followed by studies of the thermal environment (UHI/SLT). It is hoped that this article will provide scholars in this area with an understanding of the research that has been conducted and the methods used, and provide insight into future research directions.

1. Introduction

During the 20th century and since, urban population growth has been accelerating, resulting in rapid urban development, industrialization, overall population growth, and densification. Natural landscapes are being replaced by a large number of artificial surfaces, buildings, and roads, significantly changing the urban space and the urban climate [1,2]. The urban heat island (UHI) effect is accelerating due to rapid urbanization and land cover change [3].

The UHI is the phenomenon where the ambient temperature in a city is higher than that in adjacent rural areas [4]. This is caused by cushion surface changes, climate condition changes, and artificial temperature releases in the process of urban expansion under different regional climate backgrounds [5]. In addition, automobile exhaust, air conditioning, and other artificial heat sources have an important impact on the thermal environment in cities. These artificial heat sources enhance the convection and turbulence in the air and strengthen UHI effects [6].

Before the emergence of “local climate zone” (LCZ) algorithms, most studies used the temperature difference between the urban core and the suburbs to obtain the urban heat island intensity (UHII), one of the most commonly used indicators to characterize UHIs [7]. With the process of urbanization and expansion, the boundary between “urban” and “rural” becomes blurred, and the suburban comparison method is too simple, such that is impossible to accurately describe the UHI. This provides no way to accurately describe the spatial heterogeneity of the near-surface temperature caused by the urban underlying land cover described by Stewart (2011). For this reason, Stewart and Oke (2012) introduced the LCZ concept to classify “urban” and “rural” land cover more objectively, using development effects on local temperatures, and proposed the use of LCZ X – LCZ D to represent the UHI (temperature of UHI = the temperature of any LCZ (LCZ X) minus the temperature of LCZ D) [7].

In recent years, there have been increasing studies on LCZs, globally dominated by research in China [8] and the United States [9]. Other countries and regions have also carried out empirical research combined with LCZ. Although the first proposed LCZ was intended for the field site classification in UHI studies [10], it is now being used in a variety of applications. In its first decade, we reviewed LCZ applications and summarized the range of represented research topics with the help of biometric tools, identifying future trends for LCZ applications in urban research.

Oke (2006) proposed the standard “Urban Climate Zone (UCZ)” at the 13th Committee of Instruments and Methods of Observation (CIMO) conference. In 2012, he divided this type of area into different, uniform sections based on characteristics previously identified, naming this classification “Local Climate Zone (LCZ)” with Stewart [11].

The primary goal of the LCZ is to identify a standard selection of urban measurement points with a recording of UHI intensity [11]. This study shows that thermal contrasts exist across all LCZ categories, which are largely determined by building height and spacing, impervious surface fraction, tree density, and soil moisture [10]. The LCZ program aims to expand and improve on previous work on the UCZ classification, increasing the number of land categories to 17, including 10 “Built Types” categories and seven “Land Cover Types” categories. The standard grade comprises 10 building types (compact high-rise, compact mid-rise, compact low-rise, open high-rise, open mid-rise, open low-rise, lightweight low-rise, large low-rise, sparsely built, and heavy industry) and seven land cover types (dense trees, scattered trees, bush/scrub, low plants, bare rock or paved, bare soil or sand, and water), with different microclimate conditions (Figure 1). The range is defined as a region crossing hundreds of meters to 200 m, with similar physical surface properties that can affect the local climate, including sky view factor (SVF), building surface fraction, aspect ratio, impervious surface fraction, terrain roughness grade, and height of roughness elements. The index provides a general framework for UHI research, which uses the temperature difference between LCZ (∆T = LCZ X − LCZ D) to facilitate the quantification of UHI intensity. Following the LCZ’s original development, there have also been occasional efforts to demonstrate its applicability [12].

The purpose of this work was to search for and synthesize all of the available literature on LCZ in urban environments and to examine how these LCZ methods were used in the research. For this purpose, three main objectives were identified:

–

Summarize the last ten years of LCZ-related literature for summary analysis.

–

Analyze the main LCZ application themes.

–

Forecast the future LCZ research directions.

2. Review Methods

We used the following three steps for the review analysis of LCZ.

2.1. Article Screening

The articles were systematically reviewed using the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) framework [13]. Generally, systematic evaluations include the following five necessary steps: develop research issues and their scope, collect data and define sources, filter data and obtain relevant data, and finally summarize [14]. We chose “local climate zone” as the main keyword phrase in the “article title” field. Two academic research databases were selected to identify and extract the relevant scientific articles, namely, Web of Science (WoS) and Elsevier. Since the concept of “local climate zone” was proposed in 2012, our search selected all references of the literature for the period from 1 January 2012 to March 2022. The literature selection process consisted of five steps (Figure 2):

• The query process, yielding many results for two databases: 170 articles from the WoS core collection and 94 articles from Elsevier, resulting in a total of 264 articles.
• Screening alignment of the papers from the two databases, removing duplicates for further analysis, resulting in a total of 202 papers.
• Filtering both databases to identify articles within the scope of this review. Filtering the types of articles, limiting the search to research and review articles only, while excluding social material, meeting, and comment, etc. After filtering, 195 articles were obtained.
• Screening titles/abstracts/methods to identify studies that only mentioned LCZ and did not actually apply it.
• Finally, 190 related articles on LCZ were obtained.

Figure 2. Flow chart of the literature selection and review process for the WoS and Elsevier databases using the initial keywords.

Figure 2. Flow chart of the literature selection and review process for the WoS and Elsevier databases using the initial keywords.

2.2. LCZ Overall Visualization Analysis

Several aspects of the LCZ study content related to this review were used in the analyses using VOSviewer software (1.6.18), which developed by Van Eck and Waltman of The Centre for Science and Technology Studies (CWTS) in the Netherlands in 2009 is a free software. This software supports the data retrieved from the WoS and Elsevier databases, but these data cannot be analyzed simultaneously, so each database was analyzed and visualized separately. In the analysis, the data were merged with matching terms. For instance, themes such as “local climate zone”, “local climatic zones” and “LCZ” were merged with “local climate zone” to avoid confusion with more terms than actually occur.

The results were analyzed using the following criteria:

• Keyword co-occurrence of the authors in the articles selected using WoS and Elsevier.
• The most common sources of LCZ-related papers and the citation links between them are from WoS and Elsevier.
• Top reference sources for LCZ-related papers and reference links in WoS and Elsevier.

2.3. LCZ Study Classification

The classification of the article contents based on the selected papers in step 1 primarily used the following scientific research fields: LCZ mapping, measurement methods, thermal environments, and outdoor thermal comfort, and others (Figure 3).

3. Results

3.1. Descriptive Analysis

3.1.1. Geographical Patterns

There is a concentration of LCZ studies in three countries (Figure 4). China has published up to 119 articles, followed by 33 in Germany, and 28 in the United States. Globally, China is the leading country in Asia, followed by 159 related articles published in 14 countries, including India and South Korea. European countries have since published many articles, including 113 studies from 12 different countries. In contrast, Africa and Oceania have published fewer relevant articles.

The statistics are only relevant to the study site (Figure 5). Some research compared LCZ among regions (e.g., [12,15]), while others were reviews that were not counted (e.g., [16]).

3.1.2. Keywords

We identified 54 keywords in the LCZ literature and at least four simultaneously (Figure 6). The keyword “LCZ” is most commonly mentioned, together with “UHI” and “urban climate”, followed by “WUDAPT”, “urban morphology”, and “LST”.

3.1.3. Annual Distribution

The number of studies related to LCZ has increased dramatically, from 2 in 2012 to 57 in 2021, demonstrating an increasing global interest in LCZ (Figure 7).

3.1.4. Contributing Journals

The journal Urban Climate had the most published articles (43), followed by Building and Environment and Remote Sensing ranked second and third, with 22 and 16 records, respectively (

Table 1. Publication journal ranking.

	Journal	Number
1	Urban Climate	43
2	Building and Environment	22
3	Remote Sensing	16
4	IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensing	10
5	Sustainable Cities and Society	9
6	Sustainability	7
7	ISPRS Journal of Photogrammetry and Remote Sensing	6
8	International Journal of Applied Earth Observation and Geoinformation	5
9	Atmosphere	5
10	ISPRS International Journal of Geo Information	3
11	Climate Research	3
12	International Journal of Biometeorology	3
13	International Journal of Climatology	3
14	Journal of Geophysical Research Atmospheres	3
15	Landscape and Urban Planning	3

). These journals cover a wide scope of disciplines, for example, meteorology, environmental science, remote sensing technology, building science, and urban science.

3.1.5. Article Citation Frequency

The most-cited paper in the top 10 most-cited papers is the article entitled “Local climate zones for urban temperature studies” by Stewart and Oke (2012), which proposed the concept of LCZ to be the origin of LCZ (1441 citations as of March 2022) (

Table 2. Top 10 most-cited papers identified in our literature search.

	Title	Number
1	Local climate zones for urban temperature studies [11]	1441
2	Impact of urban form and design on mid-afternoon microclimate in Phoenix Local Climate Zones [9]	308
3	Mapping local climate zones for a worldwide database of the form and function of cities [17]	268
4	Evaluation of the ‘local climate zone’ scheme using temperature observations and model simulations [10]	265
5	Using Local Climate Zone scheme for UHI assessment: Evaluation of the method using mobile measurements [18]	133
6	Classification of local climate zones based on multiple Earth observation data [19]	127
7	Local Climate Zone ventilation and urban land surface temperatures: Towards a performance-based and wind-sensitive planning proposal in megacities [8]	122
8	Design of an urban monitoring network based on Local Climate Zone mapping and temperature pattern modelling [20]	91
9	Employing an urban meteorological network to monitor air temperature conditions in the ‘local climate zones’ of Szeged, Hungary [21]	83
10	Land surface temperature differences within local climate zones, based on two central European cities [22]	75

).

3.2. LCZ Mapping

3.2.1. Simple Division of City Types Using the LCZ Classification

Generating urban LCZ maps is an important part of the related research. After reviewing all retrieved articles, 57 were related to LCZ mapping, some of which focused on the generation of LCZ classification maps in each city. There are three main methods of LCZ classification: field measurements, remote sensing (RS), and geographic information system (GIS). Among them, the method based on remote sensing images—World Urban Database and Access Portal Tools (WUDAPT)—is most commonly used for the relative comparison of LCZ mapping research [17], which is a general method based on the Landsat database and the random forest (RF) classifier used in LCZ mapping, and is used in Hong Kong, China [23], Zimbabwe, Africa [24] and France [25]. The WUDAPT method was developed to promote the LCZ classification of urban climate studies at the city scale. Based on globally available satellite imagery from the Geological Survey and training samples from Google Earth (GE), the WUDAPT method can be processed in a standard workflow, applying machine learning using open-source software (System for Automated Geoscience Analysis, SAGA) to simulate human interpretations of the identification of LCZ classes [17].

A number of studies map urban LCZ using at least two of the above classification methods. Taking into account differences in urban microclimates within and between cities, the WUDAPT (Figure 8) has been developing an international database of global cities, using unanimous methodological and tool frameworks [26] (

Table 3. Basic analytical methods used in LCZ mapping.

Method	Articles	Study Site	LCZ Types
WUDAPT	[26]	London, England	All types
	[25]	Sendai, Japan	LCZ 2, 24, 3, 42, 5, 6-10, A, B, D-G
	[27]	Kyiv and Lviv, Ukraine	All types
	[28]	Madrid, Spain	All types
	[29]	Bavaria, Germany	LCZ 2, 5, 6, 8, A, B, D, G, E
	[30]	15 European cities	All types
	[31]	20 individual cities and three major economic regions in China	All types
GIS-based and WUDAPT	[32]	Hong Kong, China	All types
GIS-based	[23]	Hong Kong, China	LCZ 1-6, 9, A, C, D-G, W
	[20]	Szeged, Hungary	LCZ 2, 3, 5, 6, 8, 9
	[33]	Szeged, Hungary	Six built LCZ types
	[34]	Brno, Hradec Králové, and Olomouc, Czech Republic	LCZ 1-6, 8-10, A-G
Earth observation (EO) datasets and classifiers seemed	[19]	Hamburg, Germany	All types
Sentinel-2 data	[35]	13 cities across the world	All types
	[36]	Munich and Berlin, Germany	All types
	[37]	Shenzhen, China	All types
ECOSTRESS data	[38]	Xi’an, China	LCZ 1-6, B
Sentinel-1 Dual-Pol data	[39]	29 cities across the world	All types

).

3.2.2. Optimizing LCZ Classification

In the traditional deep learning methods such as RS and GIS, the accuracy in the urban-type LCZ (LCZ1-10) remains relatively low because RS datasets cannot provide vertical or horizontal building components in detail [40]. Building information is typically obtained from the National Geospatial Information Agency or OpenStreetMap (OSM) building footprint data [27,41]. Some novel methods can improve this issue [42,43,44]. GIS-based building datasets can be used as a primary source for LCZ classification, but their use as input data for convolutional neural networks (CNNs) is limited due to their incompleteness [32,45]. Some recent studies have applied deep learning methods to further improve LCZ classification accuracy using RS data, which makes CNNs important [35,46].

There have been many studies improving the accuracy of LCZ mapping methods, most of which have proposed an upgraded version of the WUDAPT method or have compared WUDAPT L0 data, including a team at the University of Hong Kong and others to evaluate and refine the Chinese LCZ map [31,47,48]. In Colombo, Sri Lanka, using WUDAPT to make LCZ maps for initial classification, a large number of LCZs were subdivided to better represent the local conditions of this high-density tropical city [49].

Other studies differ from the WUDAPT methods and have considered it to be insufficient in accuracy, proposing further research to improve the methods of mapping LCZ. For example, a study introduced grid-based mapping methods using 1 km, 500 m, and 250 m for LCZ identification and mapping [50]. Scholar Hidalgo proposed that although the average building height standard in France corresponds to the LCZ classification, the boundary between the open and sparsely built areas was determined using the k-means statistical method [51] (

Table 4. LCZ mapping improvement method.

Method	Articles	Study Site	Improved
WUDAPT	[30]	24 cities in Europe	Presents two strategies to expand LCZ coverage rapidly
	[35]	13 cities around the world	Proposes a multi-task learning (MTL) framework and develops an end-to-end convolutional neural network (CNN)
	[42]	Hamburg, Germany	WUDAPT + OpenStreetMap (OSM)
	[52]	15 cities that cross a series of climates	Google Earth Engine (EE) is a viable alternative approach compared to WUDAPT method
	[53]	New York, USA Nantes, France	The Delaunay triangulation versus a skeletonization
	[54]	Guangzhou, China	A block-based LCZs classification
	[55]	St. Petersburg, Russia Bamako, Mali Havana, Cuba	Simplified WUDAPT (the LCZ Generator described here further simplifies this process)
	[56]	Three cities in Belgium: Antwerp, Brussels, and Ghent	Due to the high degree of heterogeneity, neighborhood information on the LCZ mapping overcomes the difficulties of the WUDAPT community
	[57]	Manhattan, New York, USA; Atlanta, USA; Tokyo, Japan;	A fine-grained and 3D approach
	[58]	Vienna, Austria	GIS-based and WUDAPT
	[59]	Berlin, Germany; São Paulo, Brazil; Paris, France	A self-training classification framework (SCSF) based on spatial context information is proposed to address the lack of a unified framework for classification and spatial smoothing
	[60]	Xi’an, China	This method couples the World Urban Database and Access Portal Tools method
	[61]	Khartoum, Sudan	The methodology uses synthetic aperture radar (SAR) data
	[62]	Guangzhou and Wuhan, China	Combining Landsat and ASTER data to improve the LCZ mapping results
Others	[40]	Berlin, Germany; Seoul, South Korea	Provides novel methods using S2 images and incomplete building data based on a CNN classifier to classify LCZ
	[44]	11 cities around the world	This paper uses OpenStreetMap (OSM) data and multi-temporal, multispectral satellite images for the classification of LCZs
	[46]	Eight German cities	Proposes a workflow to generate LCZs using multi-temporal Sentinel-2 (S2) composites and supervised CNNs
	[48]	Guangzhou, China	Simplified LCZ method
	[49]	Colombo, Sri Lanka	Simplified LCZ method
	[50]	Nagpur, India	“Reduction-based GIS method” and “algorithm-based remote sensing” approach
	[51]	Nantes, Toulouse, and Paris, France	This paper uses GIS using administrative and 2.5 D building databases to generate the LCZs
	[53]	New York, USA; Nantes, France	Tries to translate this classification to a fine-grained level of detail
	[63]	Busan, Daegu, Daejeon, Gwangju, Incheon, and Seoul, South Korea	The study uses a multi-scale, multi-level attention network (MSMLA-Net) for LCZ classification
	[64]	Beijing, China	This paper proposes an enhanced GIS-based workflow to enable the hierarchical classification of LCZs with fewer indicators, but higher accuracies
	[65]	Wuhan, China	This paper combines neural networks and two graph convolutional networks (GCNs) into and directly using original image-generated LCZs
	[66]	Wuhan, China	This paper uses multi-source-generated LCZs
	[67]	Bandung, Indonesia	The result shows that the object-based image analysis (OBIA) provides higher accuracy in LCZ
	[68]	Basel, Switzerland	A method is proposed to bridge the gap between the classical land use/land cover (LULC) classification and the LCZ scheme
	[69]	Sydney, Australia	Examines the quality of the classification by analyzing the LST variability between LCZs using high-resolution airborne remote sensing data
	[70]	Eight metropolitan areas in the USA	Random forests and ResNets are applied for LCZ classification
	[71]	Beijing, Tianjin, and Wuhan, China	Uses freely available Sentinel-1 SAR and Sentinel-2 multispectral imagery to generated LCZs
	[72]	Shanghai, China	Provides a promising option for LCZ mapping

).

In the traditional deep learning methods, the accuracy in the urban-type LCZ (LCZ 1-10) remains relatively low because a remote sensing dataset cannot provide vertical or horizontal building components in detail. Some novel methods can improve this issue.

3.3. Measurement Methods

In addition to the LCZ studies using satellite data (in Section 3.2), there are other measurement methods, including both stationary and mobile measurements, as well as studies of LCZs using models.

3.3.1. Fixed-Point Observations

In most studies, placing measurement instruments in a fixed position for data collection and subsequent statistical analysis is a common research approach. Yang et al. simultaneously collected temperature and humidity data from 14 different LCZs in Nanjing using a fixed-point measurement method within 401 days [73,74]. LCZ air temperature data were collected at Parana using air temperature sensors with 10 built-in data loggers from 27 August to 15 December 2018 [75]. These data complemented the temperature data from five Spanish State Meteorological Agency (AEMET). The results showed that a relatively high UHI intensity has been observed in Madrid, reaching above 5 °C [76].

There have also been LCZ studies using data from local climate stations. One was located in Novi Sad, Serbia, to study the spatiotemporal dynamics of the official weather stations of the Hydrometeorological Service of the Republic of Serbia (RHMZ) and the relationship between air temperature and five air humidity variables (relative humidity, water vapor pressure, absolute humidity, specific humidity, and water vapor pressure deficit), based on two-year data [77].

In Berlin, Germany, they analyzed a 2015 annual dataset of near-surface air temperatures (AT) from the Cyber Movement Citizen Weather Station (CWS) and the surrounding environment. The data, combined with the concept of the LCZ, showed the applicability of a new dataset for urban climate research [78].

3.3.2. Mobile Measurements

Mobile measurement methods have been widely used in UHI or urban climate studies. Contrary to fixed-point observations, mobile measurement is a more cost-effective approach in investigating urban environments [79] (

Table 5. Mobile measurements used in LCZ studies.

City	Articles	Mobile Measurements	Equipment
Nancy (France)	[18]	Instrumented vehicle	PT100 probe
Nancy (France)	[80]	Instrumented vehicle	PT100 probe
Chongqing (China)	[81]	/	/
Dijon (Eastern France)	[82]	Cargo bike	HOBO U23 v2 sensor located in an M-RSA solar radiation shield and attached to a data logger
Hong Kong (China)	[83]	Light-colored Toyota bB mini MPV vehicle	TESTO™ 480 Thermometers Garmin™ GPS Map 62 s Handheld Navigator Wide-angle video camera

).

The summer air temperature was recorded in the 13 LCZs in Nancy, France, in 2013 and 2015, with recordings made every three meters in a moving vehicle. To study the effect of speed on temperature, a sensitivity test was performed and the results showed that the average temperature value was little affected by vehicle speed [18,80]. In Chongqing, the authors used an instrumented vehicle for automatic loading temperature and humidity data loggers (HIOKI 3641) and a GPS data logger (GlobalSat DG-100). Five types of LCZ were measured [81]. In Dijon (eastern France), urban temperatures were analyzed using bicycle mobility measurements to quantify the effect of urban morphology on microscale variations in temperature [82]. An evaluation was made on the local climate zone classification in high-density heterogeneous urban environments using mobile measurements [83].

Fixed-point observations can acquire more data at the same LCZ location, and in contrast, mobile measurements can acquire data from more locations with fewer instruments; thus, to classify many types, research requires more instruments for which this measurement is suitable. Scholars can choose different measurement methods according to the needs and purposes of the data within the same LCZ type.

3.3.3. Urban Climate Models

Urban climate models are most commonly used as an LCZ research method (

Table 6. Models frequently used in LCZ studies.

Models	Number of Studies	Articles
Weather Research and Forecasting Model (WRF)	4	[86,87,88,89]
Solar and LongWave Environmental Irradiance Geometry-model (SOLWEIG)	4	[90,91,92,93]
MUKLIMO	2	[94,95]
Three-dimensional Microclimate Model	1	[9]
Urban energy Balance Calculation Model (UDC model)	1	[54]
SUEWS	1	[84].
Urban Canopy Layer (UCL) Model	1	[85]
Digital Surface Model (DSM) and normalized Digital Surface Model (nDSM)	1	[96]
Ozone Chemical Transport Model	1	[97]
Surface Energy Balance over Land (SEBAL) model	1	[98]
UrbClim model	1	[99]

). Ariane Middel used model research as early as 2014, and it was also the second most-cited research article. The author used a three-dimensional microclimate model, ENVI-met, to simulate five LCZ types in Phoenix [9]. Alexander proposed that, although in recent years there have been models to describe urban surfaces and to simulate their climate effects, there is no solution to the data processing requirements of more complex urban models. Thus, they used the running Surface Urban Energy and Water Balance (SUEWS) energy balance model combined with LCZ as a solution [84]. In 2020, Liu proposed an application of the LCZ concept to quantify urban blocks, and developed the urban canopy layer (UCL) model to propose a visualization-based spatial form optimization strategy for local-scale urban blocks, developed and validated using Dynamic Local Energy Balance Model—V1.0 software [85].

3.4. Thermal Environment

Since LCZ was first proposed to facilitate further research on UHI, the number of articles in this field is the largest in all research fields.

3.4.1. Urban Heat Islands (UHIs)

UHIs refer to a higher temperature in urban areas compared to that of the surrounding environment. This temperature difference is generally larger during the day [100], and is closely related to an assessment of the thermal environment in urban areas [101]. In 2012, Oke proposed the use of the equation ∆T = LCZx − LCZD to describe and calculate urban heat island intensity. This method has already been used in a number of cities around the world. For example, Benjamin Bechtel, from Ruhr-Universität Bochum in Germany, analyzed and compared the surface urban heat islands (SUHI) of LCZs in 50 cities around the world [12]. During a period of 401 days, Yang et al. used fixed-point measurement to simultaneously collect the temperature data of 14 different LCZs in Nanjing to evaluate the Nanjing UHI [73]. However, using RS datasets on the LCZs to study the UHI is not enough, so many scholars have used other data to research the study area of UHI. Zhou analyzed Sendai’s UHI in Japan based on 12 GIS-derived LCZ classifications with geometric and land cover properties, and proposed two specific UHI mitigation strategies for Sendai [102]. In Dublin, Ireland, the LCZ air temperature was obtained using metadata from six fixed weather stations, and the LCZ was derived as Dublin’s UHI [103]. These results suggest that under conditions favorable for the development of UHIs, on average, an LCZ with high impermeability/building coverage is >4 °C higher than an LCZ with high permeability/vegetation coverage at night.

3.4.2. Land Surface Temperature (LST)

Understanding temporal and spatial variation in land surface temperature (LST) and the thermal impact of varying urban landscapes is critical for developing strategies to improve urban thermal conditions. The spatial distribution of heat in urban areas is primarily assessed via site air temperature or satellite-based LST measurements [18]. There are many studies in China exploring the relationship between LCZ and LST. For example, high-resolution remote sensing was used in Xi’an to explore the relationship between the LCZ-based 2D/3D urban form and LST inter-seasonal spatial changes in image and vector building data [38,104]. They also used MODIS to explore the urban surface thermal performance of the LSTs of three urban areas in China [105].

There have also been some related studies in the Czech Republic. In Prague and Brno, the two largest cities in the Czech Republic, a new GIS-based approach to LCZ demarcation was employed, using both LSTs from land and satellite data to explore how well LCZ categories differentiate LSTs [22]. Another paper attempts to combine LCZ and LSTs using a complex approach derived from a case study in Hradec Kralove, the Czech Republic [106].

3.5. Outdoor Thermal Comfort

Thermal comfort is a subjective evaluation of human satisfaction with the surrounding thermal environment (ISO 7730). Outdoor thermal comfort is closely related to the surrounding climate and environment, and the research in this field has been relatively rich, including such elements as heat comfort in cold areas [107,108,109,110], outdoor thermal comfort for special population groups [111,112], heat and comfort at tourist attractions [113], heat experiences in different populations and their effect on outdoor heat comfort [114], thermal comfort among outdoor plants [115,116], and the general impact of thermal comfort [117], but this article explores the articles combining OTC and LCZ to explore people’s comfort index, which aims to improve the quality of the thermal environment.

Part of the study on thermal comfort used a field measurement and a survey questionnaire to evaluate the relationship between LCZ and outdoor thermal comfort. For example, in 2014, the author Villadiego performed an OTC study in the hot and humid climate of Barranquilla, Colombia [118]. Five LCZ types were selected for physical measurements and they used the ASHRAE sensorial scale of seven symmetrical points to assess people’s subjective thermal sensation and thermal comfort among LCZs [118]. In Hong Kong (2019), Lau et al. evaluated the microclimate conditions and the subjective perception of the thermal environment in eight LCZs in sub-tropical high-density cities, including Hong Kong [119], among others.

Some research has also used models to describe LCZs. For example, Salal Rajan and Amirtham measured it at the three LCZs of Chennai, and used an “ENVI-met” model simulation to estimate the thermal comfort index (Tmrt + PET) [120]. Geletic used the MUKLIMO_3 (Mikroskaliges Urbanes KLImaMOdell in 3-Dimensionen) urban climate model in Germany to analyze outdoor thermal comfort conditions by HUMIDEX during the summer of 2015 (4–14 August), using meteorological data from 14 sites [95]. In Zhengzhou, Ren et al. used predicted mean vote (PMV) to calculate the OTC based on LCZs, and studied the correlation between LST and the vegetation-type effects of LCZ [121].

In addition, LCZ and thermal comfort were quantified by Lin Liu et al. First, nine typical urban LCZs were selected for a field survey. Outdoor thermal comfort levels with different urban spatial characteristics were assessed using a statistical method based on the fuzzy-AHP method “Quantitative”. Through physical meteorological measurements and a subjective questionnaire, the thermal “neutral” value ranges were found for the three thermal comfort indicators, (TCI)-PET, OUT-SET*, and UTCI; linear, logarithmic, and cubic models were applied, and Fi (a parameter that describes/relates to thermal comfort indicators) and TCI values were calculated for the nine LCZs. The quantitative equation between Fi and TCI is expressed in their work [122].

Another data source for a study of thermal comfort and LCZ is the Hungarian Meteorological Service (HMS). This study found that the most pleasant areas during the day were those with fewer buildings, but the opposite was true at night. This work was applied at the urban scale, thus breaking the previous limitations of LCZ in OTC, which were at the micro scale [123].

Articles on thermal comfort tended to be based on small-scale research, regardless of the method used. Several typical LCZ types were selected to use different thermal comfort indexes for outdoor thermal comfort research (

Table 7. Research on LCZ and thermal comfort.

Method	Articles	Type	City, Country	LCZ Types	Thermal Comfort Index
Simulation	[91]	SOLWEIG	Vancouver, Canada	6 LCZs	Tmrt
	[93]	LongWave Environmental Irradiance Geometry (SOLWEIG)	Brazil	LCZ 1 and 3	Mean Radiant Temperature (Tmrt)
	[95]	MUKLIMO_3 urban climate model	Brno, Czech Republic	LCZ 2-10, A-E, G	HUMIDEX
	[120]	ENVI-met	Chennai, India	LCZ 2. 3. 5. 6	Physiological Equivalent Temperature (PET)
	[121]	ENVI-met	Zhengzhou, China	All types	Predicted Mean Vote (PMV)
	[124]	BioKlima 2.6 model	Shenzhen, China	11 LCZ samples	Universal Thermal Climate Index (UTCI)
Questionnaire	[118]	Seven-point ASHRAE scale	Barranquilla, Colombia	5 LCZs	Mean Thermal Sensation Votes (MTSV)
	[119]	Seven-point ASHRAE scale	Hong Kong, China	LCZ 1-6, LCZ D, E	PET
	[122]	Seven-point ASHRAE scale	Shenzhen, China	9 LCZs	PET, OUT-SET* and UTCI
Weather station	[123]	Hungarian Meteorological Service (HMS)	Szeged, Hungary	6 LCZs	PET

).

3.6. Urban Planning and the Urban Environment

3.6.1. Urban Planning

Urbanization is determined by geographic location, local policies, and local circumstances. City managers most urgently need to fully consider climate adaptation issues in land use planning and implementation to improve climate resilience [125]. Therefore, many researchers combine LCZ with local policies with the purpose of enabling decision makers to propose better urban planning solutions and to improve the wellbeing of residents. Urbanization is one of the major drivers of global thermal environmental change and is changing rapidly and on a large scale, especially in China [126]. Kunming (China) [127] provides a simple framework to monitor rapid urban expansion and changes in urban form, and transfer/transform it to urban practitioners, enabling them to formulate sound urban planning strategies for sustainable urban development. Comparing the two high-density cities of Hong Kong and Guangzhou (China), this research provided a fast and effective way to understand the thermal environment within high-density cities and will help to better plan to improve urban thermal conditions and ensure more sustainable development [128].

The primary task of mapping LCZs in Indian cities has been shown to be very challenging without properly implemented spatial development plans, an understanding of complex urban morphology, and the issue of insufficient data. This information would provide city planners with the correct configuration of LCZ maps to understand how to concentrate on developing areas for UHI mitigation [129]. According to Perera and Emmanuel (2018), in Colombo (Sri Lanka), current planning regimes cannot address the challenges posed by individual local, regional, and global warming, and a simpler approach is proposed that uses the LCZ system for the environmental analysis of data-poor developing cities [59].

3.6.2. Urban Environments

Populations continue to move into cities, accelerating the process of urbanization. Environmental issues are the key to comfort solutions in urbanization. Numerous studies highlight the impact of respiratory particulate matter on human health [130]. A study in the English Bazar Municipality [131] attempted to correlate respiratory particulate matter (RPM) and LST in each LCZ to determine the effect of RPM on LST. Nine LCZs were identified in this study area. The results showed that LCZ3, LCZ5, LCZ6, LCZ7, and LCZD had relatively high LSTs and RPM, with important spatial seasonal variation. Particulate matter <2.5 µm in aerodynamic diameter (PM2.5) has been recognized as one of the major pollutants that reduce urban air quality [132]. Yuan Shi et al. investigated the effects of land use and landscape patterns on the spatial variation of PM2.5 in Hong Kong (China), particularly to determine the impact of regulating PM2.5 concentration levels on the types of landscapes affected.

A study in Shanghai aimed to provide a standard framework for estimating building carbon emissions within LCZs to better predict carbon emission patterns from buildings within other LCZs and to assist in energy management in different LCZs [133].

3.6.3. Others

In addition to the above-mentioned research types on LCZ, LCZ has also been increasing in popularity in other fields. Although LCZ was proposed to describe UHIs, there are also studies using LCZ on health issues, urban ventilation performance, and energy consumption. Oscar Brousse et al. were the first researchers who attempted to use WUDAPT and generate the LCZ in tropical sub-Saharan Africa, and they linked the framework of the LCZ as a tool to improve urban health with a malaria spread risk model [134,135]. A study in Shenyang (China) analyzed the wind information from 16 weather stations to demonstrate the suitability for the use of an LCZ classification scheme to represent local-scale urban ventilation performance [136].

Several articles have also appeared on energy consumption in buildings. Using three different methods to investigate the energy consumption patterns of two different building types in the LCZ in Nagpur, India, it has been shown that the recommendations of the LCZ are critical for building energy policy making [137]. Using the building energy simulations (BES) and as input to the EnergyPlus simulations, the yearly, monthly, and hourly cooling and heating energy demands (LCZ 8) were investigated and compared across all six LCZ sites. The method is used to reasonably evaluate the influence of the heat island effect on the building energy performance at the neighborhood scale [138].

4. Challenges and Future Direction

Oke suggested that the LCZ is a relatively new classification and framework [11]. The LCZ classification was proposed to facilitate the quantification of UHI, but now, researchers are primarily using it for LCZ mapping research that can be seen from Section 3.2. Indeed, UHI can be further explored, although there have been many of these studies conducted in recent years. We summarize five avenues for improvement.

First, there are two key problems in the actual implementation of LCZ mapping: a lack of data and a lack of method specification. From Section 3.1, it can be analyzed that most of the research sites are in China; therefore, additional studies on LCZ mapping in other areas that have not been studied are needed and can further increase the possibility of standardization. While numerous accuracy assessments and precision improvements have been made for LCZ classification, there is still a need to improve cartographic accuracy and to develop a unified/standard method for generating LCZ maps.

Second, improvements in LCZ mapping methods are required. There have been many methods used in LCZ applications, with most studies focusing on the use of WUDAPT and GIS to classify LCZ type. Section 3.2.2 discusses many studies that use different methods to improve the accuracy of LCZ mapping. Most of them use different deep learning or machine learning methods, including CNN (traditional and novel), GCNs, and RF. LCZs represent a generic climate study description applicable to land cover (LULC). A key point is that LCZs are universal in their application and they can be linked to measurable urban parameters of urban form and function. There have been many articles dedicated to improving the accuracy of LCZ mapping, but there is not yet a universal method. Instead of requiring more research to further improve the methodology to increase its accuracy, LCZ maps can be generated for the entire globe.

Third, the application of LCZ is becoming more extensive. It is not only used in urban planning, climatology, UHI, and other fields, but also in many new fields such as diseases and building energy. According to the statistics of this paper, LCZ mapping-related literature is the most numerous, followed by urban thermal environment (including UHI and LST), and although LST can reflect UHI to a certain degree, few articles simply discuss UHI. Since LCZ was first proposed to describe UHI, most scholars still focus on LCZ mapping innovation, in addition to the LCZ mapping standardization based on the second point noted earlier. Thus, the future research direction should increase the main research on the pure UHI theme. New breakthroughs have also been made. Therefore, innovations can be made using combined data sources. This also suggests that other fields could be developed in conjunction with LCZ to discover novel research directions.

Fourth, modeling is most frequently used as a research method. In other fields except LCZ mapping, modeling is used the most frequently. It is recommended that other research methods are used to increase the actual research in this area.

Finally, from the perspective of urban policymakers, there are policy-making aspects that need to be considered. LCZ can help to show the distribution of the thermal environment in a city, which can guide city planners to formulate corresponding urban development plans, provide a basis for urban future planning, improve the implementation of policies, and provide a more comfortable living environment for urban residents.

5. Conclusions

A total of 190 LCZ studies were collected and analyzed in our review. This research used bibliometric and thematic analyses to systematically review the local climate zones. Over the past decade, the number of studies related to LCZs has increased, demonstrating its importance in urban climate research. Overall, the results show that more than half of the published studies on LCZ are in Chinese cities. Due to its close relationship with the urban heat island effect, the local climate zone is an important way to solve the problem of climate change. Systematic reviews reveal a wealth of data sources, methodologies, and analyses used to generate and use maps of local climate zones. The most common topics that appear in the LCZ literature include LCZ mapping, measurement methods, thermal environments, and outdoor thermal comfort. Among them, LCZ mapping has the largest number of articles, followed by thermal environment. Regarding the study of LCZs, scholars focus on discussing the method of LCZ mapping, from the basic RS images and GIS methods used, and then further improve the accuracy, ignoring the fact that LCZ was proposed to describe the UHI. A discussion of UHI and research using other research methods other than modeling is encouraged for LCZ. At the same time, a further standardization of the LCZ mapping method and improvements to the accuracy of LCZ mapping will allow effective preliminary analysis for the field of urban planning, and guide policies to improve the urban thermal environment. The study of LCZ in urban areas improves the urban environment, relieves thermal stress, helps urban decision makers to enhance urban planning solutions, improves the outdoor thermal comfort of urban residents, and helps promote urban sustainability.