Identifying research progress, focuses, and prospects of local climate zone (LCZ) using bibliometrics and critical reviews

Abstract

The local climate zone (LCZ) has been an important land surface classification used to differentiate urban climate between localities. The general knowledge maps of LCZ studies are needed when LCZ-related research has attracted great attention. This study integrated bibliometrics and critical review to understand the status quo and suggest future research directions. Bibliometrics provided a statistical technique to explore large volumes of article data from the Web of Science, ScienceDirect, and Scopus databases, based on the Co-Occurrence 13.4 (COOC) software. The bibliometric results indicated a rapid increase in LCZ publications and identified the high-frequency keywords which can be clustered into two groups, including a human thermal comfort-related group and the other urban climatology-related one. From 2011 to 2020, the effects of land use and urban morphology on urban climate and heat island effects predominated the LCZ-related research. Since 2021, the research focuses had shifted to the fields of thermal environment and heatwave, due to the growing demand for human thermal comfort and heat risk reduction. Moreover, this study identified ‘Land Surface Temperature’ and ‘Heatwave’ as two focuses of LCZ-related research during the last decade. Their critical reviews demonstrated the need for additional in-depth LCZ-heatwave studies that consider the risk of human exposure. This study also recommended incorporating hydrological concerns and social issues into the LCZ plan for a more integrated LCZ research outlook. Overall, this study provides not only a comprehensive understanding of LCZ knowledge networks, but also critical details on research focuses and potential research prospects.

Highlights

• •Integrating bibliometrics and critical reviews to understand the LCZ development.
• •LCZ studies can be clustered into human thermal comfort and urban climatology groups.
• •Bi-clustering illustrates the temporal changes in the keyword clusters of LCZ studies.
• •Revealing two focuses on land surface temperature and heatwave in LCZ research.

1. Introduction

Surface characteristics on the planet are complex and cannot be fully detected only using observations and on-site measurements [1]. As a result, surface landscape classification is required to characterize and separate natural and built-up environments scientifically, which, in particular, helps people understand and optimize urban and regional settings [2]. Landscape classes are diverse using a series of spatially homogeneous regions in terms of land cover and use as well as other physical characteristics such as building density and layout [3]. The landscape has been typically classified at the macro scale based on satellite data [4], such as the CORINE Land Cover (CLC) inventory and MODIS data product, which provide long-term temporal changes in the land cover/use around the world. These satellite-based land-use data roughly detect the spatial distributions of land cover, such as urban areas, forests, and water bodies. Moreover, the need for finer separation of the urban landscape increases as urbanization accelerated. Urban geometry is a typically micro-level method to measure urban landscapes in terms of building spatial arrangements [5]. However, both satellite-based land data and urban geometric data vary depending on study areas and objectives, and as a result, are not comparable [6], limiting the generic landscape classifications [7]. The development of local climate zone (LCZ) classification can address this limitation, by providing 10 built types (LCZs 1–10) and seven land-cover types (LCZs A-G) [8]. A total of 17 LCZ classes are all standardized and each LCZ is a representative landscape unit with homogeneous land cover (e.g., impervious surface) and geometric structure (e.g., street aspect ratio) [9]. In other words, the LCZs include multiple standardized land surface characteristics which otherwise cannot be available using individual satellite-based or geometry-based landscape classes.

Due to the land-atmosphere interactions, landscape changes in the earth system have impacts on climate environments [10]. Therefore, climate-relevant information in different landscape classes is required as guidance for climate-resilient urban planning. Due to the capacity to reflect climatic differences, LCZ classification has been widely used to demonstrate and compare inter-city and intra-city climate conditions [11]. In detail, LCZs A-G showing more rural landscapes illustrate the land cover effects on climate. For example, land surfaces with higher albedo have a greater capacity to reflect long-wave radiation fluxes from the earth surface to the atmosphere, decreasing mean radiant temperature (T_mrt) and keeping the land surface cooler [12,13]. The expansion of vegetation, especially trees, contributes to reducing daily thermal stress due to its strong evapotranspiration efficiency [14]. Moreover, LCZs 1–10 with more urban landscape elements differentiate the street-level geometric characteristics such as street aspect ratio and their impacts on pedestrian thermal comfort [15,16], regardless of hourly, daily, and yearly time scales. For example, wide streets with low-rise buildings are considered to cause the highest thermal stress, whereas narrow streets with tall buildings are desired for less thermal stress, which suggests urban compactness with a higher aspect ratio for thermal mitigation [17]. Using the LCZs as the landscape representations, urban climate and thermal stress can be regulated by modifying landscape surface albedo, evapotranspiration, and urban canyon-related surface roughness simultaneously [18]. Furthermore, LCZ classification provides a local-scale (100 × 100 m²) land-climate interaction for microclimate regulation purposes [19], which complements traditional macro or mesoscale climate classification [20]. LCZ-based local data contributes to revealing spatially-varying situations for fit-for-purpose urban planning [21]. In summary, the LCZ is a prevailing landscape classification method by providing local generic representations of both land covers and surface structures.

LCZ classification has been a hotspot for surface landscape detection and urban climate issues after its original publication [8] was cited more than 2700 times according to Google Scholar records. Therefore, given the promising research outcomes in the LCZ field over the past decade, we need a systematic review to understand the LCZ development status and suggest future research directions. The LCZ review papers are lacking, among which, the critical review is dominant. For example, [20] synthesized the data sources, methods, and themes in LCZ studies. [22] reviewed the GIS stream for LCZ mapping. [23] compiled the previous source area in the built LCZs. Therefore, to map a ‘big picture’ of LCZ-related research for future works, this study adopts bibliometric analysis combined with critical reviews to cast light on a broad view of 1) the keywords that are frequently used in LCZ research, 2) the networks in which keywords are clustered and commonly co-occur, 3) the temporal changes in keywords over years, and 4) research trends and focuses related to LCZs. Bibliometrics provides mathematics and statistics to explore a large amount of article data scientifically, such as tracking the regional and temporal distributions of publications. The bibliometric analysis makes it easy to identify statistically significant information and concentrate on more popular or newer subtopics efficiently. Following the bibliometric analysis, the critical reviews display the specific research status quo, research limitations, and potential future research works. As a result, this bibliometrics-led and critical review-assisted approach can offer both broad and in-depth knowledge.

2. Methodology

Fig. 1 depicts the framework for the bibliometrics-led review of available LCZ-related publications, including data collection, data screening, and bibliometrics and critical reviews. This bibliometrics-led approach enables to achieve three research objectives: 1) detecting overall clusters of high-frequency keywords over the past decade; 2) identifying the temporal changes in high-frequency keywords; and 3) demonstrating the LCZ-related research focuses by comparing the weighted frequency of each keyword.

Fig. 1.

A Framework for data collection, screening, bibliometric analysis, and research significance.

2.1. Data sources

Considering data representativeness and accessibility, we used Scopus, Web of Science (WoS) Core Collection, and ScienceDirect as the databases and collected the initial data searched on November 19, 2022. In the WoS core set, we used TS = local climate zone (TS = topic) as the search term and filtered the related literature 224 records. By searching the literature that has the term local climate zone in the title, abstract, or keywords, we retrieved 193 records in ScienceDirect and 449 records in Scopus, respectively. This research focused on the bibliometrics of journal papers, and a total of 866 publications in the field of local climate zone were selected as the raw data.

2.2. Data screening

The article records derived from Scopus, WoS, and ScienceDirect were rearranged and merged manually into an excel file with the standard format including author, publication year, title, keyword, country, etc. Co-Occurrence 13.4 (COOC) software [24] was used, which can detect and remove duplicate records in this merged file automatically, and then yield new data after the screening process. A total of 427 records were shortlisted for bibliometric analysis in this study. Moreover, articles may use different expressions to represent the identical concept. For example, some publications used ‘local climate zone’ as a keyword, while others used ‘LCZ’, ‘local climate zones’, or ‘local climate zone (LCZ)’ instead. Also, some articles may choose meaningless or irrelevant keywords, such as ‘urban’ and ‘urban areas’, which have no bearing on the research implications. Therefore, an Excel file (examples in Appendix A) including the original keywords, the corresponding substitutes, and those insignificant keywords, was created as an input of COOC. The COOC software can merge all similar keywords and remove the insignificant keywords automatically based on this Excel sample. In this way, the bibliometrics analysis can be more accurate.

2.3. Bibliometric analysis

After data screening, COOC also offered a series of functions to map the literature networks and detect keyword development trends. For example, co-word analysis is common in bibliometrics to identify the relationships between various keywords. The keywords that co-occur more frequently in one article, are closer theoretically [25]. Co-word networks were then conducted to illustrate keyword clusters, by categorizing and aggregating keywords according to their similarity distance derived from co-word analysis [26,27]. Moreover, two-dimensional variables, such as keyword and publication year, can be clustered simultaneously using the bi-clustering algorithm where keyword clusters and publication year clusters were linked in a pairwise manner [28]. Different from the co-word networks, bi-clustering allowed inferring dominant keyword clusters over publication years.

3. Results

3.1. LCZ-related research overview

The publication changes reflected the knowledge accumulation and development status of the LCZ field. A total of 427 LCZ-related papers were distributed between 2011 and 2022. The LCZ research progressed at a slow pace from 2011 to 2015 (Fig. 2a). Since 2017, the LCZ outputs experienced sharp growth with annual publications increasing by more than 5 times over the last 5 years (Fig. 2a). The cumulative publications over the past decade presented an exponential curve (Fig. 2b), demonstrating the recent surge in interest in the LCZ discipline.

Fig. 2.

Annual (a) and cumulative (b) number of LCZ publications from 2011 to 2022. The 2022 data are for the months from January to November.

From 2011 to 2022, LCZ-related studies were mainly published in the ‘Urban Climate’ journal, particularly those from research institutions in China, Hong Kong (China), India, Germany, and France. China is a leading contributor to LCZ-related discipline, with the majority of its outputs published in journals including ‘Building and Environment’, ‘Sustainability (Switzerland)’, and ‘Remote Sensing’ (Fig. 3 and Appendix B).

Fig. 3.

Networks of dominant countries and journals of LCZ-related publications from 2011 to 2022.

Since the first LCZ paper in 2011, a total of 883 keywords have been adopted, with 745 keywords appearing only once in the literature. The keyword of ‘Local Climate Zone’ had the highest frequency (323 times), followed by ‘Urban Heat Island’ (175 times) and ‘Urban Climate’ (94 times). This study concentrated on the high-frequency keywords with the top 10 frequencies over the past decade, which were depicted in Fig. 4, showing the common topics or themes relating to the LCZ field.

Fig. 4.

Tree diagram of the top 10 most frequently used keywords from 2011 to 2022.

3.2. LCZ-related research clusters based on keyword networks

3.2.1. Co-word analysis

The co-word analysis helped to network different keywords in the selected 427 publications by summarizing the co-occurrence times of two keywords. Focusing on the high-frequency keywords, a co-word network was created based on the co-occurrence matrix (Table 1) to visualize the keywords' relationships. The more frequently two keywords co-occurred, the stronger link they had. That is why ‘Local Climate Zone’, ‘Urban Heat Island’, ‘Urban Climate’, and ‘Land Surface Temperature’ were linked with the most intensity (Fig. 5). Moreover, co-word networks demonstrated two clusters of co-occurred keywords (Fig. 5). ‘Thermal Environment’, and ‘Heatwave’ were clustered as a group related to urban heat stress and human thermal comfort. Meanwhile, the rest of keywords were grouped into another category that was primarily concerned with the urban heat island and urban climatology in relation to land use and urban morphological features.

Table 1.

Example of the co-occurrence matrix for the four keywords with the highest co-occurrence frequencies.

	Local Climate Zone	Urban Heat Island	Urban Climate	Land Surface Temperature
Local Climate Zone	0	134	63	63
Urban Heat Island	134	0	33	41
Urban Climate	63	33	0	7
Land Surface Temperature	63	41	7	0

Fig. 5.

Co-word networks for high-frequency keywords.

3.3. Temporal changes in keyword clusters

The previous sections identified 10 high-frequency keywords over the past decade, and detected their clusters based on the quantitative co-occurrence intensity. This section aimed to figure out how these high-frequency keywords in the literature were distributed over years by bi-clustering keywords and publication years. Fig. 6 visualized the interactions of keywords and publication years to reveal whether and which keywords were more or less focused in different years. From 2011 to 2020, in general, land property and urban heat island issues were the emphases of LCZ-related research, although the urban climate in relation to urban morphology was dominant between 2014 and 2016. On the other hand, a clear topic change can be observed from 2021 to 2022 when land surface temperature, thermal environment, and heatwave received a great deal of attention in LCZ analyses. It can be explained by the increased health-related demand for outdoor thermal comfort and less heat hazard.

Fig. 6.

Bi-clustering of high-frequency keywords and publication years from 2011 to 2022.

According to Fig. 6, it was hypothesized that LCZ-related research focuses have been moved to urban heat stress and thermal comfort in recent years. To examine it, the dynamics of LCZ-related research focuses should be detected. In general, the more recently a keyword occurs, the more likely it is to be a research focus. As a result, this study defined the weighted frequencies of keywords according to the order of their first occurrence (Appendix C). The keywords with higher weighted frequencies are more likely to be the research focuses. Among the top 10 high-frequency keywords, the weighted frequency of the keyword ‘Land Surface Temperature’ had a considerable increase in the last 10 years, followed by ‘Heatwave’ (Fig. 7). It implied that, within LCZ-related literature, those associated with ‘Land Surface Temperature’ and ‘Heatwave’ can be seen as the research focuses.

Fig. 7.

The trends of weighted frequencies of the top 10 high-frequency keywords.

4. Discussion

4.1. LCZ research progress

Using the keyword-year bi-clustering, the LCZ research progress can be revealed according to the yearly variations in high-frequency keywords. In 2012, in the context of climate mitigation, the LCZ scheme was developed formally in order to better understand the relationships between land surfaces and climate conditions. At the early stage of LCZ-related research, urban heat island issues related to remote sensing-based land use properties were the research focuses. Since 2014, there were increased investigations into the roles of urban morphology in urban climate. Apart from land cover and use, morphological characteristics, such as building layouts and street geometries, have great influences on urban climate variables, such as air temperature. Climate information on both land covers and urban morphology can be found via the World Urban Database and Access Portal Tools (WUDAPT) initiative. It provides communities and platforms where the climate-relevant database on worldwide cities can be digitalized, gathered, stored, and disseminated in a standardized format [29,30]. WUDAPT, as a typical remote sensing-based approach, also provides multi-scale Landsat data, and its level 0 data can be used for mapping LCZs quickly [30,31]. LCZs combined with WUDAPT, as a result, have been classified worldwide based on their unique land-cover properties, land-use patterns, and built-up structures [32,33], such as the LCZ maps in Europe [34], Africa [35,36], and Asia [37,38]. That is why the keyword ‘WUDAPT’ became dominant between 2019 and 2020 (Fig. 6).

Furthermore, since 2021, the focus of attention has been moved from the simply urban climate and urban heat island to heat-related investigations that benefit both the physical radiant environment and human thermal perception. Due to urban-climate interaction, land use changes, built-up expansions, and various human activities have influences on energy flows between the earth system and atmosphere and cause excessive heat fluxes [39]. Consequently, heatwave-based thermal discomfort and land surface temperature become the new interest in LCZ fields [21,40].

4.2. LCZ research focuses

Fig. 7 demonstrated two keywords as research focuses. Their critical reviews helped us understand the status quo and offer potential directions for future research.

4.2.1. The focus on ‘land surface temperature’ in LCZ-related research

We reviewed 75 out of 427 selected publications that used ‘Land Surface Temperature’ (LST) as a keyword and synthesized three aspects of LCZ-LST research. First, multi-temporal and multi-spatial LST analyses using LCZ classification can be found. For example, a number of studies have identified the diurnal [41], monthly [42], seasonal [43], and yearly [44] LST differences between various LCZ types. [45] also predicted different LST scenarios in the future corresponding to LCZ changes. Spatially, the global suitability of the LCZ scheme in macroclimate LST studies has been examined [46]. The LST case studies have also covered the local scale [47], city scale [48], and regional scale [49]. Second, the associations between LCZ and LST have long been investigated. Compared to natural and social factors such as population density, road density, and vegetation index, LCZ distributions have the greatest explanatory power for LST disparities [50]. [51] used correlation analysis, multiple linear regression, and spatial regression to reveal the statistical LST-LCZ correlations. Similarly, [52] used both global regressions, in terms of ordinary least squares regression (OLS) and random forest regression (RF), and spatial regressions, such as semi-parametric geographically weighted regression (SGWR) and multiscale geographically weighted regression (MGWR), to model the uniform and heterogeneous LST-LCZ relationships for local settlement improvement. Third, the LST analyses related to LCZ classification had multiple applications and implications. For example, the LCZ-LST relationships help inform land use strategies for heat mitigation [53] and tree resource allocation [54], optimize LCZ compositions [40], and enable to construct ventilation corridors for the minimum urban heat island effect [55,56].

4.2.2. The focus on ‘heatwave’ in LCZ-related research

There was a rising interest in ‘Heatwave’ especially after 2020 (Fig. 7), which is understandable in the current context of increased extreme climates and heat-related hazards and mortality. The heatwave indicators are multiple, such as net radiant fluxes [57]. For instance, [58] reported the daily cycle of the heat flux using an energy balance-based machine learning. Apart from temperature disparity due to different LCZ layouts, socioeconomic factors were added to evaluate subjective heat stress, which helped to identify households (non-)vulnerable to heatwave [59,60]. Although some studies used extremely high temperature to represent the magnitude of the heatwave [61], the composite heat indices have been applied more extensively, such as Physiologically Equivalent Temperature [62], Universal Thermal Climate Index [63], and Humidex [64]. Variabilities in heat index-based heat stress between LCZ types have been demonstrated both spatially [65], hourly [66], diurnally [67], seasonally [43], and yearly [68]. On the other hand, among 34 out of 427 selected publications that used ‘Heatwave’ as a keyword [69], were the first to combine in-situ air temperature data and population mortality, extending conventionally physical environmental assessment to human vulnerability to climate change hazards for improving the social infrastructure distributions (e.g., hospital). In 2022, [67] assessed heat exposure risk based on the combinations of high-temperature intensity, frequency, and population density for risk governance. The exposure time to extreme heat was also utilized to measure the risk of exposure to heat stress [70].

4.3. LCZ research prospects

For LCZ-LST studies, coarse and fine spatiotemporal scales, global and local techniques, and multiple LST implications all have been investigated in recent literature. Due to the limited amount of unexplored space in the LCZ-LST field, the weighted frequency of the keyword ‘Land Surface Temperature’ experienced a great drop from 2021 to 2022 (Fig. 7). Meantime, LCZ-heatwave investigations have been receiving more attention, when human wellbeing and social benefits are given more priority. In spite of this, the human risk of exposure to heatwave is under-explored. The findings showed that only three publications looked at heatwaves from a human exposure perspective, out of the 427 papers chosen for this study. As a result, heat exposure risk can be one of the LCZ-related research prospects. One direction is that urban growth patterns, shown as LCZ dynamics in terms of compositions and configurations, should be taken into account while determining heat exposure risk. Additionally, functional zone distributions and the resulting demographic changes, such as the distributions of the ages and the youth, are the influences of spatial patterns of heat exposure risk.

The bibliometric analysis in this study demonstrates the multi-dimensional application of LCZ in urban planning, including urban morphological characteristics, land use monitoring and classification, urban climatology in terms of air temperature and urban heat island, earth surface heat in relation to land surface temperature, and human wellbeing in terms of thermal regulation and heatwave mitigation. Despite being widely used, the LCZ approach has not extended beyond the urban climate domain. For example, based on the water-energy nexus, it should be feasible to introduce LCZ into urban hydrological issues, such as drought and runoff management. We argue that both climate and hydrology-relevant information in LCZs should be given equal weight. Moreover, LCZs define not only various physical spaces but also urban functions, which means that LCZ spatial patterns may shape social activities, such as human moving behavior [[71], [72], [73]]. We claim that LCZs and social dimension indicators should be further linked, which enables the environmental and social data to be interpreted using LCZs. An integrated framework linking LCZs and multi-faceted indicators allows for broader human well-being in urban settings.

5. Conclusion

This paper uses bibliometric analysis and critical review to uncover the LCZ research progress, recap the LCZ research focuses in terms of land surface temperature and heatwave, and frame the LCZ research prospects, based on a total of 427 publications between 2011 and 2022 from the WoS core set, ScienceDirect, and Scopus databases.

One of the technical advances in this study is that we use the COOC software to merge similar keywords and remove nonsense and insignificant keywords for more accurate bibliometric statistics. A rapid development trend in the LCZ field from 2011 to 2022 has been verified. The high-frequency keywords are overall clustered into an urban heat-related group for human thermal comfort as well as a surface temperature- and land property-related category in terms of urban heat island and urban morphology. Temporally, the keywords in LCZ-related research move from urban climate and urban heat island related to land property and urban morphology to land surface temperature and human health-related outdoor thermal comfort. Moreover, the terms ‘Land Surface Temperature’ and ‘Heatwave’ are identified as two research focuses over the past decade. The LCZ-heatwave investigation from the perspective of human exposure risk is suggested as a future research direction. This study also argues that the LCZ approach should be expanded beyond the urban climate field to hydrological concerns and social activities, for instance, for creating an integrated LCZ framework.

This bibliometrics-led method also has limitations. For example, some of the newest papers may be available till next year due to different publication cycles between journals. Overall, this study provides a comprehensive picture of the LCZ development status over the past decade, maps the keywords-based topic networks in LCZ fields, and detects two research focuses for suggesting LCZ-related research prospects. With the combination of bibliometric analyses and critical reviews, not only general research trends but also particular knowledge about current research focuses have been demonstrated, enabling more in-depth LCZ research efficiently.

Author contribution statement

All authors listed have significantly contributed to the development and the writing of this article.

Funding statement

Yi Zhou TAO was supported by Basic Public Welfare Research Program of Zhejiang Province [LGN19E080002].

Data availability statement

Data will be made available on request.

Declaration of interest's statement

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

Appendix A. Supplementary data

The following is the Supplementary data to this article.