Applications of local climate zone classification in European cities: A review of in situ and mobile monitoring methods in urban climate studies

2.1 Literature search and selection

A comprehensive screening process initially involved determining the keywords used to identify and include relevant articles that best aligned with the review’s research topic. Filters applied during the literature selection process within Scopus database (https://www.scopus.com; assessed on 3rd April 2024) included document types (e.g., articles and review papers). The exclusive use of the Scopus database for the literature search was justified by the findings of Lehnert et al. [9], which showed that Web of Science (WoS) and Scopus produced comparable results. Therefore, for the purposes of this research, Scopus was considered sufficient.

This review primarily focuses on articles that pivot on researching “local climate zones.” After the introduction of the LCZ framework in 2012 by Stewart and Oke [8], research within this scientific field expanded significantly over the following years. Therefore, the term was selected as the primary keyword for the literature search through the Scopus database. Additionally, articles that implemented in situ and mobile meteorological stations were included in this review. The data collected from in situ and mobile weather stations across cities in Europe are used to determine the accuracy of the abovementioned LCZs. Moreover, the second keyword established for literature selection through Scopus was decided to be the “measurement,” which allows us to select papers that are related to data collection through field measurements (e.g., in situ and mobile measurements). Given the inter- and multidisciplinary nature of LCZ with diverse methodologies applied, the keywords were chosen to eliminate articles that were not suitable to answer the objective of the review.

The time range that is used for the literature selection obtained articles from 1st January 2012 to 3rd April 2024. The year 2012 is marked as the starting point for the search because the framework of “local climate zone” was established and introduced during that year [8].

Database literature search was conducted in April 2024, consisting of several steps.

The first step in the selection process involved searching for articles containing previously identified keywords (LCZs and measurements) through Scopus by applying the “search within all fields” function, which provided 75 articles.

In the second step, filters were applied to narrow down the search within the specified time range mentioned above. The document types selected for the search included articles and review papers, while conference papers, editorials, etc., were excluded. Considering that this review focuses on cities in Europe, only this geographical area was included, while the cities in countries from other continents were excluded. Additionally, only articles in the English language were considered in this literature selection, while articles in other languages were eliminated. These non-English publications constitute a small portion of the total and often reflect content similar to studies already published in English by the same authors. Applying these filters resulted in a significantly reduced number of papers, specifically totaling 45 articles.

The third step in evaluating relevant articles involved reviewing the titles, keywords, and subsequently the abstracts of the articles. Through this examination process, it can be determined whether, among the 45 articles in the database, there might be articles that do not pertain to the European continent, focus on remote sensing or GIS, do not apply LCZs as the main subject of research, or the articles do not fulfill the criteria. Occasionally, during the search process, the system may make errors and inadvertently contain articles that do not align with the keywords specified in the filters. After a more detailed examination, it was found that one study only mentions the term “local climate zones” without encompassing measurements and data obtained in those areas, while the other articles completely correspond to the relevant publications. At the end of the third step, there were 44 studies that were assessed for eligibility.

Nevertheless, the fourth and final step involved the “snowballing method” [42], which gave good results as well. Through the review and analysis of the selected studies for this article, additional relevant studies were identified by examining the references cited by the authors. By employing this method, 17 additional articles were identified from reviewed articles that aligned with the filters, but were not initially found during the Scopus search. At the end of the selection, the database used for this review contained 61 articles. The entire process of selecting articles described above, and their illustration is made by following the PRISMA concept (Figure 1). PRISMA stands for Preferred Reporting Items for Systematic Reviews and Meta-Analyses. It was developed to improve the reporting of systematic reviews and meta-analyses by helping authors present their work in a complete, transparent, and clear way [43].

Figure 1

Procedure PRISMA chart of the literature selection using selected keywords in Scopus and applying snowballing method [43].

2.2 Inclusion criteria

The classification of the articles was achieved with the use of the two keywords “local climate zones” and “measurements,” where the keyword “local climate zone” is the primary one.

In order to visually present the analysis required for this review, we used VOSviewer (version 1.6.18.0). VOSviewer is a software that is employed to create various illustrative maps using data from multiple databases [44]. VOSviewer utilizes data from several databases, while for this purpose bibliometric data were obtained from the Scopus database. Additionally, all 61 selected articles are presented in detail through the Supplementary materials (Table S1).

The obtained outcomes are presented in the following text through annual trends in keyword application, keyword co-occurrence trends in chosen articles, journal distribution and publication trends, and article citation frequency.

2.3 Classification of applied measurements in LCZ studies

This review is based on data obtained in two ways:

(a) In situ measurements – Data obtained from in situ weather stations are an important part of understanding a thermal condition in local scale. The data obtained through these measurements include day and night oscillations of various meteorological parameters, as well as consistent data throughout the year; (b) Mobile measurements – This type of measurement includes the most diverse areas that correspond to the local- or microclimate conditions in the city. Mobile measurements can also be used in any part of the day, and can even refer to data collection in motion (bicycle, cart, etc.) to identify variations that in situ station might lack.

3.1 Bibliometric analysis

Annual trends in keyword application were analyzed in order to observe the frequency of applied keywords, as well as the combinations in which they were used and their purposes. Considering that research needs have arisen over the years, the articles were divided in two time periods. The aim was to point out variations in keyword occurrences across these two intervals from the beginning to the end of the observed period. The first period is from 2012 to 2018 and the second from 2019 to 2024. During this time range, the occurrence and appearance of keywords progressed over the years and the most used ones are mentioned in the following sentences. The analysis was performed using VOSviewer (Figure 2).

Figure 2

Analysis of authors’ keyword occurrence in selected Scopus articles (2012–2024) using VOSviewer. Circles represent the keywords used in the set of articles chosen for the review. The size of each circle indicates the frequency of the term (larger circles mean the term appears more frequently in the selected articles). The color of each circle represents the average publication year of the articles that used that keyword. Blue circles correspond to keywords used around 2014, green around 2018, and yellow indicate more recent usage. Lines connect terms that co-occur frequently in the same articles. The thickness of each line represents the strength of the co-occurrence (thicker lines indicate the terms appear together more often). The color of the lines (blue, green, and yellow) indicates the time period of co-occurrence between keywords, showing temporal relationships between frequently associated terms.

Articles that were published in the first half of the obtained period (2012–2018) used the keywords such as “local climate zones” as the most occurred keyword, as well as “urban climate” and “urban heat island.” Nonetheless, keywords that had more appearance in this time range, beside the three abovementioned keywords, are “monitoring network,” “urban meteorological network,” and “mobile measurements”. Furthermore, the keyword “air temperature” and “local climate zone” in the vast majority of cases were applied together, as well as “urban heat island.”

The first period of analyzing the keywords shows a remarkable growth in urban climate studies, where the methodology was mostly based on using monitoring networks to understand the occurrence of UHIs in LCZs and their relationships. Additionally, during this period, both in situ and mobile measurements were used, indicating a desire to present increasingly diverse data since the LCZ framework became publicly available. Furthermore, there is no noticeable use of various analytical tools such as MUKLIMO_3, regression analysis, or even LST measurements, which suggests that remote sensing and its application to LCZs was not widely adopted during the initial period of keyword analysis.

Moreover, in the second part of the following period (2019–2024) the main keywords were the same as in the first period, with the difference where “urban heat island” is more frequently used than the term “local climate zone.” In most of the studies the keywords “local climate zone” and “urban heat island” were used together in combination with “land surface temperature,” “satellite,” and “surface heat island” because the use of remote sensing data increased during the following years.

In the second period of analysis, a clear shift can be observed in the methods used to study urban climate. Unlike the first period, remote sensing has become widely used in articles focused on understanding the UHI. Moreover, in recent years, urban planning and surface measurements, as well as crowdsourcing, have contributed to the development of urban climatology and a broader understanding of this type of research, along with the adoption of alternative methods.

The most notable difference between these two periods is that in the second part of article selection, there was significantly more keywords related to citizen weather station and citizen weather network and citizen science. This reflects a broader temporal shift in research focus, as terms related to citizen science have become increasingly common, likely due to extensive EU funding in this area, where including such components is sometimes even mandatory. It can be concluded that during the last few years, the interest for the mobile measurements such as citizen stations rise which broadened many horizons and created a new spectrum of research into LCZs and urban climate. It has been noticed that in the previous years, the combination of mobile in situ measurements on one side and remote sensing and GIS on the other side increased during the second observed period.

Further, co-occurrence of keywords was analyzed. The selected articles for the review were visualized using VOSviewer (Figure 3). This concept, which is part of the bibliometric analysis used in this study, illustrates how frequently different keywords appear together. A fundamental aspect of the visualization is to highlight the application of LCZ scheme across various areas of the research field.

Figure 3

The keyword co-occurrence network in selected articles from Scopus. Each circle represents an author keyword used in the selected articles. The size of each circle indicates the frequency of that term’s occurrence (larger circles mean the term appears more frequently as an author keyword). Colors (blue, green, and red) are used to group keywords into clusters based on their co-occurrence patterns. Keywords within the same color cluster are more closely related to each other, forming thematic groups or research subfields within the dataset. The lines between circles represent connections between keywords. More frequent keyword combinations across multiple studies produce thicker lines, while less frequent combinations are shown with fainter lines. The color of a line (e.g., green, blue, or red) indicates a group of words that are often associated together.

The keyword visualization in VOSviewer (Figure 3) illustrates how often keywords co-occur within the selected articles. For example, keywords highlighted in blue, such as “heat island,” are mostly associated with terms like “crowdsourcing,” “measurement method,” “microclimate,” and others.

Therefore, the visualization of the literature selection reveals that the keywords local climate zones, urban climate, urban heat island, and air temperature stand out, exhibiting numerous links to various frequently used keywords represented as smaller nodes (circles). This connection shows the strong association with terminology from urban climate, linking the different urban networks conducting in situ and mobile monitoring.

The keyword “urban heat island” has a total link strength of 86 connections, while “local climate zones” has 67 links to several keywords presented in Figure 3. “Urban climate” and “air temperature” also show significant associations with the same keywords with the total link strength of approximately 50, while the others have less than 20 or do not have a notable linkage.

In Figure 3, three different clusters can be seen in red, blue, and green. Red cluster is focused on terms such as LCZs, UHI, LST, climate change, atmospheric temperature, and urbanization. This suggests a focus on the physical aspects of urban climate change and measuring the temperature effects of urbanization, while the green cluster is dominated by keywords like urban climate, air temperature, environmental monitoring, relative humidity (RH), and climate modeling. It appears to be more oriented toward quantitative methods, modeling, and the measurement of microclimatic characteristics in urban areas. Moreover, the blue cluster is connected to topics such as heat island, urban planning, crowdsourcing, dataset, and measurement method. This cluster places more emphasis on methodology, data management, and urban planning aspects.

There is a clear inter-cluster connection, particularly between the red and green clusters, through nodes such as air temperature, climate conditions, and LCZ. This indicates an interdisciplinary nature of the research, where physical measurements are closely linked with modeling and urban planning contexts. The blue cluster is slightly more peripheral, although it is connected to others through methodological terms, which implies that methodological innovations are only partially integrated into the broader thematic areas. The size of the nodes indicates the frequency of keyword usage where UHI, urban climate, and air temperature are dominant, suggesting their central role in the chosen literature.

There are still some limitations to this figure, such as overlapping in terms. Additionally, the red cluster is quite dense, which can make it harder to read individual terms. Applying filters based on term frequency might improve the overall readability of the map.

Journal distribution and publication trends were examined by reviewing all 61 articles selected for this study, it was found that some journals published a higher number of these articles than others. Table 1 presents a summary of these journals, organized by the quantity of articles they contributed to the review.

It is observed from the database of articles used in this research that the majority of articles, in which this subject area was most recognized, were published in the journal Urban Climate, with as many as 11 articles, which is twice as many as in the second-ranked journal, the Hungarian Geographical Bulletin. Given that a larger number of articles were published in the journal Urban Climate, it is important to highlight that this journal serves as the dominant platform where the scientific community converges around the topic. This reflects a clear trend within the field. Consequently, a significant portion of the relevant literature included in this review is published in Urban Climate, which aligns closely with the central focus of this research on urban climate, covering topics such as the UHI effect, OTC, and related issues. In contrast, other journals listed in Table 1 contain noticeably fewer articles included in this review. This is primarily because their research focus tends to diverge toward different aims, such as urban design, remote sensing, energy systems, and other related fields.

Following these are the journals Geographica Pannonica, Theoretical and Applied Climatology, and International Journal of Climatology, while the other journals have fewer than three published articles in the database.

Moreover, article citation frequency highlights the most significant publications by the number of citations and it is shown in Table 2. The most cited articles [19,45,30] focus on analyzing urban air temperature variations through diverse data sources, methodologies, and climate classification systems. They focus on mapping LCZs, evaluating the influence of urban characteristics on thermal patterns, and exploring adaptation strategies to enhance thermal comfort in European cities.

Moreover, the prominence of certain authors is highlighted in Figure 4. Authors are marked in yellow to indicate their influence, considering both primary and co-authorization. The most cited researcher is Oke T.R., with 166 citations. Unger J. and Gal T. also have more than 100 citations each, while the remaining authors have fewer than 100 citations (citation last check 13.10.2024).

Figure 4

Visualization of authors citation. Warmer colors represent areas of higher citation density, larger font size indicates authors with more citations, and the distance between names reflects the strength of their co-citation relationships.

The geographical scope of the study covers a large domain covering the continent of Europe, involving more than ten countries (Figure 5).

Figure 5

Contribution of articles per country. The quantity of collected studies for review is represented in shades of purple. Countries shown in white indicate that no articles were found, while progressively darker shades correspond to a greater number of studies retrieved from that country.

Figure 5 shows that out of the 61 studies selected for this review, as many as 10 were conducted in Germany, with research primarily concentrating on cities such as Oberhausen, Berlin, Augsburg, Freiburg, etc. Thus, the majority of research was concentrated in German cities. Following Germany, Hungary contributed nine studies, with most research carried out in Szeged. Serbia had eight studies, where Novi Sad was the most frequently researched city. Additionally, seven studies from the Czech Republic were included, with the research largely focusing on Brno and Olomouc. Research conducted in cities in Italy and France was also part of the review, accounting for six and four studies, respectively. Both Ireland and Switzerland had more than one article included.

The map of publications reveals a strong concentration of research outputs within specific regional networks, primarily in Germany, with few local networks based in Hungary, Czech Republic, Serbia, etc. This uneven geographical distribution introduces a potential spatial bias in urban climate analysis, as less-studied regions may possess climatic characteristics that are underrepresented in current datasets. Such bias could limit the generalizability of findings and highlight the need for expanded research efforts across a broader range of urban environments worldwide. This limitation also affects the applicability and robustness of the LCZ classification system, as it relies on diverse urban contexts to fully capture local climate variability.

3.2 In situ measurement-based studies

The collected data from in situ stations enables the recognition of similarities and differences among LCZ. An analysis of the chosen literature shows that most research is based on data collection from in situ stations, such as urban networks, national observatories, or CWS. Additionally, Lehnert et al. [9] and Migliari et al. [46] presented similar methodological approaches for studying urban climate phenomena, including thermal analysis based on mobile and in situ measurements within the LCZ framework.

Several studies [12,15,18,22,47,48,49,50] have focused on investigating the spatiotemporal variations of UHI in cities across LCZ. Multiple studies that used in situ measurements [10,12,18,22,51] has shown that built-up areas with high building density and impervious surfaces (e.g., LCZ 2) record the highest UHI intensities, while natural (LCZ and A) or less dense zones (LCZ 6) remain cooler, especially at night. The maximum UHI intensity is typically reached a few hours after sunset. Unger et al. [10] concluded that the greatest differences between LCZs occur before dawn, with temperature differences reaching up to 5°C during early spring. The intensity of temperature differences among LCZs can vary depending on the city, duration of the measurements, instruments used, as well as local characteristics and the spatial and temporal variability of the UHI effect. For instance, while LCZs 2, 3, and 5 exhibit the strongest summer UHI, they can still be cooler than other zones during daytime. Moreover, heating rates differ significantly, illustrating the complex and dynamic nature of urban thermal patterns. The importance of multiple factors is evident, as shown in the study by Dunjić et al. [52], where it was demonstrated that air temperature dynamics correlate with air humidity, and that the urban dry island (UDI) effect is strongest during the day and evening, especially in LCZ 2, with lower values observed in the afternoon in LCZs 5, 6, 8, and 9 in Novi Sad, Serbia. This type of study highlights the necessity of analyzing and understanding not only air temperature but also humidity, in order to capture changes in other indices such as UDI, which is essential for comprehending the UHI effect, as these phenomena are directly linked. Savić et al. [17] monitored the nocturnal UHI in Novi Sad, highlighting that LCZs 2, 5, and 6 exhibited very high or high risk values, which were associated with a higher rate of mortality. The findings show that nighttime heat in densely built areas can seriously impact health, pointing to the need for more research on how UHI varies across cities and how it is connected to mortality.

A few articles used combination of in situ data with different data collection methods. Varentsov et al. [50] combined thermal satellite images and mesoscale modeling, while Feng et al. [53,54] used a combination of MODIS observations and two urban meteorological networks to explore the relationship between surface UHI intensity and canopy heat island intensity. Boccalatte et al. [25] combined sensor data with modeling approaches at various scales (local, city, street, etc.). Núñez-Peiró et al. [22] and Unger et al. [10] combined urban networks and official stations. Combination of different measurement methods, as presented previously, provides a more complete understanding of urban climate phenomena and highlights the strengths of each technique. However, even this combination can have limitations, such as inconsistencies between data sources, differences in spatial or temporal resolution, and remaining gaps in data coverage. These challenges should be improved in future research methodologies, along with finding better ways to reduce these biases.

Some articles showed the innovative method approaches such as modeling which have improved UHI assessment accuracy. Agathangelidis et al. [55] found that UHeatEx better captures urban spatial heterogeneity in Athens compared to LCZ mapping, especially when analyzing T _min data from five LCZ 2 stations. This study shows the importance of innovative approaches in understanding urban microclimates, as it is one of the few that uses complex modeling methods to improve UHI assessment. The limitations of such models, such as the impact of spatial resolution and input data quality on the outcomes, must be carefully taken into consideration. Núñez-Peiró et al. [56] showed that using UHI intensity as a model input, rather than air temperature, led to much better results and improved modeling results. The results suggest a potential to make urban climate research more efficient by minimizing the time and expenses required for field measurement campaigns.

OTC has been one of the prime topics for articles that for their investigation used in situ measurements. Milošević et al. [57], Unger et al. [58], and Müller et al. [45] used a thermal comfort index PET (physiological equivalent temperature) to monitor different LCZs in cities such as Novi Sad (Serbia), Szeged (Hungary), and Oberhausen (Germany), respectively. Geletič et al. [59] assessed OTC in Brno during a heatwave using MUKLIMO_3 and urban climate data, evaluating HUMIDEX based on air temperature (Ta) and RH, and found that the most built-up LCZs (2, 3, 5, 8, 10) were the least comfortable. Šećerov et al. [60] provided thorough analyses of air temperature, RH, and OTC. Their study found that LCZ 2 recorded the highest temperatures and the lowest levels of thermal comfort during the measurement period, while natural LCZs (A and D) offered the most comfortable conditions on site. While these studies effectively illustrate that highly built-up LCZs (such as LCZ 2, 3, 5, 8, and 10) tend to be less thermally comfortable and often validate the expectations of the LCZ framework, the magnitude of discomfort can vary significantly depending on the duration of heat events, local adaptation measures, and uncertainties in OTC indices, all of which require further investigation.

Researchers often do not discuss how inconsistencies in defining or mapping LCZs, or variations within each zone, might affect their results or make it difficult to compare findings across studies. The study that does discuss is the one by Briegel et al. [61], who applied the human thermal comfort neural network machine learning model, highlighting both inter- and notably intra-LCZ variability, with distinct universal thermal comfort index distributions between old and new building districts despite similar characteristics. This demonstrates the importance of analyzing intra-LCZ differences due to specific characteristics; in this study, it is a building age. Guerri et al. [62] evaluated thermal stress in one of the warmest areas of Florence, Italy, focusing on the main agri-food market (LCZ 8) to test three tree-based mitigation scenarios. They used in situ microclimate monitoring, urban characterization, and spatial analysis with GIS tools (QGIS) and microclimate simulations (ENVI-met) to analyze thermal patterns. Their research shows that the LCZ framework can be used for more than just simple classification, it can also help plan targeted urban interventions. But, like many other studies based on LCZ, it does not fully address the uncertainty in the accuracy of LCZ mapping or the differences between zones, which could affect the effectiveness of microclimate modeling and mitigation.

One of the primary focuses in research on LCZs is air temperature in different urban settings. Therefore, studies on inter- and intra-LCZ air temperature differences have been conducted in several cities, such as Berlin (Germany) [63,64], Olomouc (Czech Republic) [65], Brno (Czech Republic) [26], Augsburg (Germany) [23], Szeged (Hungary) [14,66], and Novi Sad (Serbia) [11]. These investigations revealed a consistent difference between built-up LCZs during the day and night, with especially high differences observed at night. Most authors highlighted nighttime temperature differences as the most significant, while Fenner et al. [63], Lehnert et al. [65], Skarbit et al. [66], Beck et al. [23], and Fricke et al. [14] analyzed the nighttime temperatures and found that they were considerably higher in built-up LCZs compared to non-built-up LCZs, with LCZ 2 standing out, as expected. Geletič et al. [26], Lehnert et al. [65], Fricke et al. [14], Skarbit et al. [66], and Quanz et al. [64] showed that nighttime urban heat is most pronounced in LCZs 2, 3, 5, 6, 8, 9, 10, and E, with particularly high temperatures found in compact built areas, especially LCZ 2B with dense courtyards and minimal vegetation. It has been observed that even within the same LCZ class, urban and rural settings can behave differently. Urban canopies may dampen short-term weather fluctuations more effectively than exposed locations. This means that some LCZs, although conceptually warmer, can show milder thermal responses depending on vegetation, sky view factor (SVF), or enclosure. Wind also plays a significant role, as it can increase variability and reduce thermal stress. To avoid the influence of micro-scale characteristics, LCZs of different classes should not be directly compared through single points. Within subclasses, geometric layout should also be considered.

In this review, several relevant articles that used CWS data were identified, focusing on the analysis of temperature differences between LCZs in cities such as London (England), Berlin (Germany), Oslo (Norway), and Bern (Switzerland). Brousse et al. [67] analyzed air temperature in London using data primarily collected from CWS. Moreover, some studies highlight the importance of combining CWS with other methods, such as networks of low-cost instruments [68], urban networks [39], remote sensing techniques [30,69], and modeling approaches to more effectively analyze and interpret the data. Citizen weather stations have gained popularity in recent years due to their affordability, higher density, and placement in areas with high population density. They can be deployed across different locations within the same LCZ, capturing variations in parameters and providing temperature data year-round under all weather conditions, resulting in a broad and rich database. However, crowdsourced instruments have negative aspects as well, such as less representation of natural LCZs and inconsistencies caused by diverse setups. Despite these challenges, the dense and widespread nature of crowdsourced instruments allow for detailed investigation of temperature variability within LCZs, which is influenced by local urban structure heterogeneity and larger-scale urban influences. Addressing biases through comparison with nearby established stations can enhance the reliability of crowdsourced data for urban climate analysis.

In several cases, data from official stations served as the primary source for analysis [10,13,22,29,49,50,56,59,69,70]. Other studies combined official station data with additional sources, including urban network stations [10,22,59,70], EURO-CORDEX model simulations [13,29], and remote sensing technologies [48,49,71,72]. Alexander et al. [73,74] applied the LCZ framework in a different context, combining data from the meteorological station at Dublin Airport with urban energy flux measurements. Bokwa et al. [28] carried out a study on heat load in Central European cities using a combination of urban climate modeling and observational monitoring data. On contrary, Bechtel et al. [75] highlighted the importance of using remote sensing independently, without relying on data from in situ or mobile instruments, thereby enhancing the potential for alternative data collection methods. Nonetheless, Šečerov et al. [16] noted that long-term in situ measurements are essential for capturing thermal variations in different urban settings. These data can help compare temperature differences both within the same city and between different urban areas, providing valuable insights into the urban climate, especially in the canopy layer.

Using in situ measurements for investigating UHI and OTC, as well as for calculating various indices and indicators, can be an effective approach. When positioned at designated measurement points within or around LCZs, in situ measurements can improve our understanding and analysis of these values. They also allow us to gather data over specific time periods (daily, nightly, weekly, monthly, seasonally, or annually), which helps collect the necessary information for thorough analysis. Nonetheless, in situ measurements come with certain drawbacks. It is often the case that in situ stations are set up as part of a specific project and once the project ends, the instruments are neglected and measurements are no longer continued at those locations.

Moreover, in situ stations sometimes fail to cover all parts of an LCZ properly, especially considering that differences exist even within the same LCZ. Another issue is that these stations are often placed where they will not disturb people, sometimes at a height above human level, which means they do not fully reflect the actual thermal experience at ground level or the areas where people are constantly moving. In situ measurements also do not capture the short-term, moment-to-moment changes in human thermal perception, especially when walking through different parts of an LCZ, for example, under a tree, next to a building, or near a glass facade. Therefore, they may not be suitable for certain types of environments.

Combining in situ data with remote sensing and LST measurements helps improve temperature monitoring and enhances the calculation of bioclimatic indices. This approach is also visually easier to interpret. The introduction of CWS, supported and increasingly implemented through EU initiatives, is helping to expand measurement coverage, particularly in areas with high population density.

3.3 Mobile measurement-based studies

The use of mobile instruments allows for the monitoring of dynamic conditions, enabling the rapid collection of data at multiple points within or between different LCZs.

Several studies [19,20,21,27,76,77,78] have used car-based measurements for different purposes. Leconte et al. [19,20,21] emphasized that when using instruments on vehicles, data collected at speeds below 15 km/h should not be used due to heat release from other vehicles, which can affect the measurements, which is why car speeds were in a range of 15–60 km/h. Moreover, Lehnert et al. [27] excluded the measurements on traffic lights where there was influence of heat release in traffic. Leconte et al. [20] found that LCZ 2 and 5 retain more heat due to low SVFs and tall buildings, whereas LCZ 8 and 6/9 exhibit better nocturnal cooling due to more favorable urban morphology. Lelovics et al. [77] confirmed that compact and mid-rise urban areas are warmer than open and low-rise zones, while sparsely built areas showed temperatures similar to rural LCZ classes, validating the link between LCZ types and air temperature.

Skarbit et al. [71] used mobile measurement techniques, incorporating a Citizen weather stations receiver for georeferencing, global positioning system/global navigation satellite system receiver for georeferencing, remote sensing via an airborne thermal camera to collect real-time temperature data, and data from the Hungarian Meteorological Service. Moreover, Herbel et al. [79] conducted a study on detecting the atmospheric urban heat island in Cluj-Napoca, Romania, using a combination of mobile measurements taken from a car and data from in situ stations to enhance the analysis and understanding of the phenomenon.

Bicycle-based measurements are one of the well accepted forms of mobile data collection, chosen for their ability to easily navigate a variety of urban environments. In a study by Lehnert et al. [80], daytime and nighttime temperature measurements were taken to identify hot and cold spots in the city center of Olomouc, Czech Republic. The study found that temperature differences between LCZs were more pronounced at night than during the day. LCZ 2 was consistently the hottest area, making it a hotspot, while LCZ D stood out as one of the coolest locations. On contrary, Emery et al. [81] focused on evening measurements and found that areas with dense vegetation and water bodies likely aid in cooling urban environments at night, potentially reducing the UHI effect.

In a similar manner to the previous research, Rathmann et al. [82] aimed to simultaneously collect data on various relevant characteristics across different urban routes to better understand people’s physiological responses in specific climate conditions. Instead of the method used in earlier studies, this work focused on capturing variations in the thermoregulatory responses of participants during transient thermophysiological states (such as walking) under diverse outdoor conditions. The measurements were taken using a wearable sensor system, with physiological data collected via Microsoft Band 2 wristbands.

Car-based measurements capture temperature and other data while moving but these measurements alone do not show pedestrian routes even though pedestrians are the people most affected by thermal conditions and their routes are largely influenced by the urban environment. Although the variations may be minimal factors like shading and SVF can still introduce bias. Furthermore, studies have shown that LCZ classification is a more important factor than distance from the city center in explaining the UHI effect and that temperature variability within LCZ classes is linked to city size and topographic diversity.

Furthermore, mobile measurements such as those conducted by bicycle or with devices like wearable sensor systems provide better results for the surrounding environment and pedestrians. Bicycle paths are often located right next to pedestrian walkways, presenting a slightly different situation during measurements and enabling more accurate results under certain climatic conditions. In addition, bicycles and certain devices do not contribute to an increase in the surrounding air temperature, as is the case with cars, although a potential factor may be the influence of stronger wind if the cycling speed is higher than walking speed, which directly affects the perception of the environment. Moreover, bicycles and wearable devices allow better access to various locations along the route, while cars are limited to roads. Therefore, it is very difficult to determine the type of mobile measurement to be used, as it depends on the specific city, its local structure, population density, and other factors.

Nonetheless, the combination of various methods is the most important aspect of such measurements in order to fill gaps and correct biases during data analysis within LCZs. Some futuristic beliefs and ideas still lead to other types of research and innovations, as studies have shown that LCZ classification offers a better understanding of intra-urban temperature variations compared to land cover classification because it better reflects the differences in urban morphology, vegetation, and different surface characteristics on site. Therefore, there was a proposal to extend LCZs into “local wellbeing zones” to integrate climatic, physiological, and personal factors, which can help to improve strategies for mitigating urban heat stress and enhancing climate resilience.