City Digital Twins: their maturity level and differentiation from 3D city models

ABSTRACT

The emerging field of City Digital Twins has advanced in recent years with the help of digital infrastructure and technologies connected to the Internet of Things (IoT). However, the evolution of this field has been so fast that a gap has opened in relation to systematic reviews of the relevant literature and the maturation of City Digital Twins on an urban scale. Our work bridges this gap by highlighting maturity in the field. We conducted a systematic literature review with bibliometric and content analysis of 41 selected papers published in Web of Science and Scopus databases, covering five areas: data types and sources, case studies, applied technologies and methods, maturity spectrum, and applications. Based on maturity indicators, the majority of the reviewed studies (90%) were at initial to medium stages of maturity (up to element 3), most of them focused on 3D modelling, monitoring and visualisation. However, digital twins cannot be limited to 3D models, monitoring and visualisation, for they can be developed to include two-directional interactions between humans and computers. Such a high level of maturity, which was not found in the reviewed studies, requires advanced technologies and methods such as cloud computing, artificial intelligence, BIM and GIS. We also found that further studies are essential if the field is to handle the complex urban challenges of multidisciplinary digital twins . While City Digital Twins extend by definition beyond mere 3D city modelling, some studies involving 3D city models still refer to their subjects as City Digital Twins. Among the research gaps we identified, we’d like to highlight the need for near-real-time data analytics algorithms, which could furnish City Digital Twins with big data insights. Other opportunities include public participation capabilities to increase social collaboration, integrating BIM and GIS technologies and improving storage and computation infrastructure.

KEYWORDS:

• City Digital Twins
• big data
• IoT
• artificial intelligence
• city information modelling

1. Introduction

The transformation from traditional cities to smart cities has brought forward new urban development requirements that can be supported by emerging technologies (Wang et al., 2021). Such transformations hold out the possibility of improving the performance of services, infrastructure maintenance and managing socio-economic processes of urban governance. Eventually, they could lead to better living conditions for inhabitants (Caprari, 2022). Dynamic City Digital Twins employ emerging technologies and offer significant potentials to realise these possibilities (Hämäläinen, 2020).

The initial ideas of digital twins were first raised and discussed by Michael Grieves at the University of Michigan and John Vickers of NASA in 2003 in relation to product lifecycle management (Grieves, 2014, 2016, 2019). The first model (i.e. defined by Grieves) was called the “Mirrored Spaces Model”; then, in 2010, the name “digital twins” was introduced by NASA (Grieves, 2005; Shafto et al., 2012; Singh et al., 2021). Since then, research has been conducted to develop and mature digital twins using three-dimensional (3D) models (Callcut et al., 2021) of assets. Such digital replicas are used to observe, predict and maintain their target systems (Singh et al., 2021). The emergence of new technologies and methods (e.g. the Internet of Things (IoT), Artificial Intelligence (AI), big data, cloud computing and blockchain) supports the evolution and maturity of digital twins in various domains, including manufacturing, aerospace, medicine, engineering and smart cities (MarketsandMarkets, 2020) and for various tasks, such as data processing, storage, modelling and analysis (Botín-Sanabria et al., 2022).

Over the past two years, many urban studies have employed digital twins to solve urban challenges and problems, such as infrastructure management (Conejos Fuertes et al., 2020; Simonsson et al., 2021), environment monitoring (Ford & Wolf, 2020), sustainable economy (Çetin et al., 2021), social developments (including public health, wellbeing, epidemiology and others) (El Saddik, 2018; Ferdousi et al., 2022; Topping et al., 2021) and sustainable development (Allam & Jones, 2021). More recently, social development studies including infrastructure management (Ferdousi et al., 2022) have adopted City Digital Twins to optimise citizen functions and so enhance the quality of life and wellbeing. This is achieved by monitoring and modelling human behaviour and interactions with the environment and public spaces (El Saddik et al., 2019; Ferdousi et al., 2022). In addition to quality of life, City Digital Twins can facilitate sustainability, which is especially important given that sustainable infrastructure can have an impact on more than 70% of the Sustainable Development Goals (SDGs) of the United Nations. It can bring socio-economic benefits as well as environmental and ecological benefits to a society through technologies such as earth observation (remote sensing, light detection and ranging (LiDAR), and unmanned aerial vehicles (UAV)). Earth observation would help digital twins to monitor environmental changes (land cover, wetlands, rivers and underground waters), built environment changes (land use and traffic flow), and human activity (Hartley, 2022; Song & Wu, 2021).

City Digital Twins can also identify opportunities to optimise resources, as well as enhance system and economic efficiency (Dembski et al., 2020), enable urban energy management (Agostinelli et al., 2020; Francisco et al., 2020), improve energy consumption management (Zhang et al., 2018), optimise Positive Energy Districts (Shen et al., 2021) and improve Net Zero Energy Buildings (Kaewunruen et al., 2019). Yet, for all of the advances in City Digital Twins, their concept and application at a city scale is still in its infancy. Various challenges and opportunities exist in the field, and further studies are required to discern and utilise the full potential of City Digital Twins (Shirowzhan et al., 2020; Singh et al., 2021).

To the best of our knowledge, few reviews of studies in this domain have been conducted to date (Nam & Pardo, 2011). However, due to the emerging nature of the topic, a growing body of discussions, academic and technical research and reports is being published on the matter. Here, we review the most recent and relevant studies on City Digital Twins over various domains of urban governance, and we discuss their contributions.

In terms of urban infrastructure systems, Al-Sehrawy et al. researched the adaptation of digital twins in urban planning to the city scale, focusing on urban infrastructures (Al-Sehrawy et al., 2021). That study employed a meta-methodology to review literature systematically, while three other sub-methodologies were used to develop digital twins and a classification system. The research discussed how poor capturing of knowledge hinders studies and applications of digital twins in urban areas and the built environment. Pan et al. reviewed urban freight logistic literature, assessing it by bibliometric analysis, and defined a holistic and sustainable urban freight logistics framework (Pan et al., 2021). This framework illustrated the relationship between the urban freight logistics system and other contributing factors, such as socio-demographic characteristics, environment, land use, economy, urban utilities, and infrastructure.

Deng et al. presented a systematic literature review focused on urban governance and smart cities around the world (Deng et al., 2021). This study proposed new city management, operation and maintenance in digital twins. Their proposed platform was based on methods such as mapping technology, machine learning, IoT, simulation, monitoring and collaborative computing.

Monitoring the environment and disaster management is another application of digital twins in cities. Fan et al. reviewed disaster management in urban digital twins (Fan et al., 2021). Their paper explored the potential for a unified digital twins paradigm for interdisciplinary fields, integrating artificial intelligence and algorithms, and increasing the quality of monitoring and visualization of the environment and humans. Recently, Shahat et al. reviewed digital twin practices within the domain of urban studies (Shahat et al., 2021). The authors used a thematic analysis to explore current and future opportunities as well as challenges for City Digital Twins applications, and to define a research agenda.

A city is a complex system that benefits from multi-dimensional and multi-scale analyses and actions to solve its unique challenges (Grieves, 2022). Although these review papers provide valuable information about the potential of digital twins at the city scale, they mainly focus on one single application of City Digital Twins. A holistic view of this research would support decision-makers in addressing complex urban challenges, such as city administration and disaster management. For this reason, a novelty of this study is the scope of research that looks at City Digital Twins as a whole and at the city scale. More importantly, this research is the first attempt to investigate the development and maturity of City Digital Twins in different important fields such as technology, method and dataset.

This paper, therefore, fills research gaps by employing both content analysis and bibliometric analysis to assess the existing literature based on five areas, namely i) data types and sources; ii) case study; iii) applied technologies and methods, iv) maturity spectrum; and v) corresponding application. The first section reviews data types, sources and accessibility to data to identify gaps in existing datasets and future directions. We also review case studies in the existing literature to identify adopted technologies. This enhances the exploration of research gaps. Next, we assess applied technologies and methods, including an investigation of data analysis, visualization and modelling methods. In the fourth section, we review the maturity spectrum, identifying the stages of City Digital Twins presented in the literature using the maturity model introduced by Evans et al. (Evans et al., 2019). Finally, we classify studies into categories and sub-categories based on their application. These five sections support the introduction of future directions for City Digital Twins studies by addressing current trends and gaps.

Considering the main objective of this study, we aim to respond to the question of how much City Digital Twins have matured over time and if their applications have been consolidated. Thus, the paper’s contributions are as follows:

• Classification of city-scale digital twins based on their maturity stage and integration: This paper is the first attempt to classify the literature on digital twins on the “city-scale” based on maturity level and spectrum.

• A systematic review of adopted methods, technologies, tools, datasets and types in City Digital Twins: This paper proposes a detailed review of analytical technologies and methods, as well as datasets and sources at the city scale.

• Classification of City Digital Twins applications at the city scale: This work identifies the most popular City Digital Twins field of research systematically while introducing existing challenges and opportunities in each research area.

• Highlighting future directions through reviews of case studies mentioned in the literature: We highlight future directions extracted from case studies in the existing literature.

The remainder of this paper is structured as follows. Section 2 presents the preliminary digital twins definitions that exist in the literature. Section 3 introduces corresponding methods in the systematic literature review. Section 4 researches literature review findings and results, which are presented in two parts: i) bibliometric analysis, and ii) content analysis. Sections 5 and 6 provide information on the current state, challenges and limitations of City Digital Twins research and opportunities for future research.

2. Preliminary definitions

In 2003, Michael Grieves proposed digital twins as a “virtual digital representation analogous to physical items” (Grieves, 2014). In 2012, NASA characterized digital twins as “an integrated multi-physics, multi-scale, probabilistic simulation of an as-built vehicle or system that uses the best available physical models, sensor updates and fleet history to mirror the life of its corresponding flying twins” (Glaessgen & Stargel, 2012). While various definitions of digital twins have been proposed over the past decade (Barricelli et al., 2019), there is no universal agreement on what constitutes digital twins (Cimino et al., 2019; Haag & Anderl, 2019).

Digital twins in the built environment have spawned a corresponding presence in the literature over the past few years. The existing literature largely interprets digital twins as a dynamic (Hämäläinen, 2021) digital representation (Allam & Jones, 2021; Beil et al., 2020) or replica (Laamarti et al., 2020) of the real world (Allam & Jones, 2021; Huo et al., 2021; Çetin et al., 2021) or urban environments (Beil et al., 2020; Nochta et al., 2021; Pan et al., 2020; Pang et al., 2021; Schrotter & Hurzeler, 2020; Visconti et al., 2021; Çetin et al., 2021) (including their living and non-living assets, processes, and systems) (Lu et al., 2020; Mylonas et al., 2021; Nochta et al., 2021). This digital model integrates big data analytics technologies (such as artificial intelligence and machine learning) (Al-Sehrawy et al., 2021; Dignan, 2020; Lu et al., 2020) to create living digital simulation models (Al-Ali et al., 2020; Lu et al., 2020). Besides simultaneous connection (Laamarti et al., 2020) of the physical entity and seamless data transmission (Dignan, 2020; Laamarti et al., 2020; Pan et al., 2020), bidirectional communication (Al-Sehrawy et al., 2021) can be used to monitor and predict current and future situations (Ford & Wolf, 2020; Gutierrez-Franco et al., 2021; Lu et al., 2020).

Various aspects of digital twins arise from the scope of studies. For example, Al-Sehrawy et al. (2021) and Park et al. (2019) focus on the connection between physical and virtual platforms, while others use digital twin applications in the definition of areas such as decision making (Ford & Wolf, 2020; Huo et al., 2021), policy-making (Nochta et al., 2021), pandemic prediction (Pang et al., 2021) and cyber-security (Alshammari et al., 2021). Table 1 provides some definitions from different perspectives and scopes.

Table 1. Comparing digital twins’ definitions proposed in various perspectives and scopes.

Title	Scope	Definition
“Real-time traffic quantization using a mini edge artificial intelligence platform” (Broekman et al., 2021)	Freeway/smart cities/traffic	“An integrated multi-physics, multi-scale, and probabilistic simulation of a complex product and uses the best available physical models, sensor updates, etc., to mirror the life of its corresponding twin.”
“Smart cities in the time of climate change and Covid-19 need digital twins” (Dignan, 2020)	Climate change	“A digital model of a physical asset that continuously collects information (via sensors, drones or other IoT and IoT data collection tools) applies advanced analytics, machine learning, and artificial intelligence (AI) to gain valuable real-time insights about the physical asset’s performance, operation or profitability.”
“Digital twins’ conceptual model within the context of internet of things” (Al-Ali et al., 2020)	Transportation	“A simulation of a system or product mirroring the physical model into a virtual model”
“A Novel HDF-Based Data Compression and Integration Approach to Support BIM-GIS Practical Applications” (Pan et al., 2020)	Data management	“Dynamic digital reflection of the real urban environment, which should be capable of (1) containing as many cities features as possible, (2) synchronizing with its host dynamically through multiple urban data resources, and (3) forming a real-time and integrated response to the events.”
“Smart City 4.0 from the perspective of open innovation” (Yun & Lee, 2019)	Business	“The digital twins’ strategy where the real problems of the real city are solved in the virtual city through the process of smart transformation”
“Smart-guided pedestrian emergency evacuation in slender-shape infrastructure with digital twin simulations” (Yun & Lee, 2019)	Crisis management	“A Digital Twin is a high-fidelity representation of the operational dynamics of its physical counterpart, enabled by near real-time synchronization between cyberspace and physical space”
“An animated spatial time machine in co-creation: Reconstructing history using gamification integrated into 3d city modelling, 4d web and transmedia storytelling” (Matthys et al., 2021)	Animation	“A digital twin is, in origin, a digital copy of a physical object, and for some years, 3D models of cities have also been called Digital Twins.”

In this paper, and based on reviews of the digital twin literature, the authors provide a unified definition for digital twins. Digital twins go beyond predictions and prescription algorithms to provide a comprehensive structure (multi-physics, multi-scale, and probabilistic) for analysis (monitoring, modelling, simulation and prediction) in a safe environment based on complex “what-if” scenarios from expert knowledge and big data collection (real/near-real) from the existing system.

3. Methods

The present work identifies open challenges for digital twins in cities, as well as potential research initiatives. Data collection with specified criteria, bibliographical analysis and content analysis represent the three stages of the research approach. Two search engines – Scopus¹ and Web of Science² – were selected for their high coverage of relevant articles (Albuquerque et al., 2021; Sepasgozar et al., 2020). Next, a list of published studies on the given topic was compiled from these databases and categorized according to various criteria. It should be noted that bibliometric analysis and content analysis were based on papers found in these two databases. The second stage was bibliometric analysis. Finally, the selected papers were carefully read for content analysis before being classified into groups based on their data types and sources, case studies, maturity spectrum, applied technologies and methods, and applications.

3.1. Data collection with controlled criteria

Figure 1 shows the method proposed for article selection. To retrieve relevant City Digital Twins papers, we performed a search based on the title, abstract, and keywords in the selected databases in January 2022. Moreover, we used “AND” and “OR” boolean operators to make our search results more accurate, resulting in less screening time and the elimination of irrelevant papers. Our search strings included “digital twin” AND “city” OR “urban” OR “built environment”, following Ketzler et al. (2020) and Shahat et al. (2021). We used “built environment” in our keyword search to specify and limit the results to city-scale studies, rather than be restricted to the scale of buildings. In total, 508 relevant articles were identified. Also, to make the results more relevant to the study, some inclusions and exclusions based on the paper type, language, and study area were applied, and duplicates were removed (reflected in Section 3.2 as well as in Table 2 and Figure 1). Eventually, a total of 131 unique articles was selected for further analysis.

Figure 1. The flowchart of the research approach and process including paper selection, search string, inclusion and exclusion criteria, and records’ number.

Table 2. Number of publication records obtained based on analysis search strings and keywords in the database.

Database	Search String	Output numbers
Scopus	TITLE-ABS-KEY (“digital twins” AND city OR urban OR “built environment”)	360
Web of Science	“digital twins” (Topic) and city OR urban OR “built environment” (Topic)	148
Total articles	After applying filters and removing the duplicates	72
Total review articles	After selection of city-scale articles	41

3.2. Inclusion and exclusion criteria

To ensure that only relevant papers were selected for the study, filtering and data cleaning were essential. The main inclusion criteria was that articles be related to built environments’ digital twins at the city scale. Other criteria, such as the type of papers (e.g. journal papers and excluding conference proceedings) and language (i.e. English-only papers), were applied to the search results. As this is an emerging field, theoretical and empirical studies were included. Studies from domains irrelevant to City Digital Twins and urban planning, such as “physics”, “material science”, “mathematics” and “chemistry”, were excluded from the results. No time-limiting filter was applied but the final list of publications date from 2018 to 2022. This includes related open-access articles. The screening of studies revealed that some were focused on digital twins in “buildings” or “Civil engineering”, while 41 were related to city-scale digital twins. Table 2 presents a detailed summary, and Figure 1 presents the flow of paper selection.

3.3. Bibliometric and content analysis

Bibliometric analysis is a quantitative and descriptive technique for finding research patterns, analyzing information, assessing the evolution of literature during time series, identifying research areas and guiding future research (Zhao, 2017). Here, we employed bibliometric analysis to create analytical maps of co-authorships between authors, keywords, and co-authorships between countries. This was done for digital twins in city and urban literature using the VOSviewer and R bibliometrics packages (Cobo et al., 2011; Sepasgozar et al., 2020; Souza & Bueno, 2022). The approach uses metadata in the citations to calculate and rank productions, journal sources, and country collaborations. The information on countries' annual paper production in this field is given in Table 3.

Table 3. The sum of case studies per year related to the subject of study classified based on countries that the research had been done.

	Country	Frequencies
1	China	88
2	USA	57
3	UK	54
4	Spain	36
5	South Korea	28
6	Italy	14
7	Germany	13
8	Australia	9
9	India	9
10	Belgium	8

Bibliometric analysis can only describe basic information about the existing literature. This paper employed an extra level of analysis to achieve greater insight, namely content analysis. Content analysis is a research tool for analyzing the content of various types of data (such as written, verbal or visual communication messages) to extract trends and patterns, either qualitatively or quantitatively. It enables one to determine key facets and to reduce phenomena into some pre-defined categories, contributing to improved analysis and interpretation (Chan et al., 2009). In particular, we adopted qualitative content analysis to group articles into categories, following Krippendorff (2018). We screened, reviewed, and classified papers based on their adopted technologies and methods, types of data used, technologies, and applications. This approach assisted us to identify research trends and locate gaps in the literature across these facets (Zhao, 2017).

4. Results

4.1. Bibliometric analysis

The final selection of articles in the systematic review were published between 2018 and 2022 (Figure 2). A descriptive analysis of the results indicates that while this domain of research is quite new, the publication rate has grown significantly in recent years. Moreover, as the paper selection was made in January 2022, the number of published papers for 2022 is limited to a single month, and we would expect the upward trend to continue in this emerging field of research.

Figure 2. The published scientific literature on City Digital Twins by year until January 2022.

In regard to scientific production by country, China and the United States were notable contributors to the field of digital twins, with 88 and 57 articles, respectively. The UK and Spain occupied third and fourth places, with 54 and 36 articles, respectively.

4.1.1. Identification of the most relevant authors of the research (Co-authorship analysis)

Co-authorship analysis helps researchers to find the most productive authors and their collaborations in the domain. The co-authorship of 131 papers was investigated using VOSviewer. The results can be seen in Figure 3. The minimum number of authors’ documents in this analysis was set at one. The node size, arc size, and node distance represent the authors’ document, co-authorship relations, and collaboration proximity.

Figure 3. Results of co- authorship network analysis of the 131 documents reviewed related to City Digital Twins.

Papers were classified into three clusters. Three authors (i.e. “Zhou.y” “Zhang.h” “Zhao.j”) had the highest number of published papers among all 433 authors. The green nodes were closer to each other than the other two nodes, which shows the authors in this field had closer collaboration.

4.1.2. Mapping of the most frequent words

This paper analyzed keywords of the 131 studies selected for systematic review. In total, originally 1,344 keywords were considered. For the purpose of visualization, keywords cited more than three times were chosen for co-occurrence. The results are illustrated in Figure 4. The size of nodes represents the number of times the keyword occurred, while the thickness of the line shows their interconnections. Keywords relating to “urban planning”, “built environment”, and “digital twin” emerged in a cluster corresponding to articles mainly published in the second half of 2020 (as shown in green). The most frequent keywords were “digital twin”, “smart city”, “internet of things (IoT)” and “BIM”, while “urban planning” came in at seventh.

Figure 4. Results of mapping co-occurrence network analysis (plus-keywords) of the terms in the 131 documents reviewed on the City Digital Twins’ studies, identified in the developed bibliography.

4.1.3. Identification of the most relevant countries of the research

The co-authorship map of countries (Figure 5) was created with a minimum of two documents per country. As shown in the map, a total of 22 countries were identified, where the United States, China, and the United Kingdom had the highest number of publications, followed by South Korea and Spain. Although the United States and the United Kingdom had the highest rates of publication among all countries, China and the United States were verified to have the highest international collaboration in the community.

Figure 5. Results of mapping the co-authorship network analysis among countries. The minimum number of co-occurrences of keywords was two.

4.2. Content analysis

For this part of the study, open access articles (i.e. 67 articles) were read and reviewed carefully. To narrow down the results and focus on the research aim, content analysis was limited to city and neighbourhood scales, which reduced the number of articles to 41. To identify the most important challenges and trends, information in the articles was assessed according to five criteria: i) Data types and sources; ii) Case study; iii) Applied technologies and methods; iv) Maturity spectrum; and v) Application.

4.2.1. Data types and sources

In reviewing datasets in City Digital Twins studies, we considered data types, data sources, and the accessibility of the datasets. For data type, we classified the literature’s data into structured and unstructured data, following Al-Ali et al. (2020) and Salkuti (2021) (Table 4). Structured data is a hierarchical type with a definite format and length Eberendu, 2016). This type includes relational databases and organized numbers such as structured query language, access, tabled data, numbers, strings or text (words and numbers in a group format and tabled) (Salkuti, 2021). Unstructured data cannot be defined in any structure. This second data type mostly includes digital images, videos, text, audio, and other non-database data (Eberendu, 2016; Salkuti, 2021).

Table 4. The results of classifying data types used in the existing City Digital Twins literature.

Type of data	Details of data	Description	References
Structured	Sensors data	Including different types of sensors varying from traditional sensors to wearable or personal devices	Park et al. (2019), Chaves et al. (2021), Laamarti et al. (2020), Conejos Fuertes et al. (2020), Lu et al. (2020), Lenfers et al. (2021), Sun et al. (2020), Dembski et al. (2020), Schrotter and Hurzeler (2020)
	Land administration data	Including land features and related attributes such as geographical features, cadastral maps, land-use type, land cover, ownership, and others.	Park et al. (2019), Conejos Fuertes et al. (2020), Beil et al. (2020), Buckley et al. (2021), Lu et al. (2020), Al-Ali et al. (2020), Huo et al. (2021), Simonsson et al. (2021), Park and Yang (2020a), Lenfers et al. (2021), Truu et al. (2021), Orozco-Messana et al. (2021), Bass et al. (2021), Dembski et al. (2020), Schrotter and Hurzeler (2020)
	Building information	Including information and data about buildings such as surface type, area, height, numbers of windows, and others.	Buckley et al. (2021), Lu et al. (2020), Al-Ali et al. (2020), Park and Yang (2020a), O’Dwyer et al. (2020), Orozco-Messana et al. (2021), Bass et al. (2021), Han et al. (2020), Schrotter and Hurzeler (2020)
	Socio-demographic data	Including socio-demographic attributes such as population, density, age, and others, and behavioral data	Gutierrez-Franco et al. (2021), Buckley et al. (2021), Han et al. (2020)
	Historical and time-series data	Including time-series data and previous documents data	Dembski et al. (2020), Matthys et al. (2021), Gutierrez-Franco et al. (2021), Lu et al. (2020), Al-Ali et al. (2020), Simonsson et al. (2021), Orozco-Messana et al. (2021), O’Dwyer et al. (2020), Major et al. (2021)
	3D data	Including 3D features, Point clouds, OSGB files and others.	Dembski et al. (2020), Pan et al. (2020), Matthys et al. (2021), Beil et al. (2020), Al-Ali et al. (2020), Park and Yang (2020a), Visconti et al. (2021), Lu et al. (2020), Schrotter and Hurzeler (2020)
	Traffic and Transportation	Including roads networks and data and data of vehicles	Dembski et al. (2020), Gutierrez-Franco et al. (2021), Beil et al. (2020), Park and Yang (2020a)
	Climate and weather	Including data related to weather, such as temperature, pollution, and others.	Dembski et al. (2020), Truu et al. (2021)
	Energy consumption	Including data related to the energy consumption of buildings or cities.	Park et al. (2019), Gutierrez-Franco et al. (2021), Buckley et al. (2021), Lu et al. (2020), Park and Yang (2020a), Bass et al. (2021)
	Infrastructure	Including infrastructure network and its related data.	Chaves et al. (2021), Conejos Fuertes et al. (2020), Gutierrez-Franco et al. (2021), Simonsson et al. (2021), O’Dwyer et al. (2020), Sun et al. (2020), Dembski et al. (2020)
Unstructured	Text	Including interviews, workshops, official plans, official documents	Matthys et al. (2021), Schrotter and Hurzeler (2020), Hämäläinen (2021), Çetin et al. (2021)
	Image	Including satellite images, aerial images, and documented images from the past.	Dembski et al. (2020), Pan et al. (2020), Matthys et al. (2021), Lu et al. (2020), Al-Ali et al. (2020), Huo et al. (2021), Broekman et al. (2021), Schrotter and Hurzeler (2020)
	Video	Including video traffics.	Broekman et al. (2021)

Based on our review, 63% of studies (21 papers out of the 41) had adopted structured data in their analysis. This included sensor data, land administration data, building information, socio-demographic data, historical and time-series data, 3D data, traffic and transportation data, climate and weather data, energy consumption, and urban infrastructure data. Land administration data (Al-Ali et al., 2020; Beil et al., 2020; Buckley et al., 2021; Conejos Fuertes et al., 2020; Dembski et al., 2020; Huo et al., 2021; Lenfers et al., 2021; Lu et al., 2020; Orozco-Messana et al., 2021; Park & Yang, 2020a; Park et al., 2019; Schrotter & Hurzeler, 2020; Simonsson et al., 2021; Truu et al., 2021), historical and time-series data (Al-Ali et al., 2020; Dembski et al., 2020; Gutierrez-Franco et al., 2021; Lu et al., 2020; Major et al., 2021; Matthys et al., 2021; Orozco-Messana et al., 2021; O’Dwyer et al., 2020; Simonsson et al., 2021), building information data (Al-Ali et al., 2020; Bass et al., 2021; Buckley et al., 2021; Han et al., 2020; Lu et al., 2020; Orozco-Messana et al., 2021; O’Dwyer et al., 2020; Park & Yang, 2020a; Schrotter & Hurzeler, 2020), and sensors data (Chaves et al., 2021; Conejos Fuertes et al., 2020; Dembski et al., 2020; Laamarti et al., 2020; Lenfers et al., 2021; Lu et al., 2020; Park et al., 2019; Sun et al., 2020) were the most frequent types of data used in City Digital Twins studies. Although unstructured data (e.g. video, image, and text) was frequently employed in digital twins research (Marai et al., 2021; Wu et al., 2022), only several city-scale studies (Al-Ali et al., 2020; Broekman et al., 2021; Dembski et al., 2020; Huo et al., 2021; Hämäläinen, 2021; Lu et al., 2020; Matthys et al., 2021; Pan et al., 2020; Schrotter & Hurzeler, 2020) used such data in their work.

Some researchers believe that participatory methods can improve the effectiveness of digital twins and decision-making (Abdeen & Sepasgozar, 2022; Li et al., 2021). Various technologies and methods of data gathering, analysis, and decision-making are therefore adopted to increase stakeholder and community participation in city digital twins (Abdeen & Sepasgozar, 2022; Thoneick, 2021). These participatory methods and technologies include (but are not limited to) structured/semi-structured interviews, volunteered geographic information (VGI) and social media data (Abdeen & Sepasgozar, 2022). Despite the benefits of such participation for improving digital twins, few of the reviewed papers applied community and stakeholder participation through interviews (Hämäläinen, 2021; Çetin et al., 2021) and workshops (Matthys et al., 2021), or VGI (Dembski et al., 2020). Details of data types, descriptions and references are given in Table 4.

Dataset sources in the existing literature were separated into eight groups: official documents; satellites and aerial images; sensors; open sources; aerial vehicles; 3D scanners; census and surveys; and interviews and workshops. Official documents (plans, historical records, urban infrastructure documents, and official geographical databases) were used in the majority of reviewed studies (Bass et al., 2021; Buckley et al., 2021; Chaves et al., 2021; Conejos Fuertes et al., 2020; Dembski et al., 2020; Gutierrez-Franco et al., 2021; Lu et al., 2020; Matthys et al., 2021; Orozco-Messana et al., 2021; Park & Yang, 2020a; Park et al., 2019; Schrotter & Hurzeler, 2020; Simonsson et al., 2021; Sun et al., 2020; Truu et al., 2021) as the main data source. IoT data from sensing devices was another popular source, with trending applications in the smart cities literature (Alghamdi & Khan, 2021; Javed et al., 2021). Sensors were not limited to traditional types for measuring and assessing a physical phenomenon, such as weather and traffic flow, but also included personal devices to measure city data, human behaviour and activity, such as health wearables, mobile smartphones, and GPSs (Chen et al., 2019). This data is identified as the second most frequently used data source in existing City Digital Twins literature (Al-Ali et al., 2020; Chaves et al., 2021; Conejos Fuertes et al., 2020; Ford & Wolf, 2020; Han et al., 2020; Laamarti et al., 2020; Lenfers et al., 2021; Lu et al., 2020; Major et al., 2021; Park et al., 2019; Sun et al., 2020). However, only two studies (Anda, Medina, et al., 2021; Laamarti et al., 2020) used personal devices (health wearables and mobile phones), which highlights the potential of greater analysis of human behaviour. The details of these data sources are given in Table 5.

Table 5. Distribution of reviewed studies in the subject of research based on data sources used in their research.

Data sources	Description	References
Official documents	Including official plans, historical records (images and related documents), urban infrastructure documents, and an official geographical database.	Dembski et al. (2020), Park et al. (2019), Chaves et al. (2021), Matthys et al. (2021), Conejos Fuertes et al. (2020), Gutierrez-Franco et al. (2021), Buckley et al. (2021), Lu et al. (2020), Simonsson et al. (2021), Park and Yang (2020a), Truu et al. (2021), Orozco-Messana et al. (2021), Bass et al. (2021), Sun et al. (2020), Schrotter and Hurzeler (2020)
Satellites and aerial images	Including different satellite images, aerial images, and airborne imagery.	Dembski et al. (2020), Huo et al. (2021), Lenfers et al. (2021), Bass et al. (2021)
Sensors	Including different types of sensors, such as traditional sensors for physical phenomena and personal devices.	Park et al. (2019), Chaves et al. (2021), Laamarti et al. (2020)¸ Conejos Fuertes et al. (2020), Lu et al. (2020), Al-Ali et al. (2020), Lenfers et al. (2021), Sun et al. (2020), Ford and Wolf (2020), Major et al. (2021), Han et al. (2020)
Open sources	Including DWG open data and open-source websites (for geographical information, weather data, and others).	Matthys et al. (2021), Pang et al. (2021), Beil et al. (2020), Buckley et al. (2021), Al-Ali et al. (2020), Park and Yang (2020a), Orozco-Messana et al. (2021)
Aerial vehicles	Including devices such as drones and unmanned aircraft.	Lu et al. (2020), Al-Ali et al. (2020), Broekman et al. (2021), Anda, Medina, et al. (2021)
3D scanners	Including 3D scanner and Lidar.	Dembski et al. (2020), Schrotter and Hurzeler (2020)
Census and survey	Including official census and surveys	Dembski et al. (2020), Buckley et al. (2021), Truu et al. (2021)
Interviews and work shops	Including structured and semi-structured interviews.	Dembski et al. (2020), Matthys et al. (2021), Hämäläinen (2021)

Data accessibility was assessed with two parameters: i) using open data in the data-gathering process; ii) raw and processed data available due to authorial decisions. Open data is available and accessible online (over online platforms, websites, and databases) and can be gathered and corrected easily by users and machines (Neves et al., 2020). Such data can benefit the whole community (e.g. citizens, governments, and stakeholders) in various ways; for example, by improving information sharing between stakeholders, increasing efficiency by reducing data-gathering time, enabling analysis, supporting knowledge discovery in complex problems, and embracing economic opportunities through providing new jobs (Neves et al., 2020; Ojo et al., 2015; Vetrò et al., 2016). Moreover, our findings indicate that only some studies (Al-Ali et al., 2020; Beil et al., 2020; Buckley et al., 2021; Matthys et al., 2021; Orozco-Messana et al., 2021; Pang et al., 2021; Park & Yang, 2020a) have used open-source datasets, despite these benefits. These studies generally used OSM (www.openstreetmap.com) and local web-based portals (Beil et al., 2020; Dembski et al., 2020; Orozco-Messana et al., 2021; Park & Yang, 2020a) for geographical and land administration data, public health datasets (Pang et al., 2021), and socio-economic datasets (Buckley et al., 2021; Park & Yang, 2020a).

Our second parameter concerned the availability of raw and processed data, and only six papers provided this opportunity (Ford & Wolf, 2020; Hämäläinen, 2021; Lenfers et al., 2021; Lu et al., 2020; Orozco-Messana et al., 2021; Pan et al., 2020). It should be noted that sharing data can enable the sharing of knowledge and increase the validity and clarity of analyses and processes.

4.2.2. Case studies

Case studies exemplify concepts and practical applications while demonstrating the efficacy of digital twins in more interdisciplinary contexts, such as urban areas. They can be effective in identifying innovations as well as analyzing and increasing data quality, assessing results from an isolated system, identifying future research opportunities and discerning ways to improve and update results.

In this section, we review various case studies introduced in the City Digital Twins literature. What was evident overall was the need to improve existing digital twins in terms of integrity and interaction, including with GIS and BIM technological advancements, the sustainability and efficiency of digital modelling and analysis (Gutierrez-Franco et al., 2021; Huo et al., 2021; Lu et al., 2020; Schrotter & Hurzeler, 2020; Sun et al., 2020; Visconti et al., 2021), citizen engagement and collaboration with public administration in planning and design processes (Hämäläinen, 2021), and socio-economical attributes (Dembski et al., 2020; Hämäläinen, 2021; Lu et al., 2020; Sun et al., 2020).

The reviewed studies mainly concentrated on City Digital Twins simulations or work on advancing technical and methodological problems. Meanwhile, City Digital Twins visualization, which is more complex than 3D modelling of a single building, seems to have received less attention, even though urban scale visualization requires various tools, technologies, and methods that offer advanced spatial analysis. For this reason, some researchers believe it is essential to integrate software such as BIM and GIS into City Digital Twins (Pan et al., 2020; Song et al., 2017; Zhu & Wu, 2021). The details of the case study analysis are given in Table 6.

Table 6. The results of analyzing case studies mentioned in the existing City Digital Twins literature.

Case study	Purpose	Technologies	Data type	Further discussion
Cambridge, England (Lu et al., 2020)	Developing digital twins at building and city levels with a focus on the operation and maintenance field	BIM Cloud Computing	Sensors data Land administration data Building information data Historical and time-series 3D data Energy consumption data	Making practical digital twins on a large scale Working on organizational productivity Focusing on social and economic results such as optimizing city services.
Zurich, Switzerland (Schrotter & Hurzeler, 2020)	Increasing public awareness through realizing 3D spatial data Developing and discussing scenarios through visualization and analysis	GIS	Sensors data Building information data Land administration data 3D data LIDAR images	Generating better inter-connected 3D data Updating of existing data Systematic recording of urban furniture Better integration of the BIM and GIS Detection and maintenance of the infrastructure Building a 3D cadastre database.
Helsinki, Finland (Hämäläinen, 2021)	The applicability of digital twins in urban planning and innovative developments.	BIM	Text Land administration data	Transforming digital twins perspectives through socio-technical aspects
Herrenberg, Germany (Dembski et al., 2020)	Aiding in participatory and collaborative processes in digital twins.	Big Data analytics	Sensors data Land administration data Historical and time-series 3D data Traffic and Transportation Climate and weather data Infrastructure data Images	Developing an integrated toolbox and method that focuses on advancing methods concerning socio-economic data.
Barcelona, Spain (Sun et al., 2020)	This paper has analyzed the operation of the urban water cycle under the cyber-physical framework.	Real-time analysis	Sensors data Infrastructure data	Deep research in detailed interconnecting digital twins Interfacing with real systems to provide real-time control
Remscheid, Germany/Flat Iron USA/Vienna, Austria (Visconti et al., 2021)	To keep the digital twins of the case study synchronized (both directions) with the actual city’s data and information.	CityGML Data-Driven and Model	3D data	Integrating Interfaces and toolchain to support the design cycle Generalizing the synchronizing to a many-to-one model
Haapsalu, Estonia/Söderhamn, Sweden (Truu et al., 2021)	Presenting an innovative concept in pluvial flood risks reduction Creating dynamic interlinkages between land developments, the urban drainage system, and other flood-related factors	GIS	Land administration data Climate and weather	Mitigating the flood risks faced buildings Analyzing the best design of the building envelope in floods.
Changsha, China (Huo et al., 2021)	Improving previous methods of large-scale oblique photogrammetry models visualization.	GIS	Land administration data Images	Linking synchronous loading method (based on the parent-child relationship) and data structure of quadtrees and octrees Visual content and video streaming integration
Dublin (Buckley et al., 2021)	Exploring the use of an urban building energy model to examine the potential for creating a near-zero carbon neighbourhood	UMI (urban modelling interface)	Land administration data Building information data Socio-demographic data Energy consumption data	Linking archetypes to their geography using building footprints
Bogota (Gutierrez-Franco et al., 2021)	Flexible decision making, achieving social welfare, increasing management productivity, and reducing environmental impacts and costs.	GIS Artificial intelligence	Socio-demographic data Historical and time-series Traffic and Transportation Energy consumption Infrastructure data	Incorporation of a circular economy in City Digital Twins framework.

4.2.3. Applied technologies and methods

For this section, we reviewed adopted technologies and methods of analysis that were used in City Digital Twins papers (Table 7). Based on our findings, we note that some technologies, such as GIS (Buckley et al., 2021; Conejos Fuertes et al., 2020; Gutierrez-Franco et al., 2021; Huo et al., 2021; Lenfers et al., 2021; Lu et al., 2020; Major et al., 2021; Matthys et al., 2021; Orozco-Messana et al., 2021; Pan et al., 2020; Park & Yang, 2020a; Simonsson et al., 2021; Truu et al., 2021), Artificial Intelligence (include Machine Learning, Federated Learning, and Markov Model) (Al-Ali et al., 2020; Anda, Ordonez Medina, et al., 2021; Broekman et al., 2021; Gutierrez-Franco et al., 2021; Han et al., 2020; Laamarti et al., 2020; Lenfers et al., 2021; Matthys et al., 2021; O’Dwyer et al., 2020; Pang et al., 2021; Park & Yang, 2020a; Park et al., 2019), and Internet of Things (Al-Ali et al., 2020; Broekman et al., 2021; Major et al., 2021; O’Dwyer et al., 2020; Park et al., 2019), stand out as the most common and relevant technologies for monitoring, visualization, and prediction in City Digital Twins analyses.

Table 7. Classification of applied technologies in the 41 reviewed City Digital Twins literature.

Technology	References
GIS	Pan et al. (2020), Matthys et al. (2021), Conejos Fuertes et al. (2020), Gutierrez-Franco et al. (2021), Buckley et al. (2021), Lu et al. (2020), Huo et al. (2021), Simonsson et al. (2021), Park and Yang (2020a), Lenfers et al. (2021), Truu et al. (2021), Orozco-Messana et al. (2021), Major et al. (2021)
BIM	Pan et al. (2020), Lu et al. (2020), Major et al. (2021), Hämäläinen (2021)
UMI (urban modelling interface)	Buckley et al. (2021)
Cloud Computing	Lu et al. (2020)
Internet of Things	Park et al. (2019), Al-Ali et al. (2020), O’Dwyer et al. (2020), Broekman et al. (2021), Major et al. (2021)
Artificial Intelligence*	Matthys et al. (2021), Gutierrez-Franco et al. (2021), Park and Yang (2020a), Lenfers et al. (2021), Park et al. (2019), Al-Ali et al. (2020), Broekman et al. (2021), Pang et al. (2021), Laamarti et al. (2020), Han et al. (2020), O’Dwyer et al. (2020), Anda, Medina, et al. (2021)
Edge and Mini Edge Computing	Park et al. (2019), Al-Ali et al. (2020), Broekman et al. (2021)

*This category consists of AI and its related technologies that are used in selected research. Amongst which are Machine Learning, Federated Learning, and Markov Model.

Most studies (Buckley et al., 2021; Conejos Fuertes et al., 2020; Gutierrez-Franco et al., 2021; Huo et al., 2021; Matthys et al., 2021; Orozco-Messana et al., 2021; Park & Yang, 2020a; Simonsson et al., 2021) employed GIS mainly for spatial analysis; only in some cases was it adopted for urban 3D models (Matthys et al., 2021). Some studies (Carstens, 2019; Matthys et al., 2021; Zhu & Wu, 2021) concluded that GIS technology failed to provide a sufficient graphical user interface and 3D maps of sufficiently high quality. For this reason, and to achieve well-developed City Digital Twins, we believe that BIM and GIS technology integration is essential (Lu et al., 2020; Major et al., 2021; Pan et al., 2020; Visconti et al., 2021). However, only one paper among the 41 reviewed studies focused on this integration (Pan et al., 2020). Pan et al. (2020) explored a hierarchical data format-based scheme to solve the inefficiency of city information modelling storage in GIS and BIM integration. Some studies conducted research to integrate these technologies in digital twins (Gnädinger & Roth, 2021) in various dimensions, such as integrating data sources, modelling spatio-temporal statistics, developing semantic web standards, and integrating IFC and CityGML at different levels (Celeste et al., 2022; Song et al., 2017; Wang et al., 2019). In order to achieve a high level of integration, theoretical advances in technologies, tools, and methods are vital. Based on this study, further research is required to develop these aspects and answer the challenges that lie ahead in large-scale digital twins such as cities, including storage process, time management, and data type (Song et al., 2017). These challenges include providing and developing semantic data model and integrated city information datasets that include various attributes of city components (Xia et al., 2022). Also, as some researchers believe (Shkundalov & Vilutienė, 2021; Zhu & Wu, 2022), BIM/GIS integration should in future solve problems related to interoperability data challenges, such as data exchange time and flow and optimization.

Technological advances have provided new opportunities for industries and academics in data gathering, analysis, visualization, and modelling (Shahzad et al., 2022), including cloud computing, artificial intelligence, Internet of Things and big data (Lu et al. 2019; Shahzad et al., 2022). Machine learning, deep learning, federated learning and the Markov model are types of artificial intelligence (Nguyen et al., 2021; Rupali & Amit, 2017). Federated learning is an emerging field of artificial intelligence that allows all users and clients to train and retrain data while learning from shared models. This technology will help in exchanging knowledge from various sources while improving itself (Pang et al., 2021). In our reviews, we found that federated learning is an emerging technology in City Digital Twins. Pang et al. (2021) adopted this technology to increase collaboration in City Digital Twins, particularly in crisis management. In addition, cloud computing (Borodulin et al., 2017; Jiang et al., 2021) and blockchain technologies (Yaqoob et al., 2020) have provided new opportunities in digital twins studies for real-time and near real-time analysis. However, based on our city-scale digital twins review, it appears that only Hou et al. (Lu et al., 2020) have worked on cloud computing in City Digital Twins.

The next part involved analysing applied methods in City Digital Twins literature (Figure 6). Generally, real-time analysis (Broekman et al., 2021; Conejos Fuertes et al., 2020; Han et al., 2020; Huo et al., 2021; Lu et al., 2020; Sun et al., 2020), data-driven and model-driven methodology (Gutierrez-Franco et al., 2021; Han et al., 2020; Major et al., 2021; Park & Yang, 2020a; Simonsson et al., 2021; Visconti et al., 2021), and big data techniques and analytics (Al-Ali et al., 2020; Conejos Fuertes et al., 2020; Dembski et al., 2020; Gutierrez-Franco et al., 2021; Laamarti et al., 2020), were frequent methods that were applied in the literature.

Figure 6. Distribution of applied methods in 41 reviewed City Digital Twins studies at a glance.

3D city models and City Digital Twins require advanced knowledge of several fields, such as geometry, photogrammetry, digital image processing and digital analysis (Lee & Yang, 2019; Shan & Sun, 2021; Yang, 2019). 3D models can be created and shaped by means of various technologies and methods, such as image-based analysis, laser scanner-based analysis, remote sensing-based analysis and others (Park & Yang, 2020a, 2020b). In our review, we found that while image-processing methods were employed in several digital twin studies, this method was not used in the reviewed literature. In addition, photogrammetry and geometry were only used in one paper (Huo et al., 2021). In this study, Hou et al. (2021) deployed Oblique Photogrammetry models to improve visualization fluency and decrease memory use.

Another point worth mentioning is that learning loops and the consideration of human intelligence in the loop of artificial intelligence (AI), as well as digital analysis, are essential to increase system resilience and robustness, improve functionality (Rikakis et al., 2018; van der Aalst et al., 2021) and hinder failure during disruption periods. Human intelligence, therefore, should be considered as part of the data process in digital twins to provide different answers in “what-if” analyses and situations (van der Aalst et al., 2021). For this reason, it is essential to consider human intelligence and participation in methodological and learning processes. Based on our research results, four papers used interviews (structured and semi-structured), co-creation workshops, federated learning methods or held co-creation workshops (Hämäläinen, 2021; Matthys et al., 2021; Nochta et al., 2021; Pang et al., 2021; Çetin et al., 2021). More details on City Digital Twins content analysis are given in Table 1S in the Appendix.

4.2.4. Maturity spectrum

Maturity models are used to assess processes, organizations and systems against a norm and so identify the potential for improvements (Evans et al., 2019). These models, which originated in the software industry for improving software development and maintenance (Paulk et al., 1993), are a common topic of discussion in digital twin academic and professional literature (Davila Delgado & Oyedele, 2021). By defining a series of progressive phases that outline required capabilities, digital twins’ maturity models facilitate tracking, benchmarking and achieving technological advances. An overview of various maturity levels proposed in the academic publications is presented in Table 8.

Table 8. Digital twins’ maturity models and classifications that proposed in different sources.

Papers	Bojr et al. (2020)	Madni et al. (2019)	Evans et al. (2019)	Chen et al. (2021)	Shemyakina et al. (2022)
Proposed levels of digital twins Maturity	Generation #1 – monitoring platforms	1 Pre-Digital Twin	Element 0: Reality capture (for existing physical assets)	Unaware	Vertical
	Generation #2 – intelligent semantic platforms	2 Digital Twin	Element 1: 2D map/systems or 3D model (object-based only)	Identifiable	Horizontal
	Generation #3 – agent-driven socio-technical platforms	3 Adaptive Digital Twin	Element 2: Connected to persistent (static) data, metadata and BIM Stage 2	Aware	Connective
		4 Intelligent Digital Twin	Element 3: Enrich with real-time (dynamic) data	Communicative	Intellectual
			Element 4: Two-way integration and interaction	Interactive
			Element 5: Autonomous operations and maintenance	Instructive and Intelligent

In this paper, we selected the Atkins maturity spectrum (Evans et al., 2019) as our main model and extended it, following Botín-Sanabria et al. (2022) and Chen & Jupp (2021). In this model, the lowest element (element 0), which digital twins begin with, is defined as reality capture. Data in element 0 is collected through various methods such as surveying, point clouds, photogrammetry, drones, aerial vehicles and others, to build existing physical assets, as-built maps and basic geometry design.

In the next element, objects are designed and updated through realities captured in element 0. It should be noted that element 1 is only limited to surfaces and shapes of objects and does not include BIM data or metadata. In element 2, element 1 is connected to persistent data sets and empowered by various metadata such as BIM. In this element, all data is unified as a single reference point that will reduce errors, uncertainty and costs, while increasing validity. Element 2 helps engineers, project planners and managers to make faster and better decisions. In other words, it provides integrated multi-physics and multi-scale simulations and answers questions such as “what if”. The next element includes sensors, Internet of Things (IoT) devices, and real-time/semi-real, with only one-directional flow between digital and physical assets. Element 3 also needs real-time data-gathering devices (active or passive) such as sensors (Evans et al., 2019).

In element 4, the integration between different parts of digital twins develops significantly. In this element, digital twins can be linked with other twins or assets. Also, two-way interaction and integration between humans and machines forms. For this integration, sensor technologies and mechanical augmentation of the physical asset are vital. In the final element, the highest level of digital twins is created to operate and maintain the system autonomously. In element 5, complete self-governance and self-management with comprehensive oversight and transparency is achieved (Evans et al., 2019). Based on this, digital twins begin with simple data gathering, 3D maps and simple monitoring but can proceed to complete automation of the system. Figure 7 illustrates the digital twins maturity levels and the above explained five elements.

Figure 7. Digital twins maturity levels and descriptions proposed by Evans et al. (2019).

We reviewed the maturity spectrum of existing City Digital Twins to date, as shown in Figure 8. In total, 21 practical City Digital Twins cases were reviewed. We found that most of the City Digital Twins were at elements 2 or 3. Results show that 38% of reviewed papers were at element 2 and 52% were at element 3. At element 3, studies have integrated real-time data streams with the help of IoT sensors and machine learning. Among the reviewed articles that are allotted to element 3, it seems that three papers (Bass et al., 2021; Dembski et al., 2020; Lu et al., 2020) have worked on more mature digital twins than the others in this element. Although these papers had created one-directional data flow and integrated different dimensions of City Digital Twins, they were unable to identify human and machine interaction or automation in their research.

Figure 8. Results of classification of 41 reviewed City Digital Twins studies based on their maturity elements proposed by Evans et al. (2019).

Based on our findings, City Digital Twins can be found at element 4 in only one study (Sun et al., 2020). Sun et al. (2020) designed and presented City Digital Twins for real-time control of the urban water cycle. This paper created two directions of information flow – in different digital layers, and between humans and machines. Although it seems that this paper has designed and proposed more advanced digital twins than other reviewed articles, it is limited to urban infrastructure, specifically, the urban water system, and does not include other dimensions of the urban system as a comprehensive City Digital Twin.

In conclusion, city and national digital twins developments go beyond digital replicas of the physical environment and 3D models. While recent studies have advanced beyond making 3D models and visualizations only, more research and projects are required to integrate humans and machines (two-directional data flow), automation in maintenance and operation, and consideration of different city dimensions in digital twins as a whole.

4.2.5. Research applications

This section reviews and classifies papers according to their applications. To do this, we employed annotated labels (i.e. main field, sub field, application) and a three-level classification system. First, we classified papers into their corresponding main-field (shown in the inner circle), following Ketzler et al. (2020). These fields include built environment, infrastructure, energy management, I.T. and technology, environment, business and economy, public health, population movement, and institutional. Next, the papers of each class were classified into separate groups based on their corresponding sub-fields, which are reflected in the middle circle. Finally, we classified the papers in each main field based on their corresponding applications, which are represented in the outer circle. Depending on their contents, collected manuscripts could be mapped to one or more applications. Figure 9 illustrates the result of this classification. In the figure, the section’s colour intensity and size reflect the frequency of the explored research areas. The inner circle is divided by the total number of papers in each main field. Therefore, the largest section represents the most frequent papers in the relevant field among the reviewed literature. In the outer circle, the number of applications in each main field of the reviewed papers is reflected in the size of each segment and the intensity of colours.

Figure 9. Results of the content analysis on the papers of City Digital Twins. The inner circle shows the main fields of studies in City Digital Twins literature. The next circle illustrates the corresponding sub-fields, and the outer circle reflects the corresponding applications in each subfield. Different colours show separate fields. More frequently explored research areas are reflected with more intense colours and larger sections.

Integration can be defined in two ways: first, the paper provides integrated research on data and information, process and subject; second, interdependencies between city components and elements are studied as a whole. The results show that the first type of frequency seems to occur more often than the second type (Al-Ali et al., 2020; Anda, Medina, et al., 2021; Bass et al., 2021; Buckley et al., 2021; Gutierrez-Franco et al., 2021; O’Dwyer et al., 2020; Park & Yang, 2020a; Sun et al., 2020; Truu et al., 2021). In such research, the main focus is on data structure, data fusion and technical process, and papers mostly focus on one main topic, such as transportation. However, this does not provide solutions for complex urban problems, which are interdisciplinary by nature.

Another interesting point about these applications is the frequency of interoperability and data fusion in the articles. Data fusion, interoperability and management are some of the most common applications and components of digital twins, especially in the urban context, where they deal with large quantities of data. The studies employed a range of strategies to meet these challenges. Examples include: developing a five-layer system architecture in urban areas (Dembski et al., 2020; Park et al., 2019), working with different types of real-time data (Al-Ali et al., 2020; Broekman et al., 2021) and a mixed-method approach for urban 3D modelling (Dembski et al., 2020).

In light of the reviewed papers, it appears that little research has been done on decision-making (Bass et al., 2021; Gutierrez-Franco et al., 2021; Lenfers et al., 2021; Çetin et al., 2021), policy-making (Nochta et al., 2021) and policy evaluation (Marcucci et al., 2020; Orozco-Messana et al., 2021). This is the case even though various studies have increased situational awareness and monitored or tracked the situation. Examples include monitoring environmental situations and increasing awareness in energy management and consumption (Chaves et al., 2021; Park & Yang, 2020a; Simonsson et al., 2021).

5. Discussions: current state, challenges, and limitations of City Digital Twins

Despite the development of City Digital Twins over recent years, the domain is still emerging and further research is needed to overcome its limitations and challenges. Some researchers (Batty, 2018) believe that complete and ultimate digital twins of cities will never be achieved, but advances can help in the maturation process. For this reason, it is crucial to identify research gaps and the potential to advance digital twins.

The first research gap that our study has identified pertains to technical reports and the literature. For our paper, we decided to conduct a “systematic literature review”, which does not include technical reports. However, as this topic is an emerging research field, some research results are presented in technical reports (industry and universities) and still need time to appear in journals. Therefore, an excellent opportunity exists for researchers to review the literature and technical reports to find new advances and innovations in the field.

Below and in Figure 10, we outline the challenges, gaps, possible future discussions and limitations related to each section. This section can guide City Digital Twins researchers, urban planners, policy-makers and city managers as they choose future directions for research.

Figure 10. Current state, challenges, and limitations of City Digital Twins.

5.1. Data types and sources

Open-source databases and participatory data-gathering methods are considered to be important sources of data gathering in digital twins and, especially, real-time data analysis. Some open-source databases have been used in reviewed studies, while there is an opportunity to use volunteered geographic information (VGI) and social media data, especially in real-time analysis.

Human behaviour and activity analysis are essential for understanding community dynamics. So, a variety of sensors – such as personal health devices, mobile phone cellular data and personal GPSs (vehicle tracking) – could be used more often, anonymously, in city-scale digital twin studies. To fill these gaps, data and information integration across various city domains, stakeholders and scales are needed, allowing data to be processed from different sectors, scales and formats simultaneously.

5.2. Case studies

Several points can be deduced from the case studies analysis in this research. One point mentioned in the reviewed literature is the need to improve GIS and BIM in technology matters; this is not only limited to 3D modelling, but also in data type, data flow and synchronization. Another key point arising from the reviewed case studies is the value of citizen engagement and collaboration with public administration in planning and design processes, especially considering the socio-economical attributes of City Digital Twins. Human intelligence should be considered in digital twins of cities, including input data, analysis, processing and decision-making.

Such case studies and applications should be embedded in the governance structure of cities. For example, the Smart Urban Governance framework suggested by Jiang et al. (2020) provides a useful approach to shift case studies beyond technical innovations to socio-technological innovations and adoptions.

5.3. Applied technologies and methods

Human intelligence should be considered in the loop of artificial intelligence (AI), with digital analysis essential for increasing a system’s resilience, robustness and advancement. Our findings show that human intelligence can be used more in loops, and we expect this to receive greater attention in future research. However, it should be noted that human participation may increase the cost of the system, which could be considered a limitation (Nochta et al., 2021).

We also want to mention the development of advanced technologies for real-time and near real-time analysis, such as big data analysis, cloud computing, data mining and the Internet of Things.

Research potential also lies in the integration of technologies such as BIM and GIS. More research needs to be done to answer looming challenges on a semantic data model, data interoperability, and data flow.

Image processing might also proffer itself as a potential analysis method in City Digital Twins. This method could increase our knowledge of the environment, such as mapping traffic flow and human behavior monitoring.

Some challenges and limitations confront the development of these technologies such as a lack of regulations, the speed and accuracy of analysis, and cost. For example, human participation will significantly increase the quality of results but may also increase a system’s cost (Nochta et al., 2021).

5.4. Maturity spectrum

City Digital Twins are not limited to 3D models of cities. This digital replica of the physical asset ranges over various elements of maturity and can culminate in full automation and operation. On the spectrum of digital twin maturity, from element 0 to element 5, our results show that 38% of the reviewed papers were at element 2, and 52% were at element 3 (considering real-time and static data), while no studies fell at full automation level (element 5). To create a mature City Digital Twin (towards elements 4 and 5), it is necessary to use more advanced tools, technologies and methods to foster integration and two-way interaction between humans and machines.

5.5. Research application

Cities are complex systems, and solving urban challenges and problems requires interdisciplinary perspectives and approaches. For this reason, City Digital Twins demand a multi-dimensional and multi-scale approach to provide a holistic view for urban planners and decision-makers. Our review found that most of the papers here have focused on one research field and dimension, such as energy consumption, transportation or city infrastructure as isolated silos. Therefore, there exists an opportunity to take a systems approach to use City Digital Twins to plan future cities.

City Digital Twins-powered planning support systems, such as the Rapid Analytics Interactive Scenario Explorer (RAISE), provide a glimpse of what is about to unfold and give urban planners the ability to drag and drop new infrastructure such as metro stations and then calculate expected increases in property values (value uplift) in near real-time (Pettit et al., 2020). However, the use of such powerful digital twins also raises questions about their institutionalization, ethical use and moving to AI/ML-assisted urban decision-making (Rittenbruch et al., 2022). All of this needs to be carefully considered by future research, development and applications of City Digital Twin platforms.

Lessons from the past regarding such urban technology show it is critical to solve socio-economic challenges if digital twins are to become mainstream and useful for urban planners – take for example the seminar paper by Douglas Lee: Requiem for Large Scale Urban Models (Lee, 1973).

6. Conclusion

This study discussed and evaluated digital twins projects at the city scale and in urban areas. For this multi-dimensional, complex topic, we used bibliometric and content analysis to identify and address challenges while conducting a systematic literature review of the domain. Our bibliometric analysis found that the number of publications on this field has been steadily increasing since 2018. A total of 41 papers were thoroughly reviewed and evaluated. Our study also found that the concept of urban digital twins is not confined to 3D urban models but must serve as a link between human intelligence, inhabitants’ needs, technology and urban studies such as transportation, commerce and infrastructure. Results show that, although City Digital Twins have matured up to element 3, real-time/near real-time, autonomous operations and maintenance have not been achieved for digital twins of cities. We also found that more research with an interdisciplinary viewpoint is needed to address complex urban concerns. According to this study, most City Digital Twins projects focus primarily on one aspect of urban concerns, such as energy use, distribution or transportation. These programs, in fact, lack a multi-dimensional approach to addressing significant urban concerns.

Another area of future research is integrating not just real-time data, but also near-real-time data analytics algorithms to gain fast insights from large data sets collected in digital twins. Data mining, machine learning and AI algorithms are examples of this. Furthermore, human intelligence should be integrated with AI and machine learning to improve the resilience and robustness of digital twins in the event of interruptions or unforeseeable conditions. As a result, more participatory methods are required to promote human collaboration and obtain better results. Furthermore, implementing a learning circle is critical for the system’s resilience, while integrating BIM and GIS technology to aid 3D and 4D data analysis and decision-making is an obvious gap that we identified in this study. The reason for this emphasis is that a more integrated platform, containing as much data and as many attributes as possible of something as complicated and large as a city, would increase the maturity, intelligence, function and sustainability of the digital model. Such a well-functioning model – achieved by the integrated development of different technologies, mainly GIS and BIM – could increase the precision and validity of simulations, analysis and decisions made with the help of these models. Technological advances are not the only way to achieve the highest levels of integration and maturity; theoretical developments should also be provided in future research to support this integration. The technical difficulties of interoperability, data processing speed and storage offer additional study prospects. Finally, there is a need to look into how this technology can be institutionalized with ethical use and a degree of comfort with AL/ML providing decision support.

City Digital Twins technologies show much promise for assisting communities, planners and decision-makers to plan more resilient, sustainable, productive and equitable cities.

Disclosure statement

No potential conflict of interest was reported by the authors.

Data availability statement

Derived data supporting the findings of this study are available from the corresponding author [Sara Shirowzhan] on request.

Supplementary material

Supplemental data for this article can be accessed online at https://doi.org/10.1080/20964471.2022.2160156.

Additional information

Notes on contributors

Homa Masoumi

Homa Masoumi received her MSc degree in Urban planning from Shahid Beheshti University, Tehran, Iran in 2020. Currently, she is an urban planning researcher whose research interests are focus on smart cities, urban development, spatial science, risk analysis, and disaster management.

Sara Shirowzhan

Dr Sara Shirowzhan is the lecturer and the co-convener of the Smart Cities and Infrastructure Cluster in the School of Built Environment (BE) at the University of New South Wales (UNSW), Sydney, Australia. Sara’s areas of research in technologies relevant to the built environment include advanced GIS, nD BIM, sensing technologies, digital twins, and artificial intelligence for applications in smart cities, infrastructure, urban development, and construction. Sara teaches and supervises students on City Analytics, Machine learning, Object detection, GIS, Construction Informatics and BIM relevant topics at undergrad and postgrad levels at UNSW. She is currently an Editorial Board member of the journals of Sustainability (MDPI) and Buildings. Sara completed her PhD in Geomatics Engineering from the School of Civil and Environmental Engineering at UNSW.

Paria Eskandarpour

Paria Eskandarpour is undertaking a PhD at Monash University’s Emerging Technologies Research Lab. Her PhD project investigates Proptech and its consequent inequalities in urban areas from a social, political, and economic perspective. Paria holds a Master of Urban Design and a Bachelor of Urban Development. Her research interests include platform real estate; critical social and political-economic studies of urban developments, smart cities, big data and disruptive technologies in urban planning.

Christopher James Pettit

Prof. Chris Pettit is the Director of the City Futures Research Centre, inaugural professor of Urban Science, and Plus Alliance Fellow at UNSW Sydney. He is currently Chair of the Board of Directors for CUPUM (Computational Urban Planning and Urban Management) and on the International Advisory Board for the “Geo for all” initiative. He is a member of the Planning Institute of Australia’s National Plantech Working Group, the advisory board for the Centre for Data Leadership, the Committee for Sydney’s Smart Cities Taskforce and the NSW Government Expert Advisory Group for Planning Evidence and Insights.

Notes

1. www.scopus.com.

2. www.webofscience.com.

Appendix

Table 1S. A detailed content analysis of 41 reviewed studies on the subject of research.

Topic	country	methods	data	Theory/ practice /review			Case study	Theme		Application
				T	P	R		Main field	Sub field
“A digital twin uses classification system for urban planning & city infrastructure management” (Al-Sehrawy et al., 2021)	United Kingdom	Systematic literature review	Previous research			*	-	Built environment	Urban planning	-
“A Novel HDF-Based Data Compression and Integration Approach to Support BIM-GIS Practical Applications” (Pan et al., 2020)	China	Data reorganization Data transformation Data aggregation	CityGML, pictures, IFC, analytical data		*		Zhong Chun	IT and Technology	Data management	Interoperability, data fusion,
“A Socio-Technical Perspective on Urban Analytics: The Case of City-Scale Digital Twins” (Nochta et al., 2021)	United Kingdom	Workshops and interviews	Data from interviews	*			Cambridge	Built environment	Urban planning	Policy making, evaluation, citizen’s collaboration
“A systematic review of a digital twin city: A new pattern of urban governance toward smart cities” (Deng et al., 2021)	China	Systematic literature review	Previous research			*	-	Institutional	Urban governance	-
“AI-based physical and virtual platform with 5-layered architecture for sustainable smart energy city development” (Park et al., 2019)	South Korea	5-layer architecture	Land-use data, real-time data		*		Korea	Infrastructure	Electricity	Visual platform
“An animated spatial time machine in co-creation: Reconstructing history using gamification integrated into 3d city modelling, 4d web and transmedia storytelling” (Matthys et al., 2021)	Belgium	Workshops Co-Creation Analyzing the Lost Historical Architecture and City Environment Modelling Storytelling	Text (from workshops), documents, Open 3D Data		*		Ghent	IT and Technology	Gaming and animation	Visualization, interoperability
“Application Study in the Field of Solutions for the Monitoring Distribution Transformers of the Overhead Power Grid” (Chaves et al., 2021)	Brazil	Monitoring system for overhead transformers	Numeric data		*		Sao Paulo, Brazil	Energy management	Energy consumption and efficiency	Monitoring, real-time analysis, network dynamic
“An ISO/IEEE 11073 Standardized Digital Twin Framework for Health and Well-Being in Smart Cities” (Laamarti et al., 2020)	Canada	Data analytics on the cloud	Sensor data		*		-	Public health	Health and well-being	Monitoring and analysis
“Building and exploiting a Digital Twin for the management of drinking water distribution networks” (Conejos Fuertes et al., 2020)	Spain	Hydraulic model Machine Learning and A.I. algorithms Simulate hypothetical scenarios and optimization tools	Sensor data, geographical data		*		Valencia, Spain	Infrastructure	Water system	Operation, management, network dynamics
“Cybersecurity for digital twins in the built environment: Current research and future directions” (Alshammari et al., 2021)	United Kingdom	Systematic literature review	Previous research			*	-	I.T. and Technology	Cybersecurity	-
“Collaborative city digital twin for the COVID-19 pandemic: A federated learning solution” (J. J. Pang et al., 2021)	China	Federated Learning	Spatial and real-time data		*		-	Public health	Crisis management	Data processing, citizen’s collaboration
“Circular digital built environment: An emerging framework” (Çetin et al., 2021)	Netherlands	Developing Framework Workshop with Expert Literature review Map digital technologies framework.	Previous research, interview		*		-	Business and economy	Economy	Participation of stakeholders, decision-making
“Data-driven methodology to support long-lasting logistics and decision making for urban last-mile operations” (Gutierrez-Franco et al., 2021)	United States	Forecasting techniques, clustering, data mining, probability distributions Quantitative Modeling Simulation and Experiments Deep reinforcement learning	Historical data, spatial data		*		Bogota	Infrastructure	Urban transportation system	Decision-making, integration, monitoring
“Detailed Streetspace Modelling for Multiple Applications: Discussions on the Proposed CityGML 3.0 Transportation Model” (Beil et al., 2020)	Germany	CityGML 3.0 Transportation Model	Open data, geospatial data		*		New York	Infrastructure	Urban transportation system	Software, data fusion, integration
“Designing an Energy-Resilient Neighbourhood Using an Urban Building Energy Model” (Buckley et al., 2021)	Ireland	Urban Building energy models (UBEMs) Urban Modelling Interface (UMI)	Archetypes, GIS database		*		Dublin	Energy management	Energy consumption and efficiency	Simulation, integration, data fusion
“Developing a Digital Twin at Building and City Levels: Case Study of West Cambridge Campus” (Q. Lu et al., 2020)	United Kingdom	In-depth analysis RVT model	BIM object, historical data, environment and energy data		*		West Cambridge Campus	Built environment	Urban planning	Monitoring, data fusion
“Digital Twin Conceptual Model within the Context of Internet of Things” (Al-Ali et al., 2020)	United Arab Emirates	End-to-end conceptual model	Sensor data, Historical data, real-time data		*		-	Infrastructure	Urban transportation system	Integration, analyzing, interoperability, designing a software
“Digital Twins From Smart Manufacturing to Smart Cities: A Survey” (Mylonas et al., 2021)	Greece	Systematic literature review	Previous research			*	-	Built environment	Urban planning	-
“Digital Twins: A Critical Discussion on Their Potential for Supporting Policy-Making and Planning in Urban Logistics” (Marcucci et al., 2020)	Italy		Previous research	*			-	Infrastructure	Urban transportation system	Decision-making, Behavior modeling
“Efficient visualization of large-scale oblique photogrammetry models in unreal engine” (Huo et al., 2021)	China	Data organization method	Geometric attributes, OSGB file		*		Changsha, China	Built environment	Photogrammetry and remote sensing	Visualization, data management, interoperability, real-time analysis
“Experiences from city-scale simulation of thermal grids” (Simonsson et al., 2021)	Sweden	Co-Simulation Physical and Data-Driven Models	Geographical data, piping and instrumentation, historical data		*		Sweden	Energy management	Energy consumption and efficiency	Simulation, monitoring, interoperability
“Future (post-COVID) digital, smart and sustainable cities in the wake of 6G: Digital twins, immersive realities and new urban economies” (Allam & Jones, 2021)	France	Critique paper	-	*			-	Business and economy	Economy	Integration
“GIS-enabled digital twin system for sustainable evaluation of carbon emissions: A case study of Jeonju city, south Korea” (Park & Yang, 2020a)	South Korea	Mapping Machine learning algorithms Spatial statistics basis function (RBF) kernel algorithm Backpropagation neural network (BPNN) Hot spot analysis	3D building data, spatial data, energy, waste and vehicle data		*		Jeonju City, South Korea	Energy management	Energy consumption and efficiency	Integration, monitoring, interoperability
“Improving Model Predictions-Integration of Real-Time Sensor Data into a Running Simulation of an Agent-Based Model” (Lenfers et al., 2021)	Germany	Real-time data correction through agent-based methods and modeling Research and Simulation with Multi-Agent	Real-time data, sensors data, R.S. data		*		Hamburg, Germany	Infrastructure	Urban transportation system	Simulation and decision-making, real-time analysis
“Integrated decision support system for pluvial flood-resilient spatial planning in urban areas” (Truu et al., 2021)	Estonia	A dynamic combination of 1D hydraulic modelling Climate scenario making	Geographical data, cadastral, land cover and digital elevation maps, weather data, city plans		*		Haapsalu (Estonia) and Söderhamn (Sweden)	Environment	Natural disaster management	Data management, integration, designing a software
“Integration of an energy management tool and digital twin for coordination and control of multi-vector smart energy systems” (O’Dwyer et al., 2020)	United Kingdom	Integration of the sustainable energy management and system simulation (framework/model/simulation)	Temperature data, solar radiation, building simulation metrics, historical data		*		Greenwich, London	Energy management	Energy consumption and efficiency	Simulation, integration,
“Neighbourhood modelling for urban sustainability assessment” (Orozco-Messana et al., 2021)	Spain	LCA analysis (life cycle)	Historical data, cadastral information, orthophoto, ownership data, height, buildings characteristics		*		Benicalap in Valencia (Spain).	Environment	Natural disaster management	Policy evaluation, simulation
“Model-driven engineering city spaces via bidirectional model transformations” (Visconti et al., 2021)	Austria	Bidirectional transformations Support model-based development Synchronizing among city spatial models and graph-based analyzable models	CityGML data		*		Remscheid (Germany), Flat Iron (N.Y., USA), Vienna (Austria)	I.T. and Technology	Computer science	Interoperability
“Potential Energy, Demand, Emissions, and Cost Savings Distributions for Buildings in a Utility’s Service Area” (Bass et al., 2021)	United States	Automatic Building detection Energy Model creation (AutoBEM) Simulation	Satellite, LiDAR, tax-assessor data, energy consumption, land-use layer		*		Chattanooga, Tennessee	Energy management	Energy consumption and efficiency	Simulation, integration, decision-making
“Real-time control of urban water cycle under cyber-physical systems framework. Water” (Sun et al., 2020)	Spain	Simulating Virtual reality through MATLAB Supervising Management Framework of Multilayer CPS-based Optimizing algorithm of Integrated MPC-Based	Sensor data		*		Barcelona	Infrastructure	Water system	Virtual reality, integration of different fields, analyzing and monitoring, Network dynamics
“Real-time traffic quantization using a mini edge artificial intelligence platform” (Broekman et al., 2021)	South Africa	Quantitative proof of concept Real-time traffic analysis platform	Video data		*		Pretoria, South Africa	Infrastructure	Urban transportation system	Real-time analysis and platform, Interoperability
“Smart City 4.0 from the perspective of open innovation” (Yun & Lee, 2019)	South Korea	Systematic literature review	Previous research			*	-	Business and economy	Business	-
“Smart city for sustainable urban freight logistics” (Pan et al., 2021)	France	Systematic literature review Bibliometric analysis	Previous research			*	-	Infrastructure	Urban transportation system	-
“Smart cities in the time of climate change and Covid-19 need digital twins” (Dignan, 2020)	United Kingdom	-	-	*			-	Environment	Natural disaster management	Behavior modeling, simulation,
“Smart Cities with Digital Twin Systems for Disaster Management” (Ford & Wolf, 2020)	United States	-	ICT sensor		*		-	Environment	Natural disaster management	Integration, simulation, monitoring, decision making
“The Digital Twin of the City of Zurich for Urban Planning” (Schrotter & Hurzeler, 2020)	Switzerland	Report GIS	Official documents, Sensors, Building information, Land administration, 3D data, LIDAR images		*		Zurich	Built environment	Urban planning	Visualization, citizen engagement, policy evaluation, integration
“Smart-guided pedestrian emergency evacuation in slender-shape infrastructure with digital twin simulations” (Han et al., 2020)	China	CA-based model data-driven approaches through machine learning	Behavioral dynamics, sensor data		*		-	Population movements	Crisis management	Behavior modeling, simulation, monitoring
“Synthesizing digital twin travellers: Individual travel demand from aggregated mobile phone data” (Anda, Medina, et al., 2021)	Singapore	Markov models framework designing clustering strategy	Mobile phone data		*		-	Infrastructure	Urban transportation system	Data fusion, tracking, integration
“The Use of a Data-Driven Digital Twin of a Smart City: A Case Study of angstrom Alesund, Norway” (Major et al., 2021)	Norway	Framework of data-driven digital twin	GIS data, BIM, demographic data, and (IoT) data		*		Ålesund, Norway	I.T. and Technology	Data management	Citizen engagement, localized information systems, interoperability, visualization
“Urban development with dynamic digital twins in Helsinki city” (Hämäläinen, 2021)	Finland	A qualitative exploratory research strategy Semi-structured interviews content analysis	Interviews from smart city practitioners		*		Helsinki and Pietarsaari	Built environment	Urban planning	Integration, interoperability, and participation of stakeholders
“Urban Digital Twins for Smart Cities and Citizens: The Case Study of Herrenberg, Germany” (Dembski et al., 2020)	Netherlands	Mixed method approach 3D modeling Mathematical models of street network Urban mobility simulation Simulation of Wind flow Simulation patterns of people’s movement	Geographic data, geo-referenced space syntax data, emission data, traffic simulation, weather and climate, historic empirical, real-life data, sensor data, social data, images		*		Herrenberg, Germany	Built environment	Urban planning	Simulation, data fusion, Behavior modeling, integration, citizens’ engagement