Georeferencing Building Information Models for BIM/GIS Integration: A Review of Methods and Tools

Abstract

With the rise of urban digital twins and smart cities, the integration of building information modeling (BIM) and geospatial information systems (GISs) have captured the interest of researchers. Although significant advancements have been achieved in this field, challenges persist in the georeferencing of BIM models, which is one of the fundamental challenges in integrating BIM and GIS models. These challenges stem from dissimilarities between the BIM and GIS domains, including different georeferencing definitions, different coordinate systems utilization, and a lack of correspondence between the engineering system of BIM and the project’s geographical location. This review critically examines the significance of georeferencing within this integration, outlines and compares various methods for georeferencing BIM data in detail, and surveys existing software tools that facilitate this process. The findings underscore the need for increased attention to georeferencing issues from both domains, aiming to enhance the seamless integration of BIM and GIS.

1. Introduction

Building information modeling/Geographic information system (BIM/GIS) integration is gaining popularity due to its diverse applications and advantages within architecture, engineering, and construction (AEC) and geospatial domains. Over the past decade, researchers have firmly established the essential role of BIM/GIS integration in shaping smart city designs and developments. Although significant progress has been achieved in various aspects during past years, some challenges and complexities persist in achieving the seamless integration of BIM and GISs. The primary impetus lies in the disparities between the BIM and GIS domains. While the immediate scopes of BIM and GIS differ, their integration can benefit from various applications. The georeferencing concept is one of the dissimilarities between the BIM and GIS domains, making the integration process difficult [1,2,3]. GIS utilizes a coordinate reference system to integrate heterogeneous spatial data such as various types of raster and vector data or external data sources. On the other hand, BIM designs the management information of all phases in its local reference coordinate system, which is different from GIS. Consequently, to take full advantage of integrated 3D data of BIM in GIS environments, BIM data need to be appropriately georeferenced. Moreover, the seamless transformation of coordinate systems between BIM and GIS contributes to enhanced interoperability, which presents a pivotal facet of BIM/GIS integration [4,5,6,7,8,9].

With the advent of urban digital twins and smart cities, the integration of BIM and GISs has gained more attention from researchers. While notable progress has been made, challenges persist in fully addressing georeferencing issues within BIM models. These challenges primarily stem from the inherent differences between the BIM and GIS domains. First, the concept of georeferencing is entirely different in BIM and GISs. The geospatial sector, having extensive experience, has developed varying standards, algorithms, and technologies for georeferencing, whereas these aspects differ within the BIM domain [10]. Although various methods and tools exist for georeferencing geospatial data, BIM software tools and users are relatively new to this topic [11]. Second, BIM designers or software developers do not store and use georeferencing information, which can be stored in the Industry Foundation Classes (IFC) data format. Also, accessing the georeferencing attributes in developed tools is not perfectly optimized [12]. Third, misinterpretation of the georeferencing capabilities inherent in existing data formats, such as IFC, contributes to the unresolved nature of the georeferencing challenges [4]. Fourth, disparities in schema export, a necessity for maintaining interpretability amid various BIM data formats (e.g., IFC and Revit), often result in the omission of georeferencing information, occasionally due to software limitations during the exporting process [6]. Finally, BIM elements have been created in a local Cartesian coordinate system (LCCS), an engineering system, while the GIS users need a geographic coordinate system (GCS) (global scene) and projected coordinate system (PCS) (local scene). Another dissimilarity is the difference between their coordinate systems and the lack of relations to convert them easily [13,14]. The main reason for the BIM designers to be interested in using close-to-origin Cartesian coordinates is that BIM software works better in this system with small and more manageable numbers rather than the precision of floating point numbers [15]. These factors underscore the complexity of addressing georeferencing challenges within BIM/GIS integration.

Although many studies address general aspects of BIM/GIS integration, such as geometric and semantic transformations, the topic of georeferencing BIM models has received much less focused and systematic attention [1,2,7,9,11,12]. In georeferencing BIM models, comprehensive review papers are crucial in synthesizing existing knowledge, identifying gaps, and offering insights for future research directions. However, it has come to our attention that, to date, there has been a notable absence of a dedicated review paper addressing georeferencing BIM models in the existing literature. This absence presents a significant knowledge gap that needs to be addressed.

Considering these complexities and research gaps, this paper aims to address the challenges of georeferencing within BIM/GIS integration through a comprehensive systematic literature review. Reviewing both academic and industry developments and highlighting the need for greater collaboration between the two will enhance understanding and provide valuable insights to guide future research and practices. Moreover, highlighting the current state of georeferencing issues, identifying current limitations in standards, and proposing directions for improvement can provide industries and companies with helpful information and capabilities to facilitate integration into GIS domains in future works.

The rest of the paper is organized as follows. Section 2 outlines the review methodology. Section 3 employs bibliometrics to quantify existing research in this domain. Section 4 delves into the review results based on defined objectives. Section 5 discusses our findings, and lastly, Section 6 concludes the paper regarding the pivotal role of georeferencing within the broader BIM/GIS integration landscape.

2. Review Methodology

Various review methods are available in research, tailored to different needs and goals. This study employs the systematic literature review (SLR) approach, recognized for its comprehensive nature compared with other review types. According to [16], conducting an SLR involves a sequence of steps: (1) defining research objectives; (2) selecting a suitable literature database; (3) formulating precise search queries; (4) establishing criteria for selecting relevant papers; (5) analyzing the included studies; and (6) presenting the findings. Subsequent sections will delve further into these steps, providing a comprehensive understanding.

2.1. Research Objectives

This study aims to explore the concept of georeferencing within the context of BIM/GIS integration from multiple dimensions. It first seeks to establish a foundational understanding of georeferencing by examining its concepts in both GIS and BIM, highlighting key differences and synergies. Furthermore, this study explores how various types of BIM systems approach georeferencing, identifying the specific challenges and best practices associated with spatial alignment. It also critically assesses the georeferencing capabilities of the IFC data standards, evaluating their strengths and limitations in supporting effective BIM/GIS integration. Additionally, the research examines the collaboration between industry and academia, shedding light on how partnerships in this area can drive advancements in georeferencing techniques and methodologies. Lastly, the study reviews and compares current software solutions for georeferencing BIM models, providing insights into their functionality and effectiveness in real-world applications. Through this multifaceted exploration, the research aims to uncover emerging trends, offer practical recommendations, and contribute to the future development of georeferencing within the BIM/GIS integration framework. This exploration relies on an SLR conducted using the PRISMA 2020 protocol as a guiding search approach.

2.2. Eligibility Criteria

A straightforward approach to streamlining the SLR study involves establishing effective paper selection criteria. With the above research objectives, eligibility criteria can be set to identify the most pertinent papers for the SLR. Given the novelty and recent trend of the topic, a specific time limit is not imposed. The selection process employs the following criteria:

• Articles must be in English and published in journals or conference proceedings.
• Articles should distinctly address at least one of the research objectives.
• The articles should focus on georeferencing BIM data rather than utilizing pre-existing georeferenced data.

2.3. Identification of Studies

For the literature study, we have focused on two critical databases, Web of Science and Scopus, which were chosen as the primary sources for this research. After selecting databases, the essential part of the research is the query term used for the search engines. The topic of this study is two main words: georeferencing and BIM. The point is that many articles used the whole term for BIM (“Building Information Model/Modelling”). For the georeferencing term, it is noted that the following words have been used: “geo-referencing” and “georeferencing”. After investigating all of the possibilities in different words and phrases, the following query terms have been finalized:

• Web of Science: (Topic (BIM OR “building information modelling” OR “building information model”) AND Topic (“geo-referencing” OR “georeferencing”)).
• Scopus: (TITLE-ABS-KEY (BIM OR “building information modelling” OR “building information model”) AND TITLE-ABS-KEY (“geo-referencing” OR “georeferencing”)).

The first query included 44 papers; the second included 52 papers up to December 2024. Also, this review has been enriched by including gray literature to bring sufficient depth to our review and analysis. In addition to peer-reviewed academic literature, this review incorporates a range of grey literature to provide a broader perspective on georeferencing within BIM/GIS integration. This includes sources such as websites, blogs, web articles, and YouTube videos, which offer practical insights, emerging trends, and industry-driven perspectives that may not yet be captured in formal academic publications. Including these sources ensures a more comprehensive understanding of the current knowledge and practice in this evolving field.

2.4. Article Screening

In the selection process for the final papers, the following steps have been considered, and the results have been summarized in Figure 1:

• Step 1: Retrieve documents with the query term from the databases.
• Step 2: The journal and conference papers in English have been selected in this step.
• Step 3: The duplicate articles from two databases have been removed.
• Step 4: After skimming the paper’s abstract, articles that fit into the scope of study have been selected for the final papers based on inclusion criteria.
• Step 5: Full-text reading has been conducted in this step to select the final papers. The selection is based on the inclusion criteria discussed in the previous section.

3. Bibliometric Analysis (Meta-Analysis)

3.1. Time Series Analysis

At the beginning of any research, it is essential to determine if the research field is of interest in academia and what the development tendency is. The bibliometric analysis reveals this information based on the article’s publication year. Figure 2 shows the number of publications related to georeferencing BIM models between 2016 and 2023. Until 2018, the topic was not so dominant, and only two papers mentioned the georeferencing of BIM models in 2007. The main reason is that georeferencing of BIM models is one of the steps in the BIM/GIS integration process. It is reasonable that during the initial years of research on BIM/GIS integration, most studies focused on integrating BIM and GIS from geometry and semantic aspects. After making solid progress in the main concepts of integration, the georeferencing of BIM models is considered another issue that needs to be addressed by the studies. Although the overall number of publications has not been significant in the last decade, the figure shows a rising tendency that can be considered proof of a growing interest in georeferencing BIM models that leads to an increasing amount of research in this field. This allows us to conclude that there is growing academic interest in this field of research. As a result, the number of studies has experienced a significant increase after 2017. However, the number of studies on georeferencing BIM models is still low compared with similar studies in the BIM field. This proves the lack of attention paid to this field of research in the BIM literature. The primary trend shows that this topic is prevalent nowadays, which indicates that future research in this field is full of opportunities and challenges.

3.2. Journal and Conference Analysis

Another essential aspect that researchers find interesting is where research papers about the georeferencing of BIM models are being published. The data we have includes a total of 42 documents, and they fall into three main categories: 22 papers in scientific journals, 18 presented at conferences, and 2 published as books. Among the journals, the Automation in Construction and ISPRS International Journal of Geo-Information journals stand out with the most papers, having published seven and three articles on this topic. A complete list of journals can be found in

Table 1. Distribution of the selected papers among different journals.

Source	Number of Documents
Automation in Construction	7
ISPRS International Journal of Geo-Information	3
Transactions in GIS	1
Journal of Information Technology in Construction	1
Computers, Environment, and Urban Systems	1
Applied Sciences	1
Construction Management	1
Remote Sensing	1
International Journal of Digital Earth	1
Industry 4.0 for the Built Environment	1
Journal of the Chinese Institute of Engineers	1
Urban Science	1
Journal of Geodesy, Cartography and Cadastre	1
Frontiers in Built Environment	1

. On the conference front, the “ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences” leads the way with 11 articles.

Table 2. Distribution of the selected papers among different conferences.

Source	Number of Documents
ISPRS Annals of the Photogrammetry, Remote Sensing, and Spatial Information Sciences	11
IABMAS Conference 2018	1
2018 Baltic Geodetic Congress (BGC Geomatics)	1
International Conference on Innovative Production and Construction (IPC 2017)	1
17th International Conference on Computing in Civil and Building Engineering	1
24th CIB-W78 Conference: Bringing ICT Knowledge to Work	1
ASCE International Conference on Computing in Civil Engineering 2019	1
The 54th International Conference of the Architectural Science Association (ANZAScA) 2020	1

also lists conferences and the number of documents published in each.

3.3. Keyword Analysis

The most frequently occurring keywords were analyzed to determine the main focuses of investigations in papers related to georeferencing BIM models. This analysis gives insights to answer the research questions regarding the existing methods for georeferencing BIM models and involved data formats, projects, and tools. The analysis is based on the number of keywords in the papers. The co-occurrence analysis helps convey the main contents of the documents and the range of investigated areas in a field. It also provides a picture of a field and development trends in the research area. This type of analysis is notable in bibliometric studies, is based on the results of the existing situation, and allows the current scope of interests to be determined.

By investigating the keywords of selected publications, it has been realized that georeferencing is one of the main aspects of the BIM/GIS integration process. As expected, the terms “Geographical Information System” and “Building Information Models” have been used frequently in the keywords of studies. Interestingly, most studies concentrated on the IFC data rather than the Revit format. The possibility of manual georeferencing of Revit models in Autodesk Revit software can be considered as a reason for the researchers to focus on how the georeferencing information can be stored and utilized in the IFC data format, which is an open standard format and can be a significant step to achieving interoperability between BIM and GIS. Also, a noticeable number of publications were related to the GeoBIM Benchmark 2019 [17,18,19]. Figure 3 illustrates the word cloud of keywords created by the wordclouds.com website (accessed on 1 January 2025).

Using VOSviewer software (version 1.6.20), the co-occurrence map of all keywords with a frequency higher than two has been produced. The results of VOSviewer indicate that 86 keywords have been used in selected articles. The keywords such as “Building Information Modeling”, “Industry Foundation Classes”, and “Geospatial Information System” have the highest number of occurrences at 34, 31, and 14 times, respectively. Some of these connections are stronger due to the similarities between keywords; for example, there is no difference between “Building Information Modelling”, “Building Information Modeling”, and “BIM”, so as a result, we merged similar terms together. Figure 4 illustrates the co-occurrence map of keywords.

3.4. Country and Organization Analysis

The data in Figure 5 provide a snapshot of the geographic distribution of studies covered in the included research, comprising 42 papers. Among the countries featured, Australia and Germany emerge as the leads with the highest number of papers, accounting for approximately 43% of the total, followed closely by the Netherlands, contributing about 19%. Italy, Canada, Sweden, and China comprise the remaining countries, constituting 19% of the research output. These statistics highlight an apparent concentration of research activity in a select group of countries, with Australia, the Netherlands, and Germany being at the forefront. The main reason for the high number of publications in the Netherlands and Germany is the GeoBIM Benchmark project, which was a Scientific initiative funded by the ISPRS (ISPRS Scientific Initiative 2019) and the European Spatial Data Research Association (EuroSDR). Many studies have been published regarding the results of this project, as its second task was about options for georeferencing IFC data. Further analysis of the topics and methodologies covered within these papers may shed light on the reasons behind this distribution and its implications for the field.

Table 3. The most active organizations with two or more articles.

Organization	Number of Documents
Delft University of Technology (The Netherlands)	8
Curtin University (Australia)	5
Technical University of Munich (Germany)	3
Bauhaus-Universitat Weimar (Germany)	2
University of New South Wales (UNSW) (Australia)	2
University of Melbourne (Australia)	2
Royal Institute of Technology (Sweden)	2

depicts the organizations that have more than two articles related to the georeferencing of BIM models. Table 3 shows that the Delft University of Technology has the highest number of documents (8). In second and third place are the following organizations with five and three documents: Curtin University and the Technical University of Munich. The following organizations had two papers: Bauhaus-Universitat Weimar Royal Institute of Technology, the University of New South Wales (UNSW), and the University of Melbourne. Intriguingly, only two organizations have published more than two articles in Germany. This suggests that a broader range of organizations within Germany has contributed to the research on this topic, especially considering that seven documents originate from Germany.

4. Results

The georeferencing of BIM has become increasingly important in ensuring spatial accuracy and effective integration with GIS. Georeferencing enables BIM models to be placed within a global coordinate system, facilitating their alignment with other geospatial data for urban planning, infrastructure development, and construction management. This section explores the critical concepts of georeferencing in both GIS and BIM, highlighting the different BIM types and their georeferencing needs. The capabilities of the IFC standard for supporting georeferencing are examined, along with the ongoing collaboration between industry and academia, to address the challenges of integrating BIM with geospatial systems. Furthermore, an overview of software tools for georeferencing BIM models is provided.

4.1. Georeferencing Concepts in GIS and BIM

Geospatial technology has precise definitions of georeferencing, offering diverse methods to address this matter within its domain. In its simplest sense, assigning geodetic coordinates to geographical or geospatial entities encapsulates the essence of georeferencing. The complexities of georeferencing within the context of BIM/GIS integration predominantly stem from the BIM domain. Unlike the well-established and consistent definition of georeferencing in GIS, the BIM landscape lacks a unified understanding. Consequently, diverse interpretations have emerged within the BIM community, often diverging from the established georeferencing concept in GIS. This divergence has sometimes led to misunderstandings and misguided attempts to address this challenge within the BIM framework [4]. Based on the geospatial context, all of the features of a GIS geodatabase must have accurate geographic location and related information. Georeferencing is the ability to accurately locate the GIS features in geographic space [20]. Georeferencing must be unique and persistent through time. The former means there should be only one location associated with a given georeferenced object, which is shared among everyone who works with that information. The latter emphasizes the consistency of the georeferencing information of data. Changing the georeferencing information frequently can be confusing and expensive for updating all related records [21]. A comparison of georeferencing definitions across various studies is presented in

Table 4. Georeferencing definitions in studies from the BIM and GIS domains.

Domain	Study	Georeferencing Definition
GIS	[22]	“Georeferencing is aligning geographic data to a known coordinate system so it can be viewed, queried, and analyzed with other geographic data”
	[23]	Formal georeferencing: related to coordinate reference system, which refers to the exact location of geographical objects
	[23]	Informal georeferencing: colloquial references to geographical objects such as name (approximately)
	[6]	Linking the coordinates of the BIM/IFC file with its real-world coordinates
	[13]	Obtaining geographic coordinates
	[5,19]	Transferring the object’s geometric coordinate to its right geographical location
	[24]	A process to relate the internal coordinate system (or local coordinate system) of a digital map or aerial photo to a ground system of geodetic coordinates
	[25]	Geo-positioning an object using a correspondence model derived from a set of ground control points (GCPs) for which both ground and image coordinates are known, where a correspondence model means the functional relationship between ground and image coordinates
BIM	[26]	Assign geographic coordinates to an image or misplaced vector data to display them correctly
	[10]	A coordinate transformation between two systems, where one of them is related to geodetic datum
	[26]	Assigning the right reference system to BIM data
	[27]	The process of placing an asset on the surface of the Earth
	[7]	The process of associating a map or raster image with a spatial location
	[28]	The mapping from the Cartesian coordinates of the model to a georeferenced coordinate reference system

, clarifying subtle variations in interpretation. Although these definitions are similar in their purpose, they may lead to different research efforts to address georeferencing issues [4].

While GIS offers a well-established framework for georeferencing, ensuring accuracy and consistency over time, the BIM domain faces challenges due to its varying interpretations and less formalized approach. These discrepancies underscore the need for a more unified understanding of georeferencing in BIM, especially when integrating with geospatial systems. Bridging this gap will be essential for achieving seamless collaboration between GISs and BIM, paving the way for improved project coordination and data interoperability across both fields. By reviewing the various definitions of georeferencing from both the GIS and BIM domains, it can be concluded that georeferencing is a process of transforming the local coordinate system of BIM models to a geographic coordinate reference system or assigning coordinates of a geographic reference system to BIM if no local coordinates exist. This ensures that the data can be correctly aligned, visualized, and analyzed in relation to real-world locations. Effective georeferencing is unique, consistent, and persistent over time, preventing misalignment issues and ensuring interoperability across different software platforms.

4.2. BIM Types and Georeferencing

In the field of surveying and laser scanning, terms such as as-designed, as-built, as-constructed, and as-is are sometimes used interchangeably, which can lead to confusion. However, each term has a specific meaning in the context of BIM and construction, and it is essential to distinguish them clearly. There is a connection between the various types of BIM models and the stage of the defined project. At the first step of the project, before starting the construction, the designer teams of BIM models will design the “as-designed BIM”, which includes initial design and specifications before construction begins. After progress in the project, based on pre-defined timelines, “as-constructed BIM” models will be generated to track the building state during construction, showing deviations from the design concept. The main benefit of this BIM model is resolving the conflicts of design and construction situations during the project. As-built BIM documents the completed state of the building post-construction, including changes made during the process. As-is BIM represents the current state of an existing building, especially when the original design documentation is unavailable [29]. Figure 6 shows the definition of various types of BIM models and the traditional computer-aided design (CAD) and BIM examples for the provided definitions. Figure 7 illustrates the connection between the different types of BIM models.

While many BIM models used for internal building management or design purposes may not initially include georeferencing, georeferenced BIM models are increasingly important for applications that involve integration with GIS or urban planning. The growing trend of creating georeferenced BIM models from the start helps bridge the gap between the BIM and GIS domains. Some methods enable the creation of georeferenced BIM models from the very beginning. For example, using advanced surveying techniques such as GNSS or LiDAR during data capture can directly embed spatial coordinates into the BIM model, facilitating seamless integration with GIS data. LiDAR and GPS technologies have been used to create georeferenced BIM models in the following areas: developing green infrastructure BIM models [30], developing a historical building information modeling (HBIM) system for rehabilitation of the site [31], automated unmanned aircraft system (UAS) based condition assessment of bridges [32], and image-based deformation monitoring of bridges [33]. Utilizing advanced surveying techniques early in modeling, such as LiDAR or GPS, addresses many georeferencing challenges. These methods ensure initial spatial accuracy, effectively bridging BIM and GIS domains. Their application in various sectors, as cited, underscores the value of integrating geospatial data from the onset.

Scan-to-BIM is a process that converts laser-scanned point clouds into BIM, which are then used by development, design, and construction teams. The process starts with a laser scanner, which can be mounted on a drone, a stationary tripod, or carried by a person around the site. The scanner captures points in space corresponding to the site’s geometry, including walls, doors, and other features, by recording their x-, y-, and z-coordinates. The final product is a point cloud enriched with metadata, which teams can use for project planning and revisions [34]. The BIM reconstruction (as-is/as-built BIM) begins with acquiring a point cloud using laser scanning technology. Next, the following steps have to be conducted to generate the final as-is/as-built BIM, sometimes known as scan-to-BIM [35]: data acquisition and preprocessing, segmentation and classification, and 3D modelling and BIM integration.

While traditional scan-to-BIM workflows often result in non-georeferenced models, advancements in technology now allow for direct georeferencing during data acquisition. Using mobile mapping systems, UAV-based LiDAR, and topographic scanners equipped with GNSS, point clouds can be accurately georeferenced in real time, minimizing the need for manual alignment in later stages. Sometimes the BIM users will transfer the georeferenced points clouds into the local coordinate systems. As a result, it is essential to preserve the georeferenced point clouds or at least a metadata document, which can be used later to perform georeferencing. Georeferencing point clouds are critical to accurately positioning and contextualizing data within a known coordinate system. Here are the standard methods used for georeferencing point clouds [36]:

• Direct georeferencing in the field: This method georeferences the point cloud at the time of capture using high-precision sensors. Some topographic scanners, mobile LiDAR systems, and unmanned aerial vehicle-based LiDAR (UAV-based LiDAR) equipped with global navigation satellite system (GNSS) and inertial measurement unit (IMU) can record accurate coordinates and orientation data in real time. Because the georeferencing happens during data collection, there is no need for additional alignment steps later. This makes the process faster and ensures spatial accuracy right from the start, which is especially useful for large-scale mapping, infrastructure surveys, and environmental monitoring.
• GCPs: If the point cloud is not georeferenced, GCPs with known coordinates, such as physical markers or identifiable features, are added. Survey-grade instruments then measure these GCPs, and the point cloud is aligned and georeferenced based on their coordinates.
• Register with existing models or plans: If georeferenced models, plans, or datasets exist for the area, the point cloud can be aligned to them using identifiable features. This alignment can be performed manually or with specialized software, indirectly georeferencing the point cloud based on the known data.
• Registration in a previously georeferenced point cloud: If an existing georeferenced point cloud covers the same area, the new point cloud can be registered to it. Common features in both clouds are used for alignment, allowing the new point cloud to inherit the georeferencing from the already referenced cloud.

BIM community published a report about 3D scans to georeferenced BIM models, conducted from an interview with Stephane Hanich, project manager at Bouygues Bâtiment Île-de-France Public Works. Stephane Hanich mentioned that: “Georeferencing has enabled buildings to be precisely positioned within the campus and vis-à-vis other buildings, networks, and property boundaries. This method offers accuracy and reliability that is very difficult to achieve in other ways. The deliverable is easily integrated into our 3D model, can detect clashes earlier, and can make more rigorous studies. It is an undeniable gain of time. Not to mention that the 3D scan can be viewed on the cloud as a 360° photo in which we move. Whether for meetings in the study phase or for exploitation, it is a great communication tool” [37]. When georeferenced scanning was initially introduced, it was still in its early stages, and only the latest versions of certain software could generate a georeferenced BIM model. However, the technology has since advanced significantly.

While various BIM types such as as-designed, as-constructed, and as-is/as-built models offer valuable insights throughout a project’s lifecycle, their georeferencing is often overlooked. This lack of georeferencing, particularly in scan-to-BIM workflows, creates challenges in integrating BIM models with the geospatial domain. However, advanced techniques like LiDAR, GPS, and GCPs provide solutions to bridge this gap, enabling the creation of georeferenced BIM models. As geospatial technology continues to advance, the potential for seamless BIM and GIS integration becomes more attainable, enhancing project accuracy and cross-domain collaboration.

4.3. Georeferencing Capabilities of the IFC Standard

The IFC is an international standard (ISO 16739-1:2018) and an open, object-based data model developed by buildingSMART [38]. It promotes interoperability in the AEC industry, particularly within the realm of BIM. IFC’s primary function is to enable the exchange of construction models, primarily encompassing 3D representations of buildings. The International Alliance for Interoperability initially developed this widely recognized open standard for BIM, and it has since been taken over and maintained by buildingSMART to facilitate enhanced information interoperability among stakeholders [39].

As research on BIM/GIS integration has grown, georeferencing BIM data have gained significance in achieving seamless integration between BIM and GIS. Consequently, newer versions of IFC have incorporated certain georeferencing concepts. To delve into the process of georeferencing within the IFC data format, it is essential to understand the concepts of coordinate systems and georeferencing levels in the IFC first. In the following, we comprehensively explain various relevant classes connected to georeferencing, elucidating their relationship to corresponding concepts in the GIS domain.

The IFC utilizes a system of relative placement known as the placement hierarchy. This hierarchy positions an object (child) with a higher class (parent). The placement hierarchy encompasses elements, stories, buildings, sites, and projects [4]. The project entity can include several connected or disconnected Sites. The most superficial level of georeferencing information can be stored in the IfcPostalAddress entity, which has the following attributes to represent the address of the project site: InternalLocation, AddressLine, PostalBox, Town, Region, PostalCode, and Country. IfcPostalAddress can be referenced by IfcSite and IfcBuilding, which have the attributes of SiteAddress and BuidlingAddress, respectively, with the type of IfcPostalAddress. Every Site can store a single geographic reference point in the form of Latitude, Longitude, and Elevation of the World Geodetic System (WGS84) or European Petroleum Survey Group:4326 (EPSG:4326) for absolute placement in exchange with GIS world. The Site’s attributes are RefLatitude, RefLongitude, RefElevation, LandTitleNumber, and SiteAddress. RefElevation is relative to sea level, and it is impossible to define its datum. As outlined in the IFC standard, the geographical reference provided could pertain to either the precise location of the origin of the local placement of the IfcSite or serve as an approximate position intended solely for informational use [19].

One step further, the IfcSpatialStructureElement entity is a general concept of all spatial elements (such as site, building, building story, and space), and IfcLocalPlacement can define their relative placement. For the site, the IfcLocalPlacement can be absolute (i.e., world coordinate system (WCS)), defined by the IfcGeometricRepresentationContext entity. This entity has the following attributes: WorldCoordinateSystem and TrueNorth. The first attribute can be used to offset the project coordinate system from the global point of origin, and the latter attribute defaults to the positive direction of the Y-axis of the WorldCoordinateSystem. Considering IfcSite and IfcGeometricRepresentationContext, it is possible to georeference IFC models, but these values are often set to zero, the wrong location, or rough approximations. This issue is further complicated by the inconsistent definitions of the positive direction for longitude in IFC2x3 and IFC4 [19].

The most complete entity containing georeferencing information is IfcMapConversion, which has been added to IFC version 4. It can be used to transform from an engineering coordinate system (ECS) into a coordinate reference system (CRS); however, it cannot deal with the projection of a map from the geodetic coordinate system. Eastings (E), Northings (N), OrthonogalHeight (O), XAxisAbscissa (XAA), XAxisOrdinate (XAO), and Scale (S) are the attributes of IfcMapConversion. Salheb et al. (2020) developed a methodology to convert City Geography Markup Language (CityGML) to the IFC data format, which is needed to transfer the georeferenced data from global coordinates to local coordinates [40]. They mentioned that the lack of supporting various coordinate systems can be a problem that has been addressed in the latest version of IFC. Georeferencing has two main steps: (1) establishing spatial reference and (2) obtaining coordinate transformation parameters and transforming coordinates. Based on the mentioned attributes, the coordinate transformation can be calculated using the affine transformation [4,11].

There are some studies investigating the capabilities of georeferencing in the IFC data format, which has some problems, including misinterpretation of the IFC geometry structure, the misunderstanding of IFC entities, and the misuse of IFC entities [10,13,19]. They considered the IfcSite as the highest geometry container of the IFC, whereas the IfcProject is the highest container [4]. The main issue of this consideration is that the geographic reference point of IfcSite was mistakenly interpreted as the origin of the engineering system (project-level local coordinate system (LCS)), which leads to neglecting the relative placement of IfcSite and IfcProject. Figure 8 shows the georeferencing entities and their relation.

In 2019, Ref. [10] proposed six levels for storing georeferencing attributes, from level of georeferencing (LoGeoRef) 10 to LoGeoRef 60, based on the entities described above, which has been illustrated in detail in Figure 9. The higher levels are more complex and contain more information. However, higher levels do not include lower-level data. These levels have been defined without focusing on infrastructure [28]. Nonetheless, it is essential to clarify that these levels do not indicate the precision or quality of georeferencing but primarily pertain to how georeferencing data are structured and stored. In certain instances, like that of LoGeoRef30 and LoGeRef40, their accuracy can be identical as they are intended to represent the same values despite variations in their storage methods within the IFC file [15]. As described in the IFC specification, all objects with a spatial context are derived from the class IfcProduct, which has two attributes: ObjectPlacement and Representation. A placement can be absolute, relative, or in a defined grid [15]. The absolute placement is with respect to the WCS of the project, which IfcGeometricRepresentationContext defines. The objects in IFC can be placed relative to their superior objects. They supposed that the most superior object in IFC is IfcSite, which can be placed with respect to the project WCS. However, it cannot directly refer to the geodetic CRS; the IfcCoordinateOperation element can assign IfcSite to the geodetic CRS [41].

In a study by [4], the georeferencing capabilities of the IFC data format were investigated, and the authors concluded that in defining these levels of georeferencing, the IFC entities have been misused because IfcSite is not the highest object container. This involved two specific instances of misuse: (a) utilizing the relative placement attribute (represented by an IfcAxis2Placement3D object) of IfcSite to store its position relative to a CRS instead of the project-level LCS as specified in LoGeoRef30, and (b) employing the WorldCoordinateSystem attribute (indicated by an IfcAxis2Placement3D object) to record the relative placement with respect to a CRS by LoGeoRef40. The consequence of this misapplication of these IFC elements is the generation of extensive coordinate values, which can potentially lead to unforeseen issues when used with BIM tools, as highlighted in the reference. BIM models containing such misapplied elements become incompatible with BIM tools, rendering them unusable for their intended purposes. Figure 10 illustrates the mapping between these proposed levels for storing georeferencing in the IFC data format.

Ref. [15] investigated 57 IFC models derived from real-world practices to scrutinize the storage of georeferencing information within these models. This examination was prompted by the observed misalignment between the standards outlined in the IFC model specifications and the actual georeferencing data encountered in practice, resulting in several noteworthy issues. The analysis encompassed two critical aspects: the location where georeferencing information was stored and the quality of the stored data. The former aspect was assessed based on [10] georeferencing levels, with the observation that this information could be stored multiple times and that there was a corresponding absence of a validation system for ensuring conformance to defined levels. A significant issue was identified in the lack of support for coordinate reference systems within the IFC2x3 version. Concerning the latter aspect, the storage quality was examined regarding North, East, and elevation data, with particular attention being paid to the direction of rotation, all of which were assessed and reported independently.

Table 5. The result of the quality assessment for the IFC model conducted by [12].

Quality of Georeferencing Info.	Number of IFC Models	Notes
No information	1	For only one case, there was no information at all.
Wrong	24	Misleading information, completely wrong. The reason can be related to the default value for the software (mostly Revit) to use the Boston location for the models.
Far approximation	7	Random points in the country where the building was supposed to be built or designed.
Close approximation	2	Random points in the city where the building was supposed to be built or designed. The designer probably provides the information for energy-related assessments but cannot integrate it with the geoinformation.
Accurate	9	Sufficiently accurate information to integrate with the geospatial data.

reports the detailed results of the inspection with respect to georeferencing information.

Ref. [43] transformed the 3D geometric information of CAD objects expressed by the IFC model into the Geography Markup Language (GML) model. The main focus of this study was on transformation from the swept solid models to the Boundary-representation (B-Rep) models, which included transferring coordinates from the local system to the real-world system. Their proposed method extracted the coordinates of vertices of the objects in the local coordinate system and transferred them to a real-world coordinate system by using affine transformation to generate the GML object model. They did not mention how they accessed the target coordinates in the real-world system. Ref. [44] used an instance-based method to generate the mapping rules between IFC and CityGML. One of the major transformations of data from IFC to CityGML was the transformation of the local placement system in IFC to the world coordinate system in CityGML, in which they utilized the method proposed by [43] to overcome it.

Ref. [45] proposed a transformation path for converting IFC data models into CityGML using the Feature Manipulation Engine (FME) software (Version 2016.1) to support urban energy modeling on web visualization applications. One of the major steps in this study was handling the georeferencing of the IFC model during the transformation. The first step was defining a projected coordinate system for the IFC model to bring it from a 3D Cartesian coordinate system (CCS) into a 3D GCS-based model in CityGML. FME provides the CRS_Transformer, LocalCoordinateSystemSetter, Scaler, 3DRotator, and Offsetter transformation to support this translation. However, the important point to consider is the fact that in FME, the IFC files are read with a bounding box determined by the extent of the model, which is always perpendicular to the x- and y-axis of the drawing canvas. As a result, the rotation has to occur before the translation transformation in FME. After that, the model shifted to the correct location.

Ref. [46] investigated the possibility of using IFC models in the geospatial environment to support the fire response management process and indicated the importance of geolocating the IFC models in the geospatial environment correctly. They suggested using the information obtained from the IFC attributes such as IfcSite and IfcGeometricRepresntationContext or utilizing a geospatial object as a template to extract some ground control points. Ref. [26] proposed an economical approach to georeferencing the IFC model of a bridge using affine transformation. The process involved several steps: initially, selecting specific control points to establish connections between the IFC model and its corresponding representation in the GIS realm. Subsequently, the transformation parameters were computed, enabling the rectification of the model through affine transformation. Lastly, the z-values of individual vertices were adjusted utilizing a scaling factor. The critical point to achieve good accuracy depends on the number and location of the control points. Manual intervention and testing only on a simple bridge model are the limitations of this approach that need to be addressed in future works.

Implementing georeferencing at the application level (software) and the exchange level (suitable exchange formats) can lead to achieving interoperability of BIM and GIS. Refs. [47,48] focused on the exchange level, specifically the IFC data format version 4. They presented the current georeferencing capabilities of IFC4 and proposed extensions to address its limitations, including two new entities supporting geographic coordinate reference systems (GCRSs) and rigid transformation of BIM geometries. The current capabilities have been described in this section. The most important requirement of the BIM models is the possibility of using the IFC models without any manual intervention. As a result, they focused only on LoGeoRef50, the only level providing information about the CRS. They developed three solutions to provide complete georeferencing information in the IFC data format as follows:

• CRS.
• No breaking changes: The goal is to achieve the possibility of introducing a geographic CRS, which needs to modify the definition of IfcCoordinateReferenceSystme. GeodeticDatum and VerticalDatum have been moved to IfcProjectedCRS as they are unnecessary in a generic CRS. The new optional attribute WellKnownText has been added to IfcProjectedCRS. Also, a new entity, IfcGeographicCRS, can model the geographic CRS by having the following attributes: GeodeticDatum, PrimeMeridian, and Unit.
• Breaking changes: The WellKnownText (WKT) attribute can be introduced to the abstract IfcCoordinateReferenceSystem entity, and the definition of IfcGeographicCRS remains the same.
• Coordinate operations: With the introduction of new children to the abstract IfcCoordinateReferenceSystme, the usage of IfcMapConversion shall be restricted in the schema only to allow IfcProjectedCRS entities being referenced by the attribute TargetCRS. Thus, the SourceCRS referring to an IfcRepresentationContext uses projected coordinates with elevation.

In the beginning, BIM/GIS integration focused on integrating building models with GIS to support the design and development of smart cities. With significant progress in this field, the users are considering integrating infrastructure BIM data with GIS. Infrastructure BIM data are related to large-scale projects, such as roads, railroads, bridges, etc., which may involve several countries. The most important factors that should be considered in the georeferencing process in this type of project are the curvature of the Earth, aspects of the geodetic CRS, and the distortion implied by them. The curvature of the Earth, irregularities in gravity, and map projection play an essential role in defining the geometric context of the BIM models and calculating coordinates for surveyors of infrastructure projects [47,49]. Although the IFC can embed georeferencing information to model infrastructure projects, multiple sites should be used to control the effect of the curvature of the Earth, while the current IFC models can only support one instance of IfcMapConversion [7].

Infrastructure BIM data are commonly designed in a LCCS, assuming the Earth is flat. As a result, designers consider the constant scale of their projects. On the other hand, the georeferenced digital terrain models (DTMs) to a map projection will be used as a basis data for infrastructure design. The DTM is not a 1:1 representation due to scale variation, which causes a discrepancy between the designed infrastructure BIM and DTM [14]. To address this, IFC 4 has been extended with IFC alignment to support alignment geometries such as roads and railroads, which are used for infrastructure design. Ref. [14] investigated the geographic capabilities of IFC alignment extension. The IFC alignment project team proposed two steps to georeferencing infrastructure BIM models: rotating the engineering system to align the y-axis with cartographic north and translating the origin of the engineering system using the known Easting, Northing, and Height. As the IFC standard supports map projections identifiable by an EPSG code, the scale distortion depends on the geographic location of the project and the well-known map projections available for the area. Countries such as the United Kingdom, which use a single transverse Mercator projection zone, will be more affected by this problem [14].

IFC version 4 supports the geodetic CRS (gCRS) by referring to an identifier of the EPSG database. However, referring to custom gCRSs is impossible, which can be problematic for large infrastructure projects. Jaud et al. (2019) proposed a novel approach to expand the IFC schema with the WKT notation to handle the custom gCRS in the Brenner Base Tunnel (BBT) project [50]. Possibilities of parametric descriptions of any gCRS and acceptance by GIS made WKT the best option to store the custom gCRS information in the new entity IfcWellKnownTextCRS. In addition, Ref. [49] proposed an additional class deriving from IfcCoordinateOperation with the name of IfcBilinearConversion containing an array of tuples to link between the coordinates in the geodetic CRS defined with IfcProjectedCRS and the corresponding coordinates used in the BIM model. The important note is to define a new class IfcGeodeticPoint to separate it from the Cartesian point defined in the schema as IfcCartesianPoint.

4.4. Collaboration of Industry and Academia

The Open Geospatial Consortium (OGC) and buildingSMART International published a paper as an output of their integrated digital built environment joint working group [51]. This report focuses on a variety of challenges and addresses the need to develop better integration at a high level between BIM and GIS standards, IFC, CityGML, and LandInfra. The differences between their CRSs are considered one of the main challenges. CityGML’s use of gCRSs provides benefits such as easier indexing in GIS software, quicker spatial queries, and smoother integration of different datasets. However, a drawback is that changing an object’s global position requires updating every geometric feature. On the other hand, the IFC format use local cartesian CRSs, which must be transformed into a geodetic CRS for integration with other geolocated data, like CityGML models. IFC models often lack precise geolocation, making automated spatial integration challenging. Additionally, differences in how network topologies are represented and inconsistencies in object identification between standards highlight the importance of accurate real-world positioning to infer relationships, identify duplicates, and resolve conflicts between objects [52].

BuildingSMART published several technical reports about user guides for georeferencing in IFC in January 2020 [53]. Initial discussions with the industry identified the lack of knowledge with regard to georeferencing and the strong interest in this subject in the user base in Australia. Their objective was to work with a large site, but they also needed to be able to bring data in from smaller sites. The concept that defines the cadastre or terrain locates built assets and geographic features, etc., and it is represented in IFC2x3 and IFC4 by the IfcSite entity. This is important because some users and members were not clear about whether the data were associated with the IfcProject or IfcSite. Their suggestion was that for every small site within a large project, a land surveyor should be used to establish the two sets of coordinates, which should not be too close together, and the 2D Helmert transformation parameters that will be captured in the map conversion settings should be calculated and its parameters stored in the entity IfcMapConversion in IFC4 or in special property sets in IFC2x3. A strong recommendation was to allow for the storing of the 2D Helmert paired coordinate points, which would allow surveyors to carry out transformations to assess the accuracy of the process.

Table 6. The 2D Helmert transformation and CRS attributes of IfcMapConversion and IfcProjectedCRS entities.

	Attributes	Data Type	Definition
IfcMapConversion	Eastings	IfcLengthMeasure	The translation in X between the two coordinate systems
	Northings	IfcLengthMeasure	The translation in Y between the two coordinate systems
	OrthogonalHeight	IfcLengthMeasure	The translation in Z between the two coordinate systems
	XAxisAbscissa	IfcReal	The X component of the rotation between the two coordinate systems
	XAxisOrdinate	IfcReal	The Y component of the rotation between the two coordinate systems
	Scale	IfcReal	The scale in X, Y between the two coordinate systems
IfcProjectedCRS	Name	IfcLabel	Name by which the coordinate reference system is identified. NOTE: The name shall be taken from the list recognized by the EPSG.
	Description	IfcText	Informal description
	GeodeticDatum	IfcIdentifier	Name by which this datum is identified. The geodetic datum is associated with the CRS and indicates the shape and size of the rotation ellipsoid and this ellipsoid’s connection and orientation to the actual globe/Earth. It needs to be provided if the Name identifier does not unambiguously define the geodetic datum as well
	VerticalDatum	IfcIdentifier	Name by which the vertical datum is identified. The vertical datum is associated with the height axis of the CRS and indicates the reference plane and fundamental point defining the origin of a height system.
	MapProjection	IfcIdentifier	Name by which the map projection is identified.
	MapZone	IfcIdentifier	Name by which the map zone is identified.

shows the 2D Helmert transformation and CRS attributes, which will be stored in IfcMapConversion and IfcProjectedCRS, respectively [54].

A detailed demonstration scenario has been generated by buildingSMART to show how to set up georeferencing in a building or linear infrastructure model. Figure 11 illustrates the process with an explanation.

Dion Moult (architect/software developer) and John Mitchell (co-founder of buildingSMART Australia) proposed three scenarios to address the geolocation in IFC2x3 and IFC4. The main issue in IFC2x3 is the fact that there is no geolocation entity like ProjectedCRS and IfcMapConversion in IFC4. As a result, they proposed defining two new sets (the IfcPropertySet is a container that holds properties within a property tree) called EPset_ProjectedCRS and EPset_MapConversion to store the same attributes. These two sets must be associated with the relevant top-level IfcSite entity. The next issue was the confusion over optional fields in the IfcProjectedCRS. They proposed EPSG:### number in the IfcProjectedCRS. The name field must match a code in the Official EPSG Registry, specifically aligning with the XPath ProjectedCRS/identifier in the GML export. This would allow for retrieval of the corresponding WKT. Additionally, Moult and Mitchell suggest modifying the documentation for optional fields as follows [56]:

• MapProjection: Required if the Name identifier does not clearly define the map projection and the CRS is 3D.
• MapZone: Required if the Name identifier does not clearly define the map zone and the CRS is 3D.
• MapUnit: Required if the Name identifier does not clearly define the map unit and the CRS is 3D.

To resolve ambiguity with fields like VerticalDatum and GeodeticDatum, Moult proposed that their values should match the XPath of VerticalDatum/identifier and GeodeticDatum/identifier, respectively. This approach ensures a single, unambiguous value and aligns with the XPath convention used for ProjectedCRS/identifier. Using identifiers enhances data integrity by avoiding the issues associated with aliases [56].

Dion Moult (21 May 2019) explained the IFC CRS similar to the level of georeferencing that [10] proposed. Moult tried to investigate the mentioned coordinate systems in Revit software and figured out how they work. Revit provides an IfcProject that includes an IfcGeometricRepresentationContext, but it lacks an IfcMapConversion, which is crucial for recording CRS, data, projections, and related information. The IfcProject’s WorldCoordinateSystem is fixed at (0, 0, 0) and does not change regardless of Revit’s origin placements. The RefLatitude and RefLongitude are derived from the Revit project’s “Location” settings and are unrelated to the Survey Point or Project Base Point. Additionally, Revit does not distinguish between buildings, sites, or sub-sites, meaning all contents are placed within a single site and building. This limitation is important because, without a CRS conversion, users often input local CRS coordinates into the Project Base Point and Survey Point. Given that these values are typically large, the IfcSite’s local placement can end up significantly offset in space [57].

The collaboration between industry leaders like buildingSMART and OGC, along with academic contributions, has led to significant advancements in addressing georeferencing challenges within BIM and GIS integration. Through technical reports, standards development, and practical demonstrations, these efforts aim to streamline the use of geospatial data in BIM models and overcome the limitations of current technologies and standards, such as IFC. Continued collaboration will be crucial in developing more robust solutions for georeferencing, enhancing the accuracy of digital models, and fostering greater interoperability between BIM and GIS domains.

4.5. Software for Georeferencing BIM Models

Analyzing the georeferencing BIM models must be performed from two perspectives: (1) the capabilities of existing data formats to handle the georeferencing information of BIM models; and (2) the performance and support of software tools for georeferencing BIM models. The former has been investigated in detail through the previous sections. The word cloud of included studies almost does not include any specific software tools, which indicates that few researchers have mainly been focused on investigating the capabilities and performance of various software tools for georeferencing BIM models. The rest of this part will investigate in detail the existing papers regarding the software tools for georeferencing BIM models.

4.5.1. GeoBIM

In 2019, GeoBIM Benchmark, a scientific initiative funded by ISPRS and EuroSDR, investigated the most relevant issues in managing CityGML and IFC within the existing software. The second task of the project was dedicated to exploring the available tools and procedures for georeferencing BIM models (exported IFC models), which contained eight responses, including a completed questionnaire, screenshots of results, and problems encountered in each software package tested. The project focused on two main points of view. The first one focused on the capabilities of the tools for georeferencing, and the second aspect was related to the exported IFC models (both IFC2x3 and IFC4 versions) to check for the possibility of achieving LoGeoRef levels based on the definitions that have been proposed by [10] and editing georeferencing information in tools [12,17,18].

The total number of 31 software tools has been categorized into six groups: GIS programs; 3D viewers; extract, transform, and load (ETL) tools; 3D modelling (CAD); analysis software; and BIM programs. Generally, there is little control over how georeferencing information is stored in the final model. In summary, just 30% of the tools effectively utilize the georeferencing details, utilizing accurate global coordinates from either LoGeoRef30 or LoGeoRef40. Minimal alterations were noted in the evaluations using the IFC4 dataset, which does not employ the extended IFC4 georeferencing capabilities in any case. However, in most instances (90%), the model’s orientation is maintained correctly, although it was challenging to confirm whether the additional True North information was being interpreted [12].

Table 7. The list of examined software tools in the GeoBIM Benchmark project [12].

	GIS	3D Viewer	ETL	3D Modelling (CAD)	Analysis Software	BIM Software
Software Name	Bentley map enterprise	BIM Vision	FME	FreeCAD	Acca PriMus-IFC	Autodesk Civil 3D
	ArcGIS Pro	FZKViewer		Bentley MicroStation + TerraSolid Lexocad	RFEM	DDS-CSD
		eveBIM viewer		BricsCAD Ultimate		Solibri office
		STR Vision IFC Viewer		Blender + IFC plugin		Tekla structures
		Solibri anywhere				Acca us.BIM viewer+
		RDF IFC viewer				Simplebim
		FME data inspector				Infraworks
						ARCHLine.XP
						Allplan
						AutoCAD Architecture
						ACCA Edificius
						BIMserver
						Revit
						Vectorworks
						ArchiCAD

shows the names of included software tools in this project.

The following results have emerged from the project, which has been concluded based on participants’ reports [17]: First, the tested tools allow for the manual georeferencing of the imported IFC models, but they cannot read and write all georeferencing elements in IFC. Second, the various interpretations of georeferencing options in IFC by tools make it more challenging to manage georeferencing consistently. Third, there was no difference between the performance of georeferenced and non-georeferenced models in explored tools as the tools probably stored the georeferencing information as values and still worked with the local coordinate system of the models. Finally, in some cases, setting the CRS and height reference system was possible. However, the supported list of the CRS was different in various tools.

Regarding the exported IFC2x3 version, the maximum LoGeoRef levels belong to eveBIM and FME_Script tools, which achieved level 30 for three axes: North, East, and Height. FZK Viewer stored the information up to level 30, which only considers North and East values; it did not store the correct direction. In the case of Revit, the georeferencing data were accurately recorded for just one model. To be precise, the North and East coordinates were incorporated into the IfcCartesianPoint linked with IfcSite (similar to LoGeoRef 30), while the Height information was placed in the RefElevation attribute of IfcSite (similar to LoGeoRef 20). Finally, FME_Desktop exported the IFC models to LoGeoRef 20 by storing only RefLatitude and RefLongitude [17]. FME_Script, Revit, and FZK Viewer have investigated the IFC models exported to the IFC4 version. The first two tools did not support any georeferencing elements of IFC4 (IfcMapConversion and IfcProjectedCRS), but FZK Viewer can modify the values of IfcProjectedCRS in some cases [17]. Based on the GeoBIM Benchmark results, it is possible to reach a LoGeoRef higher than 30 for the North, East, and Height values, although the TrueNorth attribute stores the rotation in some cases based on the LoGeoRef40 definition. Also, few tools implement the IfcMapConversion and IfcProjectedCRS classes allowed by IFC version 4 due to the higher usage of the IFC2x3 version in practice [17].

4.5.2. Autodesk Revit

The Revit format, made by Autodesk, is a unique data style widely used in BIM in the AEC sector. It is known for its powerful 3D modeling capabilities, allowing architectural, structural, and mechanical, electrical, and plumbing (MEP) aspects to come together for holistic building design and documentation. Revit’s popularity comes from its capacity to help create detailed 3D models that capture not just the shape of a building but also essential details about its parts, materials, and how it works. Because of this, Revit has become a central tool in modern AEC practices, making it easier for different experts to work together on construction projects with better efficiency and accuracy. Although Revit has been developed for the AEC domain with local coordinate systems, it can store the location of models to perform georeferencing to a geodetic coordinate reference system [58].

The discrepancy between the coordinate systems used in these two fields is one of the problems in achieving BIM/GIS integration. Understanding these coordinate systems can play an important role in addressing this problem. By understanding these coordinate systems, transferring between coordinate systems can be achievable. A coordinate system is one of the indispensable parts of the geospatial context, and various coordinate systems with different features have been well described and utilized in this field. On the BIM side, local coordinate systems have been primarily used in designing projects. Coordinate systems can be categorized into horizontal and vertical coordinate systems, described in detail. Figure 12 illustrates various types of coordinate systems in the geospatial field.

Data can be located across the surface of the Earth using horizontal coordinate systems. This coordinate system can be of three types: geographic, projected, and local coordinate system.

• GCS: A geographic coordinate system can define the real location of features on a model of the Earth’s surface. The datum represents the Earth model in GCS. Because the Earth’s surface is not perfectly smooth or round, many different data are designed for different parts of the world.
• PCS: The projected coordinate system is related to the visualization of features of the Earth model on the flat map. The projection is the mathematical algorithm that defines how to present the round Earth on a flat map. In addition to a projection, a PCS includes a geographic coordinate system.
• LCS: This type of coordinate system can define a coordinate system of large-scale (small-area) projects. Most CAD or BIM projects used LCS. Although the coordinate system is sometimes mentioned as the world coordinate system in BIM or IFC standards, it is the LCS.

Locating the relative height of the data layers is performed by the vertical coordinate system, which has two types: gravity-based and ellipsoidal systems. Gravity-based vertical coordinate systems reference a mean sea level calculation, while ellipsoidal coordinate systems reference a mathematically derived spheroidal or ellipsoidal volumetric surface.

Autodesk Revit has three different coordinate systems, and having a clear understanding of them can help perform georeferencing on BIM data. The coordinate systems are: (1) the internal coordinate system (ICS), (2) the survey coordinate system (SCS), and (3) the PCS.

• ICS: This coordinate system provides the basis for positioning all of the elements in the Revit model. The internal origin is a reference point for all elements within the Revit model. In the first step of creating a model, the Project Base Point and Survey Point are placed at the internal origin. Notably, the location of the internal origin (startup location) is fixed and cannot change, and the model geometry must be within a 16 km radius of it. By default, importing or exporting a CAD or Revit file will be performed relative to this point.
• SCS: This coordinate system is related to specifying the location of elements on the Earth’s surface and has close meaning to GIS, global, or projection coordinates. Notably, the curvature of the Earth can be considered in this system in case of large-scale projects. The origin has been shown with “Survey Point” in Autodesk Revit software. This point is usually used to represent the origin of a “shared coordinates” system among multiple linked Revit or CAD files. That means its location is most useful when exporting and importing files. Also, the coordinates of this point can be shared when exporting/importing files.
• PCS: This coordinate system is for the project and determines the position of objects relative to a specific point near the model. The origin point is “Project Base Point”, a reference point for measurements across the site. “Project North” shows the direction of the north in the project, and “True North” is the real-world north direction. The angle between these two can be set in Project Base Point attributes.

In addition to the various points and origins in Autodesk Revit software, the “Project Location” menu provides tools that have been considered for geolocation purposes, including specifying the geographic location for the project (“Location”), managing coordinates for linked models (“Coordinates”), and controlling the position of the model on the site (“Position”). The “Location” tool within this menu defines the project’s address using parameters such as the city’s name or postal code. However, it is crucial to note that this method is more related to representing the project location in terms of latitude and longitude format rather than embodying the proper concept of georeferencing. This example vividly illustrates the diverse ideas of georeferencing within the realms of BIM and GIS. Defining the postal code of a building, while convenient for some purposes, cannot be considered a comprehensive georeferencing solution. Importantly, this address-based georeferencing method may not be suitable for precise spatial analysis or interoperability with GIS, emphasizing the need for more robust georeferencing approaches in scenarios requiring higher accuracy and compatibility.

Considering the above definitions, it can be concluded that defining the SCS in Revit files can be an approach to performing georeferencing of BIM data. However, it should be investigated whether the georeferencing information of Revit files could be preserved after exporting it into other formats such as IFC or reading it in different software such as ArcGIS Pro (V3.4). Ref. [41] categorized the georeferencing of Revit files into three methods: using the survey and project-based points, using coordinates of the CAD dataset, and using two known control points. No standardized procedure exists to set the coordinate system within the Revit files uniquely. The Project Base Point defines the origin of the project coordinate system, and the Survey Point defines the origin of a “shared coordinate system”. Ref. [6] investigated the possibility of performing georeferencing on BIM data. The study has found some relations between the location attributes of Revit files and the IFC georeferencing classes, but the georeferencing process could not be conducted fully automatically. It was mentioned in the Revit file that the exported IFC file had preserved the updated longitude and latitude values of the Survey Point. However, some geometric and semantic issues have occurred while exporting and importing IFC files. Also, the proposed approach is manual and tedious and cannot be implemented on a large-scale project with a big number of buildings.

Paul Wintour published an article about the process of georeferencing Revit models based on survey or GIS datasets such as georeferenced CAD files. To prepare a Revit site model, start by creating a circle at a known universal transverse Mercator (UTM) point in AutoCAD and reposition it to (0,0,0) to align with the “Project Local” coordinate system. This ensures that the geometry is closer to Revit’s internal origin. In Revit, confirm that the length units are consistent, set the view to True North, and link the CAD file using the “Auto–Origin to Internal Origin” option. Next, set the site coordinates (“Select Coordinates at a Point” option of “Coordinates” in the “Project Location” tool of the “Manage” tab) by selecting the Survey Point and entering the corresponding coordinates, ensuring that the Project Base Point remains aligned with the internal origin [59].

Ref. [60] presented a method proposed by the GEOBIMM project to integrate BIM and GIS in the underground network. The georeferencing of BIM data were considered in their study. The geographic information had been assigned to the Survey Point in the Revit file to specify the coordinates of the entire model. Ref. [19] investigated the GeoBIM project results and explained the georeferencing BIM models in Autodesk Revit software. The results of both studies indicated that the exported IFC files contained the correct georeferencing information after performing the proposed solution. Ref. [61] presented a prototype for integrating historical BIM with GIS. Providing a known cartographic coordinate and azimuth by GNSS points made georeferencing Revit models possible in Autodesk Revit software. An additional projection file (.prj) was used to visualize the georeferenced BIM models in the ArcGIS Pro environment.

Ref. [62] explored the potential applications of BIM-IoT-GIS integrated digital twins for post-occupancy evaluations. Their proposed methodology included three parts; the main steps were georeferencing BIM models and validating them. It was essential to verify that the BIM model and sensors had been georeferenced correctly and imported to the correct location in the GIS environment. They georeferenced the Revit model using Project Base Point and Survey Point. Google Earth Pro has provided the value of the coordinates for those points. Finally, after importing the model into ArcGIS Pro, it was placed in its correct location using the Define Projection tool. Due to the lack of established methods for validating BIM/GIS integration, the georeferencing part has been validated by comparing the coordinates of Project Base Point and Survey Point in both Revit and ArcGIS Pro tools.

The second method links a CAD dataset to the Revit project. The coordinates of the linked project become the shared coordinates of the Revit project, and its origin is the origin of the shared coordinates within the Revit project. The limitation of this method is that it is impossible to change the internal coordinates of the Revit project [41]. The last method is the most native way of georeferencing, which transforms the Revit project geometry into a CRS using the two known control points. This function is only available within the Autodesk Point Layout (APL) plug-in [41]. However, it is also possible to apply georeferencing to the BIM model by combining functions such as rotate, shift, and scale. Ref. [63] published the practical experience regarding the processing of 3D models of buildings by three methods: mesh processing, CAD, and HBIM. It is suggested that MicroStation software be used to perform georeferencing of CAD data and Revit software for HBIM models by applying a combination of functions such as rotate, shift, and scale.

4.5.3. ArcGIS Pro

ArcGIS Pro supports BIM files (IFC2x3, IFC4, and Revit [versions between 2016 and 2022]) as ArcGIS BIM file workspaces. BIM data need to have a valid ESRI coordinate system (PRJ file) and optional coordinate transformation information (WLD) file, which has been defined as follows:

• Projection (.prj) file: Defined coordinate system, data, and map projection will be stored in a text file in ArcGIS Pro. It can be created for BIM data with the same name as the Revit file (<filename>.prj). The universal projection file (esri_CAD.prj) defines the coordinate system for all BIM data in the same folder.
• A world file (.wld3): A world file (WLD3) is a text-based (ASCII format) file including four points to describe a coordinate transformation (linear affine) by defining the offset, scale, and rotation using two vectors within the assigned spatial reference. A universal world file (ESRI_CAD.wld3) defines offset control points for all CAD or BIM files stored in the same folder.

ArcGIS Pro provides the user with the “Validate Position” option to check whether the BIM data have an assigned coordinate system and if the BIM data have been georeferenced or if georeferencing is necessary. Revit, IFC, or 3D CAD files are georeferenced in an ArcGIS Pro scene using the Georeference tab tools. In case of the need to reposition the BIM model after confirming that it has a valid projection file defined for the BIM file, the BIM feature layers can be repositioned using the Locate, Move to Display, Move, Rotate, and Elevate To Ground tools on the Geoprocessing toolbar on the BIM Data tab on the ribbon. While repositioning the BIM model, it is possible to move, scale, and rotate the model based on any points the user wants. Accessing the coordinates of Project Base and Survey Points can make the georeferencing process more straightforward. Figure 13 illustrates the process of georeferencing BIM models in ArcGIS Pro software.

Ref. [64] designed an IFC-based database schema and proposed a methodology to transfer BIM data into the proposed schema. Georeferencing BIM data was one of the steps performed in ArcGIS Pro, which defined the coordinate system of the area using the “Define Projection” tool and located the BIM model in the correct position using the “Georeference” tool. In this study, Revit is the chosen BIM software, and because ArcGIS Pro is capable of reading Revit files, georeferencing is conducted using these files instead of IFC files. ArcGIS Pro does not support file formats from other BIM software like ArchiCAD or SketchUp. In scenarios where different BIM software is used, IFC files are recommended. ArcGIS Pro’s Data Interoperability extension, which utilizes FME, can be employed to georeference these IFC files. Specifically, the FME Hub 3DAffineWarper transformer is used for georeferencing using four established points in the model, and the CoordinateSystemSetter transformer is applied to assign the correct coordinate system for the georeferencing process.

4.5.4. OpenBuildings Designer

In OpenBuildings Designer, georeferencing is essential for positioning building models accurately within real-world coordinates. The software uses high-resolution BIM models, requiring careful alignment with civil data to avoid errors caused by differing coordinate systems. Below is a step-by-step guide to effectively georeferencing the BIM models [65].

• Understand the solids working area: OpenBuildings Designer’s solids working area is set to 2.6 miles (4.2 km) by default. Ensure that all elements are modeled within this area to avoid warnings and tool malfunctions.
• Open site survey file: Begin by opening a site survey file with an existing GCS. Verify the GCS through the GCS dialog box under the Drawing Aids tab.
• Create building GCS master file: Create a new file for the project’s GCS (e.g., “Project GCS Master”). Reference in the architectural model and set the orientation to coincident.
• Attach civil file and align: Attach the civil site survey file as a reference and observe its location. Because the civil file uses real-world coordinates, it will likely be positioned far from your building model. Move and rotate the survey file to align it with your building model.
• Apply Helmert transformation: After aligning the files, apply a Helmert transformation to account for the movement and rotation of the survey file. In the GCS dialog box, select From Reference and choose the survey file to calculate the transformation.
• Apply GCS to other files: Once the GCS is set in the master file, reference in other files (e.g., structural, architectural) and push the GCS transformation to them using the To Reference option.
• Verify with BIM models: Test the georeferencing by attaching the BIM models to the survey file and check if they appear in the correct geographic location using the AEC Transform Orientation option.
• Reverse workflow: To validate, reference the survey data to a BIM model, and select Geographic Reprojected to confirm the correct alignment without further adjustments.
• Google Maps and Earth integration: Once georeferencing is applied, you can view your model’s location in Google Maps or export it as a Google Earth KMZ file to see the model in a real-world terrain context.

4.5.5. ArchiCAD

As the IFC4 schema meets the Model View Definitions (MVDs) of geolocation by filling the described attributes of entities in Section 3.3, it is possible to write these attributes using ArchiCAD software (versions 23 and 24). Firstly, it is necessary to locate the “latitude” and “longitude” position of the project at ArchiCAD’s origin by opening the “Project Location” dialog from Options > Project > Preferences > Project Location. The good point is the inserted location that can be checked by clicking the “Show in Google Maps” option of the software. Next, the ArchiCAD Survey Point object must be placed in the project at a known point. After placing the Survey Point and selecting it, open the “Object Selection Settings” dialog. Under the “Survey Point Settings” tab, go to “Geo Referencing Map…” and input the corresponding values of IfcMapConversion and ProjectedCRS [66]. The main issue is when exporting to IFC, you cannot export a unit for the coordinate reference system. Additionally, the export process may apply the True North rotation while also retaining the rotation in the geolocation properties, resulting in the rotation being defined twice. This leads to an incorrectly geolocated model that requires patching. When such a file is imported with a double-defined rotation, the map conversion may be ignored, which can cause issues with correctly geolocated files from other software [67].

4.5.6. Tekla

Tekla of Trimble, another AEC software for designing BIM models, also supports the georeferencing of BIM models by defining the coordinate of Base Point in the model. Project Base Points allow you to use another coordinate system needed for interoperability and collaboration by defining Project Base Points for IFC exports and reference model imports [68]. The “Base Point” dialog box can be used from File > Project Properties > Base Point path, and the following attribute can be assigned to any point in the model as the Base Point. This option obtains the location of Base Point and assigns new values based on the user input.

Table 8. The attributes of the “Base Point” dialog box.

Name, Description	Name and Description of the Base Point
Coordinate System	Name of coordinate system you are using
East Coordinate (E)	East coordinate that represents the X-coordinate related to the civil origin
North Coordinate (N)	North coordinate that represents the Y-coordinate related to the civil origin
Elevation	Elevation that represents the Z-coordinate related to the civil origin
Latitude, Longitude	The latitude and longitude of the base point to be used in the IFC export
Location in the Model	A location for the base point in the Tekla Structures model. The distance is measured from the model origin.
Angle to North	The angle between the Y and North direction.
Project Base Point	Select the check box if you want to set a coordinate system as the Project Base Point.

shows the attributes of the “Base Point” dialog box [69].

After defining the Project Base Point, it is possible to export the model into IFC. When we export using IFC, it is possible to export based on the shared coordinate system above, which is compatible with the design models. The exported IFC models can then be imported into Navisworks and/or Newforma Konekt. There are three options in the Export to IFC dialog box for “Location by…”: Model Origin, Work Plane, and Base Point. By choosing the last one, the model will be exported into the given coordinate system for the base point. The file will not be properly positioned if it is imported into Revit, as Revit uses an origin-to-origin system [68,70].

Linking the generated/exported IFC files from various software into other AEC/Geospatial software is an important issue that is related to the interoperability of BIM/GIS integration. However, linking IFC files from Tekla Structures into Revit does not work properly, even though the IFC shows the correct geographical location and links correctly into other software such as Solibri and Navisworks. The Revit only provides the option of “Origin to Origin” linking IFC files. To ensure proper alignment between Tekla and Revit, start by identifying the Tekla origin and the national grid coordinates of the Revit origin. In Tekla, set up a new coordinate system by entering the offset and location details to match Revit’s coordinates. Additionally, input the “Angle to North” to ensure that the IFC file aligns correctly with other platforms like Navisworks or Solibri. Finally, export the IFC from Tekla using this new coordinate system, then in Revit, rotate the linked IFC by the agreed angle to achieve proper alignment [71,72].

4.5.7. Other Approaches

As mentioned, georeferencing methods have been widely developed in the GIS field. One of the common approaches is calculating the transformation matrix between two maps or objects (for example, the footprint of buildings, satellite images, etc.) using least square fitting. The most common algorithms for georeferencing include affine transformation, similarity transformation, and projective transformation [26]. With the assumption of the availability of 2D georeferenced footprint of buildings from GIS sources, it is possible to transform the BIM models to georeferenced based on the accurate footprint.

Ref. [26] performed the affine transformation for georeferencing a simple bridge. They only needed a simple linear translation and shift, achievable by affine transformation. The parameter that can impact the accuracy is the selection of control points (number and location). The performance of this method on various types of BIM data (such as buildings, railroads, roads, etc.) should be validated. Ref. [73] presented a system architecture for structuring and manipulating BIM, 3D geospatial information, point clouds, and time series data obtained from sensors. They used the 2D convex hull of BIM and footprint coming from open street maps (OSMs) and determined the affine transformation between them using the approximative rigid matching and close-form absolute orientation method. The precision of their method depended on the 2D map feature used.

In 2020, Ref. [5] developed an automatic framework to conduct georeferencing on BIM data using 2D georeferenced footprints. The other inputs are the polygon of BIM corresponding to its Z-projection and 2D convex hull. After finding the transformation matrix using the affine method, the BIM data will be georeferenced. Although the proposed method was tested with several buildings, some limitations still exist. First, their shape-matching approach would provide a highly accurate alignment for basic shapes like rectangles, squares, or circles. Still, it would not be able to guarantee the correctness of the final orientation due to the similarity of the convex hull of such shapes in all orientations. Second, the Euclidean transformation was highly dependent on points r and k, which were selected as the farthest points on the BIM polygon. However, it can be a problem with round-shaped buildings and cases in which significant difference exists in their convex hull (significant protrusion on one shape is missing on the other, which can be expected in complex BIM models). Third, the other essential prerequisite was the availability of a 2D georeferenced footprint. In some cases, the BIM models are in the design phase, and it is impossible to have a georeferenced footprint. Fourth, the manual removal of extra building elements of BIM models after vertical projection is still needed in the method. Finally, the precision of their results was not good in some cases, which was caused by the significant difference in the convex hull of their BIM and map polygons (footprints). The BIM polygon was created by vertical projection on the XY plane, which can still contain more geometry details than the georeferenced polygon. In 2022, Ref. [8] developed an integrated BIM/GIS web-based platform for a mega construction project. The proposed methodology included geometric visualization, coordinate transformation, and rendering. Regarding the transformation of BIM models, a method for georeferencing BIM models has been proposed that is exactly similar to the method of Diakite and Zlatanova.

Ref. [74] developed a cross-platform web application using open-source technologies to facilitate the collaboration and editing of IFC documents in the BIM domain, including georeferencing of IFC documents using orthophoto maps. They compared the following five BIM platforms based on the most relevant aspects of their proposed solution: Autodesk Revit, Autodesk Viewer, BIMvision, eveBIM, and usBIM.viewer+. The results showed that only Autodesk Viewer and BIMvision platforms could not perform georeferencing of the IFC model. The developed platform included viewer/editor applications that focused on the representation and visualization of IFC models, visualization of IFC model elements data, movement in the 3D scene, and viewing of IFC model revisions. One of the specific features of the editor is the georeferencing of the model following the IFC standards and editing geometry (translation, rotation, and scale of elements and the IFC model). The authors did not provide more details about the georeferencing process, but based on the figures, it is possible to change the coordinates of the center point of the model and the elevation of the base of the model.

Ref. [75] developed an open-source software, IfcTerrain, based on ETL, to convert DTM to the IFC format. The IfcTerrain converter offers the user four options to configure the exported IFC file: the type of object format, the type of geometric representation of the terrain, the georeferencing, and the annotation with metadata. The georeferencing part of the software has been developed based on the classification LoGeoRef. The possibility of storing the georeferencing information at the highest level (loGeoRef50) can facilitate the comprehensive calculation of georeferencing by BIM and geospatial software. Ref. [43] developed the IFC2GML tool to transfer the IFC models of CAD objects into GML models, which included transferring the local coordinates system of objects into the real-world coordinate system using the affine transformation. This tool supports functionalities such as selecting building objects from the IFC file, displaying related information such as the definition of the coordinate system, and transferring the objects into the GML model.

5. Discussion

The evidence from these studies implies that georeferencing issues need more attention from both domains to facilitate the BIM and GIS integration process. Although considerable progress has been made in obtaining one unique definition and standard for georeferencing in both fields, a significant attempt is needed to overcome this gap. Most studies have solved their georeferencing problems from the perspective of project-specific purposes, and a unique and comprehensive approach to address this issue is still needed. The main findings and outcomes of this study are discussed as follows, with necessary gaps related to knowledge and practice, interpretation, standardization, and implementation being identified.

5.1. Automation of the Georeferencing BIM Models Process

By investigating the existing open IFC data models, it is obvious that most published models do not include sufficient georeferencing information. This emphasizes the necessity of increasing awareness of BIM designers regarding the geospatial concepts. However, this increased awareness can be highly beneficial during the design phase, both before starting the project and throughout its construction. While these improvements address challenges in the design stage, we must also consider the existing reality of as-built BIM models for many buildings. As noted in the literature, most as-built BIM models are not georeferenced. It can be challenging or costly to request a surveyor to provide GCPs for as-built models from scratch.

On the other hand, based on the literature review, it is evident that georeferencing as-built BIM models can be achieved using various methods when 2D georeferenced footprints are available. Nevertheless, there is currently no fully automated approach, and most existing methods still require significant manual intervention. Additionally, these methods have some limitations in terms of accuracy and efficiency. Given the growing availability of open footprint datasets, it becomes clear that developing a fully automatic georeferencing method and incorporating georeferencing information directly into IFC models is essential. Such advancements would not only address the current gap in georeferencing as-built BIM models but also facilitate the seamless integration of BIM and GIS, ultimately contributing to the development of more accurate and efficient digital twins.

In the context of urban digital twins, real-time integration of georeferenced data is crucial to maintaining spatial accuracy and consistency as real-world conditions change. Automating real-time georeferencing would support efficient data fusion from multiple sources, ensuring that digital twins accurately reflect the physical environment. As digital twins become increasingly important in smart city planning, facility management, and infrastructure monitoring, automated and dynamic georeferencing methods are vital to maintaining spatial accuracy and operational efficiency.

5.2. Georeferencing Issues in IFC Models

Reviewing the studies of georeferencing IFC files indicates various interpretations in previous research. Different studies have different ideas about how to do this correctly. This reflects a clear interpretative gap in the literature with practical consequences. This confusion happens because IFC files are complicated and have detailed information. Some researchers had various understandings of how these files use coordinate systems and where things are placed in them. This study explained how different parts of IFC files relate to coordinate systems. However, we think it is a good idea for researchers to look more closely at the IFC standard itself. A detailed study on how IFC files handle georeferencing would be helpful. It could clear up the confusion, set some standard rules, and make it easier for people working with IFC files.

• Multiple Buildings on a Single Site

It is important to dig deeper into the IFC standard to understand it better. For instance, the ambiguity surrounding classes and hierarchy relations needs attention. In IFC4, it is possible to store coordinates for a point referenced by IfcBuilding and IfcSite. However, a scenario where multiple buildings exist on a single site complicates matters. One suggestion is to define separate points for each building and site to facilitate georeferencing for individual structures. Additionally, expanding the definition of ground points within the IfcProject class could enhance georeferencing accuracy and encompass the entire project, ensuring georeferencing for all buildings and sites within the project based on their relative local positions.

• Large-Scale Projects with Multiple Sites

Another scenario is related to large projects where we can have different sites in a single project. In these cases, it is required to use multiple sites to handle the effect of Earth curvature on construction works. However, there is only one instance of IfcMapConversion for storing georeferencing information, which cannot be sufficient in this case. Although it has been mentioned in the literature to divide large projects into smaller sites, it does not provide practical criteria or guidelines on how to determine the division points or how to maintain consistency across contiguous sub-models. On the other hand, the recommendation to establish separate transformations for each site might introduce inconsistency and complexity, especially if the transformations are not managed uniformly. One approach could be to introduce a higher-level georeferencing entity within the IFC schema that links multiple IfcMapConversion instances under a single project. This would allow each site to have its own transformation parameters while maintaining a cohesive and consistent georeferencing structure for the entire project.

• Ambiguities in IFC Georeferencing Hierarchies

Regarding the current capabilities of the IFC data format to store the georeferencing information [10], categorized them into six levels (LoGeoRefs). However, it is possible to store complete georeferencing information in LoGeoRef50, and there are some issues that need to be addressed in this standard. First, the presence of multiple options for storing georeferencing information within IFC files can be misleading, as not all levels provide sufficient detail to achieve complete and accurate georeferencing. This inconsistency may result in confusion and improper application in practice. We suggest the following approaches to address this issue: (1) the straightforward approach is removing the extra entities and attributes of IFC and implementing unique and comprehensive classes and entities that can store georeferencing information, ensuring consistency and accuracy. (2) the challenging way can be developing an approach to produce the required attributes for each level of georeferencing from other levels and embedding them into the IFC file. This method needs to examine the possibility of extracting the direct and indirect relationships between the entities and attributes of the IFC data format and probably utilizing some external data such as 2D georeferenced footprints.

Second, any elements in IFC can have ObjectPlacement to another IFC element, which can have a different map conversion. This can create a problem in the consistency of georeferencing. For example, in LoGeoRef30, the IfcSite can store ObjectPlacement with respect to CRS, and the model will be partially georeferenced. In this case, if attributes of higher level of georeferencing such as LoGeoRef40 or LoGeoRef50 populated, there will be problems regarding the consistency of georeferencing as IfcSite will interpret both placements and it cannot properly georeferenced.

• Lack of Interconnectivity in Georeferencing Levels of IFC

There is a noticeable lack of studies focusing on establishing the interconnections between different levels of georeferencing (LoGeoRef) within the IFC framework. The LoGeoRef classification system introduced by [10] outlines six levels of georeferencing, ranging from basic location data (LoGeoRef10) to comprehensive georeferencing information (LoGeoRef50). However, these levels are often treated independently, with limited research exploring how attributes from lower levels can inform or generate higher levels. For instance, consider a scenario where an IFC file contains only LoGeoRef10 and LoGeoRef20 information. In this case, it remains unclear how to systematically calculate and populate the attributes required for a higher level, such as LoGeoRef50, based on the available data. This lack of a structured approach can lead to inconsistencies and ambiguities when integrating BIM and GIS data. Developing a methodology to derive comprehensive georeferencing data from minimal input levels would significantly enhance the clarity and consistency of georeferencing practices within the IFC framework. Such an approach could include algorithms that infer higher-level attributes from basic georeferencing information, leveraging spatial relationships, metadata analysis, or external geospatial datasets. Addressing this gap would facilitate more reliable georeferencing in complex BIM projects and promote uniformity across different georeferencing levels.

5.3. Academic and Industry Perspectives on Best Practices

Based on the systematic analysis of existing literature, standards, and software tools, this review identifies several key knowledge gaps that remain unaddressed in the georeferencing of BIM models. These gaps reflect limitations in both theoretical understanding and practical implementation, and they continue to hinder the seamless integration of BIM and GIS. (1) It has become evident that within the realm of BIM, there is no universally agreed-upon definition for “georeferencing”. This lack of consensus presents a significant knowledge gap when effectively incorporating accurate georeferenced BIM data into the GIS field. (2) The lack of understanding spatial referencing systems, a fundamental aspect well established within the GIS context, needs to be addressed within the BIM domain. (3) The BIM designers often fail to retain or save the georeferencing information of scan-to-BIM models generated by LiDAR technology. In some cases, the BIM designers will transfer the georeferenced point clouds into local coordinates during the design phase. To bridge these gaps, it is important to educate BIM designers to increase their understanding of these three pivotal concepts. Leveraging LiDAR data to create georeferenced BIM models represents a significant opportunity to advance the field. Also, BIM models will not integrate with GIS during the design phase, making it essential for BIM designers to consider the mentioned concepts during their BIM design stage. It is imperative to prioritize knowledge dissemination among BIM users to address this issue and enhance the seamless integration of BIM and GIS.

Besides some ambiguity in the IFC data format, several studies have proposed some extra entities and extensions to improve the georeferencing capabilities of the IFC data format. Ref. [10] proposed six LoGeoRef for categorizing the existing capabilities of the IFC data format. Although various researchers and projects (GeoBIM Benchmark Project) have already made strides using this structure as a guide, it is surprising that these levels have not been formally recognized within the IFC’s official documentation. This represents a clear standardization gap, where widely recognized practices have not been institutionalized within the schema. Even though there are some ambiguities regarding the entities, attributes, and relationships between LoGeoRefs in IFC as discussed in this section, adopting these levels into the standardization can still be beneficial. Standardizing these levels within the IFC would pave the way for a consistent and globally understood approach to georeferencing. Beyond its fundamental significance, such standardization would provide a clear roadmap for software developers and increase the accuracy of georeferencing. With a universally accepted structure in place, developers could seamlessly incorporate these georeferencing levels into their applications. Also, the users from both domains can populate required georeferencing information into the IFC models after performing georeferencing based on these adopted levels. Such standardization would streamline data exchange processes and improve interoperability between BIM and GIS tools.

When evaluating software tools for their performance in georeferencing and BIM/GIS integration, our review highlights the notable strengths of three particular tools: Autodesk Revit, ArcGIS Pro, and FME, each offering distinct advantages. Autodesk Revit emerges as a strong contender due to its capability of georeferencing Revit models and exporting them into IFC files with an extensive array of options and features. This functionality is a critical asset in integrating BIM and GIS data. However, it is worth noting that Autodesk Revit does not provide direct support for CRS. A significant challenge is that BIM designers often overlook the Project Base Point or Survey Point within Revit software. It is advisable to make it mandatory to specify the location of these points in project delivery documents or as a note. Another recommendation is to align the internal origin and these points at one corner of the building. Such an alignment would greatly simplify the georeferencing process, particularly for users of other software tools like FME and Edificius, who would benefit from knowing the exact location of these origins. On the other hand, ArcGIS Pro demonstrates proficiency in reading and visualizing Revit models while offering georeferencing capabilities. One of the standout features of ArcGIS Pro is its support for a wide variety of GCS and PCS, making it versatile in handling spatial data integration tasks. One feature that can be implemented in ArcGIS Pro is the capability of exporting the georeferenced BIM models to Revit or IFC data formats. Lastly, FME stands out by providing users with diverse transformer tools that can help modify georeferencing attributes within IFC data files. This flexibility in data transformation can significantly facilitate the adjustment and refinement of georeferencing information, enhancing the interoperability between BIM and GIS datasets.

The remaining software tools in the previous analysis appear to have room for improvement, particularly in their capacity to edit and update georeferencing attributes within IFC files and to support coordinate reference systems. While these tools may offer some valuable features, they have not yet reached a level of georeferencing proficiency that matches the capabilities of Autodesk Revit, ArcGIS Pro, and FME. Several potential explanations could account for this finding. (1) Many of these tools were primarily developed by BIM designers, whose understanding of georeferencing may differ from that of GIS experts. This perspective disparity could lead to limitations in the tools’ georeferencing functionality. (2) Achieving a higher LoGeoRef in certain tools can be a complex and time-consuming process. Georeferencing tasks often require intricate adjustments and precise calibration, which may not be streamlined in some software. (3) It is essential to acknowledge that this evaluation might not entirely reflect the capabilities of all of these software tools. As the users were selected with various levels of experiments for investigating each tool, there might be nuances and advanced features that were not explored in this review, potentially affecting the results.

6. Conclusions

This review has systematically examined the concept of georeferencing within BIM/GIS integration, highlighting key challenges, current practices, and available tools. Through an in-depth analysis of the literature, IFC schema capabilities, and georeferencing workflows, this study has revealed several persistent gaps that continue to hinder seamless BIM/GIS integration.

One of the major findings is the lack of a standardized definition of georeferencing in the BIM domain, which creates challenges for interoperability with GIS systems. Additionally, the overlooked potential of LiDAR data and the frequent loss of spatial metadata during BIM model creation highlight significant knowledge and practice gaps. Increasing awareness among BIM professionals regarding the importance of geospatial metadata and coordinate reference systems is essential for improving integration outcomes.

To address these challenges, future developments should prioritize the formalization of georeferencing standards within the IFC schema—particularly through the structured adoption of frameworks like LoGeoRef. Technical efforts should also focus on resolving current ambiguities in the spatial hierarchy and coordinate system definitions within IFC files.

From a practical standpoint, the evaluation of software tools indicates that while platforms like Autodesk Revit, ArcGIS Pro, and FME offer useful features for georeferencing, limitations remain. Enhancing CRS support in Revit, improving IFC export options in ArcGIS Pro, and expanding transformation capabilities in FME could significantly improve interoperability. Developers should also consider embedding LoGeoRef concepts into their platforms to support consistent and structured georeferencing across BIM and GIS domains.

One promising future direction is to address the knowledge gap among BIM designers so that future design-phase BIM models will include the necessary georeferencing information. As this awareness improves, the focus can gradually shift towards georeferencing as-built BIM models, where the challenge remains more significant. Currently, we have access to a large amount of open data, including footprint open databases, that has already been georeferenced. Therefore, an effective future approach could be to develop automated methods for georeferencing as-built BIM models using these georeferenced footprint open databases.

By investing in standardized practices and automated solutions, the BIM/GIS community can significantly advance the seamless integration of spatial information, ultimately enabling more accurate and efficient digital twin applications.