Modelling underground cadastral survey data in CityGML

In underground environments, survey elements such as survey points and observations provide the information required to define legal boundaries. These elements are also used to connect underground legal spaces to a geodetic survey network. Due to the issues of current 2D approaches for managing underground cadastral data, prominent 3D data models have been extended to support underground land administration. However, previous studies mostly focused on defining underground legal spaces and boundaries, with less emphasis on survey elements. This research aims to extend CityGML to support underground cadastral survey data. The proposed extension is based on the survey elements elicited from underground cadastral plans, which is then implemented for an underground case study area in Melbourne, Australia. This extension integrates underground survey data with legal and physical data in a 3D digital environment and provides an improved representation of survey elements, facilitating the management and communication of underground cadastral survey data.

New underground developments such as tunnels require drafting multiple pages of survey plans to show vertically stratified legal spaces in the subsurface areas (Strack, 2021).Current practices use 2D survey plans, crosssectional diagrams, and textual notations (e.g., elevation information) to define the legal extent of underground assets.The difficulties with interpretation make survey plans barely usable for the public.This has inevitably raised the question: what is the extent of subsurface rights?(Strack, 2021).To answer this question, it is first necessary to define the ownership of underground space clearly and secondly model it precisely and correctly.The first part is related to the legal aspects of Underground Land Administration (ULA), but the second part is a technical matter (Saeidian et al., 2023).The development of a 3D land administration model is a technical part that studies the process of adding 3D legal objects in a data model (Hassan & Abdul Rahman, 2011).The 3D models derived from such data models can improve the communication and interpretation of land administration data in underground areas.Consequently, there is a growing trend towards adopting 3D data models in the realm of land administration (Asghari et al., 2021).
Surveying is the main form of capturing underground data for land administration (Aien, 2012).The cadastral surveying process involves defining, identifying, demarcating, measuring, and mapping new/changed legal boundaries (Grant et al., 2020).In the current practice, survey plans are used to define the spatial extent of underground legal spaces and boundaries as well as relevant attributes and relationships.In addition to underground legal data (legal spaces and boundaries), the original cadastral surveying data in the field is also recorded and communicated using the same 2D survey plans or separate documents.This surveying data is a summary of a cadastral survey work that provides the information required to define legal boundaries and connect underground legal spaces to a geodetic survey network.Survey documents provide various types of survey measurements and their attributes.
For example, there is a range of survey points such as control points, traverse points, and boundary points.In addition, there exist several types of survey observations such as traverse, radiation, and boundary observations.Elevation information is also provided in the survey plans and documents.
A 3D data model supporting digital land administration should provide entities to define not only underground legal data but also the cadastral survey elements.A 3D data model is the basis to create a 3D integrated digital model.It can potentially provide an effective approach to managing and communicating ULA data components including survey elements (survey data), the geometric and semantic information about the physical reality of underground assets such as utilities and tunnels (physical data) as well as legal spaces and boundaries (legal data) in subterranean spaces.However, the existing 3D integrated data models mostly focus on underground physical and legal data, with less emphasis on survey data elements.
Although some data models such as LandInfra and LADM are rich in defining survey elements and legal data, prominent 3D physical data models such as IFC and CityGML have also their use cases and benefits for ULA as discussed in some studies (Atazadeh et al., 2022;Saeidian et al., 2023).Therefore, these data models have been extended to model underground legal data in order to provide a 3D integrated digital model which defines the physical reality of underground assets and their corresponding legal spaces and boundaries.Since IFC and CityGML are limited in defining survey elements, the proposed integrated data models cannot fully support survey data.In this regard, an IFC-based integrated model was enriched with survey data by Atazadeh et al. (2021).
However, there is a knowledge gap in exploring the potential of CityGML to support survey data elements.
CityGML is a leading standard for 3D city modelling, which is used widely in the geospatial domain.Several cities apply the CityGML data structure to manage and communicate their 3D city models which serve a wide range of applications from land use planning to cadastre (Lippold, 2022).In this regard, a range of CityGML Application Domain Extensions (ADEs) has been developed in several domains (Biljecki et al., 2018).Some studies have also investigated and suggested CityGML for land administration purposes (Halim et al., 2021;Nega & Coors, 2022;Saeidian et al., 2023aSaeidian et al., , 2023b;;Siew et al., 2021).However, this data model does not support cadastral survey elements in the current version (CityGML 3.0) and the studies also ignored defining these elements in the proposed

| CURRENT S TANDARDS FOR MODELLING UNDERG ROUND S U RV E Y DATA
This section reviews the current 3D data models which include surveying data elements.The review specifically focuses on assessing the capability of these data models in terms of modelling underground survey measurements.The 3D data models assessed in the study include LADM (ISO, 2012), ePlan (ICSM, 2010), LandInfra (Scarponcini et al., 2016), as well as those contributions that enriched IFC (ISO, 2013) and CityGML (Kolbe et al., 2021) standards.Table 1 presents the entities provided by these data models to define different underground cadastral survey elements (the next section explains these survey elements in detail).
LADM is a leading international 3D data model in the land administration domain.In recent years, several studies have worked on developing LADM-based country profiles for underground areas (Dželalija & Roić, 2022;Janečka & Bobíková, 2018;Kim & Heo, 2017;Radulović et al., 2019;Ramlakhan et al., 2023;Saeidian et al., 2022;Silva & Carneiro, 2020;Yan et al., 2019Yan et al., , 2021)).However, these studies mostly focused on the legal spaces attached to underground assets.In this area, Soffers (2017) used LADM as a template to design a data model in order to link survey elements to legal boundaries in the Netherlands.For modelling cadastral survey elements, Kalogianni et al. (2021) also proposed a refined survey model for LADM considering the interoperability between LADM and LandInfra standards.Since LADM focuses on land administration, it provides some feature classes for modelling cadastral survey elements as presented in Table 1.For example, the LA_Point class from the Surveying and Representation sub-package defines cadastral survey points.The LA_BoundaryFaceString feature class is also provided for defining boundary lines and curves.However, this entity does not define some critical attributes of observations such as bearings and distances.In addition, this data model does not support other survey observations such as radiation observations.In LADM, the elevation information can be defined for points, but there is no specific class to define survey surfaces (see the next section for more information about these surfaces).Since the geometries of survey observations and surfaces are lines, curves, or surfaces, it is possible to geometrically represent these elements using the LA_BoundaryFaceString and LA_BoundaryFace feature classes.However, this mapping approach does not consider semantics and relevant attributes.It makes ambiguity for users such as surveyors who want to reuse survey data using a 3D digital model.LADM is also a conceptual data model without any encodings.This standard is limited to the legal aspects and suggests external classes for modelling underground physical objects such as pipelines and cables (Lemmen et al., 2015).Finally, it should be mentioned that the current version of LADM is being revised/extended and the new version is expected to cover more information related to surveying and data acquisition approaches as well as accuracies such as new attributes and code lists for surveying techniques and platforms (Kalogianni et al., 2021;Lemmen et al., 2019;Van Oosterom et al., 2019).
ePlan is another cadastral data model developed in Australia based on LandXML.As seen in Table 1, this data model provides several entities to define different survey elements.For example, the Points and Observation packages provide entities for defining survey points and observations, respectively.However, the survey elevation information is limited to points, and elevation surfaces are not defined.In addition, similar to LADM, this data model faces the challenge of addressing physical aspects and does not support physical data.An integrated 3D model requires to include not only underground cadastral survey elements and legal information but also information about the physical reality of underground assets (Saeidian et al., 2022c).Finally, the capabilities of ePlan are limited in terms of supporting the requirements of three-dimensional land administration (Aien, 2012).
LandInfra is another standard developed to model both land and infrastructure information.This data model provides the Survey package with three sub-packages (Equipment, SurveyResults, and Observations) for information related to observations, processes and their results gathered during survey works (Scarponcini et al., 2016).
Therefore, LandInfra has several entities that can be used to model survey elements as presented in Table 1.For example, the SurveyMark class can be used to define survey points.The SurveyObservation class and its subclasses also cover a wide range of survey observations such as angular and distance observations, total Stations observations, level observations, GNSS observations, point clouds, and image observations.The LevelObservation class can also be used to represent elevation information.The deltaHeight attribute of this class can be used to define TA B L E 1 The entities provided by some well-known data models to define underground cadastral survey elements.the elevation.The entities provided by LandInfra for modelling survey data need to be customised based on the jurisdictional cadastral surveying requirements (e.g., adding the required attributes).In addition, while LandInfra has been suggested as a potential model for a 3D cadastre (Bydłosz & Bieda, 2020), there are still a limited number of studies that have utilised it for land administration purposes.
LADM, ePlan, and LandInfra are rich in defining survey elements as presented in Table 1.In particular, LandInfra provides several entities to define these elements.However, these data models need to be customised based on jurisdictional requirements (e.g., required enumerations and attributes for the elements).In addition, they need to be enriched to fully support all underground cadastral survey elements.In the case of LADM, it is at a conceptual level and requires to be encoded in another data structure.Furthermore, LADM and ePlan do not support physical aspects.This research considers a 3D integrated model that supports not only survey elements but also a wide range of underground data components such as underground physical assets and their corresponding legal spaces and boundaries.In other words, this research aims to develop a data model for modelling survey elements, but this data model should be a part of a 3D integrated underground data model to meet the requirements of a 3D digital environment for managing and communicating the physical, legal, and survey elements in underground areas.
Previous studies described the applications and benefits of a 3D integrated model (Aien et al., 2015;Saeidian et al., 2023).Saeidian et al. (2021) considered three approaches for developing a 3D integrated data model.
The first approach is to develop a new 3D data model for all data requirements (components) which is a timeconsuming and costly approach.The second approach is to interlinkage the data models that are rich in modelling specific data component(s).For example, LandInfra is rich in survey data, LADM is a well-known conceptual standard for defining legal data, and IFC and CityGML are prominent data models for 3D modelling of physical objects at the building and city scales, respectively.However, interlinking these standards requires addressing geometric conversion and semantic interoperability (Atazadeh et al., 2017b;Saeidian et al., 2021).For example, some studies worked on the interoperability between LandInfra and LADM (Kalogianni et al., 2021;Lemmen et al., 2021;Stubkjaer et al., 2018), the integration of IFC and LADM (Oldfield et al., 2017;Ramlakhan et al., 2023), CityGML andLADM (Góźdź et al., 2014;Gürsoy Sürmeneli et al., 2022;Li et al., 2016), CityGML andIFC (Hajji et al., 2021;Rashidan et al., 2021), andLADM, IFC, andCityGML (Mi, 2019;Sun et al., 2019) in the land administration domain.In addition, jurisdictions may be reluctant to use more than one data model to store, manage, and communicate all data components since this would potentially increase the cost and time for buying, establishing, and training data models.The third approach is extending existing data models to cover all data requirements.For example, some studies extended IFC (Atazadeh, Kalantari, Rajabifard, Ho, et al., 2017;Atazadeh, Kalantari, Rajabifard, Ho, &Champion, 2017) andCityGML (Halim et al., 2021;Nega & Coors, 2022;Saeidian et al., 2023aSaeidian et al., , 2023b;;Siew et al., 2021) to create a 3D integrated data model and reported this as a viable approach to develop a 3D integrated model.Although these studies developed a 3D integrated data environment, they only considered physical and legal data without incorporating survey data elements into the developed integrated model.
Available data models such as CityGML, IFC, LADM, and LandInfra have their use cases in the land administration domain (Atazadeh et al., 2022).For instance, LandInfra focuses on infrastructures, IFC is more appropriate for use cases that require a building-scale model (e.g., condominium registration), and for use cases such as creating 3D digital property maps for an entire jurisdiction and planning and constructing large-scale tunnels and utilities, a city-scale model like CityGML is required.The selection of the approach and the data model(s) to develop a 3D  The next section describes these elements in detail.

| UNDERG ROUND C ADA S TR AL SURVE Y DATA REQU IREMENTS
Underground cadastral survey elements are defined based on the knowledge gained from investigating current practices.In several jurisdictions such as the Netherlands (Soffers, 2017) and Victoria (Atazadeh et al., 2021)  Surveyor-General Victoria, 2021Victoria, , 2022)).

| Survey points
The most important survey points are control points, reference points, traverse points, and boundary points.
All cadastral surveys need to be connected to at least two control points in Victoria (Surveyor-General Victoria, 2021).GNSS would be used to connect cadastral surveys to the datums (Surveyor-General Victoria, 2021).
However, for underground areas, this technique is not possible because of the line-of-sight barriers.Therefore, traversing needs to be done between control points and cadastral surveys through the entrances of underground structures or before burying them.Control, reference, and traverse points are used for the traversing.There are two types of control points in Victoria: permanent marks (PMs) and primary cadastral marks (PCMs).Victoria

| Survey observations
The most important survey observations are traverse, radiation, connection, and boundary.The traverse observation refers to the traversing between control, reference, and traverse points.The radiation observation connects control, reference, and traverse points to the legal space corners (LandVictoria, 2019).Consequently, the legal spaces connect to a geodetic survey network.Boundary observations are also the measurements used to define legal surveyed boundaries (for more information about surveyed boundaries see [Saeidian et al., 2023a]).Figure 3 shows some examples of traverse, radiation, and boundary observations in the AFR of the underground tunnel of Figure 2a.
All legal objects require to be connected to the adjacent lands.A primary parcel needs to be connected to the surrounding road or crown unclosed abuttal parcels.A secondary interest must also be connected to primary parcels by sharing a corner or a special observation (connection observation) from one of its corners to a primary parcel corner (LandVictoria, 2019).For more information about underground primary parcels and secondary interests (see Saeidian et al., 2023b).Figure 4 shows a few examples of connecting underground primary parcels to roads and connecting underground secondary interests to primary parcels.

| Elevation information
In addition to the survey points and observations, the elevation information is also provided in plans and AFRs.This survey information is frequently used in underground plans.Figure 6 shows some examples of the elevation information in the plan of the underground tunnel of Figure 2b.As shown, the reduced level (RL) values of the ground/site and some parts of the legal space (crown allotment) associated with the tunnel are provided.These RLs specify the elevation of certain parts of the legal space and ground/site relative to the AHD.In cross-sectional diagrams, each text specifying the elevation information defines a surface on which all points have the same elevation relative to AHD.A 3D model can define this survey information as surfaces rather than textual notations.

| THE DE VELOPED CIT YG ML E X TEN S I ON
CityGML is an OGC standard for the 3D modelling of natural and man-made objects in urban environments.This standard provides the entities for the spatio-semantic representation of urban objects and defines topologies, The last edition of CityGML (version 3.0) is introduced in UML diagrams, making it independent of any platform.Each CityGML ADE module needs to be provided in UML with its namespace (Kolbe et al., 2021).Figure 7 shows the conceptual schema of the SurveyElement module of the VicULA ADE provided in UML.The feature class ElevationSurface defines the elevation information.It should be noted that survey points also have elevation information as shown in Figure 7.However, the ElevationSurface class defines elevation surfaces as described in the previous section.This is a requirement that is very common in underground AFRs and plans and has not been well covered in the previous data models.
CityGML 3.0 provides various geometries such as primitive geometries, spatial aggregates and composites.
It uses the ISO standards such as "ISO 19107:2003 Spatial Schema" to define the geometries.These geometries The SurveyElement module of the CityGML VicULA ADE.
provided for spaces and space boundaries (the AbstractSpace and AbstractThematicSurface classes and their subclasses).However, the survey elements are not physical/logical spaces or space boundaries.Therefore, geometries are used directly for the feature classes of the SurveyElements module in order to spatially represent them.
As shown in Figure 7, the primitive geometries defined by the ISO 19107 standard such as GM_Point, GM_Curve, and GM_surface are used in the SurveyElement module.
CityGML 3.0 defines four levels of detail (LODs) for both interior and exterior objects from a highly generalised model to a highly detailed model, supporting different use cases/applications.The ADE mechanism also provides the possibility of defining LODs.CityGML 3.0 uses LODs solely for spatial representations, and not for semantics (Kolbe et al., 2021).Therefore, using the LOD concept for underground cadastral survey elements (points, lines, curves, and surfaces) is not logical.As shown in Figure 7, the defined geometric representations in the SurveyElement module do not have any specific LODs (the same approach has been used in the relief module of the CityGML CM).Therefore, they can be used regardless of LODs. It

| Y PIN G
To test the feasibility and viability of the developed data model, it needs to be implemented for an underground case study.The survey data of the case study is stored in a geography markup language (GML) file using the developed schema (the XML encoding), which is compliant with standards.This means that software tools can process this application schema and work directly with the ADE data sets, including writing, reading, visualising, and querying.The implementation uses FME Workbench and FZK Viewer to perform these tasks on the GML file derived from the 3D data model.As a 3D integrated model, this GML file is able to store: • survey elements: These elements are modelled within the proposed schema of the SurveyElement module of the VicULA ADE, which is the contribution of this study.
• underground legal spaces: These spaces are modelled within the proposed schema of the UndergroundParcel module of the VicULA ADE (Saeidian et al., 2023b).
• underground legal boundaries: These boundaries are modelled within the proposed schema of the UndergroundBoundary module of the VicULA ADE (Saeidian et al., 2023a).
• underground physical objects: These objects are modelled within the current schemas of the CityGML modules (OGC, 2021).
The schema of the SurveyElement module imports all other schemas (Figure 9).Therefore, the GML file derived from this schema is able to define underground legal spaces and boundaries and physical assets along with the survey elements.
To test the feasibility of modelling underground cadastral survey data and legal and physical objects in realworld underground areas using the proposed data model, the cadastral plan of an underground tunnel (shown in Figure 4b) and its AFR (shown in Figures 2a and 3) are used.This case study contains information about different survey points and observations.However, this case study does not have elevation surfaces.Therefore, a synthetic elevation surface is considered for this case study to test if the developed data model can define elevation surfaces.
Using the FME Workbench software tool, all survey, legal, and physical data are stored in a single GML file based on the SurveyElement module XSD that imports all other modules (XSDs) as shown in Figure 9.This tool was able to read the developed schema and write the data according to the schema, which shows that the developed schema is compliant with standards.Figure 10 shows snippets of this GML file and the information defined by the VicULA ADE (survey elements and legal spaces and boundaries) and CityGML modules (physical objects).In this and the following figures, some elements (above the lines), especially geometries, are hidden due to visualisation limitations.
The created GML file contains the geometries and attributes of different types of survey elements.Figure 11 shows this information for a survey point, a survey observation, and an elevation surface.
In the implemented prototype, all survey, legal, and physical data are converted into a single GML file containing 3D geometries, attributes, semantics, and relationships.The implemented model of the case study (the GML file) was then visualised using the FZK Viewer.The developed schema for the SurveyElements module and the implemented prototype (the GML file) were provided to this viewer.The viewer was able to read the GML file and visualise it, showing the viability and feasibility of the developed integrated model.
Figure 12 shows the visualisation of the 3D integrated underground model of the case study.In this figure, the synthetic elevation surface is not shown due to visualisation transparency.This 3D integrated model represents the tunnel and the legal space associated with it along with the survey elements.Different survey points and observations are modelled based on the AFR of the tunnel (Figures 2a and 3).Some survey points and observations are located on the ground surface to connect the legal space of the tunnel to the control point (the first point on the right).
In addition to the 3D geometries, the model also provides various attributes of the survey elements.
Figure 13 shows some survey elements and their attributes modelled, including a survey point (boundary point) located in the corner of the legal space of the tunnel (Figure 13a), a survey observation (radiation observation) connecting this boundary point to a reference point in the survey network (Figure 13b), and an elevation surface (Figure 13c).
The implemented prototype shows that the developed data model can successfully incorporate different underground cadastral survey elements in a 3D integrated model.It is able to define the geometries, attributes, and semantics of these elements.The proposed application schema and the 3D integrated model derived from it have some benefits compared to the previous data models and the current practice, which is discussed in the next section.models mostly modelling legal spaces and boundaries.A few studies enriched existing data models like BIM and LADM for modelling survey elements (Atazadeh et al., 2021;Kalogianni et al., 2021;Soffers, 2017).

| D ISCUSS I ON
However, these studies only focused on buildings and land parcels and did not investigate underground cases and requirements.Although, most survey elements are common between above-ground and underground such as survey points and observations, the importance of some survey elements for underground is observed that are less considered in the previous data models.For example, elevation information (elevation levels) is frequently provided for underground areas.Therefore, this research proposed a new element to define elevation surfaces.
The new feature class models the geometry and attributes of these surfaces in a 3D digital model instead of textual notations as shown in Figure 13c.In addition, the investigation of underground cases has revealed some challenges in the current practices, which are discussed in more detail below.
In the current practice, survey data are stored and managed in a separate document (e.g., AFR in the context of Victoria, Australia).For example, the case study of this research includes an AFR for cadastral survey elements (Figures 2a and 3) and a cadastral plan for the legal space associated with the tunnel (Figure 4b).Since it is a digital model, it is possible to define validation rules and users can interact with the model whereas in 2D AFRs, this is impossible since the survey data is provided in a static format.
The integrated 3D model also avoids providing survey data in several 2D sheets.These 2D sheets frequently refer to each other.The user needs to put them together and interpret the survey work.For example, the AFR of this research's case study has two sheets.Figure 14 shows some parts of the first sheet with several references to sheet 2. In contrast, the proposed 3D model is a single integrated model without any references to several sheets (as shown in Figure 12).For complex cases, the number of sheets and references is even greater.
The proposed 3D digital model provides an improved visualisation of survey elements compared to 2D AFRs.
As a result, interpreting and communicating this model becomes much easier.In this 3D model, the connection of underground legal spaces to a geodetic survey network is clearly visible as shown in Figure 12.In contrast, the current practice involves interpreting several 2D pages of AFRs and plans that contain various object symbology and textual notations to understand the survey work.For example, bearings, distances, and other information about observations are written as textual notations in the AFR, making the drawing very crowded and it is difficult to find relevant notations of observations.On the other hand, in the developed model (as shown in Figure 13), all information is defined as attributes and the user can retrieve this information for any element needed.Figure 15 shows another example in which the textual notation of a traverse observation clarifies that the observation has been conducted through the tunnel between the reference points on both sides of the tunnel.This textual notation makes the drawing crowded but is necessary to interpret the survey work in the 2D AFR.In the 3D integrated digital model, in addition to the survey elements, legal and physical data are also provided.Therefore, there is no need for such textual notations as shown in Figure 15.
This 3D digital model also enables queries.For example, in the current practice, if a surveyor wants to find a specific survey point or observation, all parts of the AFRs need to be searched manually to find the survey element.On the contrary, using the digital model, the user can easily find a specific survey point by executing a query based on the names/IDs.A clear model of cadastral survey elements is crucial for surveyors who want to reuse AFRs.

F I G U R E 1 4
The first sheet of the AFR for this research's case study with several references to the second sheet.
Control points are located on the ground surface, and underground legal spaces require to be connected to a geodetic network through these control points.In addition, underground legal spaces are connected to roads located on the ground surface.As underground structures are buried, such traversing and connections (surveying observations) can be done before burying the assets or a traversing needs to be performed between the points located on the ground surface and underground through the entrances of underground structures.In the AFR of this research's case study, there is no information about how the surveying has been done.Also, the elevation of reference and traverse points are not provided.Therefore, it is impossible to know which points and observations are underground, which is another limitation of current 2D practices.On contrary, in a 3D model, the user can simply distinguish this as all objects have 3D coordinate values (x,y,z).However, it is necessary to capture the elevation of the points.In this study, we assumed that all points and observations are located on the ground surface, except the points on both sides of the tunnel and the observations between them, the tunnel, and the legal space of the tunnel (Figure 16).
The implemented prototype proves that the developed data model can successfully define the survey elements like points, observations, and elevation surfaces in a 3D environment.It shows the viability and feasibility of the model from the data modelling perspective.However, some challenges were faced during implementation.
The implementation of elevation surfaces (creating these surfaces) can be challenging for a large-scale area considering the earth's curvature.In this study, since the case study was small, the synthetic elevation surface was modelled as a flat surface without considering this parameter.
This research focused on modelling the spatial elements and their semantics and attributes in the final document of a survey work (AFRs).Therefore, surveying processes were not investigated.For example, surveyors may measure a survey element several times or do a traversing between a control point on the ground surface and a point in an underground structure, calculate the direct bearing and distance between them, and provide this direct bearing and distance in the AFR (as a computed bearing and distance), not the original traversing.Another example is Figure 4c where an underground primary parcel (underground lot) is connected to a road located on the ground surface.There is no information about these processes in AFRs.It can be investigated in detail in future studies.In this regard, it is necessary to investigate cadastral surveying processes for underground areas.It is also necessary to investigate whether such information requires to be modelled based on the current regulations and possible applications of the model.

| CON CLUS IONS
The development of a 3D data model is a fundamental step towards 3D digital land administration of underground areas.This model requires covering different types of underground data including cadastral survey elements as well as physical and legal datasets.In our previous studies (Saeidian et al., 2023a(Saeidian et al., , 2023b)), the CityGML VicULA ADE was developed to integrate the physical data of underground assets with the legal spaces and boundaries.
In the current practice of Victoria, it is also necessary to provide survey information along with legal spaces and boundaries.Therefore, to enrich the CityGML VicULA ADE with survey information, this research developed the SurveyElement module.
The SurveyElement module can import other VicULA ADE modules and CityGML modules in order to integrate underground survey, physical, and legal data in a 3D digital environment.To test and demonstrate the feasibility and viability of the CityGML VicULA ADE, a prototype was implemented for an underground tunnel.
This prototype showed that the proposed extension has the capability to model various underground cadastral survey elements including different types of 3D survey points, observations, and elevation information in a realworld context.
The approach proposed in this study can be practical for augmenting CityGML with survey data.The developed data model can integrate all 3D ULA objects such as different types of underground cadastral survey elements and underground legal and physical objects homogeneously in terms of geometry and semantics.The proposed methodology was applied to Victoria to showcase its feasibility and benefits.However, this methodology can be replicated in other jurisdictions by adjusting the jurisdiction-specific data requirements in underground cadastral surveys.It should be noted that creating an integrated 3D model for ULA requires addressing some challenges such as data availability, differences in vertical and/or horizontal datums, and differences in formats.In addition to these technical challenges, institutional and legal barriers may be significant obstacles to utilising the proposed extension.

F I G U R E 1 6
The location of survey points.
CityGML-based integrated data models.Thus, a new study is required to investigate the possibility of modelling survey data elements in CityGML.This research aims to address this knowledge gap by investigating and extending the CityGML standard in order to incorporate underground cadastral survey data into a 3D integrated environment.As shown in Figure 1, this study adopts a case study driven research methodology.The first step is selecting a case study.The cadastral survey highly depends on the regulations and varies in every jurisdiction.This study focuses on the Victorian jurisdiction of Australia.Underground cadastral survey plans in this jurisdiction are investigated to extract survey elements and identify requirements.In the second step, the conceptual model of CityGML 3.0 is extended based on the requirements.The result is a conceptual ADE for CityGML 3.0 provided in a Unified Modelling Language (UML) diagram to model underground cadastral survey elements.This conceptual ADE is then encoded in Extensible Markup Language (XML) in step 3, followed by implementing it for a real-world underground area.
To develop an integrated city-scale model by adopting the third approach (enriching a base data model to support ULA data components),Saeidian et al. (2023aSaeidian et al. ( , 2023b) ) proposed a new ADE for CityGML 3.0 to represent underground physical assets and legal spaces and boundaries.As discussed in the previous section, in the current practices in different jurisdictions, survey elements also need to be provided along with the legal spaces and boundaries (for example as the Abstract of Field Records (AFR) file in Victoria).These elements are fundamental for connecting underground legal spaces to a geodetic survey network.This research aims to identify different types of cadastral survey elements in underground survey documents and enrich the CityGML 3.0 data model to support these elements and represent them along with underground legal spaces and boundaries and physical objects in a 3D integrated model.This development also benefits the CityGML standard since no study explored the potential of this standard for modelling survey data and developed an ADE in this context.The developed ADE can expand the functionality of CityGML for use cases that require survey data in a 3D city model.In order to accomplish this aim, it is necessary to identify underground cadastral survey elements in the current practice.
Survey points have two/three coordinate values (horizontal and/or vertical coordinates).These points are coordinated relative to the Australian national datums.Horizontal coordinate values are based on the Map Grid of Australia (MGA) which is the Universal Transverse Mercator (UTM) projection of the Geocentric Datum of Australia (GDA), and vertical coordinates (elevations) are based on the Australian Height Datum (AHD) which is the datum of mean sea level as determined by the National Levelling Adjustment (Chief Parliamentary Counsel, 2018; has established a network of survey control points that covers the whole state, known as PMs.These control points have 3D coordinate values (horizontal coordinates and elevation) aligned to the national datums(Surveyor- General Victoria, 2022).PCMs are also permanent survey marks that can be connected to during a cadastral survey to meet the requirements of regulations(Surveyor-General Victoria, 2021).Some points of the traversing are important for the surveyor (reference points) and the rest points are traverse points.Finally, boundary points are used to specify legal boundaries.Survey points can be provided in both AFRs and plans.Figure2shows some survey points in an AFR and a cadastral plan of underground tunnels.In the field, survey points can be marked by different monuments (e.g., peg, plaque, and survey nail) specified in the AFRs and plans.Survey points can have different attributes including name, description, point type, point state, monument type, horizontal datum, vertical datum, easting, northing, and elevation.The point type defines the type of point including control points (PM or PCM), reference points, traverse points, and boundary points.The point state can also be existing, proposed, and destroyed.
Traverse and radiation observations are straight lines defined by bearing and distance measurements.However, boundary and connection observations can be either straight lines or arcs.For underground cases, arcs are very common, especially for underground tunnels and utility easements (Figure5).The name, observation type, and start and end points (their IDs/names) are the expected attributes for line and arc survey observations.F I G U R E 2Examples of survey points: (a) the AFR of an underground tunnel; and (b) the cadastral plan of an underground tunnel.The observation type can be traverse, radiation, boundary, and connection.In addition to these attributes, other expected attribute values for lines are bearing, distance, and the types of bearing and distance.Other expected attributes for arcs are also the bearing and length of the chord, the radius, length and type of the arc, and the direction of rotation from the start to the end (rot).The distance, bearing, and arc types can be adopt dimension, computed, derived, or measured.F I G U R E 3 Examples of traverse, radiation, and boundary observations in the AFR of the underground tunnel of Figure 2a.F I G U R E 4 Examples of connecting underground legal objects: (a) connecting an underground secondary interest (easement) to a primary parcel (lot); (b) connecting an underground primary parcel (the underground crown allotment of Figure 2a) to roads (the start and end points of the straight boundary lines highlighted are on the roads); and (c) connecting an underground primary parcel (lot) to a road.
attributes, and appearances.There are different kinds of thematic modules (e.g., Building, Tunnel, Construction, Transportation, and LandUse) in the CityGML Conceptual Model (CityGML CM) as well as the Core module that is mandatory to be implemented for any application(Kolbe et al., 2021).However, CityGML does not have any specific module for modelling underground cadastral information including underground legal spaces and boundaries as well as associated survey data elements.CityGML provides the ADE mechanism for enriching this 3D model according to the application-specific needs.Saeidian et al. (2023aSaeidian et al. ( , 2023b) ) developed an ADE for CityGML 3.0 named VicULA ADE to represent underground legal spaces and boundaries in Victoria.In these studies, two modules (UndergroundParcel and Un-dergroundBoundary) have been developed for modelling underground legal spaces and boundaries.This study enriches the VicULA ADE by developing another module named SurveyElement to model underground cadastral survey elements described in the previous section.

F
I G U R E 5 Examples of arcs: (a) legal spaces of utilities (easements); and (b) the legal space of a tunnel (crown allotment).The AbstractSurveyElement class is the top class of the module with some subclasses for ULA survey points, observations, and elevation information.It is a subclass of the AbstractFeature class of the CityGML Core module.Therefore, the AbstractSurveyElement class and its subclasses (all survey elements) inherit the properties of the AbstractFeature class such as the "description" and "name" attributes.The SurveyElement module also has four F I G U R E 6 Examples of the elevation information in the cadastral plan of the underground tunnel of Figure 2b.enumerations for the type and state of points, the type of observations, and the type of bearings, distances, and arcs according to the requirements listed in the previous section.As described in the previous section, survey observations can be lines or arcs.The line and arc observations have some similar and some different properties as listed in the previous section.Therefore, two feature classes have been created for them to model different attributes.These classes are the subclass of an abstract class named SurveyObservation.This abstract class defines similar properties including attributes and the relationship between survey observations and points.As mentioned in the previous section, only boundary and connection observations can be arcs.Therefore, an object of the ArcObservation feature class can only be created if the observation type is a boundary or connection.As shown in Figure 7, the Object Constraint Language (OCL) has been used to apply this constraint on the ArcObservation feature class.
is necessary to develop an encoding for the UML conceptual model of anyADE (Kolbe et al., 2021).Similar to the conceptual data model of the latest version of CityGML (version 3.0), the conceptual data model of the developed ADE is created using UML diagrams that can be implemented using any database or file-based schemas (encodings) based on available software tools.This research developed an XML schema for the developed ADE; however, the developed data model is not limited to the XML encoding.The XML schemas of all CityGML modules (OGC, 2021) and the UndergroundParcel(Saeidian et al., 2023b) and UndergroundBoundary(Saeidian et al., 2023a) modules of the VicULA ADE are available.Therefore, this study developed another XML schema for the SurveyElement module as the new module of the VicULA ADE.Figure8shows some parts of the developed XML encoding for the SurveyElement module derived from the UML conceptual model shown in Figure7.F I G U R E 8 Some parts of the developed XML encoding (XSD) for the SurveyElement module.
This research developed an integrated 3D underground data model by proposing a new CityGML ADE at the conceptual (UML) and encoding (XML schema) levels to model underground cadastral survey data.Previous data F I G U R E 9 The use of VicULA ADE and CityGML modules in the 3D underground integrated model: (a) at the conceptual level; and (b) at the XML encoding level (importing all schemas of the VicULA ADE modules, CityGML modules, and GML by the SurveyElement module XSD).
However, legal    and survey data are strongly linked to each other since the legal boundaries are delineated based on the survey measurements.Therefore, in many cases, such as reconstruction of legal boundaries, subdivisions, and consolidations, surveyors must check the integrity of both legal and survey data by accessing siloes of documents or data repositories to confirm validity and reusability of these datasets.In other words, they must consolidate all survey and legal information from these documents or repositories in order to interpret them accurately.Combining all survey and legal data from various files is a cognitively complex challenge.On the other hand, a 3D integrated model can provide a coordinated representation and management of survey and cadastral datasets.The 3D digital model developed in this study provides the capability to integrate survey data elements with underground legal F I G U R E 1 0 Snippets of the GML file of the case study derived from the SurveyElement module XSD that imports all other modules (XSDs).spacesand boundaries in a common 3D data environment, which facilitates the interpretation and communication of underground cadastral survey data (see Figure12).In a 3D integrated digital model, survey measurements and legal spaces and boundaries can easily be checked by surveyors to identify possible inconsistencies.F I G U R E 11 Some survey elements and their properties in the GML file of the case study.F I G U R E 1 2 The visualisation of the 3D integrated underground model of the case study that has survey elements (survey points shown in cyan and survey observations shown in red), underground legal space and boundaries, and physical assets.F I U R E 1 3 Some underground cadastral survey elements and their attributes; (a) a survey point; (b) a survey observation; and (c) an elevation surface.

F
Communicating the survey data of a traverse line in 2D AFR by the 3D integrated model and CityGML 3.0.