Extending CityGML for IFC-sourced 3D city models

Abstract

Differences in the scope and intent of the contrasting IFC and CityGML data formats entail that converting the former to the latter results in loss of information. However, for some use cases it is beneficial to keep also particular information from IFC that is not native to CityGML, and achieving that requires mechanisms such as the CityGML Application Domain Extension (ADE). We develop an ADE to support retaining relevant information from IFC. Besides being driven by the particular source of the input data (IFC), this multipurpose ADE is shaped after a discovery process that involved examining potentially applicable use cases in Singapore, doubling as an extension that is adapted to a set of use cases and the local geographic context. We implement the conceptual work by generating an enriched dataset (with an automatic conversion from IFC to CityGML), visualising it, and discuss its added value in a use case.

Full text

Extending CityGML for IFC-sourced 3D city models
Filip Biljeckia,∗,Joie Lima,James Crawfordb,Diana Morarub,Helga Tauschera,Amol Kondea,
Kamel Adouanea,Simon Lawrenceb,Patrick Janssenaand Rudi Stouffsa
aNational University of Singapore, Singapore
bOrdnance Survey, United Kingdom
ARTICLE INFO
Keywords:
CityGML
IFC
ADE
BIM
3D city model
Singapore
ABSTRACT
Differences in the scope and intent of the contrasting IFC and CityGML data formats entail that con-
verting the former to the latter results in loss of information. However, for some use cases it is ben-
eficial to keep also particular information from IFC that is not native to CityGML, and achieving
that requires mechanisms such as the CityGML Application Domain Extension (ADE). We develop
an ADE to support retaining relevant information from IFC. Besides being driven by the particu-
lar source of the input data (IFC), this multi-purpose ADE is shaped after a discovery process that
involved examining potentially applicable use cases in Singapore, doubling as an extension that is
adapted to a set of use cases and the local geographic context. We implement the conceptual work by
generating an enriched dataset (with an automatic conversion from IFC to CityGML), visualising it,
and discuss its added value in a use case.
This is the Accepted Manuscript version of an article published by Elsevier in the journal Automation in Construction in 2021, which is available at:
https://doi.org/10.1016/j.autcon.2020.103440. Cite as: Biljecki F, Lim J, Crawford J, Moraru D, Tauscher H, Konde A, Adouane K,
Lawrence S, Janssen P, Stouffs R (2021): Extending CityGML for IFC-sourced 3D city models. Automation in Construction, 121: 103440.
©2021, Elsevier. Licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International (http://creativecommons.org/licenses/by-nc- nd/4.0/)
1. Introduction
The conversion of detailed architectural models stored
in IFC to semantic 3D city models in CityGML is a topical
subject [1–3]. Because CityGML is a data model designed
for the geospatial world, much information during the con-
version from IFC is inherently lost, such as energy-related
features [4,5]. Such loss of information is not necessarily
a disadvantage because usually there is no need to generate
an equally detailed counterpart in the geospatial domain, as
use cases — especially those focusing on the urban scale and
covering multiple buildings — would not have much benefit
or would even suffer from excessively rich datasets [6].
Nevertheless, in certain situations such as energy simula-
tions and indoor navigation, it is beneficial to preserve a sub-
set of information that is not possible to store with the stan-
dard data model of CityGML. For extensibility, CityGML
provides the Application Domain Extension (ADE), a mech-
anism to extend the standard data model, which among other
advantages provides means to absorb rich information from
IFC.
This paper stems from a research effort on the conver-
sion of IFC to CityGML to integrate data originating from
the architectural and construction domain in the geospatial
environment serving different stakeholders in the govern-
ment sector in simulations and analyses across multiple dis-
ciplines. In our work we develop an ADE that facilitates
the conversion from IFC to CityGML. This topic is impor-
tant as while the idea of bridging the two disparate domains
is gaining currency in academia and industry, there are still
∗Corresponding author
filip@nus.edu.sg (F. Biljecki)
ORCID(s): 0000-0002-6229-7749 (F. Biljecki);
0000-0003-0411-8770 (H. Tauscher); 0000-0002-9255-8006
(A. Konde); 0000-0002-2013-7122 (P. Janssen);
0000-0002-4200-5833 (R. Stouffs)
many unanswered questions, especially related to data for-
mats. We contribute to the body of knowledge by present-
ing an ADE that leverages on the comprehensive information
provided by the IFC schema but balances usability and sim-
plicity, and is influenced by practices pertaining to a local
geography and users.
We investigate multiple aspects related to this subject.
Most importantly, and in a generalised way facilitating repli-
cation, we tackle the research question: how to design a
specification for 3D city models that takes advantage of the
rich architectural source, and at the same time adheres to lo-
cal practices and supports a selection of use cases?
We also wrote this paper to serve as a guide when de-
signing an ADE and specifications in similar projects, i.e.
we cover the entire and extended workflow of designing a
data specification to its implementation in a dataset and soft-
ware, which is uncommon in literature: Section 2gives back-
ground information, describes the ADE mechanism, and re-
lated work; Section 3introduces the workflow we followed
and explains the rationale behind certain choices we made.
Section 4describes the multi-purpose localised (application-
driven and context-driven) ADE we have designed, named
IfcADE, along with more details about the selected use cases
and their identified properties. In Section 5we implement
the specification: we convert an IFC dataset to a CityGML
ADE-enabled counterpart that conforms to the presented spec-
ification and visualise it in two ways (desktop and web) por-
traying CityGML information beyond the standard model,
and discuss the application in a use case: we give a demo
of an indoor navigation instance that takes advantage of the
availability of the features stored according to the ADE. Sec-
tion 6discusses the limitations of the work, caveats, and
points of improvement in future.
Our work focuses on Singapore and three use cases (with
indoor navigation being the focal point in the implementa-
tion and examples), and besides providing value in giving an
insight into this development, we believe that from the scien-
Biljecki F et al.: Preprint submitted to Elsevier Page 1 of 18

Extending CityGML for IFC-sourced 3D city models
(a) Input IFC model. (b) Output CityGML models (two variants).
Figure 1: Example of a conversion from IFC to CityGML within our project (following the methodology presented across multiple
recent papers [4,19–22]). Source of the architectural dataset: Building and Construction Authority (Singapore).
tific point of view our work is sufficiently generic that it can
be scaled to other use cases and geographies. Besides mixing
both local and use case contexts, other contributions of our
holistic paper involve enhancing related work and covering
the entire development life cycle — from design to imple-
mentation, visualisation, and utilisation in a use case, topics
that are usually not covered.
2. Background
2.1. Conversion from IFC to CityGML
The two prominent data formats used in architecture, en-
gineering and construction, and the geospatial world, respec-
tively, are the buildingSMART standard Industry Founda-
tion Classes (IFC) [7,8] and the Open Geospatial Consor-
tium (OGC) standard CityGML [9,10].
Some of the benefits of bridging IFC and CityGML (and
in general BIM and GIS) (Figure 1) involve taking advantage
of GIS software/use cases that cannot be carried out with
BIM datasets, integration of multiple sources of data (e.g.
visualisation of architectural models in a geographic envi-
ronment, and carrying out use cases that require data on mul-
tiple scales), leveraging more detailed data (both geometri-
cally and semantically) in the geospatial domain, bypassing
often expensive and tedious aerial and ground surveys (ben-
efiting maintenance of data), supporting complex analyses
of the data such as environmental and planning analysis, as
well as enabling carrying out spatial analyses on buildings
that are yet to be constructed [11–13]. There is a large num-
ber of use cases requiring indoor geometry and rich seman-
tics, e.g. 3D cadastre [14], illuminance analyses [15], and
routing [16–18], and as such they may benefit from 3D city
models that are sourced from IFC.
Bridging these two disparate worlds does not only in-
volve considering different data formats, but also different
mindsets, use cases, stakeholders, and dealing with differ-
ent lineages of data, translating the architectural view of the
world to the geospatial [23,24]. Besides ‘slimming down’
an IFC model from the spatio-semantic perspective (i.e. gen-
eralisation to obtain lightweight 3D city models facilitating
use cases at the urban scale), the process of the conversion
also involves adapting both the semantics and geometry to
a different structure (e.g. translating the geometric represen-
tations and semantic classes). While there may be alterna-
tive approaches to BIM-GIS interoperability, such as keep-
ing separate files with establishing links between them or
having a joint database, the generally accepted technique is
the conversion from one format to another, predominantly
in the geospatial direction. Separate CityGML and IFC files
would add a layer of complexity, as one would need to de-
velop specific translation programs for specific applications.
Many projects have been carried out in this domain and
have been extensively described [5,25–35]. This paper is
part of the project on the conversion from IFC to CityGML
within the frame of Virtual Singapore, being preliminarily
described in [4,19,20,36]. In this paper, we follow up on
the published work by focusing on one of the pillars of the
project: the extension of the CityGML data model comple-
menting the conversion and fitting use cases requirements
and local context.
2.2. CityGML ADE
CityGML provides a generic domain-independent data
model for the spatio-semantic modelling of topographic fea-
tures in 3D in the application area of cities and landscapes.
At the expense of keeping things simple, it might be consid-
ered limited and might not suit a large number of use cases
and situations. For this reason, the CityGML data model is
often extended using the ADE mechanism. An ADE enables
augmenting the CityGML data model, extending it beyond
its scope benefiting use cases and particular scenarios. A
particular ADE can be specified with an XML schema def-
inition file (XSD) and/or with Unified Modeling Language
(UML) [37–39].
In the context of this research an ADE allows capturing
additional information on top of the native CityGML data
model, eventually enabling to keep a set of information we
are interested in (Figure 2), i.e. it generally allows:
Biljecki F et al.: Preprint submitted to Elsevier Page 2 of 18

Extending CityGML for IFC-sourced 3D city models
•Introducing new classes/features. For example, mod-
elling solar panels as features with their own geometry
(e.g. extent of the panels) and attributes (e.g. photo-
voltaic capacity).
•Adding new attributes to existing classes/features. For
example, including the number of units in a build-
ing, an attribute that does not exist in the standard
CityGML data model.
•Introducing/extending code lists of attributes. For ex-
ample, adding new building types that do not fit in
the existing code list provided by the CityGML data
model. These enable modelling freedom and customi-
sation of attributes, e.g. for aligning them to a national
standard.
•Introducing non-standard geometries to existing class-
es/features. For example, the Noise ADE enables mod-
elling noise barriers as lines [9,40].
Many ADEs have been developed around the world for a
variety of purposes [37]. For example, these include ADEs
developed for a taxation use case [41], enabling metadata
according to international standards [42], adapting the data
model to store ancient Chinese roofs [43], extending CityGML
for a national context developing a national 3D standard [44],
improving the storage of terrain in CityGML [45], and facil-
itating the integration with other data models [46]. These
few examples indicate the diversity of purposes ADEs are
developed for.
Following the examination of the capabilities of the ADE
concept, considering different options, and the literature re-
view, we deem that developing an ADE is a convincing choice
in facilitating the conversion from IFC to CityGML and pre-
serving information valuable for applications. However, most
of the related work focuses on one aspect, such as facili-
tating a single use case. In our paper, we go beyond that,
as we combine multiple aspects developing an ADE that is
(i) shaped after an input dataset (we examine the IFC data
model and translate subsets that may be deemed relevant
in the geospatial domain), (ii) serving specific use case(s)
(given that we need to preserve features that are usually not
available in the geospatial world but might be useful), and
(iii) regarding a national context (i.e. local architecture, prac-
tices, and stakeholders).
Some disadvantages of the ADE mechanism are that in
each domain there may be a specialised data format that is
more suitable for each respective purpose and that most of
the existing software does not support working with ADE-
enabled datasets out-of-the-box. Those would have to take
into account either a particular extended schema, which is
not always an easy task [47], or the extension mechanisms
in general, which is even more difficult. We did overcome
some of these disadvantages with certain modelling choices
and by implementing the work (Section 5).
An ADE is technically part of CityGML, but for the sake
of clarity in this paper we refer simply to the basic CityGML
data model as CityGML and everything else as extension-
s/ADE. For more information on the ADE concept the reader
is referred to the review paper [37], which includes 44 exam-
ples of ADEs, modelling guides [39], and the latest adopted
version of the CityGML standard [9].
2.3. Singapore
As it is the case with many other geographies, Singa-
pore has certain particularities that warrant the development
of an extension to CityGML. For example, localisations of
CityGML have been made in the Netherlands, Sweden, and
Turkey to adapt to the local context and to align to existing
government practices [44,48,49].
Examples of distinctions from other countries usually
prominent in the 3D city modelling world are that Singapore
is located in Southeast Asia (i.e. CityGML has been devel-
oped by mostly European stakeholders inevitably conform-
ing to their geographies and architecture), it has a tropical
climate (having an impact on the selection and importance
of use cases, and consequently possibly a set of required fea-
tures that in other geographies are not acquired routinely),
and it is a city-state (the different organisation of the govern-
ment might reflect a set of GIS users and use cases different
from other countries). An example of an architectural char-
acteristic in Singapore is the presence of void decks, a pub-
licly accessible sheltered space on the ground floor of res-
idential buildings playing an important social and environ-
mental role [50], a feature that might be beneficial to model,
but is not provisioned for in CityGML.
Urban digital twinning and simulation in Singapore are
largely carried out under the auspices of the Virtual Sin-
gapore programme spanning multiple research projects and
have been described in several research papers focusing on
improving acquisition, modelling, and usability of 3D city
models in the local context [51,4,20,52–54]. CityGML
has already been subject to extensions to enhance the model
adapting to the local context, e.g. the CityGML Vegetation
module accommodating additional properties about trees that
are characteristic to the government agencies in Singapore
involved in managing and using such information [52].
Finally, it is relevant to note that the Government sup-
ports the adoption and implementation of BIM. It mandates
the submission of BIM data for new constructions for regu-
latory approval [55], potentially providing a valuable source
for acquiring and maintaining 3D city models on a large
scale.
2.4. Overview of existing ADEs serving the
conversion from IFC
While several papers present methodologies to convert
IFC to CityGML (Section 2.1), only a few develop ADEs
to conserve rich information [26,25,34,31], and not all of
them are focused on buildings.
De Laat and van Berlo [25] introduced the GeoBIM ex-
tension, enabling the preservation of IFC semantics and prop-
erties in CityGML without a particular use case in focus.
About 10% of features from IFC have been considered im-
portant for general geospatial applications, and those have
Biljecki F et al.: Preprint submitted to Elsevier Page 3 of 18

Extending CityGML for IFC-sourced 3D city models
IFC
CityGML
ADE
Subset of IFC that can be directly
converted to CityGML
Subset of IFC that is not
considered relevant for the
geospatial context and therefore
it can be disregarded
Subset of IFC that should be
converted but it is not natively
supported by CityGML,
requiring an extension
Subset of CityGML that cannot be
sourced from IFC and/or it is
populated from other sources
Figure 2: The motivation for an ADE in the context of the IFC to CityGML conversion. Given the substantial difference between
the BIM and GIS domains, and their corresponding flagship formats IFC and CityGML, only subsets of IFC are relevant for the
conversion to CityGML (the shaded lines indicate conceptually the correspondence among the two formats and an extension,
and suggest the relation between different bits of data involved in workflows such as ours). However, not all are possible to be
‘captured’ by CityGML by default. Therefore an extension has to be created.
been included in the ADE. An example is the introduction of
the attribute OverallWidth to the CityGML Door fea-
ture, mapped from the IFC class ifcDoor.
The work of Kutzner et al. [56] focuses on extending
CityGML to accommodate utility networks, developing the
Utility Network ADE. In their early work [31], researchers
investigated sourcing relevant features from BIM. A rele-
vant conclusion is that a direct 1:1 mapping between IFC
and Utility Network ADE is feasible in most cases. Another
ADE not focused on buildings is the PANTURA ADE [34],
concentrating on bridges and with a particular use case in
mind (analysis of the disturbance induced by bridge con-
struction).
The Semantic City Model ADE has been introduced by
Deng et al. [26]. The key novelties of their work is the sup-
port for multiple levels of detail, and modelling the topolog-
ical relations between different features.
While these developments are valuable and provide a
solid basis for our work, and with having in mind avoiding
the duplication of existing work in the scientific community,
nevertheless, we have decided to develop a new and inde-
pendent ADE for multiple reasons: previous ones have been
developed forolder versions of CityGML, they do not adhere
to our local context and our use cases, and an independent
ADE allows for flexibility, freedom, and tailoring the data
model without compromises.
3. Methodology and considerations
In this section we describe our methodology of devel-
oping data specifications for CityGML that takes advantage
of the rich IFC information model, while being tailored to
local practices and enabling a selection of use cases, which
were chosen in consultation with the project stakeholders.
We describe considerations for the design (Section 3.1), the
discovery of requirements (Section 3.2), and the rationale
behind the selection of features (Section 3.3).
3.1. Designing an ADE and basic principles
As it is the case with designing specifications, designing
an ADE is not an exact science, and it is subject to choices
that may be deemed subjective. This is especially the case
when envisioning use case scenarios, composed of a mix of
assumptions. Different users may also have different priori-
ties even if they are part of the same use case [57].
First of all, because we are dealing with a multiplicity
of purposes (e.g. more than one use case), a question that
arises is whether it would be beneficial to design multiple
extensions or a single one. Therefore, we identify three ap-
proaches for the design of an ADE (Figure 3):
Approach 1: Involves developing independent ADE data
models for each use case. Therefore, each of the developed
ADEs will have their data model structure with customised
and new data packages, classes, and attributes. Thus, they
will inevitably have an overlap, since multiple use cases may
require the same subset of information. Most ADEs nowa-
days belong to this category as they are developed to suit a
particular use case and domain, e.g. EnergyADE [58] and
Cultural Heritage ADE [59].
Approach 2: A single ADE data model accommodating
the data requirements for all use cases into a single struc-
ture. There are ADEs catering to multiple use cases, such
as Dynamizer ADE [60], and they are largely found in the
Biljecki F et al.: Preprint submitted to Elsevier Page 4 of 18

Extending CityGML for IFC-sourced 3D city models
Approach 1
CityGML CityGML CityGML
Core ADE
Single ADE
A base ADE with
single-purpose extensions
Independent single-purpose
ADEs
Approach 2 Approach 3
One multi-purpose ADE
A B C
A B C
Figure 3: Three different options in designing an ADE that serves multiple purposes, i.e. use cases A, B, and C.
domain of national geographic information standards (e.g.
[44,61]).
Approach 3: An ADE data model with a shared core struc-
ture to which there are attached additional ADEs, i.e. new
data packages or classes that are representative for each of
the use cases (e.g. new data package with new classes/at-
tribution for use case A, new data package with new class-
es/attribution for use case B and so on). This approach is in
a way related to the cross-domain building models discus-
sion by Knoth et al. [62], and it is akin to developing ADEs
extending existing ADEs [63,64].
In this work we choose the second option for easier main-
tenance and because it turned out (as it will become evident
in the next section) that the resulting data model is not so
large to have the necessity of partitioning it among multi-
ple entities, and because there is an overlap between require-
ments of use cases (which is another advantage of our ADE
as it provides support for multiple uses while not growing
in complexity). Another advantage of developing a single
ADE is that it benefits software developers since they would
not need to consider multiple data models.
Even though this single extension provides features that
are necessary for one use case but are irrelevant for another,
it is important to note that while collecting the data it is not
mandatory to populate all features in all circumstances (i.e.
the same way CityGML supports storing roads and vege-
tation, that does not mean that all datasets have to contain
them).
3.2. Discovery of use cases and local practices
We have analysed the content of IFC, i.e. what poten-
tially useful information IFC can provide that are not in-
cluded by default in CityGML. While this is partially a sim-
ilar process and starting point as related work (Section 2.4),
the quality of this analysis was improved significantly through
conducting discovery sessions focusing on use cases with
geospatial information domain experts and practitioners —
for example, from urban planning and design, energy mod-
elling, and land administration. Although in some cases,
the intended information encoding/format these practition-
ers work with may not be CityGML, the discovery process
provides the research opportunity to explore in-depth the par-
ticularities, relative importance, and nuances between infor-
mation requirements from each of these geographic contexts
(examples follow).
It is worth mentioning that the discovery process con-
ducted in this research has been partly influenced by a sepa-
rate 3 year discovery and learning process, led by geospatial
domain experts from the UK’s national mapping agency —
Ordnance Survey. This process, carried out between 2015
and 2018, analysed information use and information require-
ments across practitioners and domain experts from more
than 60 organisations across the UK’s public sector, to cre-
ate an evidence based, simplified classification framework
for geospatial use cases. This research identified clear cus-
tomer centric user stories focused around current and future
user needs in the geospatial industry and beyond.
The outcomes from these separate pieces of work are
complementary: a simple classification framework for or-
ganising the information requirements we identify from geospa-
tial use cases; whereas the research adds how these require-
ments can be fulfilled using IFC as a potential information
source. The relevant information from both have been com-
bined in the design of the ADE, which is generic enough to
be applied to different geographic contexts.
3.3. Selection of features/attributes
Selecting relevant features to extend CityGML with is
the central point of the work. While we do not particularly
distinguish between the different features in the ADE, this
topic is important to keep in mind, especially for the im-
plementation of sourcing procedures, that is, the conversion
from IFC to CityGML. These are shaped after the three as-
pects mentioned in the previous section. On one dimension,
we have:
0. Features that can be disregarded because they do not
need to be translated into CityGML in our case, e.g.
furniture.
1. Features that can be accommodated in the CityGML
standard data model by default, e.g. walls.
Biljecki F et al.: Preprint submitted to Elsevier Page 5 of 18

Extending CityGML for IFC-sourced 3D city models
CityGML
ADE
Direct (1:1) Processing
IFC
Relevant subset Out of scope
1a 1b
2a 2b
0
Other sources
3
Figure 4: Different groups of features populating the CityGML
data model from IFC, and a part from other sources. The
annotations are explained in the text.
2. Features that require an extension in order to be pre-
served, e.g. elevators, since they are not explicitly avail-
able in the standard CityGML data model.
On another dimension, there are:
(a) Features that are conceptually available in IFC so they
can be translated in a direct 1:1 (as-is) mapping from
IFC to CityGML or to the ADE. This includes attributes
such as building function (available in CityGML) and
information about the material of a wall surface (it
requires an extension to CityGML, but once such is
available its sourcing should not be complicated).
(b) Features that can be derived after a degree of process-
ing (or manually) because they are not directly pro-
vided by IFC, but are obtainable from the existing set
of information in the IFC. For example, the number of
storeys above ground and the number and location of
access points for a room are not available explicitly in
IFC, but could be derived.
The combination of these dimensions and aspects gives
four principal groups of features, and in our work we have
encountered instances of all of them: (1a) features that are
directly mapped from the IFC model to the core CityGML
model; (1b) features that are mapped after bespoke/addi-
tional processing during conversion into the core CityGML
model; (2a) features that are directly mapped from IFC to
the ADE model; and (2b) features that are mapped after ad-
ditional processing during conversion into the ADE. Figure 4
conceptually illustrates the categories (and extends Figure 2),
while Figure 5gives a practical example. These are impor-
tant to keep in mind for the implementation of the conver-
sion.
During our work, we realised that while IFC provides the
vast majority of information required for use cases, a small
number of features/attributes required for the use cases can-
not be sourced from IFC because it is either not available in
the schema or it is rarely present in datasets, and will thus
require supplementing it with additional data sources (cat-
egory 3). For example, energy demand estimations would
well benefit from data from IFC. However, it may require
additional information such as the number of occupants of a
building and the surrounding vegetation, which are usually
sourced elsewhere.
4. IfcADE
In this section, we present the ADE we created, which we
name IfcADE, and present its genesis and motivation for the
key design decisions. For the IfcADE, we enhance features
from the CityGML 3.0 Building and Construction modules.
The version 3.0 of the standard is not yet adopted. Thus,
we rely on recently available proposals [66–68] and the con-
ceptual model available on Github 1. Even though the new
version of the standard is yet to be finalised and adopted, we
have opted to focus on this version since it provides some
advantages over the version 2.0 adopted in 2012, such as an
improved consideration for indoor features (e.g. introduc-
tion of storeys and multiple levels of detail), and it will likely
be adopted in the very near future [68]. Since we do not ex-
pect the final version of the standard to introduce breaking
changes, and instead of continuously aligning the ADE to
each new update of the data model, we plan to update the
developed ADE after the final version of the CityGML 3.0
data model is adopted.
4.1. Description of use cases
We elaborate three use case scenarios with varying in-
formation requirements according to their geospatial context
and application. We intend to demonstrate how the combina-
tion of different subsets from the standard CityGML model
and the IfcADE offers the advantage for including enhanced
semantic properties extracted from IFC — so that this can
be replicated by others to build data models for use in other
geospatial use cases. Singapore is often the subject of vari-
ous related spatial analyses in research [69,70], thus, as long
as it does not compromise the current design, the ADE en-
ables future enhancements that support new use cases. Fur-
thermore, the data model also allows to regard additional
IFC features that have not been included in this version, but
in future may turn out to be worth including.
While a focus of the paper is also on adapting the data
model to the local context, it is intertwined together with the
use cases, and thus a separate discussion is not possible. For
example, when considering use cases, we added features that
are arising from the local context, e.g. code lists of building
types in Singapore and the presence of void decks.
4.1.1. Energy
Energy analyses are a topical subject in 3D GIS [71–
74], for example — estimating the energy demand for heat-
ing buildings and analysing energy-efficient refurbishment
of buildings on a district level. Researching cooling behaviour
of households is a particularly important topic in research in
the city-state [75].
Therefore we have extended CityGML to provide infor-
mation to carry out general energy analyses. For example,
1https://github.com/opengeospatial/CityGML-3.
0CM
Biljecki F et al.: Preprint submitted to Elsevier Page 6 of 18

Extending CityGML for IFC-sourced 3D city models
Figure 5: Illustration of the relation between IFC and CityGML showing examples of categories in Figure 4. Source of the IFC
dataset (left) used to generate the illustration: Open IFC Model Repository, Department of Computer Science, University of
Auckland [65].
the ADE accommodates information on volumes of rooms,
airconditioning systems, and photovoltaic panels. Some parts
of the extension have been inspired by the EnergyADE [58],
and because this topic has been well researched and devel-
oped in related work, this use case will not be elaborated
further in this paper.
4.1.2. Urban planning and liveability
Urban planning is a versatile use case with blurry bound-
aries and a wide range of applications. Providing an appro-
priately simplified model, which can be enhanced with use-
ful semantic information from IFC may offer advantages in
the types of visualisation and analysis that can be achieved,
and aid answering questions such as:
•How much greenery is visible from specific windows
and viewpoints of a building, and how does that affect
valuation?
•What is the impact on the outdoor thermal comfort
(e.g. facade material reflectivity)?
•How will a newly constructed building impact existing
nearby units?
•What is the total floor area of all constructions in a
certain plot?
•What is the area covered with vegetation on residential
buildings?
The IfcADE caters to a few possible use cases in this do-
main, and there is much potential for new use cases to take
advantage of some of the existing features. For example,
with the increased availability of IFC datasets, in future it
might be possible to investigate what proportion of the in-
frastructure and buildings within a region or city are acces-
sible across certain types of mobility restriction.
For further reading in the use of 3D city models in ur-
ban planning and related domains the reader is referred to
papers [76–84]. We have looked into these papers to under-
stand the use cases and their requirements. For example, we
added the number of occupants of a building [79], and floor
area [80] in the developed data model.
4.1.3. Multi-modal routing
The third use case of routing provides an example of how
enhancements to CityGML may provide support to a variety
of applications. It is also the use case that we selected to
focus on in the implementation (Section 5.3). Indoor acces-
sibility, routing, and navigation within a building, or even
indoor/outdoor accessibility between buildings, are increas-
ingly popular topics in recent research [85], and examples
that could benefit from data sourced from BIM. This use case
is especially important in the context of urban models, facil-
itating navigation in the urban environment (i.e. navigation
from a space in one building to a space in another building).
There are some potential overlaps to be considered with In-
doorGML [86], but this is outside of the scope and so is not
explored in more detail in this paper.
Biljecki F et al.: Preprint submitted to Elsevier Page 7 of 18

Extending CityGML for IFC-sourced 3D city models
In addition to indoor routing from point A to point B
within a building, examples of the types of applications iden-
tified within the indoor routing and navigation use case are
as follows:
•Effective citizen services — navigation: routing and
navigation between locations within a building. For
example, how to navigate to rooms or spaces in a build-
ing that are sufficiently large, or meet the specific needs
or profiles of users — the presence of particular ther-
mal, lighting, and ambient conditions.
•Effective citizen services — accessibility and mobil-
ity: providing choice and optimisation for the type of
routes considering different mobility and accessibility
requirements — for example, barrier-free accessibil-
ity taking into account routes that are wheelchair ac-
cessible or having automatic doors. Information such
as this may be relevant in the context of Singapore’s
(and more broadly to other developed nations) ageing
population and challenges of improving user experi-
ence in accessing or navigating through spaces within
buildings. Therefore, information such as door ac-
cess types could enhance existing use cases. Further-
more, this context-aware application is also extend-
able to other building typologies, and for other access
and mobility requirements — elevators with a specific
cargo capacity and sufficiently large door widths, in
an industrial setting for example. Some of these have
been described in detail in various research publica-
tions [87,88].
•Asset management — facility management: naviga-
tion and routing that takes into account the names and
locations of specific service locations (the locations
of a sensor, fuse boxes, or specific utilities and ser-
vices). The locations of particular walls and rooms as
well as their materials construction and colour, could
also be interesting, in cases where damage needs to be
reported and/or fixed. Therefore facilitating planning
and response to required maintenance activities.
•Protection of life — emergency response: locating the
nearest or particular emergency equipment — for ex-
ample, automated external defibrillator (AED), and how
to adapt the design of spaces to improve access to AEDs.
There is also an opportunity to provide information
services to first response teams of medical profession-
als who may need to navigate through unfamiliar lo-
cations or provide situational coordination to others
during critical or emergency situations.
•Protection of life — evacuation: analysing the opti-
mal evacuation routes in case of fire in specific parts
of a building. Optimising route planning could take
into consideration corridor width (which may impact
min/max time to evacuate people safely along certain
escape routes), the size of doors, the direction of their
opening including push-pull clearance which could af-
fect the flow of people.
•Asset Management — security: for example, how to
avoid or take into account certain areas within a build-
ing with restricted security clearance or access, which
could affect navigation routes.
•Understanding commercial risk — asset and facility
management: for example, a commercial company might
need to evaluate the risks and impacts in case of a
disaster/emergency with regards to their owned space,
staff, and assets within the building. Having the right
information within an ADE model could help them
analyse where to best place their assets within the build-
ing (so that there is a minimum impact caused by flood-
ing, stampedes, or riots) or how to best route their staff
and move any objects in case of emergency — in order
to minimise company losses. Likewise, this could ap-
ply to utility companies who would want to place their
assets in the building in the most feasible place and ef-
ficient for their planned service and maintenance (this
could be where its most appropriate for their network
of assets such as cables/pipes or the nearest point to
an entrance/exit door so that the maintenance engineer
can access it quickly).
These context-aware scenarios may require specific at-
tributes that are not provided by CityGML by default. There-
fore we extend the data model and provide a template to
source them from IFC. For further reading, the reader is re-
ferred to some of the many papers published in scientific lit-
erature on this topic [89–92,16,93–96].
4.2. Modelling decisions, mapping associations
and design
The development of the data model is shaped after a set
of decisions. In this section, we elaborate on the key ap-
proaches.
One of the most crucial modelling decisions we made is
not to use sub-classes, even though these are one of the stan-
dard ADE mechanisms [37,39]. Instead, to keep a neat or-
ganisation and to not compromise the standard data model,
all additions are directly derived from GML features. We
prefix such features with ‘Ifc’ except for a few classes con-
taining information frequently originating from other sources.
An advantage of such an approach is to make use of the
standard data model and leave it intact as much as possible:
instances of existing concrete CityGML classes are not af-
fected by the ADE additions. This facilitates software that
does not understand the ADE, to ignore the respective at-
tributes, while still recognising the original CityGML classes
and their attributes. For example, a room in CityGML has a
standard class BuildingRoom with a few attributes. We
have detected several additional attributes (e.g. whether it is
accessible to wheelchair users) that should be stored, facil-
itating particular use cases. One approach would be to de-
rive a new class from the existing class BuildingRoom
inheriting the existing attributes and adding in the new ones.
However, instead, our enhancement is designed as an inde-
pendent class IfcBuildingRoom that is associated with
Biljecki F et al.: Preprint submitted to Elsevier Page 8 of 18

Extending CityGML for IFC-sourced 3D city models
the standard feature BuildingRoom and accommodates
the additional set of attributes.
We identify the following entity types from IFC that are
of relevance in this context: IfcMaterial,IfcMate-
rialSet,IfcBuilding,IfcRoof,IfcRoofType,
IfcSlab,IfcTransportElement,IfcSpace,If-
cLightFixture,IfcSolarDevice,IfcDoor,IfcWin-
dow,IfcStorey,IfcBuildingStorey, and IfcRamp.
We map them either to corresponding feature types in CityGML
or create additional types for features that do not exist in
CityGML, such as elevators.
Examples that are useful for use cases and that we intro-
duced in the IfcADE model, in particular for the implemen-
tation of the indoor navigation use case which in focus of the
implementation section of this paper, are: (i) IfcBuild-
ing and its corresponding attributes such as: Building-
Type (accompanied by a new code list), buildingName
(in Singapore a building often has a name that in some cases
is used in place of it its address), and voidDeck (another
local particularity); (ii) IfcBuildingRoom with additional at-
tributes indicating different types of access (e.g. whether a
room is accessible from outside, whether it is accessible for
the public); (iii) IfcElevator with corresponding attributes
(e.g. liftNumber); and (iv) IfcDoor including infor-
mation on its dimensions, passage, purpose, and direction
of opening.
To be more specific, we give an example of the selected
set of attributes that is preserved by focusing on the case of
the modelling of windows, which in 3D city models are in-
cluded for a variety of reasons. In the context of our use
cases windows are important in all three of them: in en-
ergy they are valuable in estimating the heat/cooling loss, in
navigation they are important for planning evacuation routes
as windows can serve as an alternative escape, and in live-
ability it can be useful to assess the equity of views from
apartments. The CityGML standard data model allows mod-
elling windows, enabling the applications mentioned above.
However, even though windows are described as openings
in CityGML, it is not prescribed in which sense they are
‘open’, whether they establish a passage or a line of sight
or both, and such information cannot be stored in CityGML
by default (unless using generic attributes). While such in-
formation may not be necessary for the first and second use
cases, the absence of such information is an issue with the
navigation use case inhibiting the planning of escape routes
in which extended information on windows may be impor-
tant [97]. Therefore, we have extended the standard CityGML
data model in modelling windows, introducing a boolean at-
tribute navigable indicating whether the window is nav-
igable or not.
Another modelling decision to highlight is the considera-
tion of both boundary surfaces and volumetric elements, en-
abled by the new version of CityGML regarding IFC-sourced
3D city models. We have enabled the preservation of at-
tributes in either case of boundary surfaces and constructive
elements (e.g. see the feature IfcWall), depending on the
spatio-semantic modelling decisions. It may be worth noting
that this would not be possible with single inheritance. We
are aware that the simultaneous use of both options can lead
to redundancies and contradictions in the model. However,
in this case we prefer a flexible solution over enforcement of
good modelling practice through the schema.
The partial duplication of some features from the stan-
dard data model is a deliberate design choice to reduce am-
biguity and smooth the way for the implementation. For ex-
ample, the room name and room number are enabled in the
ADE as two new and distinct additions, even though the cur-
rent version of the CityGML 3.0 proposal allows modelling
both concurrently in the same feature, but the approach as
it is may add ambiguity and inhibit software implementa-
tion since it may not be clear which is which. With our
approach, these information are streamlined in two separate
attributes, reducing ambiguity and easing software imple-
mentation. This approach is also useful to make the lineage
clear that this piece of information originates from IFC, in
case there are multiple data sources. Our work indicates that
having multiple use cases nowadays may require data inte-
gration from multiple sources, so choices such as this one
regard that situation.
In order to support the localisation of the specification
and reflect the use cases, we have also introduced new code
lists that categorise possible values. For example, the build-
ing typology is in line with the categorisation in Singapore.
4.3. UML
Following the presented methodology and study, we have
designed 14 classes containing 63 attributes, and 5 new code
lists. The UML of the IfcADE is included in Figure 6, as an
enhancement to features found in the Building and Construc-
tion modules of CityGML 3.0.
Biljecki F et al.: Preprint submitted to Elsevier Page 9 of 18

Figure 6: The UML diagram of the developed IfcADE (the added classes are in beige with the pertaining code lists in green),
extending features from the standard CityGML Building and Construction modules (in yellow).

Extending CityGML for IFC-sourced 3D city models
5. Implementation and demonstration
We have verified the feasibility of the ADE by generating
(Section 5.1) and visualising (Section 5.2) an instance of the
schema. Furthermore, while the focus of this paper is the
design of the data model, and thus the implementation of
use cases is out of scope, we also showcase an early demo
of the use case of indoor navigation (Section 5.3) to give a
better understanding of the benefit of the work.
5.1. Generation of the data
We have used the IFC dataset of a large non-residential
building in Singapore (Figure 1), and converted it to CityGML
with a selective conversion process that was designed to source
the required information from IFC and output them to CityGML
according to the developed ADE [4]. The following excerpt
from an auto-generated instance of the ADE shows addi-
tional attributes for a building and a room, contained within
the extended features IfcBuilding and IfcBuildin-
gRoom, respectively:
1<cityObjectMember>
2<bldg:Building gml:id="ifc-0
PcOs3Gr9BD8TsNt7WYaK3">
3<ifc:ifcBuildingProperty>
4<ifc:IfcBuilding gml:id="gml-0
PcOs3Gr9BD8TsNt7WYaK3">
5<ifc:noOfStoreys>11</ifc:noOfStoreys>
6</ifc:IfcBuilding>
7</ifc:ifcBuildingProperty>
8<bldg:buildingRoom>
9<bldg:BuildingRoom gml:id="ifc-3
rdtWOORP9xx22XnTyVHz8">
10 <ifc:ifcBuildingRoomProperty>
11 <ifc:IfcBuildingRoom gml:id="gml-5
f069b48-09fe-4d69-bff0-1
ca8e5f445ec">
12 <ifc:roomName>ANCILLARY OFFICE</
ifc:roomName>
13 <ifc:roomHeightMin uom="m">3.6</
ifc:roomHeightMin>
14 <ifc:roomNumber>133</ifc:
roomNumber>
15 <ifc:roomHeightMax uom="m">3.6</
ifc:roomHeightMax>
16 <gml:name>133</gml:name>
17 </ifc:IfcBuildingRoom>
18 </ifc:ifcBuildingRoomProperty>
19 <gml:name>133</gml:name>
20 <spaceType>closed</spaceType>
21 <!-- GML geometry -->
22 </bldg:BuildingRoom>
23 </bldg:buildingRoom>
24 <!-- ... -->
25 </bldg:Building>
26 </cityObjectMember>
27 </CityModel>
5.2. Visualisation of the generated dataset
As an ADE is an extended data model, most CityGML
software packages either ignore or cannot handle these datasets
containing features beyond the standard schema. We have
visualised the dataset generated above in two ways: (i) in a
cross-browser web-viewer customised to support the devel-
oped ADE [98] (Figure 7), and (ii) in the FME Data Inspec-
tor by loading the XSD representation of the ADE (Figure 8;
this work doubles as a validation that the resulting ADE is
properly developed). Both these examples reveal the addi-
tional features/attributes enabled by the ADE and show the
same feature exemplified in the previous section.
5.3. Use case: indoor navigation
The utilisation of data in a use case and development of
software enabling use cases is left as a separate topic for fu-
ture work. However, in this section, we briefly give a sneak
peek of the value the enriched data will enable in use cases.
We have worked on the use case described in Section 4.1.3:
indoor navigation, using a solution currently being devel-
oped by Ordnance Survey. Our ADE enables having room
numbers and room names stored separately with less ambi-
guity (i.e. roomNumber; see Figure 6), and in this demo
we have used these attributes, instead of storing them within
the range of the standard data model.
However, because the software architecture of the in-
door navigation system does not support CityGML, we had
to convert the data to another format. This is not far from
a real-world scenario, as CityGML is conceived as an ex-
change format, and it is not always intended to be used di-
rectly in a software. An example of routing between two
rooms with a particular room number, using the same input
IFC dataset as in the previous sections, is shown in Figure 9.
While in this example we are not directly using CityGML
because the software does not support it, the content of the
resulting dataset is akin to the one we would obtain, contain-
ing additional features and resulting in the same outcome.
At this stage, we have developed the use case within a
single building, and its extension to a larger scale (routing
on the urban level between buildings) is a plan for future
work. At the moment, this work is in progress giving here a
sneak peek to suggest the implementation, and future work
will also involve making use of other information such as
wheelchair access.
6. Discussion, lessons learned, and limitations
6.1. Sourcing corresponding information from
IFC
There are multiple points that need to be stressed when
it comes to limitations of sourcing data from IFC.
First, the process of generating a dataset according to the
ADE is inhibited by the availability of features in real-world
IFC datasets. Conceptually the vast majority of the informa-
tion required for this ADE and the use cases can be sourced
from IFC. However, as it is the case with CityGML, IFC
datasets in practice are not always rich with semantics, pos-
sibly leaving some required features unpopulated in practice.
Second, there might be a misclassification in the seman-
tics (e.g. it is not always stored properly, preventing to take
advantage of it), and different conventions on the spatio-
semantic aspects may result in different interpretations dur-
ing the production of the IFC-sourced CityGML datasets [36],
which is also an issue on the geospatial end [99–101].
Biljecki F et al.: Preprint submitted to Elsevier Page 11 of 18

Extending CityGML for IFC-sourced 3D city models
Figure 7: Viewing the generated file in a custom-built web-viewer [98]. The left side of the image shows the ADE attributes.
Third, many features cannot be mapped 1:1 from IFC.
Some features should be acquired either manually or require
substantial processing. For example, the number of storeys
above ground of a building and its total height including
basement: as straightforward as these appear to be, they are
not explicitly available in the IFC model, and they might
not always be trivial to compute, as IFC geometry can be
complex to work with [102]. The building height is also
subject to different interpretations [103,104]. Therefore the
availability of different features is highly dependent on the
workflow/implementation. On the other hand, while the dis-
advantage of some of these attributes is that processing is
required, after implementing such rules the availability of
such features is virtually always warranted (e.g. it is always
possible to calculate the floor area of a room, no additional
semantic information is needed in the input IFC dataset).
It is a general information modelling question to trade-
off defining a procedure to calculate a value against storing
the calculation result explicitly in a data model. IFC allows
both, with derived attributes to hold calculation procedures
and base quantities to store values derived from geometry
explicitly. Although the set of typical base quantities and
included procedures may be reasonable for typical BIM use
cases, specific requirements of geospatial use cases may not
be covered.
Another challenge is the different structure of geometry
and semantics between the two formats, entailing different
interpretation on how to preserve certain features, rendering
some portions of the work inevitably subjective. For exam-
ple, a wall in IFC is represented in geometry and semantics
differently than in CityGML.
Finally, there are limitations to the IfcADE because an
analysis that requires specific information from IFC that has
not been converted to CityGML + IfcADE will not be avail-
able.
6.2. Lack of compliant software and use case
support
At the present time of the submission of this paper, ver-
sion 3.0 of CityGML is under development and therefore at
the moment there is not much software support for it. On top
of that, one of the disadvantages of developing an ADE is
that it is normally not supported by existing software. While
we have thoroughly discussed use cases with different stake-
holders and made an effort to tackle the implementation of
the ADE, one of the limitations of the work is that it is dif-
ficult to realise the implementation of the ADE in use cases
fully.
Much of the lessons learned in this research is related to
use cases. It is often difficult to gather the specific require-
ments of a use case, inhibiting the development of a sup-
porting data model. The problem boils down to one thing
– software support, which is still lacking in 3D use cases.
Such a limitation does not allow checking whether a series
of use cases is served appropriately by a data model.
Software vendors often wait for the need to be expressed
by their users before actually putting solutions in place within
their software (along with the chicken-and-egg dilemma: whether
there is a lack of software because there is no data, or there is
no data because there is no software to work with it?). There-
Biljecki F et al.: Preprint submitted to Elsevier Page 12 of 18

Extending CityGML for IFC-sourced 3D city models
Figure 8: Viewing the generated file in the FME Data Inspector, revealing multiple room attributes enabled by the ADE (bottom
panel).
fore, this shows there is a need for awareness being built for
both users, stakeholders, and system suppliers in order to
support datasets enriched with additional information.
Finally, as it is the case with developing data specifica-
tions, it is a challenge to balance simplicity and capacity.
Tailoring and adding more features for particular use cases
may add value and even enable additional applications, but
at the same time it increases complexity and may discourage
software developers to provide an implementation.
7. Conclusion
BIM datasets are promising to become a valuable source
for 3D city models and their maintenance. Therefore it is
important to adapt different aspects on the geospatial end,
such as the development of enriched data models easing to
bridge the two paradigms.
We have presented a cohesive CityGML ADE that is
shaped after three orthogonal aspects: (i) regarding the in-
put data and lineage (BIM/IFC); (ii) considering the local
context (Singapore); and (iii) catering to multiple use cases
(energy, urban planning, and navigation).
We described a rationale behind the development of an
ADE, and stress on how we carried out a discovery process
to reach the presented design supporting transfer of infor-
mation between open 3D data standards IFC and CityGML.
Therefore we believe that our work can be replicated else-
where. Besides the investigation of use cases and the devel-
opment of a specification, we implement the work rounding
up the complete cycle: we (i) generated a dataset according
to the developed specification; (ii) visualised the dataset with
enhanced additions in two ways (desktop and web); and (iii)
demonstrated the potential of using the dataset for a partic-
ular purpose: indoor context-aware routing highlighting the
benefit of the enhanced tailored specification.
While forgoing the rich details from IFC is usually justi-
fied to maintain the generation of simple and lightweight 3D
city models, we show that there is a benefit in keeping a cer-
tain subset of information and carrying it over to CityGML.
The selection of such a subset was shaped after a detailed
discovery process with stakeholders and analysing require-
ments of use cases. A byproduct contribution of our work
is that we look into details in some practical use cases with
real-world stakeholders and describe them.
While this is not the first ADE that was designed in a re-
search project on BIM-GIS interoperability, it is in several
ways a contribution, as the ADE offers support for a breadth
of applications while not compromising simplicity, and the
paper includes also topics uncommon in related work: it
demonstrates an implementation, it describes use cases and
the selection of features, and it reports usually undocumented
deliberations — we carefully considered different modelling
options, which we noted down and argued the pros/cons, po-
tentially assisting fellow researchers with guidance in similar
projects.
Although for a single application domain there may be
a more suitable solution such as a dedicated standard, the
advantage of the IfcADE is the range of applications that can
Biljecki F et al.: Preprint submitted to Elsevier Page 13 of 18

Extending CityGML for IFC-sourced 3D city models
Figure 9: Implementation of indoor navigation making use of room numbers/names, as an example of additional attributes
defined by our extension.
be supported while retaining the data in the same format.
As much as the extended data model pertains to specific
use cases, some of the features may be considered rather
generic and useful to multiple use cases. Therefore it might
be beneficial to add some of these features in the standard
CityGML data model, rather than having them in an ADE
(e.g. information about the access of rooms). Vice-versa,
the ADE has to be aligned to the final adopted version of
the standard. After our ADE was designed, some of the at-
tributes have been added to the data model, but we plan to
wait for the adopted version of the standard and synchronise
the two.
We deem this work mature, however, opportunities for
future work are many as this ADE is intended to be a con-
tinuous development by stakeholders, following use case re-
quirements, and developments related to software support
and available datasets.
First, we plan to enhance the specification according to
additional use cases. There are other use cases or applica-
tions that could be of interest in this particular geographic
context to adapt the ADE to, such as outdoor thermal com-
fort [105], real estate valuation [70], mitigating noise [106],
conforming to local standards assessing building performan-
ce [107], and aligning to 3D cadastre developments in Sin-
gapore [108]. In its present form, the developed ADE pro-
vides a well-founded basis for additional use cases because
it potentially has an overlap with some of them, e.g. it al-
ready contains data on the material and other information
that might be useful for managing cultural heritage [109,
110].
Second, a topic that would be interesting to pursue is the
diverse selection of semantics according to different levels
of detail of the indoor of 3D city models: there have been
several research papers focusing on the topic which might
be beneficial to align to [111–114,67,115]. In our imple-
mentation we did deal with two representations (Figure 1)
and both support the ADE, but there is room for additional
representations.
Third, a potential important research direction is the ex-
tension of other CityGML modules, such as transportation
and vegetation. In this work, we have focused on buildings,
as the most prominent topographic feature in our context.
However, current and future use cases may also require other
modules of CityGML. They are continuously subject to pro-
posed extensions [116], and these developments will be im-
portant to follow in the context of smart cities and 3D data
infrastructure.
Finally, we plan to investigate the translation of the ex-
tension to other data formats and encodings, such as CityJ-
SON [117], potentially allowing more efficient utilisation of
the datasets on the web.
Acknowledgements
We appreciate the building dataset provided by the Build-
ing and Construction Authority (Singapore). We gratefully
acknowledge the input of Singapore Government agencies
and their continuous communication and interest in our project:
Singapore Land Authority, Government Technology Agency,
Housing and Development Board, Building and Construc-
tion Authority, Urban Redevelopment Authority, and others.
Suggestions and contact with Alex Babington, Tim Man-
Biljecki F et al.: Preprint submitted to Elsevier Page 14 of 18

Extending CityGML for IFC-sourced 3D city models
ners, Oliver Snowden, Tony Joyce, Jeremy Morley, and Carsten
Roensdorf (Ordnance Survey), and Kavisha Kumar (TU Delft)
are gratefully acknowledged. The comments of the review-
ers are appreciated and improved the paper.
This material is based on research/work supported by
the National Research Foundation under Virtual Singapore
Award No. NRF2015VSG-AA3DCM001-008.
References
[1] X. Liu, X. Wang, G. Wright, J. Cheng, X. Li, R. Liu, A State-of-
the-Art Review on the Integration of Building Information Model-
ing (BIM) and Geographic Information System (GIS), ISPRS In-
ternational Journal of Geo-Information 6 (2) (2017) 53. doi:
10.3390/ijgi6020053.
[2] F. Noardo, L. Harrie, K. Arroyo Ohori, F. Biljecki, C. Ellul, T. Kri-
jnen, H. Eriksson, D. Guler, D. Hintz, M. A. Jadidi, M. Pla,
S. Sanchez, V.-P. Soini, R. Stouffs, J. Tekavec, J. Stoter, Tools
for BIM-GIS Integration (IFC Georeferencing and Conversions):
Results from the GeoBIM Benchmark 2019, ISPRS International
Journal of Geo-Information 9 (9) (2020) 502. doi:10.3390/
ijgi9090502.
[3] H. Wang, Y. Pan, X. Luo, Integration of BIM and GIS in sustainable
built environment: A review and bibliometric analysis, Automation
in Construction 103 (2019) 41 – 52. doi:10.1016/j.autcon.
2019.03.005.
[4] R. Stouffs, H. Tauscher, F. Biljecki, Achieving Complete and Near-
Lossless Conversion from IFC to CityGML, ISPRS International
Journal of Geo-Information 7 (9) (2018) 355. doi:10.3390/
ijgi7090355.
[5] T. W. Kang, C. H. Hong, A study on software architecture for ef-
fective BIM/GIS-based facility management data integration, Au-
tomation in Construction 54 (2015) 25 – 38. doi:10.1016/j.
autcon.2015.03.019.
[6] Y. Deng, J. C. P. Cheng, C. Anumba, A framework for 3D traffic
noise mapping using data from BIM and GIS integration, Structure
and Infrastructure Engineering 12 (10) (2016) 1267 – 1280. doi:
10.1080/15732479.2015.1110603.
[7] buildingSMART, Industry foundation classes, Tech. Rep. Version
4, Addendum 2, buildingSMART International Limited (2016)
[cited 15 August 2017].
URL http://www.buildingsmart-tech.org/ifc/
IFC4/Add2/
[8] ISO, Industry Foundation Classes (IFC) for data sharing in the con-
struction and facility management industries (2013) [cited 13 April
2019].
URL https://www.iso.org/standard/70303.html
[9] Open Geospatial Consortium, OGC City Geography Markup
Language (CityGML) Encoding Standard 2.0.0, Tech. rep. (2012)
[cited 13 April 2019].
URL http://www.opengeospatial.org/standards/
citygml
[10] G. Gröger, L. Plümer, CityGML – Interoperable semantic 3D city
models, ISPRS Journal of Photogrammetry and Remote Sensing
71 (2012) 12 – 33. doi:10.1016/j.isprsjprs.2012.04.
004.
[11] Z. Zhu, L. Zhou, C. Zhang, B. Lin, Y. Cui, M. Che, Modeling of
Macroscopic Building Evacuation Using IFC Data, ISPRS Interna-
tional Journal of Geo-Information 7 (8) (2018) 302 – 25. doi:
10.3390/ijgi7080302.
[12] Y. Chen, E. Shooraj, A. Rajabifard, S. Sabri, From IFC to 3D Tiles:
An Integrated Open-Source Solution for Visualising BIMs on Ce-
sium, ISPRS International Journal of Geo-Information 7 (10) (2018)
393. doi:10.3390/ijgi7100393.
[13] Y. Deng, V. J. L. Gan, M. Das, J. C. P. Cheng, C. Anumba, Inte-
grating 4D BIM and GIS for Construction Supply Chain Manage-
ment, Journal of Construction Engineering and Management 145 (4)
(2019) 04019016. doi:10.1061/(asce)co.1943-7862.
0001633.
[14] J. Stoter, H. Ploeger, R. Roes, E. van der Riet, F. Biljecki, H. Ledoux,
D. Kok, S. Kim, Registration of Multi-Level Property Rights in 3D
in The Netherlands: TwoCases and Next Steps in Further Implemen-
tation, ISPRS International Journal of Geo-Information 6 (6) (2017)
158. doi:10.3390/ijgi6060158.
[15] R. Southall, F. Biljecki, The VI-Suite: a set of environmental
analysis tools with geospatial data applications, Open Geospatial
Data, Software and Standards 2 (2017) 23. doi:10.1186/
s40965-017- 0036-1.
[16] T.-A. Teo, K.-H. Cho, BIM-oriented indoor network model for in-
door and outdoor combined route planning, Advanced Engineering
Informatics 30 (3) (2016) 268–282.
[17] U. Isikdag, S. Zlatanova, J. Underwood, A BIM-Oriented Model
for supporting indoor navigation requirements, Computers, Environ-
ment and Urban Systems 41 (2013) 112 – 123. doi:10.1016/j.
compenvurbsys.2013.05.001.
[18] H. Tashakkori, A. Rajabifard, M. Kalantari, A new 3D indoor/out-
door spatial model for indoor emergency response facilitation,
Building and Environment 89 (2015) 170 – 182. doi:10.1016/
j.buildenv.2015.02.036.
[19] H. Tauscher, R. Stouffs, An IFC-to-CityGML Triple Graph Gram-
mar, in: eCAADe, Vol. 1, 2018, pp. 517 – 524 [cited 13 April
2019].
URL http://papers.cumincad.org/data/works/
att/ecaade2018_265.pdf
[20] A. Konde, H. Tauscher, F. Biljecki, J. Crawford, Floor plans
in CityGML, ISPRS Ann. Photogramm. Remote Sens. Spa-
tial Inf. Sci. IV-4/W6 (2018) 25 – 32. doi:10.5194/
isprs-annals- iv-4- w6-25-2018.
[21] J. Lim, H. Tauscher, F. Biljecki, Graph transformation rules for
IFC-to-CityGML attribute conversion, ISPRS Ann. Photogramm.
Remote Sens. Spatial Inf. Sci. IV-4/W8 (2019) 83 – 90. doi:
10.5194/isprs-annals- iv-4- w8-83-2019.
[22] H. Tauscher, Creating and maintaining IFC–CityGML con-
version rules, ISPRS Ann. Photogramm. Remote Sens. Spa-
tial Inf. Sci. IV-4/W8 (2019) 115 – 122. doi:10.5194/
isprs-annals- iv-4- w8-115-2019.
[23] Y. Song, X. Wang, Y. Tan, P. Wu, M. Sutrisna, J. Cheng, K. Hamp-
son, Trends and Opportunities of BIM-GIS Integration in the Archi-
tecture, Engineering and Construction Industry: A Review from a
Spatio-Temporal Statistical Perspective, ISPRS International Jour-
nal of Geo-Information 6 (12) (2017) 397. doi:10.3390/
ijgi6120397.
[24] J. Zhu, X. Wang, P. Wang, Z. Wu, M. J. Kim, Integration of BIM
and GIS: Geometry from IFC to shapefile using open-source tech-
nology, Automation in Construction 102 (2019) 105 – 119. doi:
10.1016/j.autcon.2019.02.014.
[25] R. De Laat, L. Van Berlo, Integration of BIM and GIS: The Devel-
opment of the CityGML GeoBIM Extension, Advances in 3D Geo-
Information Sciences, Springer Berlin Heidelberg, 2011, pp. 211 –
225. doi:10.1007/978-3- 642-12670- 3_13.
[26] Y. Deng, J. C. P. Cheng, C. Anumba, Mapping between BIM and 3D
GIS in different levels of detail using schema mediation and instance
comparison, Automation in Construction 67 (2016) 1 – 21. doi:
10.1016/j.autcon.2016.03.006.
[27] S. Donkers, H. Ledoux, J. Zhao, J. Stoter, Automatic conversion of
IFC datasets to geometrically and semantically correct CityGML
LOD3 buildings, Transactions in GIS 20 (4) (2016) 547 – 569.
doi:10.1111/tgis.12162.
[28] M. El-Mekawy, A. Östman, I. Hijazi, A Unified Building Model
for 3D Urban GIS, ISPRS International Journal of Geo-Information
1 (3) (2012) 120 – 145. doi:10.3390/ijgi1020120.
[29] C. Ellul, G. Boyes, C. Thomson, D. Backes, Towards Integrating
BIM and GIS—An End-to-End Example from Point Cloud to Anal-
ysis, Vol. 15 of Advances in 3D Geoinformation, 2017, pp. 495 –
512.
Biljecki F et al.: Preprint submitted to Elsevier Page 15 of 18

Extending CityGML for IFC-sourced 3D city models
[30] A. Geiger, J. Benner, K. H. Haefele, Generalization of 3D
IFC Building Models, 3D Geoinformation Science, Springer In-
ternational Publishing, 2015, pp. 19 – 35. doi:10.1007/
978-3- 319-12181- 9_2.
[31] I. Hijazi, M. Ehlers, S. Zlatanova, T. Becker, L. Van Berlo, Ini-
tial Investigations for Modeling Interior Utilities Within 3D Geo
Context: Transforming IFC-Interior Utility to CityGML/UtilityNet-
workADE, in: T. Kolbe, G. König, C. Nagel (Eds.), Advances in 3D
Geo-Information Sciences, Springer, Berlin, Heidelberg, 2011, pp.
95 – 113. doi:10.1007/978- 3-642-12670- 3_6.
[32] U. Isikdag, S. Zlatanova, Towards Defining a Framework for Auto-
matic Generation of Buildings in CityGML Using Building Infor-
mation Models, in: J. Lee, S. Zlatanova (Eds.), 3D Geo-Information
Sciences, Springer Berlin Heidelberg, 2009, pp. 79 – 96. doi:
10.1007/978-3- 540-87395- 2_6.
[33] S. Jusuf, B. Mousseau, G. Godfroid, J. Soh, Path to an Inte-
grated Modelling between IFC and CityGML for Neighborhood
Scale Modelling, Urban Science 1 (3) (2017) 25. doi:10.3390/
urbansci1030025.
[34] R. Sebastian, M. Böhms, P. van den Helm, BIM and GIS for low-
disturbance construction, in: Proceedings of the 13th International
Conference on Construction Applications of Virtual Reality, 2013,
pp. 469 – 479 [cited 13 April 2019].
URL https://repository.tudelft.nl/view/tno/
uuid:1ed10694-572f- 433c-95be- f726c7704086
[35] T. Gilbert, S. Barr, P. James, J. Morley, Q. Ji, Software Systems
Approach to Multi-Scale GIS-BIM Utility Infrastructure Network
Integration and Resource Flow Simulation, ISPRS International
Journal of Geo-Information 7 (8) (2018) 310. doi:10.3390/
ijgi7080310.
[36] H. Tauscher, R. Stouffs, Extracting different spatio-semantic struc-
tures from IFC using a triple graph grammar, in: Intelligent and
Informed, Proceedings of the 24th International Conference of the
Association for Computer-Aided Architectural Design Research in
Asia (CAADRIA) 2019, 2019, pp. 605–614 [cited 13 April 2019].
URL http://papers.cumincad.org/data/works/
att/caadria2019_038.pdf
[37] F. Biljecki, K. Kumar, C. Nagel, CityGML Application Domain
Extension (ADE): overview of developments, Open Geospatial
Data, Software and Standards 3 (1) (2018) 13. doi:10.1186/
s40965-018- 0055-6.
[38] L. Van den Brink, J. Stoter, S. Zlatanova, Modeling an application
domain extension of CityGML in UML, Int. Arch. Photogramm. Re-
mote Sens. Spatial Inf. Sci. XXXVIII-4/C26 (2012) 11 – 14. doi:
10.5194/isprsarchives-xxxviii- 4-c26- 11-2012.
[39] L. Van den Brink, J. Stoter, S. Zlatanova, UML-Based Approach
to Developing a CityGML Application Domain Extension, Trans-
actions in GIS 17 (6) (2013) 920 – 942. doi:10.1111/tgis.
12026.
[40] A. Czerwinski, S. Sandmann, L. Elke, Stöümer, Sustainable SDI
for EU noise mapping in NRW – best practice for INSPIRE,
International Journal for Spatial Data Infrastructure Research 2 (1)
(2007) 1 – 18.
URL http://www.ikg.uni-bonn.de/uploads/tx_
ikgpublication/071105_EC_GI_GIS_fullpaper_
Czerwinski.pdf
[41] V. Çağdaş, An Application Domain Extension to CityGML for im-
movable property taxation: A Turkish case study, International Jour-
nal of Applied Earth Observation and Geoinformation 21 (2013) 545
– 555. doi:10.1016/j.jag.2012.07.013.
[42] A. Labetski, K. Kumar, H. Ledoux, J. Stoter, A Metadata ADE
for CityGML, Open Geospatial Data, Software and Standards 3 (1)
(2018) 42. doi:10.1186/s40965- 018-0057-4.
[43] L. Li, L. Tang, H. Zhu, H. Zhang, F. Yang, W. Qin, Semantic 3D
Modeling Based on CityGML for Ancient Chinese-Style Architec-
tural Roofs of Digital Heritage, ISPRS International Journal of Geo-
Information 6 (5) (2017) 132. doi:10.3390/ijgi6050132.
[44] L. Van den Brink, J. Stoter, S. Zlatanova, Establishing a national
standard for 3D topographic data compliant to CityGML, Interna-
tional Journal of Geographical Information Science 27 (1) (2013)
92 – 113. doi:10.1080/13658816.2012.667105.
[45] K. Kumar, H. Ledoux, J. Stoter, A CityGML extension for han-
dling very large TINs, ISPRS Ann. Photogramm. Remote Sens.
Spatial Inf. Sci. IV-2-W1 (2016) 137 – 143. doi:10.5194/
isprs-annals- iv-2- w1-137-2016.
[46] L. Li, J. Wu, H. Zhu, X. Duan, F. Luo, 3D modeling of the
ownership structure of condominium units, Computers, Environ-
ment and Urban Systems 59 (2016) 50 – 63. doi:10.1016/j.
compenvurbsys.2016.05.004.
[47] Z. Yao, C. Nagel, F. Kunde, G. Hudra, P. Willkomm, A. Donaubauer,
T. Adolphi, T. H. Kolbe, 3DCityDB - a 3D geodatabase solu-
tion for the management, analysis, and visualization of seman-
tic 3D city models based on CityGML, Open Geospatial Data,
Software and Standards 3 (1) (2018) 208. doi:10.1186/
s40965-018- 0046-7.
[48] H. Eriksson, T. Johansson, P.-O. Olsson, M. Andersson, J. Eng-
vall, I. Hast, L. Harrie, Requirements, Development, and Evalu-
ation of A National Building Standard—A Swedish Case Study,
ISPRS International Journal of Geo-Information 9 (2) (2020) 78.
doi:10.3390/ijgi9020078.
[49] S. A. Aydar, T. Yomralıoglu, E. D. Ozbek, Modeling Turkey
National 2D Geo-Data Model as a CityGML Application Domain
Extension in UML, International Journal of Environment and
Geoinformatics 3 (3) (2016) 1 – 10 [cited 13 April 2019].
URL https://dergipark.org.tr/tr/download/
article-file/294320
[50] O. G. Ling, T. T. Tan, The Social Significance of Public Spaces in
Public Housing Estates, Public Space Design, Use and Manage-
ment, NUS Press, 1992, pp. 69 – 81 [cited 13 April 2019].
URL http://www.worldcat.org/title/
public-space- design-use- and-management/oclc/
930609565
[51] K. Soon, V. Khoo, CityGML modelling for Singapore 3D na-
tional mapping, Int. Arch. Photogramm. Remote Sens. Spa-
tial Inf. Sci. XLII-4/W7 (2017) 37 – 42. doi:10.5194/
isprs-archives- xlii-4- w7-37-2017.
[52] L. Gobeawan, E. S. Lin, A. Tandon, A. T. K. Yee, V. Khoo,
S. N. Teo, S. Yi, C. W. Lim, S. T. Wong, D. J. Wise, P. Cheng,
S. C. Liew, X. Huang, Q. H. Li, L. S. Teo, G. S. Fekete, M. T.
Poto, Modeling trees for Virtual Singapore: from data acquisi-
tion to CityGML models, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLII-4/W10 (2018) 55 – 62. doi:10.5194/
isprs-archives- xlii-4- w10-55-2018.
[53] R. Van Son, S. Jaw, J. Yan, V. Khoo, R. Loo, S. Teo, G. Schrot-
ter, A framework for reliable three-dimensional underground utility
mapping for urban planning, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLII-4/W10 (2018) 209 – 214. doi:10.5194/
isprs-archives- xlii-4- w10-209-2018.
[54] J. Yan, S. Jaw, R. Van Son, K. Soon, G. Schrotter, Three-
dimensional data modelling for underground utility net-
work mapping, Int. Arch. Photogramm. Remote Sens. Spa-
tial Inf. Sci. XLII-4 (2018) 711 – 715. doi:10.5194/
isprs-archives- xlii-4- 711-2018.
[55] S. Ho, A. Rajabifard, Towards 3D-enabled urban land administra-
tion: Strategic lessons from the BIM initiative in Singapore, Land
Use Policy 57 (2016) 1 – 10. doi:10.1016/j.landusepol.
2016.05.011.
[56] T. Kutzner, I. Hijazi, T. H. Kolbe, Semantic Modelling of 3D Multi-
Utility Networks for Urban Analyses and Simulations, International
Journal of 3-D Information Modeling 7 (2) (2018) 1 – 34. doi:
10.4018/ij3dim.2018040101.
[57] K. Wong, C. Ellul, User requirements gathering for a national 3D
mapping product in the United Kingdom, ISPRS Ann. Photogramm.
Remote Sens. Spatial Inf. Sci. IV-4/W6 (2018) 89 – 96. doi:10.
5194/isprs-annals- iv-4- w6-89-2018.
[58] G. Agugiaro, J. Benner, P. Cipriano, R. Nouvel, The Energy Applica-
Biljecki F et al.: Preprint submitted to Elsevier Page 16 of 18

Extending CityGML for IFC-sourced 3D city models
tion Domain Extension for CityGML: enhancing interoperability for
urban energy simulations, Open Geospatial Data, Software and Stan-
dards 3 (1) (2018) 139. doi:10.1186/s40965- 018-0042- y.
[59] I. Prieto, J. L. Izkara, F. J. D. d. Hoyo, Efficient Visualization of the
Geometric Information of CityGML: Application for the Documen-
tation of Built Heritage, Vol. 7333 of Computational Science and Its
Applications – ICCSA 2012, 2012, pp. 529 – 544.
[60] K. Chaturvedi, Z. Yao, T. H. Kolbe, Integrated management
and visualization of static and dynamic properties of seman-
tic 3D city models, Int. Arch. Photogramm. Remote Sens. Spa-
tial Inf. Sci. XLII-4/W17 (2019) 7 – 14. doi:10.5194/
isprs-archives- xlii-4- w17-7-2019.
[61] P. O. Olsson, T. Johansson, H. Eriksson, T. Lithén, L. H. Bengts-
son, J. Axelsson, U. Roos, K. Neland, B. Rydén, L. Harrie, Un-
broken digital data flow in the built environment process – a case
study in Sweden, Int. Arch. Photogramm. Remote Sens. Spatial
Inf. Sci. XLII-2/W13 (2019) 1347 – 1352. doi:10.5194/
isprs-archives- xlii-2- w13-1347-2019.
[62] L. Knoth, J. Scholz, J. Strobl, M. Mittlböck, B. Vockner, C. Atzl,
A. Rajabifard, B. Atazadeh, Cross-Domain Building Models—A
Step towards Interoperability, ISPRS International Journal of Geo-
Information 7 (9) (2018) 363. doi:10.3390/ijgi7090363.
[63] K. Kumar, H. Ledoux, T. J. F. Commandeur, J. E. Stoter, Modelling
urban noise in CityGML ADE: case of the Netherlands, ISPRS Ann.
Photogramm. Remote Sens. Spatial Inf. Sci. IV-4-W5 (2017) 73 –
81. doi:10.5194/isprs-annals- iv-4- w5-73-2017.
[64] L. L. o. Scholtenhuis, X. d. Duijn, S. Zlatanova, Representing geo-
graphical uncertainties of utility location data in 3D, Automation in
Construction 96 (2018) 483 – 493. doi:10.1016/j.autcon.
2018.09.012.
[65] Open IFC Model Repository, House Sr. Dubal Herrera (Publishable
under cc-by-3.0 licence) (2018).
URL http://openifcmodel.cs.auckland.ac.nz/
Model/Details/314
[66] T. Kutzner, T. H. Kolbe, CityGML 3.0: Sneak Preview, in: 38.
Wissenschaftlich-Technische Jahrestagung der DGPF und PFGK18
Tagung in München—- Publikationen der DGPF, Vol. 27, 2018, pp.
835 – 839.
[67] M. O. Löwner, G. Gröger, J. Benner, F. Biljecki, C. Nagel,
Proposal for a new LOD and multi-representation concept
for CityGML, ISPRS Ann. Photogramm. Remote Sens. Spa-
tial Inf. Sci. IV-2/W1 (2016) 3 – 12. doi:10.5194/
isprs-annals- iv-2- w1-3-2016.
[68] T. Kutzner, K. Chaturvedi, T. H. Kolbe, CityGML 3.0: New Func-
tions Open Up New Applications, PFG – Journal of Photogram-
metry, Remote Sensing and Geoinformation Science (2020) 1 –
19doi:10.1007/s41064-020- 00095-z.
[69] P. P. Yang, S. Y. Putra, W. Li, Viewsphere: a GIS-based 3D visibil-
ity analysis for urban design evaluation, Environment and Planning
B: Planning and Design 34 (6) (2007) 971–992. doi:10.1068/
b32142.
[70] S.-M. Yu, S.-S. Han, C.-H. Chai, Modeling the value of view in high-
rise apartments: a 3D GIS approach, Environment and Planning B:
Planning and Design 34 (1) (2007) 139 – 153. doi:10.1068/
b32116.
[71] S. Paiho, J. Ketomäki, L. Kannari, T. Häkkinen, J. Shemeikka, A
new procedure for assessing the energy-efficient refurbishment of
buildings on district scale, Sustainable Cities and Society (2019)
101454doi:10.1016/j.scs.2019.101454.
[72] Y. Chen, T. Hong, X. Luo, B. Hooper, Development of city buildings
dataset for urban building energy modeling, Energy and Buildings
183 (2019) 252 – 265. doi:10.1016/j.enbuild.2018.11.
008.
[73] S. M. Murshed, S. Picard, A. Koch, Modelling, Validation and
Quantification of Climate and Other Sensitivities of Building En-
ergy Model on 3D City Models, ISPRS International Journal of
Geo-Information 7 (11) (2018) 447 – 22. doi:10.3390/
ijgi7110447.
[74] M. Bizjak, B. Žalik, G. Štumberger, N. Lukač, Estimation and opti-
misation of buildings’ thermal load using LiDAR data, Building and
Environment 128 (2018) 12 – 21. doi:10.1016/j.buildenv.
2017.11.016.
[75] A. Salvo, Electrical appliances moderate households’ water demand
response to heat., Nature communications 9 (1) (2018) 5408. doi:
10.1038/s41467-018- 07833-3.
[76] T. Agius, S. Sabri, M. Kalantari, Three-Dimensional Rule-Based
City Modelling to Support Urban Redevelopment Process, ISPRS
International Journal of Geo-Information 7 (10) (2018) 413. doi:
10.3390/ijgi7100413.
[77] G. Herbert, X. Chen, A comparison of usefulness of 2D and 3D
representations of urban planning, Cartography and Geographic
Information Science 42 (1) (2015) 22 – 32. doi:10.1080/
15230406.2014.987694.
[78] M. Ranzinger, G. Gleixner, GIS datasets for 3D urban planning,
Computers, Environment and Urban Systems 21 (2) (1997) 159 –
173. doi:10.1016/S0198-9715(97)10005- 9.
[79] R. Trubka, S. Glackin, O. Lade, C. Pettit, A web-based 3D visuali-
sation and assessment system for urban precinct scenario modelling,
ISPRS Journal of Photogrammetry and Remote Sensing 117 (2016)
175 – 186. doi:10.1016/j.isprsjprs.2015.12.003.
[80] M. Brasebin, J. Perret, S. Mustiere, C. Weber, 3D urban data to
assess local urban regulation influence, Computers, Environment
and Urban Systems 68 (2018) 37 – 52. doi:10.1016/j.
compenvurbsys.2017.10.002.
[81] F. C. Ahmed, S. P. Sekar, Using Three-Dimensional Volumetric
Analysis in Everyday Urban Planning Processes, Applied Spatial
Analysis and Policy 8 (4) (2015) 393 – 408. doi:10.1007/
s12061-014- 9122-2.
[82] K. W. Chen, L. Norford, Evaluating Urban Forms for Comparison
Studies in the Massing Design Stage, Sustainability 9 (6) (2017) 987.
doi:10.3390/su9060987.
[83] A. Koltsova, B. Tunçer, G. Schmitt, Visibility Analysis for 3D Urban
Environments, Proceedings of the 31st International Conference on
Education and research in Computer Aided Architectural Design in
Europe, 2013, pp. 375 – 384.
[84] S. Yu, B. Yu, W. Song, B. Wu, J. Zhou, Y. Huang, J. Wu, F. Zhao,
W. Mao, View-based greenery: A three-dimensional assessment of
city buildings’ green visibility using Floor Green View Index, Land-
scape and Urban Planning 152 (2016) 13 – 26. doi:10.1016/j.
landurbplan.2016.04.004.
[85] J. Chen, K. C. Clarke, Indoor cartography, Cartography and Ge-
ographic Information Science 1 (114) (2019) 1 – 15. doi:10.
1080/15230406.2019.1619482.
[86] K.-J. Li, G. Conti, E. Konstantinidis, S. Zlatanova, P. Bamidis, OGC
IndoorGML: A Standard Approach for Indoor Maps, in: Geograph-
ical and Fingerprinting Data to Create Systems for Indoor Position-
ing and Indoor/Outdoor Navigation, Elsevier, 2019, pp. 187–207.
doi:10.1016/B978-0- 12-813189- 3.00010-1.
[87] C. Nagel, Spatio-semantic modelling of indoor environments for
indoor navigation, Ph.D. thesis, TU Berlin (2014).
URL https://www.dgk.badw.de/fileadmin/user_
upload/Files/DGK/docs/c-727.pdf
[88] A. A. Khan, Constraints and concepts for the support of different lo-
comotion types in indoor navigation, Ph.D. thesis, Technische Uni-
versität München, München (2015).
URL https://mediatum.ub.tum.de/?id=1233285
[89] A. Vanclooster, N. V. d. Weghe, P. D. Maeyer, Integrating Indoor
and Outdoor Spaces for Pedestrian Navigation Guidance: A Review,
Transactions in GIS 20 (4) (2016) 491 – 525. doi:10.1111/
tgis.12178.
[90] M.-P. Kwan, J. Lee, Emergency response after 9/11: the potential
of real-time 3D GIS for quick emergency response in micro-spatial
environments, Computers, Environment and Urban Systems 29 (2)
(2005) 93 – 113. doi:10.1016/j.compenvurbsys.2003.
08.002.
[91] L. Liu, S. Zlatanova, A "door-to-door" Path-Finding Approach
Biljecki F et al.: Preprint submitted to Elsevier Page 17 of 18

Extending CityGML for IFC-sourced 3D city models
for Indoor Navigation, ISPRS Ann. Photogramm. Remote Sens.
Spatial Inf. Sci. II-4 (2011) 45 – 51. doi:10.5194/
isprsannals-ii- 4-45- 2014.
[92] K. Kim, J. P. Wilson, Planning and visualising 3D routes for indoor
and outdoor spaces using CityEngine, Journal of Spatial Science
60 (1) (2014) 179 – 193. doi:10.1080/14498596.2014.
911126.
[93] J.-C. Thill, T. H. D. Dao, Y. Zhou, Traveling in the three-dimensional
city: applications in route planning, accessibility assessment, loca-
tion analysis and beyond, Journal of Transport Geography 19 (3)
(2011) 405 – 421. doi:10.1016/j.jtrangeo.2010.11.
007.
[94] B. Elias, Pedestrian Navigation - Creating a tailored geodatabase for
routing, 2007 4th Workshop on Positioning, Navigation and Com-
munication, 2007, pp. 41 – 47. doi:10.1109/wpnc.2007.
353611.
[95] J. Lee, A Three-Dimensional Navigable Data Model to Support
Emergency Response in Microspatial Built-Environments, Annals
of the Association of American Geographers 97 (3) (2007) 512 –
529. doi:10.1111/j.1467-8306.2007.00561.x.
[96] M. Aleksandrov, C. Cheng, A. Rajabifard, M. Kalantari, Modelling
and finding optimal evacuation strategy for tall buildings, Safety Sci-
ence 115 (2019) 247 – 255. doi:10.1016/j.ssci.2019.02.
017.
[97] A. A. Diakité, S. Zlatanova, K. J. Li, About the subdivision of
indoor spaces in IndoorGML, ISPRS Ann. Photogramm. Remote
Sens. Spatial Inf. Sci. IV-4/W5 (2017) 41 – 48. doi:10.5194/
isprs-annals- iv-4- w5-41-2017.
[98] J. Lim, P. Janssen, F. Biljecki, Visualising detailed CityGML and
ADE at the building scale, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLIV-4/W1-2020 (2020) 83–90. doi:10.5194/
isprs-archives- xliv-4- w1-2020-83- 2020.
[99] G. S. Floros, C. Ellul, E. Dimopoulou, Investigating interoper-
ability capabilities between IFC and CityGML LOD 4 – retain-
ing semantic information, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLII-4/W10 (2018) 33 – 40. doi:10.5194/
isprs-archives- xlii-4- w10-33-2018.
[100] H. Eriksson, L. Harrie, J. M. Paasch, What is the need for build-
ing parts? – A comparison of CityGML, INSPIRE Building and a
Swedish building standard, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLII-4/W10 (2018) 27 – 32. doi:10.5194/
isprs-archives- xlii-4- w10-27-2018.
[101] K. Arroyo Ohori, F. Biljecki, A. Diakité, T. Krijnen, H. Ledoux,
J. Stoter, Towards an integration of GIS and BIM data: what are the
geometric and topological issues?, ISPRS Ann. Photogramm. Re-
mote Sens. Spatial Inf. Sci. IV-4-W5 (2017) 1 – 8. doi:10.5194/
isprs-annals- iv-4- w5-1-2017.
[102] K. Arroyo Ohori, A. Diakité, T. Krijnen, H. Ledoux, J. Stoter, Pro-
cessing BIM and GIS Models in Practice: Experiences and Rec-
ommendations from a GeoBIM Project in The Netherlands, ISPRS
International Journal of Geo-Information 7 (8) (2018) 311. doi:
10.3390/ijgi7080311.
[103] F. Biljecki, H. Ledoux, J. Stoter, G. Vosselman, The variants of an
LOD of a 3D building model and their influence on spatial analyses,
ISPRS Journal of Photogrammetry and Remote Sensing 116 (2016)
42 – 54. doi:10.1016/j.isprsjprs.2016.03.003.
[104] INSPIRE Thematic Working Group on Buildings, D2.8.III.2 IN-
SPIRE Data Specification on Buildings – Technical Guidelines,
Tech. rep. (2013) [cited 13 April 2019].
URL https://inspire.ec.europa.eu/id/document/
tg/bu
[105] N. Nazarian, T. Sin, L. Norford, Numerical modeling of outdoor
thermal comfort in 3D, Urban Climate 26 (2018) 212 – 230. doi:
10.1016/j.uclim.2018.09.001.
[106] W.-J. Zhao, E.-X. Liu, H. J. Poh, B. Wang, S.-P. Gao, C. E. Png,
K. W. Li, S. H. Chong, 3D traffic noise mapping using unstruc-
tured surface mesh representation of buildings and roads, Applied
Acoustics 127 (2017) 297 – 304. doi:10.1016/j.apacoust.
2017.06.025.
[107] R. Bozovic-Stamenovic, N. Kishnani, B. K. Tan, D. Prasad,
F. Faizal, Assessment of awareness of Green Mark (GM) rating tool
by occupants of GM buildings and general public, Energy and Build-
ings 115 (2016) 55 – 62. doi:10.1016/j.enbuild.2015.
01.003.
[108] J. Stoter, S. Ho, F. Biljecki, Considerations for a contemporary
3D cadastre for our times, Int. Arch. Photogramm. Remote Sens.
Spatial Inf. Sci. XLII-4/W15 (2019) 81 – 88. doi:10.5194/
isprs-archives- xlii-4- w15-81-2019.
[109] R. Yaagoubi, A. Al-Gilani, A. Baik, E. Alhomodi, Y. Miky, SEH-
SDB: a semantically enriched historical spatial database for doc-
umentation and preservation of monumental heritage based on
CityGML, Applied Geomatics 1 (5) (2018) 41. doi:10.1007/
s12518-018- 0238-y.
[110] E. Colucci, V. D. Ruvo, A. Lingua, F. Matrone, G. Rizzo, HBIM-GIS
Integration: From IFC to CityGML Standard for Damaged Cultural
Heritage in a Multiscale 3D GIS, Applied Sciences 10 (4) (2020)
1356 – 20. doi:10.3390/app10041356.
[111] R. Boeters, K. Arroyo Ohori, F. Biljecki, S. Zlatanova, Auto-
matically enhancing CityGML LOD2 models with a correspond-
ing indoor geometry, International Journal of Geographical Infor-
mation Science 29 (12) (2015) 2248 – 2268. doi:10.1080/
13658816.2015.1072201.
[112] B. Hagedorn, M. Trapp, T. Glander, J. Döllner, Towards an Indoor
Level-of-Detail Model for Route Visualization, 10th International
Conference on Mobile Data Management: Systems, Services and
Middleware, 2009, pp. 692 – 697. doi:10.1109/mdm.2009.
118.
[113] S. Kemec, S. Zlatanova, S. Duzgun, A new LoD definition hierar-
chy for 3D city models used for natural disaster risk communication
tool, Proceedings of the 4th International Conference on Cartogra-
phy GIS, 2012, pp. 95 – 104.
[114] H.-Y. Kang, J. Lee, A Study on the LOD(Level of Detail) Model
for Applications based on Indoor Space Data, Journal of the Korean
Society of Surveying, Geodesy, Photogrammetry and Cartography
32 (2) (2014) 143 – 151. doi:10.7848/ksgpc.2014.32.2.
143.
[115] L. Tang, L. Li, S. Ying, Y. Lei, A Full Level-of-Detail Specification
for 3D Building Models Combining Indoor and Outdoor Scenes, IS-
PRS International Journal of Geo-Information 7 (11) (2018) 419 –
20. doi:10.3390/ijgi7110419.
[116] A. Labetski, S. v. Gerwen, G. Tamminga, H. Ledoux,
J. Stoter, A proposal for an improved transportation model
in CityGML, Int. Arch. Photogramm. Remote Sens. Spa-
tial Inf. Sci. XLII-4/W10 (2018) 89 – 96. doi:10.5194/
isprs-archives- xlii-4- w10-89-2018.
[117] H. Ledoux, K. Arroyo Ohori, K. Kumar, B. Dukai, A. Labetski,
S. Vitalis, CityJSON: A compact and easy-to-use encoding of the
CityGML data model, Open Geospatial Data, Software and Stan-
dards 4 (2019) 4. doi:10.1186/s40965- 019-0064-0.
Biljecki F et al.: Preprint submitted to Elsevier Page 18 of 18