Open Data Integration in 3D CityGML-based Models Generation

Mcdonnell Ara

ujo Maieron

and Jos

e Palazzo Moreira de Oliveira

Informatics Institute, Federal University of Rio Grande do Sul, Av. Bento Gonc¸alves, 9500, Porto Alegre, Brazil

Keywords:

Smart Cities, Data Integration, 3D Models, CityGML.

Abstract:

Facing the increasing complexity of large urban centers caused by population growth and the dynamic nature

of cities, their managers seek to optimize services and infrastructures in terms of scalability, environment, and

security to adapt to demand, making their cities smarter. Therefore, these new modern centers’ administrators

should apply smart governance techniques to manage the physical and data infrastructure and seek alignment

with the global open data initiative. As a point of intersection between physical and data infrastructure, 3D

models of cities have been playing an important role in people’s daily lives, being a fundamental element for

several applications. In this context, CityGML, a semantic model for 3D data representation adopted by several

cities, appears as a possible solution for modeling. This paper presents an approach of integrating open data

in the semi-automatic generation of 3D models based on CityGML, “enriching” semantic information about

the instances with the association with the OpenStreetMaps database. A case study was performed using data

provided by the Municipality of Porto Alegre, BR. The model generated in CityGML goes through semantic,

geometric, and schema level validations, proving the proposed approach’s feasibility.

1 INTRODUCTION

Facing the increasing complexity of large urban cen-

ters caused by population growth and the dynamic

nature of cities, their managers seek to optimize ser-

vices and infrastructures in terms of scalability, en-

vironment, and security to adapt to demand, making

their cities smarter (Khatoun and Zeadally, 2016). Al-

bino et al. (2015) cites several components of smart

cities, among them the most relevant are: smart econ-

omy, smart people, smart governance, smart mobility,

smart environment, and smart living, directly related

to aspects of urban life such as industry, education, e-

democracy, logistics, infrastructures, efﬁciency, sus-

tainability, security, and life quality.

All components can be associated in some way

with the city’s physical and data infrastructure. For

instance, smart mobility is related to the available

transport modes, and smart governance needs inter-

operable platforms and databases to provide online

services and shareable data, using a standardized

semantic-based data model that uniﬁes data format

and provides shared meaning to them.

As a point of intersection between physical and

data infrastructure 3D city models are essential. In

https://orcid.org/0000-0002-7098-7061

https://orcid.org/0000-0002-9166-8801

its review, Biljecki et al. (2015) cites several studies

related to the applications of 3D city models, among

them traditional ones such as urban planning, 3D reg-

istration, routing and also more speciﬁc studies such

as radio wave propagation, irradiation estimation, and

estimation of noise pollution propagation. Thus, it

is notable that 3D models should be considered an

essential part of the database of smart cities (Prandi

et al., 2013).

The inclusion of semantic aspects enhances the

capacity of the models. In studies related to disas-

ter management and emergency response, for exam-

ple, knowing the building’s function is necessary, as

strategic structures and hospitals require special at-

tention in this context (Gr

oger and Pl

umer, 2012).

Also, it is considered a good practice of governance

the adoption of international modeling standards by

the data providers to avoid heterogeneity and allow-

ing interoperability.

As a possible 3D model solution that covers both

the semantic aspects and the heterogeneity issue,

there is CityGML

. It is a shared semantic informa-

tion model for representing 3D urban objects man-

aged since 2008 by the OGC (Open Geospatial Con-

sortium) (Gr

oger et al., 2012). CityGML has been

adopted worldwide, and several cities, in line with the

https://www.ogc.org/standards/citygml

Maieron, M. and Moreira de Oliveira, J.

Open Data Integration in 3D CityGML-based Models Generation.

DOI: 10.5220/0010383201670174

In Proceedings of the 23rd Inter national Conference on Enterprise Information Systems (ICEIS 2021) - Volume 1, pages 167-174

ISBN: 978-989-758-509-8

167

government’s open data initiative, make their models

in this standard available for sharing.

Despite the absence of 3D data, some cities pro-

vide several 2D geographic data to generate 3D mod-

els when aggregated with altimetry data. In gen-

eral, these public data are generated from traditional

and well-established cartographic production meth-

ods, following a series of criteria and norms in the

production process. Consequently, their geometric

precision meets the quality standards stipulated for

this type of product. On the other hand, as a rule,

these data only describe the instances geometrically,

having little or no semantic information such as name,

address, class, and function, information of great rel-

evance for speciﬁc applications.

This article presents an hybrid approach for inte-

grating ofﬁcial open data in the semi-automatic gen-

eration of 3D models CityGML based, performing

the semantic “enrichment” of the instances associat-

ing ofﬁcial data with Volunteered Geographic Infor-

mation (VGI) data from OpenStreetMaps (OSM).

The expectation is that such an approach can

help city administrators who seek alignment with

smart city guidelines to generate their valid 3D se-

mantic models with reuse and integration of existing

data, taking advantage of the positive aspects of each

database, reducing costs in that process. In addition,

the approach is formalized in the form of workﬂows,

facilitating their reimplementation in different con-

texts.

This paper is organized as follows: Section 2

presents the main concepts. Section 3 describes the

proposed approach. Section 4 describes a study case,

commenting on primary results. Section 5 contains

the related work, and section 6 concludes the paper.

2 MAIN CONCEPTS

Data integration consists of combining data sets ob-

tained from different sources and providing a uniﬁed

version of them. It is an old and pervasive issue, found

in several scientiﬁc and governmental sectors (Lenz-

erini, 2001). As this is a broad subject, this section

only addresses issues related to spatial data.

Mohammadi et al. (2006) deals with the issue in

the context of implementing spatial data infrastruc-

tures and consolidates the technical aspect variables

to be analyzed in the integration process as follows:

Computational heterogeneity (standards and interop-

erability), Topology, Semantics, Reference system

and scale, Data quality, Data model and Metadata. In

these circumstances, international standards must be

followed, providing common concepts, such as (In-

ternational Organization for Standardization, 2013),

which deﬁnes six elements of spatial data quality de-

scribed below:

• Positional Accuracy: Comparison of the geo-

graphic coordinates of a feature with the geo-

graphic coordinates of the object it represents in

the real world within the same reference system;

• Thematic Accuracy: Accuracy of classiﬁcations

and themes associated with speciﬁc locations or

objects, that is, the class of a pixel in a land cover

image or label of a vector feature;

• Completeness: Presence or absence of features,

attributes or relationships regarding the speciﬁca-

tions of the ﬁnal product;

• Temporal Quality: Quality of temporal attributes

such as date of collection, date of publication, fre-

quency of updating or time validity;

• Usability: Alignment of data with the needs and

requirements of the end-user.

CityGML was designed as an open data model

based on XML, being an application scheme of

GML3 (Geography Markup Language 3), an exten-

sible international standard for sharing and coding

data spatial data issued by OGC and ISO TC 211

(Gr

oger et al., 2012), having an adequate structure

to present city models with semantic information.

These semantic features allow users to perform func-

tions with the metadata provided by CityGML. The

modules in CityGML reﬂect the appearance, spatial

and theme characteristics of an object and with the

model, common deﬁnitions of the attributes and re-

lationships between basic entities of a 3D city model

were achieved, taking into account semantic and geo-

metric/topological aspects.

In CityGML, it is possible to model the instances

in ﬁve different levels of detail (LoD). The LoD pro-

vides the adequacy of the amount and reﬁnement of

information data for speciﬁc end-user applications.

One of the implementations to maintain seman-

tic integrity in CityGML is the inclusion of codelists

with discrete values for ﬁlling in the attributes of the

features, thus avoiding common mistakes such as in-

correct typing or creation of values outside the at-

tribute domain. Such lists are speciﬁed in a schema

external to CityGML, and it is only referenced in the

model.

An example of codelist used as a reference in the

model’s technical speciﬁcations can be found at http://www.

sig3d.org/codelists/standard/

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

168

3 APPROACH DESCRIPTION

Due to each module of the CityGML model’s partic-

ularity, different workﬂows are deﬁned according to

the need for processes to achieve the desired integra-

tion.

With a realistic view, where it is not easy to ﬁnd

ofﬁcial data sources that provide detailed information

on features, such as positioning of windows, doors,

divisions of internal spaces, and other details that are

part of the features and necessary for the implemen-

tation of modules in LoD3 and 4, such levels of detail

were not included in the approach. Additionally these

details are not interesting for our application.

The workﬂow representing the proposed approach

to Building module is shown in Figure 1, with the

steps to be followed in the process of integrating

spatial data, improve semantics with OSM data and

implementing the CityGML-based model. Due to

space limitations, the workﬂows proposed for the

other modules have been omitted from this article.

It is important to note that the proposed approach is

directed to 2D input data aggregating altimetry in-

formation from the intersection with Digital Terrain

Model (DTM), representing ground data, and Digital

Surface Models (DSM), representing altimetry data of

all features above ground level like buildings, trees,

and bridges. However, there are different techniques

for generating 3D geometries in addition to this (e.g.,

photogrammetry). Still, parts of the process such as

integration with VGI data and LoD2 building genera-

tion can be used in other studies with different meth-

ods to create 3D geometries.

4 CASE STUDY

In this section, a practical application of the devel-

oped approach is carried out. Initially, the test area’s

characterization and input data are made, followed

by 3D modeling of each CityGML modules imple-

mented. Finally, the generated model’s validation re-

sults are presented at three levels: schema, geometric,

and semantics.

4.1 Test Area

Porto Alegre is the capital of the state of Rio Grande

do Sul - BR. Its estimated population is 1,488,252 in-

habitants and its total area is 495,390 km

. The city

is administratively divided into 94 neighborhoods, of

which three were chosen as the test area: Menino

Deus, Santa Tereza, and Praia de Belas. This area has

heterogeneous features, involving sports complexes,

shopping centers, parks, residential areas, and rugged

relief, having the necessary characteristics for apply-

ing the case study. The area encompassing the three

neighborhoods is 9,394 km

4.2 Spatial Data Sources

The Municipality of Porto Alegre (PMPA), through

the Municipal Department of Environment and

Sustainability (SMMAS/PMPA), provides the geo-

graphic data resulting from the 2010 cadastral map-

ping project as open data, following the guidelines of

intelligent governance concerning government data.

The ﬁles can be found in PDF and DWG formats

This database provides the 2D geometric contour of

the features. The ﬁnal cartographic product gener-

ated was classiﬁed as “Class A” according to the

levels of planimetric tolerance stipulated in BRASIL

(1984) being compatible as inputs for implementing

CityGML models up to LoD3.

The altimetry data comes from the point cloud

obtained for the same mapping project, divided into

DTM and DSM. The average density is 2 pt/m

The data used for the semantic “enrichment” of

the features were obtained from the OSM. Because

it is a global open data source and developed by col-

laborative mappers, the ﬁnal product, despite under-

going a series of validation processes, is not obtained

through formal and well-established cartographic pro-

duction methods. Therefore, positional accuracy and

the levels of generalization of the features differ from

the databases provided by the PMPA. On the other

hand, the description of the features through attributes

is much better in the OSM data than the ofﬁcial data.

Besides the categorization, OSM data is described by

tags, where the mapper can include additional infor-

mation.

4.3 3D Modeling

The primary tool used in data modeling and integra-

tion is the FME

. For some digitization steps that

were performed manually, the open-source QGIS

was used.

4.3.1 Building Module

The ﬁrst common step in the implementation of

the Building module and other modules is the pre-

processing step, where the geometric integrity of the

http://www2.portoalegre.rs.gov.br/spm/default.php?

p secao=310

https://www.safe.com/

https://qgis.org/en/site/

Open Data Integration in 3D CityGML-based Models Generation

169

Figure 1: Building module integration workﬂow.

input features is checked, and the initial classiﬁcation

of ofﬁcial data is also mapped to the corresponding

CityGML codelists.

As the ofﬁcial database does not have address

data, reverse geocoding was performed automatically

using the transformer geocoder. A script in Python

was used to adapt the returned address format to the

xAL format standards used in CityGML.

Although there is an initial classiﬁcation of build-

ings in the ofﬁcial database, 95.08% of them are

classiﬁed as generic, lacking a semantic description.

Thus, an alternative found to minimize this missing

information is integration with the OSM database.

Initially, the geometric association between the fea-

tures is made, and, later, the attributes are mapped, as-

sociating the feature attribute of the OSM with those

existing in the codelists in the CityGML model. To be

considered corresponding features, the criteria used in

the building category are the comparison of the total

area of the polygons and the distance from the central

point of the building. Features with up to 36% of area

variance and a center point, not 5m apart, are consid-

ered corresponding. This wider range was necessary

due to the difference in some aspects of mapping be-

tween the two bases. Figure 2 shows particular cases

found in the geometric integration process.

The cases of non-association between the two

Figure 2: In the mosaic, the yellow outlines represent the

OSM database, the red outlines represent the features of

the ofﬁcial database that were not associated with the OSM

base, and the blue outline the features that were related. Im-

age source: Google.

bases shown in Figure 2 are related to the difference

in quality levels and generalization between both. In

case A, the difference in temporal accuracy between

the databases leads to the non-association. Case B

shows the difference in positional accuracy between

the bases caused by the divergence of reference sys-

tems and data acquisition methods. The difference

between the levels of generalization between the fea-

tures is evidenced in case C. Finally, case D shows an

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

170

Table 1: Number of features from the ofﬁcial base with

OSM attributes associated.

Attribute class function name SAG

Features 19 147 116 14

Attribute operator website description YOC

Features 29 73 31 10

example of the association between features. Despite

the differences in generalization and positional accu-

racy, the buildings passed the comparison criteria.

From 20.093 features present in the input data,

19.405 found a geometric match. The mapping of the

values of the OSM tags to the codelists was done pri-

marily automatically using the token sort ratio pro-

cess from the fuzzywuzzy

available for Python. Unas-

sociated values were analyzed individually to comple-

ment the mapping. Although, Table 1 shows a low

number of attributes added to the features of the ofﬁ-

cial base. This shows that, despite having good geo-

metric completeness and temporal quality, the OSM

data have low thematic completeness in the study

area. Table 2 shows an OSM data thematic complete-

ness comparative in four cities. We can see that the

other cities have a better collaborative mapping cul-

ture, which would allow a more signiﬁcant aggrega-

tion of attributes in these regions. In this compari-

son, test areas that were visually similar to the test

area of this work were used, not encompassing cities

as a whole. The attribute SAG and YOC stands for

“storeys above ground” and “year of construction”,

respectively.

Regarding geometry, a ﬁlter is applied according

to the ground surface’s area and height to deﬁne the

LoD of each instance. In the Building module, fea-

tures up to LoD2 were implemented. To fully ex-

ploit the potential of CityGML within the possibili-

ties of the input data, the topological relationship be-

tween surfaces and buildings was implemented using

the GML3 Xlink mechanism.

Surfaces shared between two buildings are iden-

tiﬁed with the transformer SurfaceOnSurfaceOver-

layer, and they are assigned a new unique identiﬁer.

Thus, the shared surface geometry is implemented

only once and referenced by the two buildings that

have a hierarchical relationship, avoiding redundancy

in the implementation of geometries. A shared sur-

face must be referenced in one of the buildings, in-

verting vertice’s orientation. It is necessary so that

the visualization platforms can correctly interpret the

geometries to display them in both buildings.

Figure 3 shows the study area features imple-

mented in this module.

https://pypi.org/project/fuzzywuzzy/

Figure 3: Building module overview in the study area.

4.3.2 CityFurniture Module

The implementation of the CityFurniture module is

simpler and straightforward and does not require in-

tegration with the OSM base. An additional phase

is the implementation of implicit geometries in the

classes light pole, energy tower and bus stop based

on pre-existing 3D models in 3DS format. Thus, the

geometry of the 3DS model is referenced, based on

the geographic coordinate of the original feature.

4.3.3 Vegetation Module

The Vegetation module implementation covered the

two classes provided by CityGML, PlantCover and

SolitaryVegetationObject. As in the CityFurniture

module, the features of the SolitaryVegetationObject

classes were associated with 3DS models as implicit

geometry. Due to the high number of SolitaryVegeta-

tionObject class features in the study area, a simple

3DS model was adopted to avoid overloading the vi-

sualization platforms. The tree’s height is obtained

from the DSM intersection so that the ﬁlter can be ap-

plied according to the LoD implemented. To keep the

ﬁnal module clean and due to some inconsistencies

in the input data like vegetation polygons overlapping

other features, the PlantCover class was implemented

only as surfaces instead of solids.

4.3.4 WaterBody Module

In the ofﬁcial database, three data sets together form

the Hydrography class. The ﬁrst contains the lake fea-

tures with polygon type geometry. The next two have

linear-type geometry but describe features differently.

In one, rivers and channels are described by their cen-

tral axis and, in the other, contour’s margins describe

the same features. A manual step was performed

to enable the implementation of features with Multi-

Surface geometry, creating polygon-type geometries

based on the linear geometry of the margins. In this

step, QGIS software was used.

As in the other modules, the initial classiﬁcation

of features is mapped to CityGML codelists. As the

Open Data Integration in 3D CityGML-based Models Generation

171

Table 2: Thematic completeness comparative in OSM data.

City # of Buildings

Attribute

”name” (%)

Building

Categorization (%)

YOC (%) SAG (%)

Average tags

per building

Porto Alegre 21.807 0,8 3,4 0 0 2,14

London 8.470 12,6 58,5 0,2 17,8 4,71

Berlin 8.398 5,2 70,1 1,0 61,8 3,7

Amsterdam 22.981 0,6 57,9 97,9 0,3 5,1

parameters of the generalization ﬁlters for each LoD

were not formalized in the model’s technical speci-

ﬁcations, all features existing in the ofﬁcial database

were implemented in LoD0 and 1.

4.3.5 Transportation Module

In the same way, as in the Watebody module, a manual

digitization step was performed to generate polygon

type geometries based on the data set containing the

roads’ external contour. Thus, LoD1 implementation

is possible.

Features with line-type geometry intersect with

the MDT and are converted directly to LoD0Network.

In polygonal features, a ﬁlter is applied to identify the

elevated features, such as viaducts, to aggregate the

altimetry data from the intersection with the DSM.

The other polygonal features receive the altimetry

data from the DTM.

4.3.6 LandUse and Relief Modules

The implementation of the LandUse module is sim-

pler since most of the geometries are absorbed from

other modules. In this way, only the features that do

not belong to any other module go through the inter-

section with the MDT to aggregate the altimetry data.

After that, the attributes are converted to the codelists

of the LandUse module.

After mapping attributes, the spatial relationship

analysis between layers is performed. If surfaces

overlap and the attributes of function and use are dif-

ferent, they are accumulated.

The implementation of the Relief module is

straightforward. The feature’s base geometry imple-

mented in the other modules is placed as breaklines

and the DTM point cloud as the primary input in the

triangulation. In this way, the ﬁnal terrain model will

automatically adapt to the features’ contour, main-

taining the topological integrity between the modules.

Figure 4 presents an overview of the implemented

Relief module.

4.3.7 Validation and Data Management

Validation was performed at three levels: seman-

tic, geometric, and syntactic (or schema) to verify

Figure 4: Sample of features from Relief module.

the generated model’s integrity. The semantic anal-

ysis includes plausibility checks of attributes, allow-

ing only attributes compatible with the CityGML at-

tribute structure and type. The syntax check ensures

that the read data is conformant to the OGC standard

document for CityGML 2.0 and the geometric level

validate 3D primitives according to the international

standard ISO19107 (Wagner et al., 2013).

The ﬁrst stage of validation is done by checking

the validation option in the CityGML writer in FME

Workbench. With that, the CityGML XSD is used

for validating the implementation at the schema level.

Some aspects of semantics, such as duplicate unique

identiﬁers and ﬁlling in some attributes that are de-

pendent on others, are also veriﬁed. In this step, the

inconsistencies presented were corrected by editing

the work ﬂow in the FME.

To continue the validation, two open source tools

were used: Val3dity (Ledoux, 2018) and 3DCityDB

(Yao et al., 2018). Val3dity deals with the geomet-

ric validation of features. Only the Building mod-

ule presented inconsistencies generated when iden-

tifying shared surfaces. Thirteen errors type 101

(TOO FEW POINTS) were found generated when a

line-type feature is created when the surfaces overlap

were identiﬁed. It occurs due to a rounding error in

the 4

decimal place. The elimination of the linear

feature can be done without loss of information.

All other errors are related to the orienta-

tion of the surfaces. 60.5% of the features pre-

sented errors of type 302 (SHELL NOT CLOSED),

303 (NON MANIFOLD CASE), or 307 (POLY-

GON WRONG ORIENTATION). To correct this

type of error, invert the order of the indicated polygon,

but the validation report shows the building’s unique

identiﬁer, not the surface, making it difﬁcult to iden-

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

172

Table 3: Approach, completeness and validation compara-

tive in related works.

Rel.

work

Approach # of

Modules

Validation

Auto Semi Sch Geo Sem

(1) X 1 X X

(2) X 1 X

(3) X 8 X X X

(4) X 6 X

(5) X 3 X X X

(6) X 1

(7) X 7 X X X

tify the features for correction.

The Building module was also implemented with-

out the step of identifying shared surfaces, which is

optional. There were no inconsistencies presented by

the tool in this version, ensuring the ﬁnal model’s in-

tegrity.

Once generated, 3D models can be stored in spa-

tial databases, facilitating their maintenance and man-

agement. For this, the 3DCityDB

platform was used,

a free 3D geo-database solution to import, manage,

analyze, visualize, and export virtual 3D city models

according to the CityGML standard. When loading

the model, the data goes through a scheme validation

phase again, conﬁrming the ﬁnal model’s quality.

5 RELATED WORK

Due to the growing acceptance of CityGML as a

model for implementing semantic 3D models, sev-

eral studies have been published presenting case

studies and reporting the modeling experiences with

different input data scenarios and integration plat-

forms. Most of the works deal with the integration

of ofﬁcial open data for semi-automatic generation

of the 3D model (Agugiaro, 2016)(1)(Buyukdemir-

cioglu et al., 2018)(2) (Soon and Khoo, 2017)(3).

Kolbe et al. (2015)(4) and (Jane

cka, 2019)(5) used

an automatic method to generate the CityGML-based

model of New York City and Prague, respectively.

Other approaches suggest the automatic generation

of CityGML-based models from OSM data (Goetz,

2013) (6). A difference also noted is the number of

CityGML modules and validations implemented. Ta-

ble 3 shows a comparison of the cited studies con-

sidering the approach used, modules implemented,

and validations applied to the generated models. The

present work is indicated as the item (7).

This work has similar points with the previous

research, differentiating itself by proposing a hybrid

approach of using ofﬁcial open data and VGI data

https://www.3dcitydb.org/3dcitydb/

to mitigate the lack of thematic completeness of the

datasets available by government agencies. It is also

comparable to studies of a broader scope regarding

the number of CityGML modules implemented and

validations performed. Also, an approach for auto-

matic identiﬁcation of shared surfaces in the Building

module is applied, with the limitations presented in

Section 4.3.7.

6 CONCLUSIONS

This paper presented a hybrid approach for semi-

automatic generation of 3D semantic CityGML-based

models from ofﬁcial databases and semantic ”enrich-

ment” from OSM database integration. The FME

workspaces used in the implementation, as well as

the spreadsheets used in mapping OSM attributes to

CityGML codelists will be available at https://github.

com/mcmaieron/DWG CityGML. the workspace was

designed in alignment with the case study’s input data

but can be used in a general scenario with some adap-

tations.

Thus, this work can assist city administrators who

want to start implementing a 3D database following

standards adopted worldwide.

Due to the lack of information about the hierarchy

between features in the input data, it was not possible

to use the full potential of CityGML, such as the im-

plementation of cityGenericGroup, BuildingParts and

BuildingInstalation

Some suggestions for future work to complement

the proposed approach are:

• Possible extension of CityGML for national pur-

poses, following van den Brink et al. (2013),

adding codelists speciﬁc to the national territory

mapping context, based on the current mapping

standards;

• Add semantic validation method to the VGI used;

• automatic identiﬁcation of roof format from a

point cloud, aiming to bring the features closer to

reality;

• improve the implementation of the identiﬁcation

of shared surfaces, increasing their efﬁciency and

avoiding validation errors;

• implementation of the bridge and tunnel modules

automatically;

ACKNOWLEDGEMENTS

This work was partially funded by one author indi-

vidual grant from CNPq, Brazil and also by institu-

Open Data Integration in 3D CityGML-based Models Generation

173

tional grants from the Ministry of Education’s agency

CAPES, Brazil - Finance Code 001.This work was

only possible with the support of the Brazilian Army

and the Board of Geographic Service (DSG).

REFERENCES

Agugiaro, G. (2016). First steps towards an integrated

citygml-based 3d model of vienna. ISPRS Annals of

Photogrammetry, Remote Sensing and Spatial Infor-

mation Sciences, III-4:139–146.

Albino, V., Berardi, U., and Dangelico, R. M. (2015). Smart

cities: Deﬁnitions, dimensions, performance, and ini-

tiatives. Journal of Urban Technology, 22(1):3–21.

Biljecki, F., Stoter, J., Ledoux, H., Zlatanova, S., and

C¸

oltekin, A. (2015). Applications of 3D City Models:

State of the Art Review. ISPRS International Journal

of Geo-Information, 4(4):2842–2889.

BRASIL (1984). Decreto nº 89.817, de 20 de junho de

1984. estabelece as instruc¸

oes reguladoras das normas

ecnicas da cartograﬁa nacional.

Buyukdemircioglu, M., Kocaman, S., and Isikdag, U.

(2018). Semi-Automatic 3D City Model Generation

from Large-Format Aerial Images. ISPRS Interna-

tional Journal of Geo-Information, 7(9):339.

Goetz, M. (2013). Towards generating highly detailed

3D CityGML models from OpenStreetMap. Interna-

tional Journal of Geographical Information Science,

27(5):845–865.

oger, G., Kolbe, T. H., Nagel, C., and H

afele, K.-H.

(2012). OGC City Geography Markup Language

(CityGML) Encoding Standard. page 344.

oger, G. and Pl

umer, L. (2012). CityGML – Interopera-

ble semantic 3D city models. ISPRS Journal of Pho-

togrammetry and Remote Sensing, 71:12–33.

International Organization for Standardization (2013). Ge-

ographic information — data quality. Standard ISO

19157:2016, Geneva, CH.

Jane

cka, K. (2019). Transformation of 3D geospatial data

into CityGML – a case of Prague. Reports on Geodesy

and Geoinformatics, 107(1):41–48.

Khatoun, R. and Zeadally, S. (2016). Smart cities: con-

cepts, architectures, research opportunities. Commu-

nications of the ACM, 59(8):46–57.

Kolbe, T. H., Burger, B., and Cantzler, B. (2015). Citygml

goes to broadway. In Fritsch, D., editor, Photogram-

metric Week ’15, pages 343–356, Stuttgart. Institute

for Photogrammetry, University of Stuttgart, Wich-

mann.

Ledoux, H. (2018). val3dity: validation of 3D GIS prim-

itives according to the international standards. Open

Geospatial Data, Software and Standards, 3(1).

Lenzerini, M. (2001). Data integration. In Proceedings

of the Twentieth ACM SIGMOD SIGACT SIGART

Symposium on Principles of Database Systems, pages

233–246, New York, NY, USA. Association for Com-

puting Machinery.

Mohammadi, H., Binns, A., Rajabifard, A., and

Williamson, I. (2006). Spatial data integration. In

Proceedings of the Seventeenth UN Regional Carto-

graphic Conference for Asia and the Paciﬁc. United

Nations.

Prandi, F., De Amicis, R., Piffer, S., Soave, M., Cadzow,

S., Gonzalez Boix, E., and D’Hont, E. (2013). Us-

ing citygml to deploy smart-city services for urban

ecosystems. ISPRS - International Archives of the

Photogrammetry, Remote Sensing and Spatial Infor-

mation Sciences, XL-4/W1:87–92.

Soon, K. H. and Khoo, V. H. S. (2017). Citygml modelling

for singapore 3d national mapping. ISPRS - Interna-

tional Archives of the Photogrammetry, Remote Sens-

ing and Spatial Information Sciences, XLII-4/W7:37–

42.

van den Brink, L., Stoter, J., and Zlatanova, S. (2013). Es-

tablishing a national standard for 3D topographic data

compliant to CityGML. International Journal of Ge-

ographical Information Science, 27(1):92–113.

Wagner, D., Alam, N., and Coors, V. (2013). Geometric val-

idation of 3D city models based on standarized quality

criteria, page 197–210. CRC Press.

Yao, Z., Nagel, C., Kunde, F., Hudra, G., Willkomm, P.,

Donaubauer, A., Adolphi, T., and Kolbe, T. H. (2018).

3DCityDB - a 3D geodatabase solution for the man-

agement, analysis, and visualization of semantic 3D

city models based on CityGML. Open Geospatial

Data, Software and Standards, 3(1):5.

ICEIS 2021 - 23rd International Conference on Enterprise Information Systems

174