PREPARING THE UPDATING OF THE 
BELGIAN TOPOGRAPHICAL DATABASE ; A CHALLENGING PROJECT 
J. Henrion, J. Beyen, J. Vanommeslaeghe 
National Geographical Institute, 1000 Brussels, Belgium — (jhe, jbe, jvo)@ngi.be 
Commission IV, WG IV/1 
KEY WORDS: 
ABSTRACT: 
Photogrammetry, Updating, Three-dimensional, Database, Project 
There is more to preparing the updating of a topographical database than the conception of an appropriate data model. We needed to 
thoroughly study the available internal and external data sets. Further we needed to decide on how to get the best part from each data 
set, e.g. by writing a Visual Basic application for combining structured and identified 2D data with the z-values of the parent 3D data. 
This was the best solution for most object types. However, for some object types it was better to project the 2D data on a DEM. 
Moreover, it is very useful to know the number and kinds of modifications to be expected in the landscape. A good knowledge of both 
the available data and the modifications in the landscape and hence of the amount of work to do allows to define realistic requirements 
for the future: updating cycles, selection criteria, tolerances for absolute and relative accuracy. The confrontation of these requirements 
with the (continuously evolving) technical possibilities and with the production capacity of the available means for updating will 
determine detailed needs for development, investments and training. It is also the basis for outlining the procedures that should be 
followed in order to meet the requirements. 
1. INTRODUCTION 
In 2001, the National Geographical Institute (NGI) developped a 
strategical plan. In this context different strategical projects were 
started, one of which is called 'Seamless Geographical 
Information System of Reference' (SGISR). The objective of this 
project is to establish: 
- a complete production line for collecting, treating and 
integrating update data in the topo-geographical reference data; 
- acentralized, seamless GIS for stocking, managing, editing 
and distributing the topo-geographical reference data; 
- the tools that are necessary for the NGI's applications on basis 
of the topo-geographical reference data in the scales ranging 
between 1:10.000 and 1:50.000. 
This objective is to be reached by the end of 2005. In view of the 
complexity of this project, nine associated projects have been 
started to deal each with some specific aspects, amongst which: 
External information sources, Change detection, Generalization 
and Updating. The present paper mainly deals with the updating 
project. 
In 1988 the NGI started building extremely rich digital 
topographical reference data. These base data are to be complete 
in 2006. Most base data have been produced according to the 
following (simplified) workflow: 
a) data classification: noting down in the field all semantical 
contents of the future base map on black and white aerial 
photographs 
b) stereoplotting: drawing accurately the 3D geometry in a CAD- 
file, distinguishing about 40 different kinds of elements, i.e. a 
rough classification; the resulting ‘3D line’ is accurate and 3D, 
but not topologically cleaned 
c) topological cleaning at the expense of loosing the z-values, 
detailed identification distinguishing over 300 codes + 
structuring; result: “Topl0V-GIS’. 
A second production flow at NGI results in Top50V-GIS: 
seamless 2D data for the whole country on a 1:50.000 scale. 
50% of these data result from the generalization of Top 10V-GIS. 
The other 50% were produced separately. Top50V-GIS is being 
updated in a 5 years’ cycle on basis of field survey and 
orthophotographs. Its second edition is to be finished in 2006. 
One of the goals of the SGISR-project is to migrate from several, 
separately updated datasets to an object-oriented topographical 
reference database which is updated through one process. A 
major step to reach this goal will be to combine the best part from 
each data set into the database, by adding z-values to *Topl0V- 
GIS’ and by eventually taking into account less accurate but more 
up-to-date information from Top50V-GIS. 
The workflow for updating the reference data will differ 
considerably from the original datacollection. We will integrate 
external data that meet our quality requirements. Stereoplotting 
will be used to compare the database contents with recent images 
and to update the data. We intend to update the roads more 
frequently than the other objects and will use remote sensing for 
detection of most recent changes for roads. The in-the-field 
collection of data will be considerably different from the present 
data collection. The present in-the-field updating of the 1:50.000 
seamless database by means of pen computers, provides valuable 
information on the performance of processes with pen computer 
and GPS; it will also soon be equipped with voice-command. The 
update rate and the in-the-field collection of data will also vary 
in function of the objects to be updated. Ideally the goal is a 
yearly update of roads, a two-yearly update of buildings and a 
five-yearly general update of the information in the database. 
2. THE FUTURE DATA MODEL 
The reference-data will be structured in an object oriented way. It 
takes good knowledge of the used definitions and selection 
criteria of all objects to be able to conceive an appropriate data 
model. The main difficulty lies in the conceptual choices to be 
made, part of which were treated by our project working group: 
Interna 
for ans 
on the 
this pa 
a) Hov 
+ Whi 
One s 
compai 
produc 
(1998) 
way, re 
and bu 
methoc 
neither 
hand, f 
NMA’: 
with u 
upgrad 
The otl 
based « 
develo] 
OS-En 
among 
amoun 
For se 
*classic 
We prc 
keep re 
inform: 
intentic 
nor of 
vertex 
procedi 
for cha 
structur 
indicat 
a mini 
  
Feat 
-ID 
-geor 
-attril 
  
-geor 
-geor 
-attril 
-load 
-geon 
date 
-attril 
(opt 
  
  
The che 
follows 
Add = 
Mod = 
Del =