Roland Billen
user’s needs for 3D information ? Or is it the users who don’t know what 3D can give them ? Usually users don’t know
their 3D needs. Even if 3D needs are expressed for some applications (for example the propagation of mobile phone
signal), the most urban studies are still tackled in 2D. 3D information is still considered by most users as an extension of
2D database in which 3D is added partially. The major limitation for the implementation of 3D GIS or 3D model is the
inheritance of the old 2D representation model. Just as the biggest difference between 2D and 3D conception method
lies in the fact that the referential framework changes drastically. The way to build 2D GIS is the often the
computerisation of formalised GI such as maps. While for building a 3D GIS a thoughtful study of the phenomena is
needed. This implies to enrich the descriptive study of conception methods by basic spatial concepts. Urban
applications are classified in two categories : the first one includes applications already studied in 3D (few town
planning studies) and new applications where 3D is a crucial parameters (mobile phone propagation signal); the second
one gathers other applications. The former one allows the formalisation of 3D thinking and of 3D GIS conception
methods. The later one provides the means to enrich the step of needs evaluation in a prospective way and to improve
the conception methods.
While the urban physical environment is in 3D, the human perception of this environment is according to (Marr, 1982)
either in 2.5D or in 3D. So, one can think that the most adequate model for any geographical event must be in 3D. The
only reason not to use 3D model is either technical or conceptual. By technical we mean the availability of data, the lack
of adequate data structure and process, etc. By conceptual we mean the fact that the geographical phenomena can be
explained nearly completely without the third dimension. It is common to take only the principal variables of a
phenomena in order to make a model of it. So, if one considers that in a close future no more technical limitations will
remain, the only reason to stay in 2D world must be conceptual. The problem of the integration of 3D in the urban GIS
would then need a primary study about the principal variables of the geographical phenomena to be modelled. It should
be done during the step of needs evaluation advocated by all the conception methods.
3 EFFECTS ON THE GIS CONCEPTION METHODS
There are several conception methods for Information System (MERISE, etc.), but not so many especially for GIS
(MECOSIG, etc.). It seems reasonable to say that the 3D nature of the information must be taken into account in GIS
conception methods. But what is the influence of the introduction of 3D in these methods ? A way to answer is to look
at a simplified scheme (Figure 1) of the process of a GIS Conception method, MECOSIG (Pantazis, 1996). Note that
other methods could be used and would not present big difference in this descriptive study.
ORGANISATION
DATA FLOW
DATA
PROCESSING FUNCTIONS
INTEGRATION CONTROL
Detailled
DESCRIPTIVE LEVEL
What ? Why ? N?
Conceptuat
CONCEPTUAL LEVEL influence g
What ? Why ? of 3D 1
with formal description
la
ORGANISATIONAL LEVEL | Technical
influence
How ? Who ? When ? Where ?
a of3D ?
N
LOGICAL & PHYSICAL LEVEL NN
How ? Who ? When ? Where ?
Figure 1. Possible technical and conceptual influences of 3D in the GIS conception progress
In MECOSIG, four abstraction levels are considered : - the descriptive level; - the conceptual level; - the organisational
level; - the logical & physical level. All the preoccupation classes (the organisation, the data, the data flow, the
processing function and the integration controls) must be considered at the different abstraction levels. The process of
the conception method begin with a global consideration of the preoccupation classes and goes to a more detailed study.
Usually, the conception is more descriptive and conceptual at the beginning and becomes more organisational and
80 International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B3. Amsterdam 2000.
au. o i P fiu w umi