MULTISCALE REPRESENTATION: A DATA GENERALIZATION WORKFLOW AND
QUALITY EVALUATION EXPERIENCE IN GIS ENVIROMENT
LU. Falchi * *, G. Vacca ^ E. Falchi *
* Dipartimento di Ingegneria Strutturale, Sezione di Topografia Piazza D'Armi 09123 Cagliari -
(ufalchi,vaccag,falchi)@unica.it
KEY WORDS: Generalization, Abstraction, Database, GIS, Parameters, Representation, Standards, Quality
ABSTRACT:
In geographic data representation, users need specific tools for the representation and evaluation of the data on which the decision-
making process relies. In cartographic applications, users need to deal with maps of the same geographic region at different levels of
abstraction. In most cases one database per level of detail is maintained; a multiscale representation GIS would allow storage of all
original geographic information in a single database and enforce the consistency of the different representations through appropriate,
automatic generalization processes.
In this paper we deal with the problem of geographic data generalization, beginning from a generalization workflow by improving
specific routines.
To reach the goals of the generalization process, we used the ESRI ArcGIS 8.1 platform. Inside this environment some generalization
functions were implemented by improving appropriate routines in Visual Basic language. After a presentation of the generalization
workflow, all geometric, topological and semantic cartographic constraints governing the generalization process are specified. In the
next step these constraints will be translated into tools for assessing the need and the quality of generalization solutions. These tools
are associated with a particular data model. For that data model the constraints can be best translated to measures and generalization
algorithms and parameterized. This framework is based on constraints used to control the process and represent user requirements in
such a way that the system is capable of reaching a quality close to these requirements; it reflects the idea that more than one
acceptable solution to a given generalization problem may exist. The goal was the production of a topographie map at scales
1:50000, 1:100000 and 1:250000 from the "Carta Tecnica Regionale", (CTR, official regional cartography) of the "Regione
Autonoma della Sardegna", RAS, (Regional Administration of Sardinia), at scale 1:10000, in a GIS environment.
In accordance with official national and international standards, we provide the definition of the characteristics and associated quality
evaluation tool to be used in validating geographic data.
1. INTRODUCTION information which, when properly filtered and checked, can
represent single covers.
The use of GIS instruments for the management and analysis of In most cases it is thus necessary to intervene beforehand on the
territorial, geographic and environmental data has led users to cartographic database so as to generate the necessary themes by
collect an enormous amount of data at different levels of detail; means of an analysis of the encoding used, the geometric
despite the large storage capacity of modern information incongruities and their subsequent reorganization.
instruments, there remains the problem of their integration
owing to the lack of standardization, especially in the
generation of the appropriate metadata. One solution to this 2. CARTOGRAPHIC GENERALIZATION
problem is represented by cartographic generalization
operations performed in the GIS environment. By the term cartographic generalization is meant a set of
However, despite the widespread diffusion of territorial procedures leading to the creation of a representation of
information systems, the need for an in-depth revision of geographic information, the level of detail of which varies as a
retrieval methods of the cartographic databases required for function of the scale of analysis or visualization. The objective
their use within GIS has not yet been completely accepted by of such a technique is the production of maps at a smaller scale
companies, professionals and public administrations. than the original; for this reason, more or less complex
The organization of cartographic data has shed a pitilessly operations to vary the content of the map itself or the way in
strong light on the fact that the numeric cartography produced which it is represented as a function of the graphic requisites or
thus far is not suitable for the purpose in that it springs from the of the characteristics and importance of the objects present
use of numerical retrieval operations taken almost entirely from therein (Keates, 1989), are required. We can distinguish two
analogical ones. kinds of intervention: generalization of the information database
It is thus necessary to define the characteristics that the and cartographic generalization. The former refers to acquisition
cartographic data to be processed must have in order to be of specific information from the source data and its input into a
sufficiently suitable for use within a GIS. It is important to note main database from which new databases, with fewer details
that in any cartographic work, objects are identified by codes than the main one, are derived for analysis or applications at
which, taken singly, do not represent any theme but only a basis different scales.
for the implementation of the coverage of the GIS. And the fact Cartographic generalization, on the other hand, produces
that the covers cannot be obtained by simply joining codes graphic results or displays of the database in the form of printed
together must be underscored. In such a case, cartographic maps or digital images at different scales.
production should take place after planning the information Both cartographic generalization and database generalization
system. Despite this, the maps now produced contain reduce data complexity, the former with reference to map
72
m eek NUR M fre
— re
n I A ES
Ce
pe
W
rc