DEFINING AND DETECTING CHANGES IN URBAN AREAS 
A. Bianchin* , L. Bravin * 
* GEDDeS Laboratory, DP, University IUAV of Venice, Ca' Tron, Santa Croce 1957, 30135 Venice, Italy 
alberta@iuav.it , Ibravin@iuav.it, fax 39 041 5240403 
Commission VII. WG VII/4 
KEYWORDS: Remote Sensing, Human Settlement, Change Detection, Monitoring, Analysis 
ABSTRACT: 
The general issue developed in this paper is the consistency between the visual description of urban areas and quantitative 
description obtained through analysis techniques and landscape ecology indexes. Three experiences have been carried out: the first 
applies landscape ecology indexes to different typologies of urban structures in order to verify whether they can qualify such 
typologies. The second applies some tools of analysis to different resolution images of the same test area. The third analyses the 
results and the significance of landscape descriptors, calculated over different urban areas at different scales. Such experiences allow 
us to gain a better understanding about 1) the efficiency of these descriptors and their limitations in characterizing urban structures, 
2) behaviour and significance of descriptors when applied to different resolution, which provides evidence of scale relevance. 
1. INTRODUCTION 
The description of urban phenomena needs to explicit variables 
and categories by which we can either differentiate different 
typologies or subdivide the phenomena into the parts — also 
differentiated - they are composed of. Description does imply 
criteria and related differentiations "named" within the various 
criteria. 
Defining such categories depends on the possibility to determine 
them theoretically in relation to a given discipline, or practically 
in relation to an analysis technique. 
In general indexes and classes provided by analysis techniques 
have to become useful in order to detect new or already existent 
categories. 
Classes and indexes do have their particular definition in 
description of datasets in relation to the analysis technique (for 
example morphologic or statistical analysis) and are certainly 
useful since they offer the possibility to compare various 
datasets. Nevertheless it must be emphasised that they are only 
operational tools such as mathematical operations (as addition 
or subtraction) and they must be used within contexts that are 
homogeneous as for their significance. Moreover an analysis 
process needs a subsequent, very critical step, whereby the 
results have to be put under interpretation in order to give them 
meaning with regard to the specific phenomenon under study. 
An issue that is sometimes neglected - since it is implicit in 
stating phenomenon itself - is the “scale of reasoning”. Dealing 
with spatial information, this means the scale of the document 
by which the analysis is performed. Scale mediates relevance, 
says Racine (1981). Such statement claims that a change of 
scale implies a change of the meanings relevant to the 
description of the phenomenon (Ruas and Bianchin, 2002). 
Analysis technique is indifferent to change of scale. 
The purpose is to investigate the different meanings that the 
same index can assume when working at different scales. 
While some authors, referring to a matematical concept of scale, 
claim that scale can be treated as a continuum, we agree with 
Lacoste (1980) that there are different levels of representation, 
conceptually differentiated, which in cartography correspond to 
a change of scale - that is, to a transition from an order of 
466 
magnitude to another. An order of magnitude is an interval 
between scales within which a change of meaning does not 
occur, whereas meanings change from a given interval to the 
next. 
The general question raised here is about the consistency 
between visual description and quantitative description 
generated by analysis techniques. The question can be 
articulated as follows: 
I. Can quantitative description qualify settlement typologies 
according to the qualitative (visual) definition derived 
from urban discipline? 
2. Can quantitative description support qualitative 
description? 
3. Can quantitative description confirm qualitative 
description? 
4. How does quantitative description react to the change of 
scale? 
In order to give an answer, if partial, to these questions, 
various quantitative analysis techniques have been applied to 
binary maps of the built space generated from satellite images. 
The area under scrutiny belongs to the Veneto region and 
includes concentrated as well as diffuse urbanization. The 
analysis of such urbanization requires the definition of a set of 
variables that can qualify different spatial configurations of 
settlements. 
Hence we have worked on: 
|. the variables of density and landscape ecology that have 
been applied; 
2. the comparison of the results of their application over 
different scales and different urban contexts. 
In our first work we have applied these techniques to a range 
of settlement structures belonging to four different typologies 
and two different time periods. The aim is to verify whether 
indexes are sufficiently stable but also differentiated, thus 
allowing to qualify the typologies. 
A second work deals with the issue of scale and analyses the 
density function of the same area at different levels of 
resolution. 
Intern 
  
A thi 
analy: 
satelli 
IKON 
2. INI 
The t 
throu; 
2001) 
photo 
allow 
The fi 
Spati: 
on ge 
To cc 
and a 
The f 
| pat 
total : 
while 
2 edg 
in the 
lands 
comp 
3 mec 
4. FII 
[n th« 
four 
diffus 
  
17s 
(Freg