CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
model and GIS data model are considered together, the 
requirement of a time and object oriented data structure occurs. 
Hence, in the GIS data model of this study, each element of 
analysis for each stratigraphic unit is handled as a single layer 
and presented in vector format so as to allow accurate spatial 
analysis. Raster data format is only used when an additional 
image, such as scanned photos or inventory forms, are to be 
hyperlinked to the geographical object or when only the 
visualization of the image at the background of the vector data 
is necessary. Thereupon, the data produced in each layer in 
each AutoCAD file are transferred to a GIS project file (.apr) as 
shape files (.shp) in connection with database files (.dbs), each 
defining a geographical object with a unique spatial object 
group definition and are added to the views as separate layers 
called themes in ArcView 3.2. Different views are created for 
topographical features, elements of analysis reflecting the 
existing situation and those reflecting the urban form of 
different historic periods, in which legend and theme properties 
are defined, including minimum and maximum display scales 
according to the level of geographical space. Even though they 
can be presented in different views, the utilization of the same 
projection and coordinate system for all the themes allow them 
to be geographically referenced with each other. Just like the 
vector data, raster-formatted images forming the background of 
the spatial data, such as aerial photos, plans and drawings, are 
also placed in their exact geographical position and saved in 
georeferenced file format. 
The other component of GIS data structure is the attribute data, 
which are descriptive data stored in tables concerning the 
geographical objects. Each geographical object belonging to a 
layer is represented as a record, and each category of 
information is represented as a field in the attribute tables. In 
order to define the categories, the information types for the 
study are defined. With the aim to inform identity, reveal 
sensitivity and resolve continuity about the urban form and its 
components, for this study, necessary information types are 
defined as, information for identification, information for 
characterizing historical stratification and information for 
decision-making. Information for identification consists of 
record number, type, name, address information, current legal 
status and the sources of information concerning the element of 
analysis. Drawings, photographs, legal documents like 
registration and inventory forms concerning the element of 
analysis also supplement information for identification when 
necessary. Information for characterizing historical 
stratification encompasses period of settlement / construction, 
date of construction, provisional / known minimum and 
maximum altitudes with respect to current ground level, and 
geometric information concerning the element of analysis. Last 
but not least is the information for decision-making, which is 
the information required for conservation decisions and 
described within the legal structure of the country in concern. 
For the case of Turkey, even though the criteria for decision 
making are not defined explicitly, laws, principles and decisions 
made by the conservation councils reveal that information for 
decision-making should cover position and perception, state of 
survival and degree of knowledge on existence. Following the 
constitution of GIS data model, the attribute data entry is made 
for each geographical object as different records, as a result of 
which the information system becomes ready for different data 
manipulation and analysis functions. 
2.1.3 Stratigraphic Correlation, Analysis and Evaluation: 
Data Manipulation and Analysis: Stratigraphic correlation is 
the direct outcome of geographically referenced data model 
designed for Bergama. Hence, it is possible to visualize spatial 
data both in time oriented correlation and in object oriented 
correlation. Time oriented correlation allows visualization of 
data in sequent snapshots model, whereas, object oriented 
correlation allows visualization of relation between the same 
elements of analysis in different periods. The stratigraphic 
correlations are also visualized over the 3-dimensional digital 
terrain model (DIM) with vectoral and image coverages (figure 
2). 
Figure 2. DTM of Bergama with vector coverages, which 
represents the time oriented correlation of the 
geographical objects belonging to urban form of 
Hellenistic and Roman eras. 
Stratigraphic analyses are realized through the spatial and 
attribute data concerning geographical objects through the data 
analysis and manipulation functions of GIS. There are mainly 
three groups of analysis by using the three main operation types 
of data analysis and manipulation functions. Those are: 
• Overlay operation, which allows overlaying different 
geographical objects to provide a new layer with new 
attribute data table. In Bergama this analysis is carried on 
for: 
• Settled areas of different historic periods and existing 
town as a result of which different sensitivity zones in 
the contemporary town with different stratification and 
different degrees of knowledge on existence, state of 
survival, and position and perception are provided. 
• Components of the urban form including the edifices 
and street axis as a result of which streets, blocks and 
edifices with different stratification options are 
determined. 
• Comparison and search operations, by which attributes of 
geographical objects are compared with each other and the 
ones fitting the defined criteria are selected among them. In 
Bergama, there was a strict grid-iron system in Hellenistic 
and Roman eras according to the specialists of these 
periods. The angle of the assumed grid-iron system is still 
traceable in the streets and built up areas of the 
contemporary town, which may indicate material existence 
or trace continuity and sensitivity concerning these periods 
below the contemporary tissue. Hence, angles of street axes 
and outlines of built up areas are analyzed through the 
attribute data that store the angle information. Making an 
allowance for possible deformations of the grid-iron system 
in time, an angle range is defined which is then compared 
with the angles of existing street axes and the built up area 
outlines stored in attribute tables. The geographical objects 
having outline or axis angles within the defined range are 
selected and converted into a new theme as a result of this 
analysis (figure 3). 
• Connectivity and neighbourhood operations, which provide