CIPA 2003 XIX th International Symposium, 30 September - 04 October, 2003, Antalya, Turkey 
shape subclasses - “shape” understood not in a geometrical 
sense, since all geoglyphs were mapped as polygons, but rather 
descriptively - allow an unambiguous assignment of each 
ground drawing to the five basic shapes observed in the field: 
“lineal”, “areal”, “figural”, “biomorph”, “other”. These five 
shapes are represented by a second level subclass each, which 
in turn have third level subclasses according to the more 
specific characteristics of the geoglyphs. For instance, the 
second level class "A_GeoglyphOfLinealForm" has four 
subclasses in the third level ("A_Straight", "A_Zigzag", 
"A_Meandering" and "A_LineOthers"). This three level 
hierarchy allows a disjoint and complete partitioning of all 
geoglyph objects based on their shape. For some of these 
subclasses, methods are defined to calculate width, length, or 
orientation. The applicability of these methods depends on 
geoglyph shape; therefore they can not be defined on the 
“A_Geoglyph” level. 
Similar to geoglyph shape, the construction technique, another 
important feature of the geoglyphs, has to be modeled. To 
create the geoglyphs in the stony desert, the ancient inhabitants 
of the Nasca region used four different techniques, or 
combinations of these, in order to mark the geoglyph borders 
and interior areas: both elements of the geoglyphs can be 
cleared, heaped, intact, or furrowed (Lumbreras, 2000; Reindel 
et al. 2003). To model this, a unidirectional association from the 
class "A_Geoglyph" to the class "A_Interior" and "A_Border", 
respectively, was established. Both classes contain one attribute 
of "bag" type (multivalued data type without any order, 
allowing duplicates) which can take one or more (up to four) of 
the values "cleared", "heaped", "intact" and "furrowed". The 
cardinality of this relation is 0,1 to 1, which means one 
geoglyph can be related to exactly one bag of construction 
techniques for the interior and the border. 
The relation between the geoglyphs and geometry has also to be 
modeled. The association "consists_of assigns each geoglyph 
one or more instances of the class "G_Polygon" ("G_..." 
denoting a geometrical object). The cardinality has to be 1 
("A_Geoglyph") to 1,* ("G_Polygon"), because a geoglyph can 
consist of more than one, but at least one, polygon (Sauerbier, 
Lambers, in press). For the same reason, a method "calculate 
geogl_area()" in the class "A_Geoglyph" is established to 
calculate the area of a geoglyph based on the respective areas of 
the assigned polygons. The geoglyph size derived from this 
calculation is another important attribute for typology. Further 
methods are defined to calculate outer coordinates of the 
geoglyphs in order to enable their easy location. 
A further phenomenon that has to be included in the data model 
is stratigraphy, which is a relation between objects of the same 
class, in this case the geoglyphs. Stratigraphy describes the 
physical sequence of the ground drawings that reflects the 
chronological order of their creation (Harris 1979). The data 
model has to express three conditions present in the 
archaeological record. These cases are: 
1. Object 1 is above object 2. 
2. Object 1 is below object 2. 
3. Object 1 and object 2 are equal. 
he data model, this relation is established by an association of 
the class "A_Geoglyph" with itself, where the association 
"A_Stratigraphy" has an attribute "stratigraphy" which can 
assume one of the above mentioned values. The cardinality of 
this association is 1 to 1 because it is a relation between two 
certain objects. 
Buildings and finds associated with the geoglyphs provide 
important clues concerning human activities on or nearby the 
ground drawings. Especially ceramic finds carry also 
chronological information. Thus, a connection between the 
geoglyphs and archaeological objects like metals, lithics, 
ceramics, textiles and bones has to be modeled, such that a 
geographical and contextual assignment of the finds is possible. 
For this reason, a class "A_Objects" with two attributes 
("remark", "sample") is established and related via an 
association with the geoglyphs ("A_Geoglyph"). This relation 
has the cardinality 1 to 0,1, so that each geoglyph can be 
assigned either 0 or 1 inventory of archaeological finds. The 
class "A_Objects" again has different subclasses according to 
the different kinds of finds present in the inventory (see above), 
complemented by the subclass "A_Others" to cover also finds 
which do not fit into any of these categories. 
2.3 Integrity constraints 
Integrity constraints are needed to guarantee the consistency of 
data inside the database and to prevent wrong data input by 
users. Modem DBMS provide a variety of methods for integrity 
protection (Heuer, Saake, 2000). At the current state, dealing 
with the conceptual data model, we focus on the model-inherent 
integrity constraints, which can be defined using SQL-DDL to 
ensure type integrity, key integrity and referential integrity. A 
second level of integrity constraints and rules will have to be 
implemented during the development of the logical data model, 
using e.g. trigger concepts for cascading deletion and insertion. 
The integrity constraints that can be defined for classes in the 
conceptual data model are in particular the key and the 
referential integrity by means of setting a primary key and 
foreign keys with references to attributes of other classes. The 
class "A_Geoglyph" has as a primary key the attribute 
"Geogl_ID", which is a set of unique values identifying each 
tuple unambiguously. This includes the type integrity constraint 
that this attribute must not take a null value (=NODATA). An 
example for a foreign key could be the attribute "Geogl_ID" in 
the class "A_Polygon", which is a primary key in 
"A_Geoglyph". Using this foreign key, the method 
"calculate_area()" in the class "A_Geoglyph" can access the 
attribute "area" in the "A_Polygon" class and sum its values for 
one geoglyph. 
An important integrity constraint in our case is type integrity. It 
can be obtained by defining domains for attribute values using 
the SQL-statement constraint inside a create table expression. 
One example in our model is the construction technique which 
can assume only four certain values; another instance is 
stratigraphy with three possible values (“above”, “below”, 
“equal”). 
A further method of integrity protection is provided by SQL via 
the assertion and check-clauses. Using these two expressions, 
the user can set constraints spanning over tables and relations. 
This allows to check attribute values, e.g. if a PAP-number 
(which identifies the archaeological site the geoglyph belongs 
to) assigned to a geoglyph really exists. 
2.4 Modeling the Nasca lines with UML 
The application of UML to model the Palpa geoglyphs is a 
useful tool for structuring and integrating the different kinds of 
data and is thus an important step in data processing prior to 
analysis. The class diagram provides a clear overview of the 
acquired data, while the integrity constraints guarantee data 
consistency. 
3. DATABASE AND GIS 
The class diagram presented here is the core element of our data 
model. The work at Palpa yielded, and continues to yield, 
additional data as well that have to be integrated into the model 
in a next step, inlcuding: 
• prehispanic settlements and cemeteries, 
• administrative boundaries,