While exploiting advanced technologies and technical 
innovations, it also helps to lay the foundations of a new 
form of underwater archaeology, one in which divers 
completely cede their places to remotely controlled 
devices, given that saturation diving is expensive and 
increasingly inappropriate for careful excavations on 
deep-water sites. It is our wish that our site, transformed 
for the time being into a veritable experimental 
laboratory, will permit new technologies to work for the 
benefit of historical and archaeological research. 
2.2.1 Digital Photogrammetry 
The wreck is resting at a depth of 60 meters ; even if 
divers are able to go to that depth, the work is very 
difficult and potentially dangerous. A diver can not stay 
more than about ten minutes at this depth and to establish 
a topographic map under those conditions would be near 
to impossible. We adopted a light digital photogrammetry 
method by using a non-metric digital camera, mounted in 
a waterproof housing attached to a bar on COMEX's 
submarine Rémora 2000. For a light presentation of the 
underwater photogrammetric method you can refer to 
[Drap, Long, Durand,  Grussenmeyer,  2001-A], 
[GrandRibaudF, 2000] and more generally to [cipa-uwp, 
2002] 
2.2.2 The Photogrammetric Results 
A total of two hundred pictures have been oriented, with 
a set of rulers to put them in scale and several buoys for 
vertical reference. The orientation relative to north is 
approximated. More than two thousand points were 
recorded and were used to generate a digital terrain 
model. 
The orientation phase was made with Photomodeler 4.0 
and the orientation result were introduced in Arpenteur 
for a specific survey of the amphorae. 
We develop a special bridge from Photomodeler to 
Arpenteur in order to combine the advantage and 
specificity of both these software. Photomodeler is 
actually more efficient to manage a large set of 
photographs and on the other hand we have been 
developed a special module for amphora plotting using 
data fusion from measure and theoretical model. (see 
[Drap, Long, Durand, Grussenmeyer, 2001-A ]). 
The main goal was to plot the visible amphorae and 
fragment, in order to visualise them in the same time that 
we document the site. All the data surveyed are 
formalised in XML file which are, in fact, the only result 
of the plotting phase. We can parse the XML in order to 
generate 3D representation using VRML, PovRay or 
MicroStation formalism. 
2.3 A DEDICATED DATA MANAGEMENT 
SYSTEM 
The roots of this project reside in the link between many 
tools and the elaboration of this link through a semantic 
approach of the objects to be measured and shown. 
Measurement, representation, and management are 
articulated around a common model formalized from the 
« Object » point-of-view and implemented in JAVA 1.4. 
  
Figure 1. VRML site representation, used as an interface to the 
archaeological data formalized in the XML result 
file 
  
Figure 2. MicroStation site representation, as the VRML one, 
the graphic file is automatically generated after the 
parse of data formalized in the XML result file 
  
Figure 3. MicroStation site representation. Detail of the last 
excavation, we can see the bottom of the wreck 
(wood pieces) and several layer of amphora. 
The measurement and representation phases take advantage of 
this common model by putting in place of a default value 
mechanism allowing to take complete measurements of the 
-344—