————— 
in both direc. 
features have 
also the basis 
), but most of 
ch do not con- 
^tter condition 
rplane (Aveni, 
balloon or too 
aerial images 
  
imber of 
trol points 
(signal.) 
0 
(signal.) 
0 
(natural) 
  
  
all the excavæ 
use of lack of 
Plotters AC 
  
Beutner, Sabine 
and S9. The triangulation of the Palpa color block was performed with 88 images and 8 control points. Since we were 
not very happy with the quality of the color images we ordered a B/W photoflight to be done one year later. At that time, 
without us being in the area, the signalized ground control points were not available any more. Also, kinematic GPS was 
not used. Therefore we had to measure the B/W block without any control points and connect it via joint tie points to 
some images of the color block. We did a joint bundle adjustment of both the color and B/W images including kinematic 
GPS data (BW+C). Table 2 gives the results of triangulation. 
  
Version number of images number of control points kinematic GPS oo [Um] 
  
  
  
B/W+C 211 8 yes 13.3 
  
  
  
  
  
  
Table 2: Triangulation characteristics for Palpa 
Although the photoflight was meant to produce an image scale of 1:5 000 the actual scale turned out to be 1:7 000. The 
overall block accuracy, as indicated by o — 13.3 Hm, corresponding to 9.3 cm on the ground, was good enough for the 
purpose at hand. It was clear that at this fairly high noise level self-calibration would not help improving the results and 
was therefore not used. 
3.2 DTM Generation 
In order to generate a DTM of the Palpa area a total of 57 B/W stereo pairs had to be measured. We decided to go with 
manual measurements because we needed a very precise DTM, especially for the integration of the image texture map 
and vector data of the geoglyphs. Any deviation between vector data and texture would have been noticeable very easily. 
Our previous experiences with automated commercial image matchers convinced us that these could not give us the 
required performance. 
We used profile measurements at a profile distance of 20 m with additional breaklines. This gave us about 200 000 points 
for an area of about 2 km by 9 km. Later on we also integrated all the extracted geoglyphs as breaklines into the DTM. 
Figure 4 shows the shaded version of a wireframe model of the DTM with the plateau Cresta de Sacramento as the major 
site of geoglyphs. The flat region is the Rio Grande valley. The major excavation sites La Muña and Los Molinos are 
indicated. 
Los Molinos 
La Muña 
*Fig.7: Geoglyphs 
Fig.8: 3-D model 
Cresta de Sacramento 
  
Figure 4: Shaded DTM of the Palpa area 
The wireframe model serves as the basis for orthoimage production and visualization. We have produced wireframe 
models at various resolutions according to the different purposes. Rasterwidths range from 25 m for overview representa- 
tions down to 1 m for highresolution applications. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXIII, Part B5. Amsterdam 2000. 57