included only a single SPOT image in the adjustment we 
saw a dramatic decrease in the height precision, and to a 
lesser degree in the latitude. We saw this same increase 
when we limited the tie points to the upper left corner of 
the block in the last bundle adjustment. 
Table 2 provides a breakdown of the diagnostic point 
statistics generated for each multi-sensor triangulation. 
This provides an indication of the absolute accuracy of the 
adjusted image-to-ground projections. No significant 
pattern is seen when comparing each triangulation, and at 
first glance it appears that none of the runs would satisfy 
the United States Map Accuracy Standards for a 1:24,000 
scale map. But it is important to remember that the 
control coordinates of the diagnostic points were also 
digitized from the USGS map sheets, and have an 
accuracy themselves of >10 meters. 
Table 2: Diagnostic Point Statistics for multi-sensor 
triangulations. Expressed as Linear 9046 in meters. 
    
Table 3 contains a comparison between the the spot 
elevations measured on the stereo models from the first 
multi-sensor triangulation and the baseline spot elevations 
collected from the GPS controlled stereo models. These 
statistics show how well the SPOT controlled solution can 
be used in place of a conventional aerial triangulation. 
The two solutions match in X, Y, and Z within the 
constraints of the National Map Accuracy Standards. The 
difference between the two solutions is a fairly consistent 
shift, but we can also see that the shift varies between 
Strip 7 and Strip 9. We believe that this is caused by the 
limited overlap between the photographs in Strip 7 and the 
SPOT images. 
Table 4 shows the differences in the spot elevations 
collected from stereo models generated using the GPS 
controlled stereo models, and those generated using the 
USGS map sheet controlled stereo models. We wanted to 
make these comparisons to see how the multi-sensor 
triangulation solutions compared to a conventional aerial 
triangulation which used control collected from a less 
accurate map source. We can see that the aerial 
triangulation using control from USGS maps provides a 
slightly better solution than the multi-sensor triangulation. 
  
International Archives of Photogrammetry and Remote Sensing. Vol. XXXI, Part B3. Vienna 1996 
    
Table 3: Differences in X,Y,Z coordinates of spot 
elevations collected from multi-sensor triangulated stereo 
models and GPS controlled stereo models. Units are in 
feet. 
3.39 
32.65 
9.46 
4.12 
8.06 
5.14 
19.41 
8.46 
  
Table 4: Differences in X,Y, Z coordinates of spot 
elevations collected from the USGS map controlled stereo 
models and the GPS controlled stereo models. Units are 
in feet. 
6.26 
13.63 
10.29 
  
Table 5 shows a comparison between the spot elevations 
collected from the stereo models created from the first 
and fourth multi-sensor triangulation. We saw that the 
differences between the two solutions was insignificant. 
This was expected based on the error propagation and 
diagnostic point statistics. Table 6 shows the results of 
comparing the spot elevations collected from the stereo 
models of the third and fifth multi-sensor triangulation. 
The Circular Error is comparable to the other runs and the 
Linear Error is larger. But what is more striking is the 
much larger shift in the X and Z components than in the 
other multi-sensor triangulation results. 
   
the fi 
Units 
  
Table 
the th 
comp: 
in feet 
  
Both « 
result 
conve 
points 
soluti 
The R 
was ( 
respe 
requir 
better 
contre 
topogi 
we ac 
20.66: 
But b 
collec 
triang 
aerial 
topogr