22 
throughout the test area to provide a sound basis for the evaluation of the 
accuracy of the proposed aerial triangulation method. This vertical control was 
established by third-order levels and stadia to an accuracy of better than 
t5 feet (Î1.5 metres). 
7.1.1 Accuracy of and Determined from Horizon Camera Data and the 
Accuracy of General Determined from Statoscope Data 
To evaluate the accuracy of the orientation elements, a number of 
stereo models was oriented absolutely in the B-8 using ground control. The 
orientation elements ^ and GO of the right and left plate holder were 
registered with the L-shaped bubbles. These elements were then compared with 
the ^ and U9 determined from horizon pictures. Also, the general <£) 
determined from the ground control set-ups was compared with the general <£> 
determined from statoscope data. 
The mean square errors of the absolute orientation elements and 
GU determined from horizon camera data and of general from statoscope data 
are as follows: 
m 
^ = t4 c (based on 30 comparisons) 
(based on 32 comparisons) 
c 
muo = -o 
m 
* 
!-- 7 c 
= 17 
(based on 19 comparisons) 
The higher error in CO is due to an instability of the B-8 in the OJ direction 
of about 12 C . 
7.1*. 2 Accuracy of Aerial Triangulation 
The test area is shown on Fig. 18. Strips were adjusted separately 
to vertical ground control established only at the beginning and at the end 
(i.e., at 15-model intervals) of each strip. The residual errors in elevation 
after block adjustment, based on 97 vertical control points spread throughout the 
test area, are as follows: 
87% within Î10 feet (Î3.1 metres) 
927 0 within Î12.5 feet (Î3.9 metres) 
Î16.5 feet (Î5.1 metres) 
98% within