15.
m. TEST RESULTS
The aerotriangulation procedure was programmed in FORTRAN 62 for use
on the Cornell University Computing Center CDC l6o4 Computer. The complete
procedure can he performed in a single computer run or may he run using any
combination of the previously mentioned major steps.
Tests were performed with three major classes of data: (a) theoretically
correct fictitious data; (*) fictitious data in which randomly distributed
displacements were applied to plate coordinates; and (c) actual photographic
data.
A. TESTS WITH THEORETICALLY CORRECT FICTITIOUS DATA
7
Tests were run with the Church fictitious photographs 1-12 and with
Q
the USGS fictitious North - South strip of photos 0-7* Both sets of data
have IQfjo relief and are at scales of 1:^0,000 and 1:8000 respectively. These
tests were performed to verify the formulation of the method and computational
accuracy of the computer program. Ground control and pass point configurations
are illustrated in Figure 3*01 and discrepancies between fictitious and
calculated values are given in Table 3*01 for test cases 1 - k. These
discrepancies compare favorably with results obtained with the same data
9 10
using other theoretically accurate methods 9 . As a result of these tests
it was concluded that the equations were correct and the computational accuracy
of the computer program was adequate.
B. TESTS WITH FICTITIOUS DATA CONTAINING RANDOM ERRORS
Tests with theoretically accurate data do not provide an indication
as to how the procedure will behave when random measuring errors occur. To
verify the least squares adjustment under conditions when such errors are
present, randomly distributed displacements were applied to the fictitious
plate coordinates and tests were performed with these data. Two sets of