IV. CONCLUSIONS
As a consequence of tests with the theoretically correct Church and
U. S. G. S. data, it was concluded that the aerotriangulation procedure was
formulated correctly and that computational accuracy of the computer program
was adequate. Tests were also run using actual data and fictitious data
containing random errors. These tests are of considerably more interest
with respect to the capabilities of the method.
Examination of test cases run with Set 1 (Test Cases 5 - 10), reveals
a minimum resultant mean square error in position which is 1/900 of the
flight altitude and 1/4000 of the strip length. Discrepancies of this
magnitude are in excess of allowable errors permitted by National Map
Standards for accuracy. However, it should be noted that Set 1 has a mean
square error in the measured photo coordinates of 0.05mm. 'This amount is
larger than would be expected using modern plate coordinate measuring instru
ments and technology. Also, the coordinate values in the geocentric system,
from which the resultant mean square error was calculated, have been subjected
to the linear transformation only. Discrepancies of this magnitude are to
be expected at this stage and the secondary adjustment using three-dimensional
second-degree equations should be applied to provide a better fit to ground
control points. Further testing is required to verify the secondary
adjustment.
Results obtained using Set la would satisfy National Map Standards for
accuracy. In Test Case l4, a bridge using six transformation tie points,
the resultant mean square error in position was I/89OO of the flight altitude
and l/l5,000 of the strip length. These figures are based on unadjusted
results. Further improvement could be expected on application of the
secondary adjustment.
Considerable caution must be exercised in evaluating results of tests
with Sets (l) and (la). Even though random errors were applied to plate