pr 5 ly ir pee ee 
  
  
  
  
142 THE CANADIAN SURVEYOR May, 1960 
U.S.A. Dept. of Agriculture, Washington, D.C. 
Army Map Service, Washington, D.C. 
Coast and Geodetic Survey, Washington, D.C. 
Jack Ammann Photogrammetric Engineers, Inc., San An- 
tonio, Texas. 
Lockwood, Kessler and Bartlett, Inc., Syosset, N.Y. 
Navy Hydrographic Office, Washington, D.C. 
The distribution of equipment used was as follows: 
First-order Plotters: 
Wild A-7 ae 10 submissions 
Zeiss CB A 2:3 " 
Second-order Plotters: 
Wild A-8 .. s 10 submissions 
Kelsh (including one AMS | ploner). 9 » 
Baix 7:7 i en ss 42573 » 
Total number of submissions: 38 
ANALYSIS OF THE ACCURACY OF SPOT ELEVATIONS 
Each participant, using four horizontal and five vertical control points, was 
required to produce the elevations of 73 well-defined points distributed over the 
stereomodel. The submitted values were compared with the actual ground elevation, 
which at this stage of the project were known only to the author of this report. This 
comparison revealed that the most serious errors were caused not by any shortcomings 
in photogrammetric methods or equipment, but by the use of improper procedures. 
In particular, two sources of errors were evident: (a) errors in the absolute orientation 
of the stereomodel, and (5) known systematic errors. 
ERRORS IN THE ABSOLUTE ORIENTATION OF THE STEREOMODEL 
Absolute orientation of a model consists of scaling the model, rotating it, and 
shifting the horizontal reference plane to minimize the remaining elevation discrep- 
ancies on absolute orientation points. The accuracy of these operations is limited by 
the measuring accuracy produced on the plotter, and some of the more interesting 
aspects of it will be discussed later. The shifting of the horizontal reference plane 
consists of resetting the elevation counter and usually is carried out immediately prior 
to starting the plotting operation. 
According to some opinions, the maximum error should be included in accuracy 
specifications in topographical work. This is said to be justifiable because the eleva- 
tion error in a single point affects an area around it, since the earth surface is con- 
tinuous. In particular, when photogrammetric methods are used, a larger error at 
one of the absolute orientation points must be regarded as an indication of the possible 
existence of systematic deformation in an appreciable portion of the model. There- 
fore the correct procedure is so to set the reference plane that the maximum positive 
and negative discrepancies are equal. A number of participants failed to do this and, 
in particular, disregarded the elevation discrepancy on the central control point caused 
by the earth curvature. Actually, 12 out of 35 submissions (not including three A-9 
submissions) or 349% did not shift the reference plane properly, and assigned the 
discrepancy to a possible misreading or some similar fault at the central point (see 
Fig. 3). In consequence, positive elevation errors prevailed in all but one of the above