A Numerical-Photogrammetric Survey of Highest Accuracy 
by Dr. Ing. Robert A. Schlund and Dipl. Ing. Peter Fülscher 
In 1966 the Swiss Federal Railways required the following data for a projected auto- 
matic marshalling ramp in the marshalling yard at Muttenz - Basle: 
— An accurate plan at a scale of 1:500 of the entire lay-out covering approximately 
1300 m x 200 m, with 104 switches, 4 track brakes and 43 tracks of a length of 
approximately 40 km. The plan should also show all power and lighting masts, sig- 
nals, hydrants, cable supports etc. 
— The plan should also show the deviation angles of the switches and the radii and 
length of curves, as well as. the slopes of the tracks for which an accuracy 
of + 0.1 9/oo would be necessary. 
— For the computation of curve radii and length, together with the switch types and 
deviations the coordinates of approximately 1700 points on the track axes were to 
be measured with a mean square error of position not exceeding + 2-3.cm. 
Since the curve radii had to be computed with an accuracy of approximately 5 m, it 
was essential to obtain a mean square point error of less than + 3 cm. With a determi- 
nation of the radius from the chord 2s and the versine f of a curve, as would be done 
in a terrestrial survey without computation of point coordinates, f must be measured 
with an accuracy of approximately * 5 mm as the following computation shows: 
  
  
  
g2t(R-fy-Ri-s;-TA HR + 
2 
EY LT 
R ail ind 
AR eo ipud on 
€ S 
af f 
Fig. 1 
Example: s = 10,0 m, f » 0,25 m, The versine f must therefore be measurable 
for Af = As = 0,01 m, with the same accuracy as that of the 
AR = 8+04m coordinates.