NUMERICAL PHOTOGRAMMETRY IN
RAILWAY DESIGN AND CONSTRUCTION
Dr. J.M. Zarzycki, P. Eng,
SUMMARY *
In Canada, photogrammetry finds extensive application in the location,
design and construction of railways. It is used in all phases of railway engineering from
preliminary reconnaissance surveys to construction surveys and preparafion of as-built
plans. The combination of photogrammetry and electronics has made it possible to carry
out precise measurements and perform computations much more economically than by
ground survey and conventional computation methods.
Numerical photogrammetry, combined with electronic computations,
is applied in the following stages of railway engineering: -
(a) In the location stage, to determine position of a number of points close to the probable
location of the proposed alignment. These points are determined to minimize the field
survey work. They replace conventional traversing between the basic control points estab-
lished by Tellurometer surveys.
(b) In the design stage, to determine the cross section of the ground before construction
and to compute the design quantities, vertical curves, gradients, super-elevations, slope
stake information and related data.
(c) In the construction stage, to determine the cross section of the right of way at different
stages of construction and also to determine the cross sections of the finished road bed; to
compute progress and final pay quantities, and to provide an evaluation of the constructed
grade.
In the integrated system employed by Canadian Aero Service, the
following instruments are used : Wild RC-8 aerial camera with 6" Avigon lens, Kelsh
plotter with a digital read-out attachment, Wild A-8 Autograph with EK-3 Co-ordinate
Printer and Profiloscope, Electro Dataplotter and IBM 650 Computer.
In initial tests with the Kelsh plotter and the Wild A-8 with EK-3 the
latter instrument was found to be in our overall evaluation more economical in spite of its
higher purchase price.
To determine the accuracy of the profiles obtained photogrammetrically
elevations of 211 points were surveyed on the ground over a seven-mile long section in
Northern Quebec. About half of this was gently rolling terrain, the remainder was rough
country. The checks revealed that the mean square error of the elevations obtained photo-
grammetrically is ! .12 ft. (*0.04m). This is equivalent to 1/15,500 of the flying height.
The distribution of the errors was as follows: -
80% of points : 1 0.0 ft. (*0. 0m) to
18% of points : — * 0.1 ft. (*0. 03m) to * 0.2 ft. (*0. 06m)
2% of points : t 0.2 ft. (t0. 06m)to * 0.3 ft. (*0. 09m)
+
0. 1ft. ($0. 03m)
These accuracies were achieved in a typical production project employing standard proce-
dures. These results are, therefore, indicative of the degree of accuracy which can be
expected in practical applications with the instrumentation employed.
* for presentation at the Ninth Congress of the International Society for Photogrammetry
at London, England, in September, 1960.
