136 GRAPHICAL OR NUMERICAL PHOTOGRAMMETRY?, DISCUSSION 
work in the United States. Unfortunately, this 
paper has been in the hands of the United States 
Line since July 14th and has not yet arrived at 
the Congress. I hope that it will be here before 
the Congress is over at any event, to be in- 
cluded in the Archives. 
Mr Zarzycki has mentioned to you the pro- 
cedure used by his organisation in Canada, and 
it is closely parallel to that which is employed in 
the United States. Consequently, I do not think 
it is worth while to repeat any of the data which 
he has mentioned or the procedures which he 
has outlined. 
However, I should like to take a few minutes 
to tell you about a development which is going 
on in the States and which shows great promise 
for the future. Professor Charles Miller at 
Massachusetts' Institute of Technology has been 
experimenting with what he calls a digital terrain 
model. His research has been sponsored by the 
Massachusetts Bureau of Public Works and the 
US Bureau of Roads. This concept is an ex- 
tension of the digital cross-section technique al- 
ready described by Mr Zarzycki. 
At the present moment, cross-sections are 
taken on lines perpendicular to preselected 
ground alignment, as Mr Zarzycki has said. 
This means, if the alignment is shown, a stereo 
model must be set up again, new alignment 
sheets plotted and new cross-section data 
recorded before the computer can perform its 
work computation. Consequently, to evaluate 
ten trial lines is essentially ten times as much 
work as a single line. 
The entire system at present employed is one 
of replacement of conventional techniques by 
exactly corresponding measurements made 
photogrammetrically and computations and 
plotting performed electronically. Professor 
Miller has proposed to utilise to the fullest ex- 
tent the efficiency of photogrammetry in ob- 
taining large quantities of terrain data and the 
ability of the computer to process large volumes 
of data. In short, he intends to digitise the entire 
area covered by aerial photography. 
In order to do this, a base line is established 
at one edge of the area. This is tied to a ground 
recoverable reference system such as the States 
plain co-ordinate system. Profiles normal to this 
base line are then read at suitable intervals, and 
the x-, y- and z-co-ordinates of each point are 
stored as tapes or cards for use in the computer. 
This single set of stored terrain data could be 
obtained in advance of detailed consideration of 
the highway alignment. The same data could be 
used to evaluate numerically any chosen hori- 
zontal or vertical alignment within the digitised 
zone. 
An important consideration in the digital 
terrain model is that each observed point may 
be given one or more classifications for storage 
in the computer along with the positional co- 
ordinates. These classifications might be such 
items as type of soil, depth to bed rock, cost of 
acquisition, and so on. These classifications 
could conceivably be extracted from the stereo 
model during scanning or more realistically they 
could be supplied by another analysis which may 
be completely independent of the photographs. 
The utilisation of such a concept is limited only 
by the engineer’s imagination or the sophistica- 
tion of the programming and computation which 
he is willing and able to provide. 
Suppose that he assigns relative values to the 
directness of route, cost of acquisition, cost of 
construction, in addition he will want to specify 
such items as limiting grades, site distances, 
degrees of curvature, areas in which for one 
reason or another the highway could not go. He 
would feed this information to the computer, 
tell it to design the most efficient highway from 
here to there, push the start button and go out 
to lunch. On his return he would find the design 
completed, the numerical data printed out and 
the construction plan automatically drafted. This 
may seem visionary, but it represents consider- 
ably less of a revolution than has already taken 
place in the last decade. 
Mr A. J. VAN DER WEELE: Thank you very 
much. Is there anyone on the floor who wants to 
make some remarks or has some additional 
question on this subject? If not, I should like to 
put a question myself. As I have already indicat- 
ed in my paper, I can imagine that from a tech- 
nical point of view it is quite possible to lunch 
in such a nice way as Mr Doyle has indicated, 
having the machine do the work, but I wonder if 
in this procedure it will be possible to take into 
account the number of considerations for the 
design which can hardly be expressed in num- 
bers: the political consideration which give 
preference for a certain location to a road with 
respect to another one. There may be points 
which should be avoided, natural parks or other 
things. Could you give some indication about 
future possibilities of taking those factors into a 
consideration of the problem. 
Mr F. J. DoyLE: I do not quite agree that 
many of these considerations cannot be express- 
ed numerically. For instance, you mention areas 
in which the highway cannot go, or areas in 
which it would be preferred that the highway go 
SE S. 
IU EE IL I 
Ree, M'M ^ue. a o MN baden 
ppt PT peep 
e+ 4 IA M 0 ee NPT Pret MÀ ped 
—— 09 
NY /"