lere
na
"es-
1 in
fer-
rid
fer-
the
me
not
had
ned
urt
tter
the
res
the
so
the
ent
ns.
ery
the
ich
1Ve
oth
vin
ing
the
au
of
not
ct.
1bt
ral
of
)se
1er
CONVERGENT VERSUS VERTICAL PHOTOGRAPHY, DISCUSSION 121
It seems to me that in these questions we
have the real centre of the problem today, and
no doubt Professor Schermerhorn agrees with
me in this case, according to the statement in the
publication.
Dr SCHERMERHORN: I must ask the members
of the panel, and first of all my American
friends, to answer my question.
Mr M. M. THoMPson: I will try. The ques-
tion on the accuracy of the model involves in
our estimation two phases, the theoretical accu-
racy and the practical accuracy. The theoretical
accuracy has been discussed frequently at
great length by Von Gruber in his famous
collected essays in the 1930s. It has been dis-
cussed further by Professor Hallert in his pub-
lications emanating from Ohio State University.
We are interested not only in the theoretical but
the .practical, we have taken the mathematics
as derived by Von Gruber and carried further
by Hallert in his publications, but what actually
happens in practice? In 1954 Mr Bean pub-
lished in ,, Photogrammetric Engineering” a
brief article which was called “Use of the
E.R.55", but it had to do with its use with
convergent photography. It described the results
of a test over Los Angelos, California, where we
had a great deal of control, all street intersec-
tions over a large area; flight height, 8,000 feet,
mean square error, vertical error was 8/10 foot,
which gives you H/10,000 as the vertical
accuracy.
We have done other testing of that nature,
but that is an example. We did not go too far
with mere testing, we went ahead with actual
compilation of topographic maps. 1 would not
be talking about three or four test compilations
in the laboratory, I would be talking about
30,000 or more square miles of actual topog-
raphic maps compilation. For the vertical
accuracy, of course, we have a testing pro-
gramme which we are required by our regula-
tions to conduct. I did not have the material
available at the time I sent my memorandum in,
but on 37 projects tested for vertical accuracy
— these are projects not models, each project
may consist of a great number of models — we
found that the vertical accuracy with conver-
gent photography was of the order of 20 per
cent better than with vertical.
I would like to add at this point that I tried
to make it plain in my memorandum to Dr
Schermerhorn that we do not think convergent
photography is the answer to all problems. Very
early in our application we came to a conclusion
which we had rather suspected before, that in
terrain of very high relief it loses its advantages,
but we have found that it has great advantages
in areas of moderate and low relief.
I wish there had been time or space in Dr
Schermerhorn's report to mention the matter
of costs. We are very much interested in costs,
and thus the reason that we went to convergent
photography was purely a matter of economy.
We want to map as much of an area as we can
at as little cost as possible. The figures that we
have concerning 30,000 sq miles about equally
divided between convergent and vertical photog-
raphy so that our costs with convergent photog-
raphy are some 20 per cent better. Both the
accuracy and the cost seem to come to a 20 per
cent advantage. This applies to a ten-foot con-
tour interval at a scale of 1 : 24,000 which is one
of our principal mapping sets of parameters.
I might go on and mention the point of cali-
bration which Dr Schermerhorn mentions. It is
certainly true that that is one of the great prob-
lems, and perhaps the difficulties that we have
had with calibration problems in convergent
should make us all realise that we have been
neglecting this problem quite a lot, along with
vertical. We have had this bad film and bad
calibration from the start, and this is just some-
thing which brings it to a head. I believe that
I have tried to answer all your points.
Dr ScHERMERHORN: I thank Mr Thompson
for his statement which is really an answer to
my question. The value of it is that it is based, as
he says, on practical work and that means it is
of statistical value. That is the trouble with all
of what we call here “essais controlés”, com-
paring such a result with all the trouble we have
had with Monti di Revôira. Then I say let us
look first at what has been done by the services
and look at their results. I am very grateful that
Mr Thompson has now given the figures, which
[ lacked before.
I would like to stress one point again with
Mr Thompson. There is the need for, let us say,
calibration and regular calibration of the
cameras and testing the material we put in our
instruments. We are at present in a situation,
and have been so for a long time, where the
restitution instrument, and now still more the
comparators, are in a state to do several times
better than that of the average photographic
material we put in the instruments. I mentioned
only a few bad figures which are naturally
extreme values, but what is the sense of talking
about microns if we do not even know what is
the quality of the photographs we put in the