ssive models
he propaga-
ed strips. In
>lofs derived
n transfer of
o the next in
his, indeed,
mations per
which deter.
of the block
' of applica-
tion in the
[eheran, we
were taken
were larger
elofs in his
serles with
grades and
.97 grades.
>ly flat, dis-
levation, we
ulation sys-
of the order
n; and also
je expected,
'rrors, how-
than 0.12
the azimuth
both cases.
Ose as com-
id azimuth
, caused by
nd that for
quare error
about 0.12
t the errors
' caused by
lation to be
on — and it
n answer to
-publication
ir future —
lation may
iangulation.
or principal
practised à
ounded on
nd this has
a large ex-
ularly from
that in this
| about the
image co-
s this could
UFU —————-
RADIAL TRIANGULATION, DISCUSSION 87
be taken into account also in connection with
investigations of the quality of radial triangula-
tion.
Mr P. O. FAGERHOLM: I have been asked to
join this panel on radial triangulation and have
taken some interest in the matter, and therefore
perhaps I should spend just a few minutes on
it because there has been a lack of time for this
panel to co-ordinate its ideas. Professor Roelofs
has said part of what I intended to say, and that
is fine.
What I would like to underline especially is
the need to avoid over-estimation of the draw-
backs, that radial triangulation only gives the
x and y co-ordinates, the planimetry.
What I would like especially to say is that if
you only have one series of photography from
one height, you must choose that flying height
so that in the plotting stage you get the accuracy
you wish in the final map. If you then have to
take away a part of the margin for spatial trian-
gulation this has a pretty great influence in many
cases on the densening of the geodetic net with
photographic methods. This, compared with the
latest developments in radar profiling and good
barometric levelling in helicopters, etc, is
changing the entire picture and I think we
should realise that in our discussions very
clearly.
Another good thing with radial triangulation
is that if it is numerical or mechanical it can be
broken down in small pieces, and every one of
these procedures can be checked, either spot
checked or checked to see if something is going
wrong in the adjustment if one believes that
there is something wrong there. That is not so
easy with spatial triangulation. Professor
Roelofs mentioned that it is possible, and cer-
tainly it is true with a horizon method, and so
on, to determine the nadir distance pretty well.
That is certainly true.
I also wish to underline the fact there is no
need to know the nadir point very accurately
with the very limited nadir distances we usually
have in modern photography.
Something else I should especially like to
point out is that radial triangulation has a very
low requirement on the distribution of geodetic
control points, especially if it is used as a bloc
method. I understood the tests of Professor
Roelofs with radial triangulation to be a strip
method, and I think you get the best benefit if
you use it as a bloc method.
The automation process is pretty easy to
apply in radial triangulation, and there are many
ways to do it. One is the mechanical radial
triangulation which mostly has been used for
reconnaissance mapping, but when all the
different weak points have been taken care of
one can get very good results with a mechanical
radial triangulation. One could also call it pre-
cision radial triangulation.
I am not trying to sell mechanical radial
triangulation, but I think it should not only be
regarded as that type of method which is used
for reconnaissance mapping. It has been used
to a very great extent and there are further
possibilities for it.
Prof W. SCHERMERHORN: I believe that Mr
Fagerholm can be quite satisfied that there is
hardly any method which has been used as
much as his famous mechanical radial trian-
gulation.
Prof R. RoELOFSs: I completely agree with Dr
Fagerholm that it is much better to treat radial
triangulation as a bloc, and when I used strips
in my geodetic examples it was because I was
too lazy to study the propagation of errors
through a bloc; with a strip it is much easier.
Nevertheless, we had a practical example of a
bloc of five strips of seven photographs, that
means 30 sections, each section being two
photographs. We adjusted this bloc by the Jerie
method known to you, and the result was a
relative standard error of 18 uw on photo-scale.
Prof W. SCHERMERHORN: This last figure
may give you the impression that nobody needs
to be afraid on the precision of numerical radial
triangulation. So far, there are only very few
spatial triangulations which have obtained the
precision in planimetry after bloc adjustment as
this one which has been carried out.
I have now finished the discussion on this
subject. May I invite Dr Brandenberger, who is
in charge of the second subject — Use of Auxil-
iary Data for Carrying out Aerial Triangula-
tions — to say a few words about this. I would
call your attention to the fact that this paper
has been published in Photogrammetria 1959/
1960 Congress-number A, on page 67.