l with the Procedure y 
fference that for the se 
profile points are Used 
E 
di 
djustment of Z COOrdinates n, 
y the discrepancies in 
1s of profile points en 
| Will have different dii 
have to be kept as the t 
|Ve to be correcteq Accord 
reading discrepancies of li 
ht of last photograph i 
t Mat 
1” elevations of Selected jt 
  
S867 
  
  
ion. Except for possible le 
high, whereas general atu; 
s errors. Consequently det: 
it would be if profile elus 
al determination of this ui 
graphical of numerical solute 
-eatment of experimental it 
ich are centres of gravity 
iooth curve can be olus 
: of polygon points for enr 
will be obtained. 
to various photogranné 
erimental aerial trianguli 
ing each time centres of gii 
tion eurve. | 
nine the necessary deg’ 
an error of observation qe 
  
ELEVATIONS 
ction according to A PR 
Smoothing of A PR -indications 
— Correction. Curve 
CURVE Form 
© Corre 
Established by 
CORRECTION 
fe rere 
320 
300km 
pd 
280 
260 
eu TN 
240 
- 
220 
ene ME 
200 km 
180 
160 
= 
ae 
oo” 
140 
120 
o 
A mong E 
ze 
> 7100 km 
= 
I- 
2757 
o o7 —9 80 
"o9 40 e 
M omo 
Ro. 
> 
  
€ 
3. 
Fig. 
5 
from third differences and the mean error of observations 
referred to the smooth curve in question. In our case it 
was sufficient to repeat twice the smoothing process. 
To correct the twist of the strip about x axis horizon 
measurements from infrared oblique photographs could 
be used. Single horizon indications are not precise enough 
to be used directly for the adjustment of the twist in the 
strip. Applying, however, a smoothing procedure as 
described above, the achievement of much better ac- 
curacy is possible. 
In our method the adjustment of x-coordinates is 
greatly simplified because the scale of each model is 
determined independently by the introduction of known 
ground clearance. This fact, together with the use of 
altimeter indications in the instrumental procedure 
changes the error propagation in the transfer of the 
scale. 
From the analyses of the radar profile technique 
itis evident that errors in clearance variations are 
accidental. When introduced into the plotter, ground 
clearances are affected additionally by the errors of the 
models and by the instrumental errors. The errors which 
originate in model are functions of b, errors, partial y 
errors and convergence y errors. The accumulated value 
of these errors is of linear form. The constant part of 
the errors is eliminated automatically by the process of 
coordinate transformation. In our experimental bridging 
on 330 km distance the accumulation of the accidental 
part was so small that no further adjustment was pos- 
sible. Should however a different case occur, such as 
complicated terrain forms which limit to certain degree 
the accuracy of the APR clearances, use would be made 
of the differences between ‘photogrammetric’ clear- 
ances and the APR clearances. It is hoped that the 
bridging carried on at present time over partly hilly 
country will present such a case. 
It is obvious that the scale adjustment embraces 
both x and y-coordinates. 
The use of infrared oblique photographs in order 
to correct the transversal bend of the bridged strip is 
the subject of a separate paper?!) presented by our 
group and therefore we will limit ourselves to only a 
few remarks. 
When using straight line constructed on infrared 
oblique photographs the accuracy of y adjustment does 
not depend very much on the number of straight line 
points included in the bridging. However, the regular 
shape of the correction curve may be a certain indication 
1) U. V. Helava “Use of infrared Red Oblique Photo- 
graphs in Aerial Triangulation”, paper to be presented 
at the Meeting of Commission III. 
  
    
    
   
  
  
   
   
   
   
    
        
   
  
  
   
   
    
    
   
   
   
     
     
  
  
    
   
  
a ———— N