sted 
  
  
ıly the 
racy of 
res in- 
nainly 
f GPS 
nbina- 
datum 
lon we 
oordi- 
30 cm 
. Also 
maller 
is also 
oordi- 
  
Fig.5 Influence of ground control accuracy (oc) and of 
drift parameters on the accuracy of adjusted GPS blocks 
  
  
  
  
  
  
  
  
  
Be 1:30000 
! 6 * 21 photographs 
[cm], ^» - 10um P Hay 
Ogps — 30cm - 9p [o] 4 
z i 
Bo = Wem ^ Mo — (3) 
Jt Te ROM UM e TA = e$ (4) 
uet RP 
60 durat ee (2) 
ET 2.4 
Yr tr TU 
40 Povcocse@ 00000 °° 
jer SLT dd 
20 
; 2 3 [=] 
10 30 50 100 [em] PS cp 
Mz : " 
[cm], © i z 
e) 4 Den] 
] den aiu) 34 02 
80 4 Ag) 
A En 0.000009 D] (4) 
e fr e - cee @o0000°%® 0090999 0000000000 9999 
60 e 00000 @0c0000000000 eocc 000000000? 2 | 
1 . e up) 
40 4 ; - 0.1 
control case c no drift par. (1) 
J e -————- drift par. block (2) 1 - 
2 a eme mm — drift par. strip (3) 
0 d b © 00000000... 0€ drift par. strip (4) 
3 (cross strips) 
: 2 3 [%] 
10 30 50 190: .; [em]- «ep 
  
  
nates of ground control image points is the same (cp: = 00) 
as for image tie-points. A few cases have been run in which . 
the image coordinate accuracy of ground control points has 
been assumed to be oc p; — 3ym, as could be with signalized 
control points. 'The effects have turned out to be very minor, 
and that case is not pursued here any further. 
In fig. 5 the results are displayed which demonstrate the ef- 
fects of variation of ground control accuracy on the adjusted 
blocks, for the described specifications where especially the 
GPS camera position accuracy is kept fixed to ogps = do. 
Four different cases of ground control and datum transfor- 
mations resp. drift corrections are distinguished. The cases 
are depicted in fig. 7 and 8. 
The graphs show directly that the curves (1) behave unlike 
the others. That case refers to blocks with 4 control points 
only and no datum transformation applied. The result is evi- 
dently determined by the internal block accuracy and by the 
absolute GPS postioning. The errors of the 4 ground control 
points have practically no additional effect, even if they are 
large, if no free drift or datum parameters are applied. 
In all other cases, where the block is fitted via free param- 
eters onto the control points, the error level is considerably 
higher, even if the control points have no errors at all. It is 
the GPS errors and the internal block accuracy which raise 
the error level immediately, although the different control 
scenarios (c, a, b) react differently. The influence of ground 
control errors is superimposed on that basic behaviour of 
blocks. They represent an additional error component which 
can gain considerable effect if the errors of ground control 
points increase to large values. The relations of fig. 5 indi- 
cate, however, that the effects of ground control errors are 
very minor (in most cases < 5% ) as long as the control 
points are adequately precise, i.e. as long as oc p < 00. 
697