In version. 1 both, photogrammetry and control are of the same accuracy, in version 2 the control 
accuracy is considerably poorer. 
To show the change to the worse in the accuracy after blocktriangulation due to the version 1 and 2 
^s an example a block of 10 strips, 20 models each was used in combination with a dense perimeter 
control in planimetry and a dense screen of height control (H2: i = 3). 
Compared with errorfree control the mean square standard deviations and the maximum standard 
deviations of the coordinates of minor control points were increased at the following factors. 
ox mean oy mean oz mean ox max oy max oz max 
Version 1 1.04 1.04 1.17 ].12 1.12 1.14 
Version 2 1.55 1.58 1.49 2.50 2.40 1.44 
4. Accuracy comparisons between different block adjustment methods 
4.1 Spatial blocks versus planimetric blocks with independent models 
For the control variants P1 (dense perimeter control) and P4 (4 control points in the block corners) 
the mean square standard deviations as presented in formula (1b) and (4b) can be compared with the 
corresponding results after planimetric block adjustment only. 
In case of P1 the standard deviations ox mean and oy mean are approximately by a factor 1.1 smaller 
than the mean square standard deviation of the adjusted coordinates in planimetric blocks (see [2]). 
That is true for blocks with 5 to 25 strips. 
Applying control variant P4 to a block with 10 strips, 20 models each the improvement in the 
planimetric mean square accuracy is in the order of 1.2 (see [1]). 
It should be mentioned here that the improvements are not only founded on the more general 
spatial adjustment but also on 6 tie points per a model instead of 4 in the case of the investigated 
planimetric blocks. Hence it follows that the planimetric block adjustment gives nearly the same 
accuracy as the spatial adjustment by independent models. 
4.2 Spatial blocks with independent models versus bundle blocks 
The common base on the stochastical models for the two rigorous methods of block adjustment 
(see 3.1) makes it possible to compare the results presented in 3.2. with the corresponding results 
of the theoretical bundle block studies performed up to now. A complete comparison will be possible 
as soon as the investigation of Mr. Krack will be finished. 
The preliminary results show for instance that the mean square standard deviations ox mean and 
oy mean in a block of 200 models with dense perimeter control can be improved by a factor 1.6 
performing a bundle block adjustment instead of an adjustment by independent models. That is 
valid for 20 9 sidelap as well as for 60 95. 
Concerning the heights we have learned that a bundle block adjustment improves the accuracy 
obtainable by a model block adjustment the more the poorer the control is. 
With 20 9, sidelap and the height control distribution HI (dense perimeter control only) Krack’s 
results for a block with 15 strips and 31 photos each showed an improvement at a factor 1.8 and 1.9 
for oz mean and oz max respectively in comparison with formulas (8) and (9). With the much denser 
height control distribution H2, i — 4 (see figure 3) TALT’s results for bundle blocks are only slightly 
superior to the corresponding results with independent models: a factor 1.2 in oz mean and practicly 
the same accuracy oz max [11].