1 2 A Horizontal and 
point 
  
  
  
  
  
  
  
  
  
  
  
  
ju - vertical control 
"dm o Vertical control point 
— iu 
dm Block size: 11x15 
uti Flight direction: 
ie 
> West --- East 
—»—— End and side overlap: 
—— 60% 
— 
> 7 
3 AN 
Fig. 2 Block geometry 
This example shows that under a certain geometry gross and 
systematic errors are very difficult to be separated. Their 
wrong separation will lead to wrong detection of gross errors 
and wrong Compensation of Systematic errors , and therefore, 
will decrease the accuracy and reliability of adjusted results. 
Example 3: Wrong decision between gross error and deformation 
Fig.3 is the result of deformation analysis from project Mont- | 
salvens ( Gründig,1985 ). In epoch 2 a gross error of 50 Mgon 
was introduced in the direction observation Tig and couldn't 
be detected with the data snooping. At last the analysis shows 
us à very significant shift of 1.2 cm in point 9, Obviously 
this is a wrong decision. 
From these three examples we can say that it is neccessery to. 
extend the reliability theory of Baarda into the separability | 
theory which describes the ability of distinguishing two alter- 
native hypotheses in Gauss-Markov models. 
First in the year 1983 FOrstner made & definition of separa- 
bility and indicated that the separability is essentially de- 
pendent on the correlation coefficient between two test values. 
In his doctor thesis ( Li, 1985 ) the author derived the sepa- | 
rability and reliability under two multidimensional alternative | 
hypotheses in extended Gauss-Markov models ( also see Li, 1986). | 
With this theory it can be evaluated whether two model errors | 
are statistically distinguishable and how great the effect of | 
| 
“88. - | |