THEORY OF SEPARABILITY FOR TWO DIFFERENT MODEL ERRORS AND 
ITS APPLICATIONS IN FHOTOGRAMMETRIC POINT DETERMINATIONS 
Deren Li 
Wuhan technical University of Surveying and Mapping 
Wuhan, China 
ABSTRACT 
The well-known reliability theory ( Baarda,1967/68 ) can be 
Successfully used to Judge whether a model error can be sta- 
tistically detected and how great an effect on the adjusted 
results the non-detected model errors have. Starting from 
two different multidimensional alternative hypotheses a new 
theory of separability for two different model errors is de- 
veloped. With this theory it can be evaluated whether two 
model errors can be statistically distinguished each other 
and how great an effect on the results the non-distinguish- 
able model errors have, which is very important for the loca- 
tion of gross errors, the separation of gross errors and sys- 
tematic errors, the selection of additional parameters by the 
compensation of systematic errors and the analysis of geodetic 
deformation measurements. Some investigations about the loca- 
bility of gross errors and the separability between gross and 
Systematic errors in the photogrammetric point determination 
have been done and the main results are in this paper described 
1. INTRODUCTION 
The reliability theory of Baarda is used to evaluate the re- 
liability of Least Squares and gives us lower bounds for de- 
tectable errors or estimability of additional parameters ag 
the indicators of internal reliability and Sensitivity fac- 
tors as the indicators of external reliability. It has been 
Successfully used in the field of geodesy, photogrammetry and 
engineering survey. But this theory is based on the hypothe- 
Sis test with a single alternative hypothesis. 
In many cases the test is not against a single alternative 
but against a lot of alternative hypotheses, from which we 
486. - 
  
  
s US REESE T A E A EE ums 
OFDM EA RTE