Soft modelling of possible blunders in geodetic networks

Production of highly precise and reliable geodetic positions requires an assessment of the observed values regarding possible outliers or blunders. This is traditionally done using an iterative procedure. In each step, the observations are mutually considered as blundered. The respective error is modelled, estimated and tested for significance assuming normally distributed observations. The one observation leading to the most significant estimated error in the respective step is eliminated. The procedure stops when there is no significance. The widespread use of this strategy benefits from its theoretical background, its algorithmic efficiency and its adequacy in standard data processing. However, its main disadvantage is the neglection of multiple blunders in each step. In this study fuzzy logic is used to simultaneously model, estimate and statistically test all possible errors in the observations to overcome this problem. This strategy extending the traditional way is developed and discussed with respect to the classical case. The presentation concludes with numerical examples.