ABSTRACT
Ensuring the reliability of measurements is one of the basic targets in designing geodetic networks. The reliability of a measurement is, after precision, the second component of its quality. A level too low of reliability causes a worse detection and location of gross errors, and an excessive influence of individual measurement error on the results. At too high reliability level may indicate a low economy of the design or too low precision in a group of measurements in relation to precision in other groups. The problem of reliability is tightly related not only to a detection of gross errors of measurements but also to other targets of design, such as accuracy of calculated unknowns and homogeneity of the accuracy. In aerial triangulation, photocoordinates are the most precise and most numerous group of measurements up to 95% of all, so their precision and reliability have a fundamental influence on the accuracy of aerial triangulation results, the reliability of other groups of measurements being part of it. The reliability measures and estimates applied to aerial triangulation are quoted in the paper. The results are presented of an extensive research on the photocoordinates reliability obtained from 25 aerial triangulations made in Poland in the years 2002 to 2007 for three scales of aerial photographs: 1: 13000, 1:19000, 1:26000 and for the number of photographs in a block ranging from 136 to 3402. Blocks had natural control points and the measured projection centers. With the BINGO software the measurements were calculated anew to check the correctness of adjustment and to standardize the statistic estimates, because the aerial triangulations were made with the use of different systems of measurement and adjustment. The following results are given for each block: precision of the photocoordinates, average residual of a photocoordinate, average local redundancy numbers of a photocoordinate, average number of tie points falling into one photograph and average standard deviation of determining the ground coordinate of a tie point. In tables of the results there are given also mean values of the estimates for blocks of the same scale of photographs. The reliability analysis results are given for a single tie point photocoordinates in relation to the number of photographs on which they were measured. The results are also given of an analysis of the average reliability of photocoordinates. From the analysis it appears that only in less than half of the blocks under study the reliability of the photocoordinates measurement gained a good level. In some blocks a significant disproportion exists between the big number of measurements on the photographs and the relatively small number of tie points between strips. This indicates negligence in the use of automatic measurement and besides the absence of edition of this measurement with the aid of tools for elimination of useless and ineffective tie points and tools detecting locations where additional measurements are required. The conclusions from the research concerning the measurement precision and reliability and the accuracy of the tie points calculation were taken into consideration in a project of the new aerial triangulation standard.