Accuracy and Precision Analysis of Total Station Measurements

Abstract: The main objective of this bachelor thesis was to gain awareness of accuracy and precision in total station measurements. Experiments were designed, executed and analyzed.   The approach to the subject was to examine measurements made with various prism set-ups. Prisms are supposed to be directed to the total station with the front face in the line of sight. In the experiment the prisms were directed in various angles in order to measure in non-perfect measuring arrangement and the behavior of the precision and accuracy were noted.   The mean values and standard uncertainties were calculated from the material as well as RMS Error in the cases where reference values could be used as true values. The reference values for comparison were available due to the possibility to use a test field – a network of concrete pillars with well-defined coordinates.   To see whether the precision (the standard uncertainty) changed between a measurement sample made with correctly positioned prisms and a measurement sample towards the angled prisms, a statistical test method was used. (The test method is recommended in the international standard ISO 17123-1; Optics and optical instruments, Field procedures for testing geodetic and surveying instruments, Part 1: Theory).   The results from the experiments showed that there were larger standard uncertainties in the slope distance measurements towards extreme angled prisms than to correctly directed prisms. Surprisingly it was also larger standard uncertainty in vertical angle measurements toward correctly directed prisms than slightly angled prisms. The mean values of the horizontal angle measurement drifted sideways in the measurements made towards one extreme angle to the opposite extreme so that a total shift of around one centimeter was detected at short distances as well as up to 100 m distance.    When aiming in between the separate prisms of 360-degree targets the standard uncertainty were larger in horizontal angle measurements at a few occasions where it seemed like the autolock function had chosen two different prisms to lock on to, at separate occasions. The standard uncertainties were also larger in all vertical angle measurements when aiming in between the prisms.   The results showed clearly that even when the standard uncertainties of the measurements were of equal size, the mean values – when measuring in the different test positions – could vary substantially.