A REGRESSION STUDY ON RELATIVE GPS ACCURACY FOR DIFFERENT VARIABLES

Abstract

We know that position accuracies from relative and static Global Positioning System (GPS) measurements are degraded by a variety of physical, hardware and software effects. As a result, users often take special precautions to minimise these effects and, hopefully, improve their accuracies when planning and making measurements. In this study, the effects of the baseline length (BL) and observation duration (OD) on vertical and horizontal positioning errors for relative, static GPS measurements were experimentally evaluated. In making this evaluation, 5 min to 24 hours GPS data, collected over a 15 day period, from the Southern California Integrated GPS Network (SCIGN), were analyzed. The root mean square (RMS) of the differences between the true and measured vertical (up/ellipsoidal height) and horizontal (north and east/ latitude and longitude) positions of the included stations were computed and then used to create empirical error equations dependent upon the BL and OD. Considerations for the number of satellites (SVs) and the positional dilution of precision (PDOP) were included too by necessity. With these equations, GPS users can better estimate their vertical and horizontal position accuracies based on the known BLs, desired ODs, available SVs, and predicted PDOPs when planning a project, or have an external standard of comparison when evaluating their results.