Stochastic Modelling of GPS Phase Observations for Improved Quality Estimation

Abstract

Data quality information has been recognised as essential in assessing the fitness for use of any spatial dataset, and fundamental to enabling efficient and effective data integration through spatial data infrastructure (SDI). Missing or inaccurate data quality information can result in inappropriate use of the data with associated consequences of poor decision making, reduced utility and decreased market value. The increasing use of the Global Positioning System (GPS) as a primary data acquisition source for spatial databases highlights the significance of this problem. At present the measures of quality for GPS derived coordinates given by commercial software packages tend to be unrealistic and are more often than not optimistic. This is because not all of the systematic and random errors present in the observations are fully modelled through the standard functional or stochastic models used. This paper presents some of the current problems in identifying the quality of GPS data as derived from commercial processing software. Common GPS processing strategies are reviewed in the context of error modelling and data quality. Finally, current research activities into strategies for maximizing GPS data quality are presented.