The Research Of Systematic Errors In GPS Carrier Phase Observation

The systematic errors from different sources of GPS measurements (i.e. GPS Signals) vary and are complex. Therefore strong measures for modeling them should be established because, their presence in GPS observation can degrade the positioned accuracy. Some kinds of the systematic errors in GPS observation can be used in the scientific research, for example, ionospheric delay and tropospheric delay are important information to meteorology. This research studies the models of the systematic errors in the GPS observation and the ways to classify them, analysis of the characteristics of the residual systematic errors, and develop a method that improves the accuracy of GPS point positioning. Besides studying the current theories of the GPS and surveying data processing, the following main task have been accomplished. This paper introduces semi-parametric regression model with natural cubic spline to determined integei ambiguities, and summarizes some results using semi-parametric regression in practice. From experiment I found that the solution is precise and the range of solution searched narrows using the technology that approximate solution has been obtained first to make the amounts of the parameters and systematic errors comparative. However, the smoothing parameter is crucial and difficult. The solution is reasonable using the method that the generalized cross-validation is associated with the pre-accuracy of the observation to select the smoothing parameter. And the properties of the range between two inflexions in L-curve method have been analyzed. According to the result of ionospheric delay , ionosphere is uneven and therefore it is suitable to simulate the ionospheric delay of an area instead of computing with ionospheric model. Also the analysis show that the residual systematic error may be dividedinto two parts. The first one is a parabola highly related to elevation angle, and the second one is a fluctuating cycle of sine. Therefore the method that improves the accuracy of GPS point positioning is developed based on this feature. Finally, some suggestions are put forward to model the systematic errors and the separation of some complex errors (i.e. multipath delay) from the residual systematic errors.