No abstract

Over the past 18 months, a team in the Western Australian Centre for Geodesy at Curtin University of Technology, Perth, has been researching the optimum configurations to achieve long-range and precise GPSbased aircraft positioning for subsequent airborne mapping projects. Three parallel strategies have been adopted to solve this problem: virtual reference stations (VRS), precise point positioning (PPP), and multiple reference stations (MRS). This paper briefly summarises the concepts behind the PPP and VRS techniques, describes the development and testing of in-house software, and presents the latest experimental results of our research. Current comparisons of the PPP and VRS techniques with an independently well-controlled aircraft trajectory and ground-based stations in Norway show that each deliver precisions of around 3 cm. However, the implementation of more sophisticated error modelling strategies in the MRS approach is expected to better deliver our project's objectives.

In this paper, the potential of long-range kinematic GPS positioning with a multiple reference station (MRS) network for airborne applications is discussed. A novel method of creating Virtual Reference Stations (VRS) is proposed for post-processed airborne GPS kinematic applications, which is called the modified semi-kinematic VRS method (MS-VRS). The purpose of the VRS is to generate data from real GPS observations made by the MRS network, resembling that of a non-existing (virtual) reference station situated close to the project area, so that the commonly used methods for short-range kinematic GPS data processing can be used to determine the position of the aircraft. During the initial phase, the VRS of the MS-VRS method refers to a fixed position according to the aircraft's initial approximate position, and the corrections are applied according to the aircraft's trajectory. The MS-VRS method differs from the conventional VRS method and semi-kinematic VRS method (S-VRS) in that when the aircraft's current approximate position is more than 10 km from the initial VRS position, a new VRS is created. The MS-VRS data can be generated in RINEX format, so that it can be processed using any kinematic GPS post-processing software. Using a simulated kinematic test with static data, the MS-VRS method showed a 12.1 to 47.6 percent improvement in the three coordinate components with respect to the conventional single reference station (SRS) approach. Tests and analysis with real airborne GPS data are presented in some detail using a MRS network and flight test data in Norway. The results indicate that centimetre-level accuracy can be achieved based on the proposed MS-VRS method, which is superior to the S-VRS method, with improvements of 11.4 to 47.4 percent in terms of standard deviations of the coordinate domain.

J. Badekas reinterpreted M.S. Molodensky's three-dimensional similarity transformation as a vector solution using a centroid. The solution has since been [mis]interpreted by some others with inconsistent reference to the methods of both Molodensky and Badekas, principally relating to the translation vector and the stochastic model. This appears to have led to incorrect claims that the Molodensky-Badekas method is superior to the Helmert similarity and Burša-Wolf methods. This paper reviews the development and description of the original Badekas method, reconfirming its equivalence to the Burša-Wolf method in the forward direction, and provides an alternative solution that suits the same-formula reversal common in commercial surveying software. It is also demonstrated that the Molodensky-Badekas method has no inherent superiority over the Burša-Wolf method, has an ambiguous functional model, and nominally underestimates its parameter statistics when these are compared directly to those from the Burša-Wolf method.