IVS 2000 General Meeting Proceedings |
Per Helge Andersen
Forsvarets forskningsinstitutt (FFI) (Norwegian Defence Research Establishment)
The GEOSAT software has recently extended the analysis capability from 30 to 60 GPS stations per day. In addition, a new procedure for the generation of a priori GPS orbits for the filter has been implemented. A three day IGS precise orbit for each GPS satellite is used as observations for the determination of a GEOSAT-generated GPS orbit. In this fit six orbital elements in addition to nine solar radiation pressure (SRP) parameters are solved for. The rms of residual fit for the three-day orbital dataset is typically between 2 and 4 cm for all satellites except for 2-4 satellites where the rms is significantly larger. The estimated orbit is used to generate a priori observation residuals and observation partial derivatives to be used in the filter where all data types are combined. In the 24 hour filter solution the nine-parameter SRP model are kept fixed and only the six orbital elements and one SRP-scaling parameter and a Y-bias parameter are solved for. In this way the one-day orbits will be based on a very realistic SRP model. However, for satellites with large rms of orbital fit two additional stochastic velocity change parameters are solved for. Experiments with the IGS precise orbits show that this parameterization is sufficient in order to fit a 24 hour GEOSAT-generated GPS orbit for all satellites including the outlier satellites with an rms of 2-3 cm in each coordinate. A significant part of the 2-3 cm difference is due to the use of inconsistent values for the EOPs. EOP values are taken from IERS and not the IGS EOP estimates to be used with the precise IGS orbits. In conclusion, a highly sophisticated GPS orbit strategy has been established. The use of a large number of GPS stations in combination with a moderate number of dynamical solve-for parameters and a realistic dynamical model is expected to result in GPS orbits with a precision level of a few cm in each coordinate.
It is a fact that the position of the effective phase center of the transmitter antenna of the GPS satellites is not very well known. This leads to a scale inconsistency with VLBI and SLR of approximately 2 ppb. It is therefore a standard procedure in analysis with GEOSAT to estimate the z-coordinate of the mean position of the transmitter phase center. With GPS-data alone it is only possible to determine the phase center position relative to the position of the phase center of a reference satellite. Since the GPS data in our case are combined with VLBI and SLR data absolute positions of the phase center of all satellites can be determined.
Another possible candidate for scale inconsistency between the different techniques is the tropospheric mapping function. In order to be consistent with VLBI and GPS the SLR data are analyzed with the use of the dry NMF mapping function which is used for VLBI and GPS. The signal delay in the zenith direction is however calculated using the Marini-Murray model. This procedure is in accordance with recent recommendations given by Richard Eanes. A model for tidal geocenter motion is implemented (Watkins and Eanes, [3]).
In GEOSAT data from the different techniques are combined in batches of one day which is called an arc. The state vectors and complete variance-covariance matrices from the analyses of a number of independent arcs of space geodetic data can be combined using the CSRIFS (Combined Square Root Information Filter and Smoother) program. Four parameter levels are available and any parameter can, at each level, either be represented as a constant or a stochastic parameter (white noise, colored noise, or random walk). The batch length (i.e. the time interval between the addition of noise to the SRIF array) can be made time- and parameter dependent. More details can be found in Andersen ([1]).
Statens kartverk has recently received a copy of the GEOSAT software.
The GEOSAT software will be converted to PC/LINUX in the near future.
IVS 2000 General Meeting Proceedings |