After a period of operational processing system testing, and the selection and tuning of the models applied for the Precise Orbit Determination (POD), the operational ERS-1 POD for NOAA started in December 1991. Since then, some modifications to the operational system setup have been made in order to increase its flexibility and performance. A major improvement was the inclusion of predicted or estimated orbit manoeuvres in the force model. In this and following Sections the fully-operational setup, as adopted since June 1992 will be discussed: the computation model and constants, the time schedule, and the system performances.
The operational ERS-1 orbit determination, depicted schematically in Figure 1, starts with the retrieval of global ERS-1 quick-look SLR tracking data from the Eurolas Data Center in München. These quick-look tracking data consist of both field-generated normal points and sampled measurements. The sampled measurements are converted into 1-per-20-second normal points by SSR&T.
In addition to the quick-look tracking data, ancillary data are required to support the POD process. Vital are up-to-date so-called Solar Geophysical Data, which are obtained from the Canadian Space Geodesy Forum (Canspace), and Earth Rotation Parameters, of which weekly predictions are retrieved from the IERS Bulletins-A. Finally, information on planned orbit manoeuvres are used: the start-time, duration, and magnitude of the thruster pulses, as provided by the European Space Operations Centre (ESOC).
A third type of data consists of archived DUT/SSR&T data concerning previous orbit generation runs, including an initial state-vector, estimates for atmospheric drag, and other solved-for parameters.
All available tracking data, ancillary data, and a priori parameters are processed on an arc-by-arc basis by the DUT/SSR&T version of the GEODYN II orbit determination software, developed by NASA/GSFC.
The selected length of the orbital arcs is based on a number of operational considerations:
The arcs start on Tuesday midnight or Friday noon, and the processing takes place the following Tuesday or Friday, respectively. From the orbit difference for the 2-day overlap period of two consecutive arcs, information about the accuracy of the POD may be obtained.
The central 3.5-day period is used by NOAA to create one ERS-1 IGDR, starting and ending at a crossing of the satellite over the Antarctic. Accordingly, this part is referred to as the NOAA arc.
Output of the orbit processing are a number of estimated orbital parameters and station coordinates, and an ephemeris file, which describes the position of the satellite's nominal centre of mass in terms of geodetic latitude, longitude, and height above the GRS80 reference ellipsoid, as a function of UTC time. These positions are stored at 60-second intervals in one file (Orbital Data Record, ODR) per arc, which is transferred to NOAA , where the satellite positions are interpolated to the altimeter measurement times.
Originally, the operational orbits were based on the GEM-T2 gravity and ocean tide models, whilst semi-operational orbits were determined with the GSFC PGS-4591 model, and the Joint Gravity Models JGM-1 and JGM-2. Also, the estimation of additional orbit parameters was investigated. Recently, only the JGM-2 orbits are provided, since their accuracy surpasses by far the original GEM-T2 orbits.
The computation models and constants that are used in the DUT ERS-1 POD, are based on preliminary studies [Wakker et al., 1983; Wakker et al., 1987] and the know-how and experience of DUT/SSR&T and closely resemble the TOPEX/Poseidon POD standards. A complete description of the models and constants applied in the ERS-1 POD, representing the current state-of-the-art, including very extensive and detailed dynamic, measurement, and geometry models, is given in Table 1.