On 17 July 1994, 20:29 UTC, less than 19 hours after launch, the Grasse laser system acquired the first laser tracking data on ERS-1. The next day both the Grasse and the Zimmerwald lasers tracked the satellite. Then the other lasers stations followed and within 2 weeks a total of 13 stations in Australia, Austria, England, France, Germany, Hawaii, Italy, Japan, Peru, Switzerland, and the USSR had tracked ERS-1. However, the number of data passes taken by these globally distributed systems turned out to be rather small. For the first two weeks in orbit the 13 lasers acquired 65 data passes, or on the average 0.4 tracking pass per system per day. Later on the situation improved somewhat and during the month of August the satellite was tracked by 20 lasers, providing a total of 267 data passes. The systems at Grasse and Herstmonceux were the most productive with 33 and 38 data passes, respectively.
SSR&T started its ERS-1 orbit computation experiments on 30 July 1991. Initially, laser tracking data taken during the period 22-25 July 1991 were processed, applying the NASA Goddard Space Flight Center (GSFC) GEM-T2 earth gravity and ocean tides models [Marsh et al.1989]. The first computations immediately revealed the relatively poor tracking coverage of the satellite, which was underlined by an unrealistically good orbital fit of the available measurements and a large sensitivity of the recovered parameters for changes in the computation model. The experiments were repeated with a new data arc, which covered the period 31 July to 3 August and comprised more global tracking data. In this experiment the orbital fit of the laser ranging data decreased to more-realistic levels. However, the results from this experiment also indicated the existence of serious modelling problems, which have significant delayed the operationalisation of the orbit computation process at SSR&T . For a large part these problems resulted from the fact that the ERS-1 quick-look laser observations retrieved from DGFI were neither corrected for the centre-of-mass offset of the laser retroreflector array, which is the distance between the geometric centre of the array and the satellite's centre of mass, nor for the distance between the array's optical centre and its geometric centre. The pre-launch information available at SSR&T showed that the data would have been corrected for these effects. This required a last-minute major modification of the SSR&T software.
On 19 August 1991 the first reasonably accurate orbit was computed from the laser tracking data acquired in the period 1 July to 3 August, and solving for half-daily drag coefficients. It was found that the RMS -of-fit of those tracking data which passed the screening process was about 1.3 m. From a detailed analysis of all available results it could be demonstrated that the orbit had an accuracy of about 1.5 m in the radial direction. This accuracy was considered promising for a first experiment and an intensive model tuning activity was started to further improve the accuracy of the computed orbits.
Apart from software modifications, the major improvements came from the replacement of the GEM-T2 gravity and ocean tides models by the corresponding NASA/GSFC PGS-4591 models, and later the Joint Gravity Models JGM-1 and JGM-2, developed by NASA/GSFC, UT/CSR, and CNES.