The precise tracking of ERS-1 is performed by a global network of SLR stations. Because the measurement rate differs considerably from station to station, all observations are converted to equally-spaced 1-per-20-second pseudo measurements, or normal points. The following corrections are applied to the range measurements:
Because some stations are known to deliver SLR data of lower accuracy than others, all SLR data are weighted partially by a station-dependent a-priori standard deviation, , ranging from 1 cm for high-precision lasers to 20 cm for stations with a higher system noise or inaccurate stations coordinates. The other part of the weight represents the overall model uncertainty, , for which appropriate values are chosen, depending on the applied gravity field model, ranging from 20 cm (JGM-2) to 50 cm (GEM-T2). Finally the total measurement is defined as the root-sum-square (RSS) of and .
During the last three years, the SLR tracking of ERS-1 has varied considerably, as can be seen in Figure 3. The weekly-average number of SLR normal points per day ranges from about 80 to 200. These variations are mainly due to the weather conditions on the Northern hemisphere. A typical decrease during the winter is apparent.
In the early days, the majority of stations tracked ERS-1 predominantly on the second half of the pass, because it took a considerable amount of time to ``find'' the satellite manually, by adding corrections to the predicted orbit while the satellite was passing by. The orbit predictions were especially hampered by inaccurate drag modelling, which resulted from the high solar activity levels in early 1992.
Nowadays, the solar flux has reduced to a moderate level, thus facilitating the predictions, the satellite tracking, and the orbit computation. In Figure 3 the correlation between the solar flux and and the frequency of the orbit manoeuvers is apparent. Since the atmospheric drag decreases with the solar activity, the orbit manoeuvers to keep the satellite's ground track close to its nominal pattern become less frequent at lower solar activity levels, and have a shorter duration.
The increased amount of available SLR data and decreased solar activity during the summer and fall of 1992 has stabilised and improved the orbit determination accuracy, even though the RMS of the laser residuals has increased, as can be seen in the bottom plot of Figure 3. It is very likely that the residuals during the first few months of 1992 were unrealistically low as a result of insufficient tracking. A similar effect is observed in December 1992, during which month SLR tracking data are sparse due to unfavourable European weather conditions and the Christmas holidays.
Evidently, the RMS of the laser residuals will only be a good representation of the global orbit error when the tracking data are regularly distributed in time and provide a good global coverage. as regards tracking coverage, the period of Cycle 9 was a particularly poor period. The results of orbit precision estimates presented in this Chapter, which are based on Cycle 9 data, may thus be considered a ``worst case''.
Table 2 shows that as much as 57% of the measurements were taken by 10 European stations out of a global total of 24. Like any other repeat cycle, the Southern hemisphere is poorly represented: as few as 3 stations provided a mere 13% of the measurements. As a consequence, a local orbit fit through the SLR measurements is easily obtained over the Southern hemisphere. Although the RMS -of-fit may be low for these few stations, it provides little information on the local orbit accuracy, which will most likely be much worse than elsewhere.
Figure 4 depicts the irregularity of the SLR tracking coverage. Note that the satellite is tracked predominantly over Europe and the western United States, and on ascending (night time) passes.
Furthermore, the SLR data are far from homogeneously distributed in time. During the period of POD arcs 160-169 ERS-1 was tracked only 1.6% of the time: 12h16m by one laser, 1h12m by two lasers, 26 minutes by three, 8 minutes by four, and only 1 minute by five lasers simultaneously. Thus, many large gaps between two successive tracking periods exist, during which the orbit can be computed with considerably less accuracy (Table 3). The largest gap during the pertinent period was 29h57m !