European Space Agency


Achieving, Assessing and Exploiting the ERS-1/2 Tandem Orbit Configuration

J.M. Dow, M. Rosengren, X. Marc, R. Zandbergen*, R. Piriz, & M. Romay Merino**

Orbit and Attitude Division, European Space Operations Centre (ESOC), Darmstadt, Germany

* With Logica, London, UK

** With GMV, Madrid, Spain

Although the ERS-1 spacecraft, launched on 17 July 1991, had exceeded its specified nominal lifetime when ERS-2 was launched by an Ariane-4 on 21 April 1995, it continues to function very successfully and, thanks to the decision by the contributing ESA Member States to extend ERS-1 operations beyond the end of the ERS-2 Commissioning Phase, the 'tandem mission' has become a reality. Both spacecraft are currently orbiting the Earth in almost the same orbital plane, with a spacing between them of about one third of an orbital period.

This article reviews the requirements for the dual orbit configuration of the two spacecraft, how it was achieved, and how it is being exploited.

Introduction

The advantages of being able to fly the ESA Remote-Sensing Satellites ERS-1 and ERS-2 in parallel, at least for a limited period, have long been recognised. In fact, during the evaluation of the several hundred research proposals received by ESA in answer to the Announcement of Opportunity for the exploitation of ERS-2 payload data, the utility of a 'tandem' mission was identified as a factor common to a significant proportion of the proposals.

In particular, the rapidly expanding field of Interferometry using Synthetic Aperture Radar (INSAR) requires that the projection of the orbital path on the Earth's surface repeat to within a few hundred metres after a fixed number of days. If two radar images, separated in time by one or more cycles of the ground-track repeat period, are superimposed, the interference fringes generated in those parts of the image area in which some change has taken place can be used to detect very small displacements (at the centimetre level). Thus, for example, movements caused by earthquakes and by volcanic activity can be analysed in considerable detail using INSAR techniques.

A number of applications of ERS's Radar Altimeter data, notably the cross-calibration of the ERS-2 and ERS-1 Altimeters, require that the repetition of the ground track should be close enough for cross-track variations in the sea surface, averaged over the beam width of the Altimeter, to be essentially negligible. For optimal use of the Altimeter's height measurements for oceanographic applications, a very high orbital accuracy is required. This is achieved firstly by using (in appropriate computer software) the best possible models for the orbital motion and measurement processes, and secondly by using high-precision tracking measurements , viz. laser ranging, and in the case of ERS-2, data from the PRARE (Precise Range and Range-rate Equipment). The accurate orbits obtained also provide an excellent means of assessing the quality of the operational orbit, which is used for mission-control purposes and for the computation of fast-delivery data products from the payload instruments.

Preparing for ERS-2's launch

While ERS-1 has spent time in various orbits during the past four and a half years in order to satisfy the requirements of the different applications for the payload, it had been decided before the launch of ERS-2 that the latter should remain in a fixed repeat cycle throughout its mission. With this fixed cycle, every point on the ground track will be overflown every 35 days.

The ERS orbits are Sun-synchronous. This means that the choice of orbital height and of the orbital plane's inclination with respect to the equator are such that the precession in the orbital plane due to the Earth's equatorial bulge is equal to one revolution per year, or about 1 degree per day, the same as the apparent motion of the Sun around the Earth. The requirement set for the tandem configuration was that both spacecraft should be in a 35 day repeat cycle, with a time difference of exactly one day between overflights of the same ground point (with either spacecraft leading). Since the ERS satellites orbit the Earth approximately 14 1/3 times per day, by separating the spacecraft in their orbital planes by about one third of an orbital period the trailing spacecraft can be made to pass over the track of the leading spacecraft exactly one day later.

Owing to the unexpected launch delays which affected its sister satellite, ERS-1 was able to make two complete cycles of its Geodetic Phase orbit, in a ground-track repeat cycle of 168 days, before manoeuvres started on 19 March 1995 to bring it into the 35 day orbit. Its orbital inclination was first adjusted by means of two orbital node manoeuvres 50 min apart. Two days later, a drift was initiated, and at the same time in-plane elements were corrected, with two manoeuvres separated by 1 1/2 revolutions. The drift was stopped on 24 March by two tangential burns separated by half an orbit, giving ERS-1 the desired orbit [passing over a particular point on the equator (0.1335 °E) and making 501 revolutions in exactly 35 days (14 11/35 orbits per day)] whilst awaiting the arrival of ERS-2.

This orbit was higher than the nominal initial ERS-2 orbit, for which no constraints emanating from the planned tandem mission were placed on Arianespace. Hence, in order to reach the final tandem configuration, and particularly to achieve the in- plane separation of 31.6 min, a sequence of post-launch manoeuvres would be needed which would depend on ERS-2's exact time of launch within the assigned 10 min window. A further constraint was that the separation between the two spacecraft after ERS-2's launch should be at least 30 min, to facilitate the handling of both spacecraft by a single prime ground station. This constraint was, however, relaxed to 15 min in order to open a launch window on some days on which launch would not otherwise be possible. Consequently, the launch window was open on two days, then closed on the third, with a near repetition of this pattern every three days, due to the ground track almost repeating after three days (42 33/35 revolutions in three days).

The ERS-2 Launch and Early Orbit Phase (LEOP)

ERS-2 was launched on an Ariane-4 (flight V42) at the beginning of the 10 min launch window at 01:44 UTC on 21 April 1995. Its separation from the launcher occurred over NASA's Wallops Island station close to Washington, DC. Deployment of the spacecraft's solar-array arm took place shortly after separation, and the sequence of transitions in the Attitude and Orbit Control System from launch mode, through rate-reduction mode and coarse acquisition mode to fine acquisition mode, took place within about 5 min of spacecraft separation, suggesting very nominal attitude, and low attitude rates. By the time the first tracking data from Wallops Island and the Poker Flat (Alaska) station became available (the satellite went below the horizon at Poker Flat at 02:22 UTC), it was clear that the injection orbit provided by Ariane was extremely close to that planned, with the height of ERS-2's orbit only about 1 km too low, and its inclination just 0.003 ° away from nominal.

One consequence of this very nominal orbit was the small inclination manoeuvre needed, which was in fact about one tenth of that expected on the basis of the normal launcher injection uncertainties. Moon blinding constraints on the spacecraft earth sensor used for attitude determination meant that the inclination manoeuvre, which required a 90 ° rotation of the spacecraft before and after the burn, could not take place before 25 April. The date and time of the ERS-2 launch meant that ERS-1 was the lead satellite by some 47 min (just less than half the orbital period), and so the first manoeuvres executed on 22 April were a pair of in-plane thrusts designed to initiate the drift to provide the proper separation between the two spacecraft. Orbit calibrations showed that the manoeuvres were implemented with better than 1% accuracy.

The spacecraft attitude rates during the inclination manoeuvre are shown in Figure 1. The manoeuvre started at 02:57:08 and lasted 46 s. The large rotations are apparent and the convergence afterwards is rapid. A further pair of relatively large in-plane manoeuvres on 27 April, 50 min apart, followed by a pair of small touch-up manoeuvres a day later, provided the correct starting conditions for the Commissioning Phase, which could then begin on 28 April 1995.

ERS-2 Attitude Rates
Figure 1. ERS-2 attitude rates during the first inclination manoeuvre

The commissioning and routine-phase orbits

The Commissioning Phase orbit requirements, derived from the need to cross-calibrate the Radar Altimeters, were that the spacecraft should pass over a point on the equator at 0.1335 °E longitude in a 35 day repeat cycle, with exactly 501 revolutions per cycle. Each ground track should repeat within ± 1 km of the nominal ground track, defined using a reference orbit model, including the JGM-3 gravity model.A 1 dayseparation between the overflights of a given point by ERS-1 and ERS-2 was specified. Furthermore, between latitudes 65°N and 65°S over ocean zones, the relative tracks of the two spacecraft should be within 200 m of each other. This phase lasted 130 days, or nearly four repeat cycles.

During the Routine Exploitation Phase, which has been in progress since 7 September 1995 when the Commissioning Phase was completed, the above requirements still apply, except that the 200 m between the two ground tracks has been relaxed to 80-250 m. This leeway is needed for the SAR Interferometry, for which a minimum baseline is necessary in order to generate interference fringes. As was the case for the Commissioning Phase, the individual ground tracks of each spacecraft must repeat to within 1000 m around the equator and to within 400 m at high latitudes.

ERS-1's ground track crosses that of ERS-2 at high latitudes twice per orbit, so that the lower limit of 80 m is briefly violated there.

The histories of the ERS-1 and ERS-2 ground tracks since the beginning of 1995 are shown in Figure 2, from which it can be seen that the rather severe requirements for the Commissioning Phase were consistently achieved.

ERS Ground-track Offsets
Figure 2. ERS ground-track offsets from reference

In contrast to the inclination manoeuvres during the earlier ERS-1 mission (two per year, in spring and winter), those for the ERS-2 mission, and consequently for ERS-1 in tandem, are now being carried out at fixed phases of the repeat cycles, with a maximum of five per year. This helps both to reduce the dispersion in the ground tracks and to increase the probability of obtaining interferometric images.

Assessment of orbit accuracy

One of the main purposes of the operational orbit determination is to provide the ERS-1 and ERS-2 ground segment with the latest orbit determinations and predictions, for satellite-acquisition (at the ground stations), mission-planning and fast-delivery data-processing purposes. This determination relies on S-band tracking and fast-delivery Altimeter height data, with various corrections applied. The S- band data consist of range and range-rate measurements from the Multi-Purpose Tracking System (MPTS) installed at the Kiruna ground station in Sweden. An automatic software sequence checks the data's arrival in the ESOC computer after each planned pass, and sends warning messages to the spacecraft controller's console if anomalies are detected.

Once per day, the full sequence of programs is run to process tracking data from the last three days, and to update the orbit file, including a prediction for the next nine days. The central day of the three-day moving window provides the final orbit. Consequently, the operational orbit is available to users with just one day's delay.

In parallel with this process, precise orbit determination is performed using more complete models and all the available data: quick-look laser, S-band range and Doppler, and corrected fast- delivery altimetry from Kiruna (16's normal points). The precise orbit determination is also performed automatically outside normal working hours, including the retrieval and pre-processing of tracking data and the generation of residual and orbit- comparison plots. Solutions are being generated in 4-day arcs, with a delay of typically one week necessary to collect most of the laser tracking data.

The models used in generating both the operational and precise orbits are listed in Table 1. Comparison of the operational and precise orbits allows the accuracy of the operational orbit determination to be estimated. This is currently 2-3 m along- track, 1 m across-track and 30-50 cm radially (30 cm for ERS-2; see Fig. 3).

ERS Orbit Comparison
Figure 3. ERS operational versus precise orbit comparison


Table 1. Models used in ERS operational and precise orbit determination

Reference frame
- mean equator and equinox of J2000.0
- Station coordinates computed from a Topex/Poseidon and Lageos multi-arc solution

Dynamics
- JGM-3 (36,36) gravity model (operational), JGM-3 (70,70) (precise)
- MSIS density model (Hedin 1983); detailed CD modelling; 1 daily (operational) or sub-daily (precise) drag scale factor estimation
- luni-solar gravity
- frequency-dependent solid-Earth tides, Wahr model
- detailed ocean tide model, augmented Schwiderski (precise)
- direct solar radiation pressure model (operational), taking into account spacecraft geometry (precise)
- albedo, infrared perturbations (precise)
- modelling of manoeuvre accelerations, estimation of corrective factors
- one cycle per revolution accelerations

Measurement processing
- Hopfield tropospheric correction (S-band), Murray-Marini (laser)
- Rawer-Bent ionospheric correction (S-band)
- spacecraft transponder delay, and ground calibrations
- centre of mass correction (precise)

Altimeter data processing
- fast-delivery dry tropospheric corrections
- Rawer-Bent ionospheric correction (precise)
- sea-state bias: fast-delivery correction plus an additional percentage of the SWH (precise), currently 0.0%
- ESOC wet tropospheric correction model (precise)
- ERS-1/ESOC Preliminary Mean Sea Surface, 0.3 deg resolution. Reference ellipsoid a=6378.1367 km, f=1/298.257
- solid tide correction, including permanent tides (precise)
- NSWC ocean tide models, including ocean loading (precise)
- ESOC Dynamic Ocean Topography model to degree and order 20: previous month's solution (precise)


The quality of the orbit prediction provided by the operational orbit solution is monitored constantly, by comparing each day's predicted orbit with the final orbit determined afterwards. Statistical information on the 1-day, 3-day and 6-day prediction errors for the period May to October 1995 is presented in Table 2. The relatively high prediction accuracy reflects the low solar activity (near the minimum of the 11-year solar cycle). High levels of solar activity experienced after the ERS-1 launch (and to be expected during the Envisat mission) led to prediction errors of 1 km after a day, and up to 5 km after 3 days.


Table 2. ERS orbit prediction errors


                  ERS-1/ERS-2 (metres)
                 after 1 day         after 3 days        after 6 days
---------------------------------------------------------------------

October'   95     30/20              111/98              326/274
September' 95     13/18               50/53              148/106
August'    95     14/14               58/57              201/200
July'      95     14/12               50/51              173/165
June'      95     17/16               71/72              197/227
May'       95     23/18              105/97              202/199



The quality of the individual tracking data types is summarised in terms of the amount of data available and the root- mean-square (rms) residuals from the precise solution. The 'residuals' are the differences between the actual measurements and those computed with the best models available after the solution has been obtained. Individual statistics are plotted for the laser data (including a detailed breakdown per laser station), for the MPTS S-band tracking data, and for the Altimeter data in Figure 4d. Typical fitting accuracies are 6 cm for laser, 14 cm for altimetry, 60 cm for MPTS ranging and 2 mm/s for MTPS Doppler (30's sampling).

Laser Tracking of ERS
Figure 4a. Laser tracking of ERS during October 1995

ERS Laser Processing
Figure 4b. ERS laser processing since December 1994

Doppler Processing
Figure 4c. Kiruna range and Doppler processing since December 1994

ERS Altimeter Processing
Figure 4d. ERS Altimeter processing since December 1994

Precise orbits for ERS-1 and ERS-2 are generated at the German Processing and Archiving Facility (D-PAF) in Oberpfaffenhofen, in support of refined ERS Altimeter products. A regular comparison is made between the ESOC and D-PAF precise orbits, showing a radial consistency between both solutions at about the 9 cm level during 1995 (Fig. 5). Overlap tests made by differencing short common arcs in consecutive four-day solutions show an internal radial consistency in the ESOC orbits of better than 5 cm. The objective of providing orbits with sub-decimetre radial accuracy for ERS is therefore certainly being achieved.

ERS Orbit Comparison
Figure 5. ERS precise orbit comparison (ESOC minus D-PAF)

ERS-2 Altimeter calibration

Calibration campaigns involving near-overflights of laser stations have been the most commonly used method of calibrating a radar altimeter. The recent improvements achieved in orbit determination, together with the fact that several altimetric missions are currently flying simultaneously, have opened up new calibration possibilities. They are mainly based on relative calibration between two or more satellites. A special working group, including experts from a number of European institutes working on altimetry, has been set up by ESRIN for the purpose of cross-calibrating the ERS-2 Altimeter with respect to that of ERS-1.

As noted earlier, radial orbit errors are no longer a limiting factor in calibrating a radar altimeter using global techniques. More important limitations are the corrections to the altimeter measurements themselves. The latter need to be corrected for various propagation effects in order to determine the height of the satellite above the instantaneous mean sea surface accurately. The instantaneous height of the sea above the reference ellipsoid can be determined by applying accurate corrections to account for the geoid, tides, currents, etc. Once the orbit has been determined, the height of the satellite above a reference ellipsoid can be computed; the difference between that height and the corrected measurement is then the Altimeter bias.

Over the last few years, ERS-1 and Topex/ Poseidon data have been used at ESOC to improve the corrections applied to altimeter measurements, and models for the following corrections have been developed: sea-surface topography (Fig. 6), mean sea surface (Fig. 7 highlights the differences between the new model and the OSU 91A model used previously), wet tropospheric correction, electromagnetic bias, and ocean tides. The combination of very precise orbits and very accurate corrections to the Altimeter measurements has allowed the relative calibration of the ERS-2 Altimeter's bias to be performed with an uncertainty of only about 2 cm.

Sea-level Anomaly
Figure 6. October 1995 sea-level anomaly from ERS altimetry

Mean Sea Surface
Figure 7. Preliminary mean sea surface from ERS-1 geodetic phase

The Fast Delivery (FD) data from the ERS Altimeter have been enhanced at ESOC. All necessary corrections are applied to the data as soon as they become available, typically within 24 hours. The fit of these enhanced FD data with the precise orbits computed at ESOC is within about 13-14 cm (Fig. 4d).

To estimate the relative Altimeter bias between ERS-2 and ERS- 1, four different techniques have been considered:

The use of four different techniques permits quantification of the error involved in the bias estimation. Errors in the corrections applied to the Altimeter data will affect all global methods in almost the same way, but can be expected to influence the local calibrations in a different manner. Therefore, the difference between the value computed from the global analysis and that computed from the local analysis provides an indication of any systematic errors introduced by the corrections.

A high degree of consistency has been found between the relative Altimeter bias values obtained with each of the techniques, as is apparent in Table 3.
Figure 4d shows the Altimeter bias values found using the 'orbit solution' method. They are computed every four days with a typical delay of less than ten days.


Table 3. ERS-2 Altimeter bias calibration, relative to ERS-1


Method                            Relative              Sigma (cm)
                                  bias (cm) 
------------------------------------------------------------------

Global solution                    1.20                ± 1.90
Mediterranean solution             0.30                ± 3.30
Orbit solution                     1.40                ± 2.00
Sea-surface differences            0.90                ± 1.20


A weighted fit between all the methods considered provides the final value of the bias, as well as the associated error which represents the combination of the non-systematic errors. This then has to be combined with the estimate for the systematic error, derived from a set of local calibrations. The relative Altimeter bias has been solved for in 30 ° x 30 ° regions. The systematic error's contribution will be different in each region, and the standard deviation will therefore provide an upper limit for the total remaining or systematic error.

The final value of the relative bias determined by these methods is then 1.02 ± 2.18 cm. This uncertainty value can be considered pessimistic because there is quite a large uncertainty associated with each 30 ° x 30 ° Altimeter bias determination which is not really due to the systematic error contribution.

Conclusion

The stringent requirements set for maintaining the tandem ERS-1 and ERS-2 mission configuration are being met, except for some brief excursions around inclination corrections. Both the 'operational' and 'precise' orbit determinations for the two spacecraft are being performed with a high degree of accuracy and autonomy at ESOC. The operational solutions, based on S-band tracking from the Kiruna station and fast-delivery altimetry, have typical accuracies of 2 m overall and 30-50 cm radially, with a delivery delay of just one day. Precise orbit determination, relying additionally on quick-look laser data, is available after about 10 days (time lapse needed to guarantee the acquisition of sufficient laser data), and provides a radial orbit accuracy of 5-8 cm.

One of the major advantages of the tandem orbit configuration during the Commissioning Phase was the possibility to cross- calibrate the Radar Altimeters on ERS-1 and ERS-2 without the need for a dedicated field campaign of the sort conducted for ERS-1. Several groups have shown that both Altimeters have essentially the same measurement bias, within the approximately 2 cm uncertainty of the estimation.

The latest status of the ERS orbits and their evaluation (Figure 6 and Figure 7) by ESOC is available via our Home Page on the World Wide Web:

http://nng.esoc.esa.de


About | Search | Feedback

Right Left Up Home ESA Bulletin Nr. 85.
Published February 1996.
Developed by ESA-ESRIN ID/D.