- Email: [email protected]
Hipparcos: Revised mission overview
,ldv. Space Res. Vol. 11, No.2. pp. (2)15—(2)23. 1991 Printed in Great Britain. All rights reserved.
0273—1177/91 $0.00 + .50 Copyright © 1991 COSPAR
HIPPARCOS: REVISED MISSION OVERVIEW M. A. C. Perryman Astrophysics Division, European Space Agency, ESTEC, Noordwijk 2200AG, The Netherlands
ABSTRACT The Hipparcos astrometry satellite was launched on 8 August 1989, and after spacecraft and payload commissioning, commenced the routine data acquisition phase on 26 November 1989. Having failed to reach its planned geostationary orbit, major revisions in the mission operations had to be performed, although the impact on the final mission accuracies now appears to be less severe than originally foreseen. The final astrometric accuracies attainable by the Hipparcos mission will be influenced by the spacecraft and payload performances on the one hand, and by the data reductions procedures on the other. This paper addresses the former aspects, and will cover the overall mission performances (expected lifetime, useful fraction of the time for which useful scientific data is acquired, etc) as well as some details of the spacecraft and payload performances. The in-orbit calibration and environmental results (thermal environment, background levels, geometric, photometric and optical performances) and key operational features (real-time attitude determination quality, etc) will be presented. The subsequent stages of the data processing and final elements of the accuracy assessment are presented in accompanying contributions. INTRODUCTION The goals of the ESA Hipparcos space astrometry mission /1-5/ are to acquire the astrometric parameters (positions, parallaxes and proper motions) of about 120 000 pre-defined (‘programme’) stars to an accuracy of about 2 milli-arcsec to a limiting magnitude of about 9-12 mag (for the main experiment), as well as positions and BV photometry, to a lower precision, for the 400 000 or so stars down to a limiting magnitude of about B = 10 11 mag (the Tycho experiment). A description of the Tycho experiment and details of the expected Tycho performances are given elsewhere ‘3-7/. —
The difficulties faced by the Hipparcos mission in the early days after launch were extensively reported /8-13/. The problems were due to the non-operation of the apogee boost motor, intended to place the satellite into its foreseen geostationary position. As a result, the satellite was trapped in an elliptical orbit, with a perigee of about 200 km (subsequently raised to about 500 km) and an apogee of about 36000 km. The orbital period is about 10.6 hours, and twice per day the satellite passes through the Earth’s particle radiation belts. The present orbit, and some of the problems associated with it, is shown in Figure 1. Some of the early concerns about the short expected mission lifetime have proved to be unfounded: measurements of the solar array degradation, which was originally expected to limit the lifetime, and a comparison with theoretical predictions, suggest that a lifetime of 3 years is possible. The months after launch involved considerable modifications to the satellite operations. Compared with the intended orbit (where relatively short earth occultations caused the only significant loss of useful observing time), the present orbit created considerable problems from the point of view of ground station coverage, radiation belt passages, perturbing torques and consequent at-
JASR 11:2—C
(2)15
(2)16
M. A. C. Perryman
EXTENDED ECLIPSES
APOGEE DISTANCE 30000 Km
3 GROUND STATIONS PROViDING 80-90% 10.7 HR ORBIT
COVERAGE .e~
~
f
~ PERIGEE DISTANCE
\
~‘ ATTITUDE DEGRADATION AFTER SIGNAL PE.ACQUIsmON AND PERIGEE TRANSIT
J
500KM
OCCULTATIONS, WITH FREQUENT ATTiTUDE ADJUSTMENTS AT PERIGEE
RADIATION BELTS ENHANCE DETECTOR BACKGROUND, AND DAMAGE SOLAR ARRAYS AND OPTICAL THROUGHPUT
Fig. 1. The present Hipparcos orbit, indicdting the problems associated with it. titude perturbations, and longer earth occultations and eclipses. PROBLEMS POSED BY THE REVISED ORBIT The major problems encountered in the operation of the Hipparcos satellite in its revised elliptical orbit were the following: (1) Final orbital configuration: an early decision was required to utilise the on-board hydrazine propulsion system (originally foreseen for nutation correction and station acquisition) for a possible perigee raising manoeuvre. To minimise perturbing torques at perigee, the available hydrazine was used to raise the perigee to approximately 600 km, and a corresponding orbital period of lOh 40m. The final orbit achievable nevertheless suffers from the disadvantages of poor station coverage from a single ground station as well as the continuous interception of the Van Allen radiation belts each orbit. Following the perigee raise, the solar panels and telescope baffle covers were deployed, and the spacecraft and payload subsequently commissioned. Commissioning proceeded without major difficulties, although on a longer timescale than that originally foreseen, due to the extended periods of satellite non-visibility implied by the single ground station coverage. (2) Solar array performance and satellite lifetime: after the early pessimistic lifetime predictions, the on-board systems were configured to provide in-orbit measurements of the solar array voltage and current characteristics, and the orbital characteristics were used to estimate high-energy proton and electron fluencies responsible for the degradation of the solar arrays. In-orbit incasurements and theoretical prediction are now in good agreement, and an estimated lifetime of approximately 3 years is now considered to be realistic (see Figure 2). The probability of catastrophic component failure through radiation damage is less easy to quantify, and may ultimately limit the achievable lifetime. The other limiting payload consumable
is the cold
gas
required for the attitude control: usage is more inefficient than that foreseen for
HIPPARCOS:
0 1-
Revised Mission Overview
(2)17
____________________________________________________________
a,
•~b~.~•_•
Co
a
~•~~,___
>
. —
—
Theoretical prediction In—orbit data, panels 1-3 In—orbit data, panels 1—2 In—orbit data, panels 1—2. full
-
• o
• —-
•
~
a
spin
~.nd0Iop.r.Lion
__.
0
—
I
50
100
150
200
250
300
350
Days from mission start
Fig. 2. The measured solar array voltage versus time (solid circles) superimposed on the theoretical prediction of the voltage decay (dashed line), based on the known electron and proton environment in this orbit. After the long eclipse period, during which voltage measurements were suspended, a temperature measurement anomaly on one of the solar panels became apparent as an increased scatter of the measured points. Neglecting this panel gives the open circles; extending the measurements over one full satellite spin (solid squares) now indicates a degradation less severe than the previous predictions. A corresponding lifetime of at least three years (the dashed curve) is now indicated.
the nominal mission (due to the larger perturbing torques at perigee), and while present usage should allow operation of the satellite until early 1994, some optimisation of the attitude control system, especially during the perigee transit region, may be possible. (3) Particle background radiation: as well as degrading the performance of the solar arrays,
systematically as a function of time, the —. 1 MeV electron fluence within the Van Allen belts also contributes to the instantaneous detector background, as a result of Cerenkov emission within the payload dioptric elements. At the level of the primary image dissector tube, the relatively small (30 arcsec diameter) instantaneous field of view limits the impact of this effect to a small loss (about 0.4 mag) in the limiting magnitude of the observations during the passage of the satellite through the radiation belts. For most purposes, this is a small effect. However, the larger field of the star mapper detector, used for the real-time attitude determination as well as for the Tycho experiment, significantly reduces the amount of data that can be used by the Tycho experiment, as well as that available for the real-time estimates of the satellite attitude (see below). Figure 3 illustrates the background level observed in the star mapper detectors, and the variation in the background level in the detectors seen during a solar flare. (4) Perigee transits: grouped into the category of problems arising during the periods around perigee transits are the higher radiation background (affecting in particular the real-time attitude determinttion), the extended earth occultation (when one or other of the telescope’s two fields of view are occulted by the earth), the gaps in the ground station visibility, and the higher perturbing torques, resulting in modification to the precise attitude control strategy required to keep the satellite following its pre-defined scanning law.
(2)18
M. A. C. Perryman
~uIC~~ ~IHI MID IWT(51& TIP~
~T
2S1-~5t
..
—
•.0
•.5
—
—
—
7.0 7.5 (.0 (.5 (.0 MID I(IT~Ya. TIP~ t• OPY 273)
—
1.5
0.
Fig. 3 (a, left) The variation in the star mapper detector background level during one 10.7 hour satellite orbit, and (b, right) the changes observed in the background level during a solar flare. (Figures courtsey of P. Davies, ESOC). (5) Ground station coverage: in the nominal mission, ground station coverage (telecommanding and telemetry) was to have been provided by a single ground station (at Odenwald, FRG) resulting in continuous satellite visibility for the ground station, and an overall useful observing time (after accounting for calibration, earth and moon occultations, and all other operational overheads) of between 90-95 per cent. In the revised orbit, the Odenwald station alone would have provided some 30 per cent useful data acquisition. Fortunately, it was possible to incorporate the ESA station at Perth, Australia, and the CNES station at Kourou, French Guiana, within weeks of the apogee motor failure. This brought the satellite observability to about 80 per cent, and the useful data fraction to about 55 per cent. Further assistance by NASA has brought the Goldstone station, in California, into the network. As of mid-May 1990 the Goldstone station has brought the average satellite observability to about 93 per cent, and the useful achievable data fraction to about 60-70 per cent (see Figure 4). Very importantly, it has also reduced the present ground station coverage ‘gaps’ to a maximum of about 1.5 hours. This has improved the preservation of the satellite real-time attitude estimation. (6) Eclipse passages: unlike the geostationary orbit configuration, where maximum eclipse durations of about 5 per cent of the orbital period would have been experienced, the present orbit results in eclipse durations of up to 15 per cent of the orbital period. This placed very severe constraints on the battery charging cycles, designed for the geostationary case. Fortunately, the longest eclipse duration expected over the first 3 years of the satellite lifetime, 105 mm, was passed in mid-March 1990 with a margin of about 5 mm on the battery power. This meant that no power saving procedures were necessary, and the scientific observations were able to proceed without interruption. (7) Real-time attitude determination: the high detector background, the high perturbing torques at perigee and the long signal outages due to a combination of earth occultations and ground station ‘gaps’, have resulted in considerable difficulties in running and optimising the continuously evolving real-time attitude determination tasks. This now runs more-or-less routinely, with limited ground intervention. Investigations are, however, still ongoing to improve the situation.
l-IIPPARCOS: Revised Mission Overview
(2)19
0
o 100
— -
Maximum possible data recovery
—
Useful data (attitude converged)
120
140
160
180
200
220
240
260
280
300
320
340
360
380
400
Days from mission start
Fig. 4. The useful achievable data fraction as a function of time. The figures correspond to a 4-day period, over which the ground-station coverage configuration repeates itself. The upper line indicates the maximum possible data recovery (when the satellite is observable by one or other of the ground stations), while the lower curve indicates the fraction of useful data actually acquired (accounting for occultations, ground station outages, attitude nonconvergence, etc).
THE PAYLOAD On-board the satellite, the payload and spacecraft sub-systems have worked extremely well: the
detectors and the contributory error sources (instrumental response, chromaticty, focussing precision, thermal and geometric stability, attitude control, etc) have been typically well within their specified performance values (see Figure 5). After a lengthier than nominal calibration and commissioning period, necessitated by the above difficulties, the nominal mission commenced on 26 November 1989. Between then and the end of May 1990, some 300 Gbits of data covering some 2 million stellar observations (including numerous minor planets, Uranus, Neptune, Titan and Europa) have been acquired (see Figure 6). The excellent payload performance and background level at apogee has recently led to the inclusion of 3C 273, the only non-local group object possibly observable by the satellite, into the observing programme. The satellite and payload performances closely reflect the nominal, pre-launch predictions, which have been comprehensively reported /1/. In particular, the throughput of the entire optical/detection was within 10 per cent (actually better than) the budget prediction, and the instrumental chromaticity (the geometrical shift between stellar images of extreme star colours) is below 1 milli-arcsec. Signal modulation, geometrical distortion, and payload stability, were all within their respective specifications. A drop of between 5-10 per cent in the total optical throughput, depending on star colour, has been observed since launch, due to darkening of the optics in the relay lens assemblies as a result of the radiation environment.
THE OBSERVING PROGRAMME AND DATA ANALYSIS Fortunately, it has been possible to retain the planned observing programme essentially without modification: the impact of the present operational conditions is basically to reduce the number of
observations on a given star by some 30-40 per cent (slightly more for the Tycho experiment
(2)20
S a
—
—
M. A. C. Perryman
o • .._~ ~
—
—
1.5
~ 1.51 1.51
1.5 I FF0 L5~ FF0 CIX. CD..
o
•
—
PF0~CQ.l FF0 CD.1 .~ ~ FT0CD..I
—
—
—
—
—
—
~..._...._....__.._...
F
~-n 1115
y~ ES? I ITTED
-
~
—
—
—
~
.~,——
~
~
~_~_
—
——
7~II~T
~_/~_ri~ij~i •
-4.0
a0
-~O ~
~
~
.~O5.0
~
D.~5I0.0
~
~
rixus scru~a~
,~
~
~
091
0000
FROSt 1989.000)
Fig. 5. Focus of the payload: (a, left) shows the variation in modulation of the signal from the two fields of view that occurs as the focus of the payload varies (one grid step is 1.38 microns); (b, right) shows the evolution in the focus versus time due to moisture release of the payload structure. Regular focus steps are made to ensure that the payload is always within ±1 step of the optimum focus of the combined fields of view—currently adjustments are made, close to perigee, every 14 days approximately.
O
0
I
0.~
INCA 28734
0.00
in I-. 0 z. 0
0.0
Europa (V=5.9)
Jupiter I
0. 32~998~.~898~.
~ C1
~
L~~~de I
3289~I.3~g?5I.3~97.32%7~.328 ~
MID JNTERVRL TIME ON DRY
Fig. 6. Star mapper observations of Jupiter and its moons.
(V=5.7)
HIPPARCOS: Revised Mission Overview
(2)21
TABLE 1. Predicted standard erw~z(in milli-arcsec) of the five astrometric parameters as function of stellar magnitude. The values are sky averages for stars in different magnitude intervals. The figures are derived from the accuracy analysis of Lindegren /21/, modified as described in the text.
o~cosô o~ cos6
< 6.5
6.5—7.5
7.5—8.5
8.5—9.5
9.5—10.5
10.5—11.5
> 11.5
1.02 0.85 1.20 0.99 1.24
1.04 0.88 1.25 1.02 1.28
1.11 0.94 1.32 1.08 1.36
1.22 1.01 1.44 1.17 1.49
1.38 1.15 1.64 1.33 1.71
1.80 1.48 2.15 1.70 2.22
2.32 1.92 2.76 2.23 2.88
TABLE 2. Predicted variation of astrometric standard errors with ecliptic latitude. The table gives the factor to be applied on the sky averages in Table 1. The same factor can be used for the proper motions (~,and ~) and positional components (ci and 5~ The figures are taken directly from the accuracy analysis made by Lindegren /21/.
L81 ci
6 ir
0
—
12°
1.31 1.10 1.19
12
—
240
1.25 1.08 1.16
24
—
37°
1.08 1.04 1.09
37
—
44°
0.85 0.96 0.98
44
—
53°
0.66 0.89 0.86
53
—
64°
0.73 0.89 0.76
> 64° 0.77 0.91 0.69
due to additional perturbing effects of the radiation background). The effect of this on the final expected astrometric accuracies is summarised below. The Input Catalogue /14-15/, the basis of the star observing programme and satellite attitude control, has performed entirely nominally in all respects. Due to the global nature of the Hipparcos observations and reductions it has been neither desirable nor fortunately, necessary, to modify the observing programme. It is a great tribute to the work of the Input Catalogue Consortium to note that the observations have proceeded entirely nominally; not only for the 120 000 or so programme stars, but also for the 50 or so minor planets, planetary satellites (Titan and Europa) and the large amplitude variable stars. The broadband (H), Tycho bands (B and V) photometric measurements, and the astrometric residuals at the preliminary great-circle reduction level are entirely consistent with the ground-based compilation of their a priori values. The present orbit resulted in certain difficulties in the data interfaces between ESOC and the data reduction teams. These difficulties have now been largely overcome, and the satellite data is now passing routinely through the critical elements of the data analysis chain, up to the greatcircle reduction level. Data from 12 hours of observation (lGbit of data covering some 2000 star observations) are now processed routinely by the FAST and NDAC Data Reduction Teams, to derive star abscissae on a great circle with a precision of 5-10 milli-arcsec (further details of the data reduction procedures are given elsewhere /16-19/). Instrument parameters are calibrated with sub-milli-arcsec precision, and are very stable /20/. Comparison procedures, to verify the results obtained by the two processing teams, are progressing well. Instrument parameters are calibrated (as part of the ESOC real-time payload monitoring, as part of the FAST, NDAC and TDAC’s routine data analysis, and within the FAST Consortium’s ‘first look’ analysis facility) with sub-milli-arcsec precision, and are very stable. All reduction elements are in the process of becoming fully operational.
(2)22
M. A. C. Perryman
Sint~lationsby the data reduction teams have shown that a mission duration of at least 18 months is necessary before the various astrometric parameters become separable. Should the satellite lifetime be less than 18 months, a significant improvement to the present astrometric reference frame will be made (individual positions will estimable to about 10 milli-arcsec) but with very limited astrophysical (parallax and proper motion) information. A progressively longer mission duration will lead to improvements in the precision of all five astrometric parameters until, for a lifetime of about 3 years, and assuming the present performance levels, the astrophysicallysignificant (and nominal target) accuracies of around 2 milli-arcsec should be achievable. Tables 1 and 2 show the accuracy figures, derived in /21/ for the nominal 2.5-year mission, modified in the following way: (a) allowing for an additional loss of 10 per cent observing time, in addition to the 5 per cent earth and moon occultations and 20 per cent dead-time accounted for by Lindegren, assumed to be uniformally distributed over the celestial sphere; (b) allowing for an increase of mission elapsed time from 2.5 years to 3 years. The combination of the two effects leads to a marginal nominal improvement in the positions and parallaxes, and to a 20 per cent improvement in the proper motions due to the extension of the measurement interval. The dependency on ecliptic latitude is unchanged. An alternative approach to the accuracy assessment is made by Lindegren /16/, on the basis of the present data reduction results. Long-term accuracy predictions are now, therefore, essentially dependent on the mission lifetime. If the mission survives for about 3 years at current performance levels, the original astrometric (and hence scientific) goals (2 milli-arcsec for positions, parallaxes and annual proper motions) should still be achievable. ACKNOWL EDGEMENTS The implementation of the revised Hipparcos mission was made possible by the skill and dedication of the many members of the operation team at ESOC, under the leadership of the Ground Segment Manager, Jozef van der Ha, and the Spacecraft Operations Managers, Howard Nyc and Dietmar Heger, as well as the project teams of ESA, led by Hamid Hassan, and the industrial prime contractor, Matra, led by Michel Bouffard. The Agency is also indebted to the continued guidance of the Hipparcos Science Team, as representatives of the scientific consortia involved in the Hipparcos project, who have worked since launch under difficult circumstances. The author acknowledges particularly the Input Catalogue Consortium and its leader, Dr. C. Turon, for providing a trouble-free operational and observing catalogue (the Input Catalogue), the FAST, NDAC and TDAC data reduction consortia, and their leaders Professor J. Kovalevsky (CERGA), Dr L. Lindegren (Lund), and Professor E. Høg (Copenhagen), respectively, and the FAST Consortium who, through the work of Dr. 3. Schrijver (Utrecht), established a first-look data analysis facility which has greatly expedited the early understanding of the flight data. REFERENCES 1. M.A.C. Perryman et al., ESA SP-1l11 ‘The Hipparcos Mission, Pre-Launch Status’ (1989). 2. J. Kovalevsky, Celestial Mechanics, 22, 153 (1980). 3. J. Kovalevsky, Space Science Reviews, 39, 1 (1984). 4. 3. Kovalevsky, Manuscripta Geodaetica, 11, 89 (1986). 5. P.L. Bernacca, Astrophys. Space Sci., 110, 21(1985). 6. E. Høg, this issue. 7. E. Høg, IAU Symp. 109 ‘Astrometric Techniques’, eds H.K. Eichhorn ~ R.J. Leacock (1985). 8. R.B. Neto, Nature, 340, 491 (1989). 9. S. Dickman, Nature, 340, 581 (1989). 10. S. Dickman, Nature, 341, 3 (1989). 11. S. Dickonan, Nature, 341, 269 (1989). 12. P. Aldhous, Nature, 344, 280 (1990). 13. P. Coles, Nature, 346, 96 (1990). 14. C. Turon & M.A.C. Perryman (eds), ‘Scientific Aspects of the Input Catalogue Preparation’,
HIPPARCOS: Revised Mission Overview
(2)23
ESA SP-234 (1985). 15. J. Torra & C. Turon (eds), ‘Scientific Aspects of the Input Catalogue Preparation II’, CIRIT
(1988). 16. L. Lindegren, this issue.
17. P.L. Bernacca (ed.), ‘Processing of Scientific Data from Hipparcos’, The FAST Thinkshop (1983). 18. 3. Kovalevsky (ed.), ‘Processing of Scientific Data from Hipparcos’, The Second FAST Thinkshop (1985). 19. P.L. Bernacca & 3. Kovalevsky (eds), ‘Processing of Scientific Data from Hipparcos’, The
Third FAST Thinkshop (1986). 20. H. Schrijver, this issue. 21. L. Lindegren, ESA SP-1111 ‘The Hipparcos Mission, Pre-Launch Status’, Volume III, Chapter 18.
0273—1177/91 $0.00 + .50 Copyright © 1991 COSPAR
HIPPARCOS: REVISED MISSION OVERVIEW M. A. C. Perryman Astrophysics Division, European Space Agency, ESTEC, Noordwijk 2200AG, The Netherlands
ABSTRACT The Hipparcos astrometry satellite was launched on 8 August 1989, and after spacecraft and payload commissioning, commenced the routine data acquisition phase on 26 November 1989. Having failed to reach its planned geostationary orbit, major revisions in the mission operations had to be performed, although the impact on the final mission accuracies now appears to be less severe than originally foreseen. The final astrometric accuracies attainable by the Hipparcos mission will be influenced by the spacecraft and payload performances on the one hand, and by the data reductions procedures on the other. This paper addresses the former aspects, and will cover the overall mission performances (expected lifetime, useful fraction of the time for which useful scientific data is acquired, etc) as well as some details of the spacecraft and payload performances. The in-orbit calibration and environmental results (thermal environment, background levels, geometric, photometric and optical performances) and key operational features (real-time attitude determination quality, etc) will be presented. The subsequent stages of the data processing and final elements of the accuracy assessment are presented in accompanying contributions. INTRODUCTION The goals of the ESA Hipparcos space astrometry mission /1-5/ are to acquire the astrometric parameters (positions, parallaxes and proper motions) of about 120 000 pre-defined (‘programme’) stars to an accuracy of about 2 milli-arcsec to a limiting magnitude of about 9-12 mag (for the main experiment), as well as positions and BV photometry, to a lower precision, for the 400 000 or so stars down to a limiting magnitude of about B = 10 11 mag (the Tycho experiment). A description of the Tycho experiment and details of the expected Tycho performances are given elsewhere ‘3-7/. —
The difficulties faced by the Hipparcos mission in the early days after launch were extensively reported /8-13/. The problems were due to the non-operation of the apogee boost motor, intended to place the satellite into its foreseen geostationary position. As a result, the satellite was trapped in an elliptical orbit, with a perigee of about 200 km (subsequently raised to about 500 km) and an apogee of about 36000 km. The orbital period is about 10.6 hours, and twice per day the satellite passes through the Earth’s particle radiation belts. The present orbit, and some of the problems associated with it, is shown in Figure 1. Some of the early concerns about the short expected mission lifetime have proved to be unfounded: measurements of the solar array degradation, which was originally expected to limit the lifetime, and a comparison with theoretical predictions, suggest that a lifetime of 3 years is possible. The months after launch involved considerable modifications to the satellite operations. Compared with the intended orbit (where relatively short earth occultations caused the only significant loss of useful observing time), the present orbit created considerable problems from the point of view of ground station coverage, radiation belt passages, perturbing torques and consequent at-
JASR 11:2—C
(2)15
(2)16
M. A. C. Perryman
EXTENDED ECLIPSES
APOGEE DISTANCE 30000 Km
3 GROUND STATIONS PROViDING 80-90% 10.7 HR ORBIT
COVERAGE .e~
~
f
~ PERIGEE DISTANCE
\
~‘ ATTITUDE DEGRADATION AFTER SIGNAL PE.ACQUIsmON AND PERIGEE TRANSIT
J
500KM
OCCULTATIONS, WITH FREQUENT ATTiTUDE ADJUSTMENTS AT PERIGEE
RADIATION BELTS ENHANCE DETECTOR BACKGROUND, AND DAMAGE SOLAR ARRAYS AND OPTICAL THROUGHPUT
Fig. 1. The present Hipparcos orbit, indicdting the problems associated with it. titude perturbations, and longer earth occultations and eclipses. PROBLEMS POSED BY THE REVISED ORBIT The major problems encountered in the operation of the Hipparcos satellite in its revised elliptical orbit were the following: (1) Final orbital configuration: an early decision was required to utilise the on-board hydrazine propulsion system (originally foreseen for nutation correction and station acquisition) for a possible perigee raising manoeuvre. To minimise perturbing torques at perigee, the available hydrazine was used to raise the perigee to approximately 600 km, and a corresponding orbital period of lOh 40m. The final orbit achievable nevertheless suffers from the disadvantages of poor station coverage from a single ground station as well as the continuous interception of the Van Allen radiation belts each orbit. Following the perigee raise, the solar panels and telescope baffle covers were deployed, and the spacecraft and payload subsequently commissioned. Commissioning proceeded without major difficulties, although on a longer timescale than that originally foreseen, due to the extended periods of satellite non-visibility implied by the single ground station coverage. (2) Solar array performance and satellite lifetime: after the early pessimistic lifetime predictions, the on-board systems were configured to provide in-orbit measurements of the solar array voltage and current characteristics, and the orbital characteristics were used to estimate high-energy proton and electron fluencies responsible for the degradation of the solar arrays. In-orbit incasurements and theoretical prediction are now in good agreement, and an estimated lifetime of approximately 3 years is now considered to be realistic (see Figure 2). The probability of catastrophic component failure through radiation damage is less easy to quantify, and may ultimately limit the achievable lifetime. The other limiting payload consumable
is the cold
gas
required for the attitude control: usage is more inefficient than that foreseen for
HIPPARCOS:
0 1-
Revised Mission Overview
(2)17
____________________________________________________________
a,
•~b~.~•_•
Co
a
~•~~,___
>
. —
—
Theoretical prediction In—orbit data, panels 1-3 In—orbit data, panels 1—2 In—orbit data, panels 1—2. full
-
• o
• —-
•
~
a
spin
~.nd0Iop.r.Lion
__.
0
—
I
50
100
150
200
250
300
350
Days from mission start
Fig. 2. The measured solar array voltage versus time (solid circles) superimposed on the theoretical prediction of the voltage decay (dashed line), based on the known electron and proton environment in this orbit. After the long eclipse period, during which voltage measurements were suspended, a temperature measurement anomaly on one of the solar panels became apparent as an increased scatter of the measured points. Neglecting this panel gives the open circles; extending the measurements over one full satellite spin (solid squares) now indicates a degradation less severe than the previous predictions. A corresponding lifetime of at least three years (the dashed curve) is now indicated.
the nominal mission (due to the larger perturbing torques at perigee), and while present usage should allow operation of the satellite until early 1994, some optimisation of the attitude control system, especially during the perigee transit region, may be possible. (3) Particle background radiation: as well as degrading the performance of the solar arrays,
systematically as a function of time, the —. 1 MeV electron fluence within the Van Allen belts also contributes to the instantaneous detector background, as a result of Cerenkov emission within the payload dioptric elements. At the level of the primary image dissector tube, the relatively small (30 arcsec diameter) instantaneous field of view limits the impact of this effect to a small loss (about 0.4 mag) in the limiting magnitude of the observations during the passage of the satellite through the radiation belts. For most purposes, this is a small effect. However, the larger field of the star mapper detector, used for the real-time attitude determination as well as for the Tycho experiment, significantly reduces the amount of data that can be used by the Tycho experiment, as well as that available for the real-time estimates of the satellite attitude (see below). Figure 3 illustrates the background level observed in the star mapper detectors, and the variation in the background level in the detectors seen during a solar flare. (4) Perigee transits: grouped into the category of problems arising during the periods around perigee transits are the higher radiation background (affecting in particular the real-time attitude determinttion), the extended earth occultation (when one or other of the telescope’s two fields of view are occulted by the earth), the gaps in the ground station visibility, and the higher perturbing torques, resulting in modification to the precise attitude control strategy required to keep the satellite following its pre-defined scanning law.
(2)18
M. A. C. Perryman
~uIC~~ ~IHI MID IWT(51& TIP~
~T
2S1-~5t
..
—
•.0
•.5
—
—
—
7.0 7.5 (.0 (.5 (.0 MID I(IT~Ya. TIP~ t• OPY 273)
—
1.5
0.
Fig. 3 (a, left) The variation in the star mapper detector background level during one 10.7 hour satellite orbit, and (b, right) the changes observed in the background level during a solar flare. (Figures courtsey of P. Davies, ESOC). (5) Ground station coverage: in the nominal mission, ground station coverage (telecommanding and telemetry) was to have been provided by a single ground station (at Odenwald, FRG) resulting in continuous satellite visibility for the ground station, and an overall useful observing time (after accounting for calibration, earth and moon occultations, and all other operational overheads) of between 90-95 per cent. In the revised orbit, the Odenwald station alone would have provided some 30 per cent useful data acquisition. Fortunately, it was possible to incorporate the ESA station at Perth, Australia, and the CNES station at Kourou, French Guiana, within weeks of the apogee motor failure. This brought the satellite observability to about 80 per cent, and the useful data fraction to about 55 per cent. Further assistance by NASA has brought the Goldstone station, in California, into the network. As of mid-May 1990 the Goldstone station has brought the average satellite observability to about 93 per cent, and the useful achievable data fraction to about 60-70 per cent (see Figure 4). Very importantly, it has also reduced the present ground station coverage ‘gaps’ to a maximum of about 1.5 hours. This has improved the preservation of the satellite real-time attitude estimation. (6) Eclipse passages: unlike the geostationary orbit configuration, where maximum eclipse durations of about 5 per cent of the orbital period would have been experienced, the present orbit results in eclipse durations of up to 15 per cent of the orbital period. This placed very severe constraints on the battery charging cycles, designed for the geostationary case. Fortunately, the longest eclipse duration expected over the first 3 years of the satellite lifetime, 105 mm, was passed in mid-March 1990 with a margin of about 5 mm on the battery power. This meant that no power saving procedures were necessary, and the scientific observations were able to proceed without interruption. (7) Real-time attitude determination: the high detector background, the high perturbing torques at perigee and the long signal outages due to a combination of earth occultations and ground station ‘gaps’, have resulted in considerable difficulties in running and optimising the continuously evolving real-time attitude determination tasks. This now runs more-or-less routinely, with limited ground intervention. Investigations are, however, still ongoing to improve the situation.
l-IIPPARCOS: Revised Mission Overview
(2)19
0
o 100
— -
Maximum possible data recovery
—
Useful data (attitude converged)
120
140
160
180
200
220
240
260
280
300
320
340
360
380
400
Days from mission start
Fig. 4. The useful achievable data fraction as a function of time. The figures correspond to a 4-day period, over which the ground-station coverage configuration repeates itself. The upper line indicates the maximum possible data recovery (when the satellite is observable by one or other of the ground stations), while the lower curve indicates the fraction of useful data actually acquired (accounting for occultations, ground station outages, attitude nonconvergence, etc).
THE PAYLOAD On-board the satellite, the payload and spacecraft sub-systems have worked extremely well: the
detectors and the contributory error sources (instrumental response, chromaticty, focussing precision, thermal and geometric stability, attitude control, etc) have been typically well within their specified performance values (see Figure 5). After a lengthier than nominal calibration and commissioning period, necessitated by the above difficulties, the nominal mission commenced on 26 November 1989. Between then and the end of May 1990, some 300 Gbits of data covering some 2 million stellar observations (including numerous minor planets, Uranus, Neptune, Titan and Europa) have been acquired (see Figure 6). The excellent payload performance and background level at apogee has recently led to the inclusion of 3C 273, the only non-local group object possibly observable by the satellite, into the observing programme. The satellite and payload performances closely reflect the nominal, pre-launch predictions, which have been comprehensively reported /1/. In particular, the throughput of the entire optical/detection was within 10 per cent (actually better than) the budget prediction, and the instrumental chromaticity (the geometrical shift between stellar images of extreme star colours) is below 1 milli-arcsec. Signal modulation, geometrical distortion, and payload stability, were all within their respective specifications. A drop of between 5-10 per cent in the total optical throughput, depending on star colour, has been observed since launch, due to darkening of the optics in the relay lens assemblies as a result of the radiation environment.
THE OBSERVING PROGRAMME AND DATA ANALYSIS Fortunately, it has been possible to retain the planned observing programme essentially without modification: the impact of the present operational conditions is basically to reduce the number of
observations on a given star by some 30-40 per cent (slightly more for the Tycho experiment
(2)20
S a
—
—
M. A. C. Perryman
o • .._~ ~
—
—
1.5
~ 1.51 1.51
1.5 I FF0 L5~ FF0 CIX. CD..
o
•
—
PF0~CQ.l FF0 CD.1 .~ ~ FT0CD..I
—
—
—
—
—
—
~..._...._....__.._...
F
~-n 1115
y~ ES? I ITTED
-
~
—
—
—
~
.~,——
~
~
~_~_
—
——
7~II~T
~_/~_ri~ij~i •
-4.0
a0
-~O ~
~
~
.~O5.0
~
D.~5I0.0
~
~
rixus scru~a~
,~
~
~
091
0000
FROSt 1989.000)
Fig. 5. Focus of the payload: (a, left) shows the variation in modulation of the signal from the two fields of view that occurs as the focus of the payload varies (one grid step is 1.38 microns); (b, right) shows the evolution in the focus versus time due to moisture release of the payload structure. Regular focus steps are made to ensure that the payload is always within ±1 step of the optimum focus of the combined fields of view—currently adjustments are made, close to perigee, every 14 days approximately.
O
0
I
0.~
INCA 28734
0.00
in I-. 0 z. 0
0.0
Europa (V=5.9)
Jupiter I
0. 32~998~.~898~.
~ C1
~
L~~~de I
3289~I.3~g?5I.3~97.32%7~.328 ~
MID JNTERVRL TIME ON DRY
Fig. 6. Star mapper observations of Jupiter and its moons.
(V=5.7)
HIPPARCOS: Revised Mission Overview
(2)21
TABLE 1. Predicted standard erw~z(in milli-arcsec) of the five astrometric parameters as function of stellar magnitude. The values are sky averages for stars in different magnitude intervals. The figures are derived from the accuracy analysis of Lindegren /21/, modified as described in the text.
o~cosô o~ cos6
< 6.5
6.5—7.5
7.5—8.5
8.5—9.5
9.5—10.5
10.5—11.5
> 11.5
1.02 0.85 1.20 0.99 1.24
1.04 0.88 1.25 1.02 1.28
1.11 0.94 1.32 1.08 1.36
1.22 1.01 1.44 1.17 1.49
1.38 1.15 1.64 1.33 1.71
1.80 1.48 2.15 1.70 2.22
2.32 1.92 2.76 2.23 2.88
TABLE 2. Predicted variation of astrometric standard errors with ecliptic latitude. The table gives the factor to be applied on the sky averages in Table 1. The same factor can be used for the proper motions (~,and ~) and positional components (ci and 5~ The figures are taken directly from the accuracy analysis made by Lindegren /21/.
L81 ci
6 ir
0
—
12°
1.31 1.10 1.19
12
—
240
1.25 1.08 1.16
24
—
37°
1.08 1.04 1.09
37
—
44°
0.85 0.96 0.98
44
—
53°
0.66 0.89 0.86
53
—
64°
0.73 0.89 0.76
> 64° 0.77 0.91 0.69
due to additional perturbing effects of the radiation background). The effect of this on the final expected astrometric accuracies is summarised below. The Input Catalogue /14-15/, the basis of the star observing programme and satellite attitude control, has performed entirely nominally in all respects. Due to the global nature of the Hipparcos observations and reductions it has been neither desirable nor fortunately, necessary, to modify the observing programme. It is a great tribute to the work of the Input Catalogue Consortium to note that the observations have proceeded entirely nominally; not only for the 120 000 or so programme stars, but also for the 50 or so minor planets, planetary satellites (Titan and Europa) and the large amplitude variable stars. The broadband (H), Tycho bands (B and V) photometric measurements, and the astrometric residuals at the preliminary great-circle reduction level are entirely consistent with the ground-based compilation of their a priori values. The present orbit resulted in certain difficulties in the data interfaces between ESOC and the data reduction teams. These difficulties have now been largely overcome, and the satellite data is now passing routinely through the critical elements of the data analysis chain, up to the greatcircle reduction level. Data from 12 hours of observation (lGbit of data covering some 2000 star observations) are now processed routinely by the FAST and NDAC Data Reduction Teams, to derive star abscissae on a great circle with a precision of 5-10 milli-arcsec (further details of the data reduction procedures are given elsewhere /16-19/). Instrument parameters are calibrated with sub-milli-arcsec precision, and are very stable /20/. Comparison procedures, to verify the results obtained by the two processing teams, are progressing well. Instrument parameters are calibrated (as part of the ESOC real-time payload monitoring, as part of the FAST, NDAC and TDAC’s routine data analysis, and within the FAST Consortium’s ‘first look’ analysis facility) with sub-milli-arcsec precision, and are very stable. All reduction elements are in the process of becoming fully operational.
(2)22
M. A. C. Perryman
Sint~lationsby the data reduction teams have shown that a mission duration of at least 18 months is necessary before the various astrometric parameters become separable. Should the satellite lifetime be less than 18 months, a significant improvement to the present astrometric reference frame will be made (individual positions will estimable to about 10 milli-arcsec) but with very limited astrophysical (parallax and proper motion) information. A progressively longer mission duration will lead to improvements in the precision of all five astrometric parameters until, for a lifetime of about 3 years, and assuming the present performance levels, the astrophysicallysignificant (and nominal target) accuracies of around 2 milli-arcsec should be achievable. Tables 1 and 2 show the accuracy figures, derived in /21/ for the nominal 2.5-year mission, modified in the following way: (a) allowing for an additional loss of 10 per cent observing time, in addition to the 5 per cent earth and moon occultations and 20 per cent dead-time accounted for by Lindegren, assumed to be uniformally distributed over the celestial sphere; (b) allowing for an increase of mission elapsed time from 2.5 years to 3 years. The combination of the two effects leads to a marginal nominal improvement in the positions and parallaxes, and to a 20 per cent improvement in the proper motions due to the extension of the measurement interval. The dependency on ecliptic latitude is unchanged. An alternative approach to the accuracy assessment is made by Lindegren /16/, on the basis of the present data reduction results. Long-term accuracy predictions are now, therefore, essentially dependent on the mission lifetime. If the mission survives for about 3 years at current performance levels, the original astrometric (and hence scientific) goals (2 milli-arcsec for positions, parallaxes and annual proper motions) should still be achievable. ACKNOWL EDGEMENTS The implementation of the revised Hipparcos mission was made possible by the skill and dedication of the many members of the operation team at ESOC, under the leadership of the Ground Segment Manager, Jozef van der Ha, and the Spacecraft Operations Managers, Howard Nyc and Dietmar Heger, as well as the project teams of ESA, led by Hamid Hassan, and the industrial prime contractor, Matra, led by Michel Bouffard. The Agency is also indebted to the continued guidance of the Hipparcos Science Team, as representatives of the scientific consortia involved in the Hipparcos project, who have worked since launch under difficult circumstances. The author acknowledges particularly the Input Catalogue Consortium and its leader, Dr. C. Turon, for providing a trouble-free operational and observing catalogue (the Input Catalogue), the FAST, NDAC and TDAC data reduction consortia, and their leaders Professor J. Kovalevsky (CERGA), Dr L. Lindegren (Lund), and Professor E. Høg (Copenhagen), respectively, and the FAST Consortium who, through the work of Dr. 3. Schrijver (Utrecht), established a first-look data analysis facility which has greatly expedited the early understanding of the flight data. REFERENCES 1. M.A.C. Perryman et al., ESA SP-1l11 ‘The Hipparcos Mission, Pre-Launch Status’ (1989). 2. J. Kovalevsky, Celestial Mechanics, 22, 153 (1980). 3. J. Kovalevsky, Space Science Reviews, 39, 1 (1984). 4. 3. Kovalevsky, Manuscripta Geodaetica, 11, 89 (1986). 5. P.L. Bernacca, Astrophys. Space Sci., 110, 21(1985). 6. E. Høg, this issue. 7. E. Høg, IAU Symp. 109 ‘Astrometric Techniques’, eds H.K. Eichhorn ~ R.J. Leacock (1985). 8. R.B. Neto, Nature, 340, 491 (1989). 9. S. Dickman, Nature, 340, 581 (1989). 10. S. Dickman, Nature, 341, 3 (1989). 11. S. Dickonan, Nature, 341, 269 (1989). 12. P. Aldhous, Nature, 344, 280 (1990). 13. P. Coles, Nature, 346, 96 (1990). 14. C. Turon & M.A.C. Perryman (eds), ‘Scientific Aspects of the Input Catalogue Preparation’,
HIPPARCOS: Revised Mission Overview
(2)23
ESA SP-234 (1985). 15. J. Torra & C. Turon (eds), ‘Scientific Aspects of the Input Catalogue Preparation II’, CIRIT
(1988). 16. L. Lindegren, this issue.
17. P.L. Bernacca (ed.), ‘Processing of Scientific Data from Hipparcos’, The FAST Thinkshop (1983). 18. 3. Kovalevsky (ed.), ‘Processing of Scientific Data from Hipparcos’, The Second FAST Thinkshop (1985). 19. P.L. Bernacca & 3. Kovalevsky (eds), ‘Processing of Scientific Data from Hipparcos’, The
Third FAST Thinkshop (1986). 20. H. Schrijver, this issue. 21. L. Lindegren, ESA SP-1111 ‘The Hipparcos Mission, Pre-Launch Status’, Volume III, Chapter 18.