Adv. Space Res. Voi. 11, No. 2, pp. (2)79-(2)83. 1991 027J-1177/91 $0.00 + .50 Printed in Great Britain. Ali rights reserved. Copyright © 1991 COSPAR A COMPARISON BETWEEN TWO STRATEGIES OF DATA REDUCTION FOR THE HIPPARCOS PROJECT B. Bucciarelli,*-** M. G. Lattanzi,****** F. Migliaccio,t F. Sansòf and M. Sarasso** *Space Telescope Science Institute (ST Sci), Battimare MD, U.S.A. **Osservatorio Astronomico di Torino, Italy ***ESA, Astrophysics Division, SSD, at ST Sci, U.S.A. tlstituto di Topografia, Fotogrammetria e Geofisica, Politecnico di Milano, Italy ABSTRACT Two different strategies for reducing HIPPARCOS data are directly compared and their respec- tive impact on thè final precision of thè HIPPARCOS Catalog is evaluated. These strategies are thè FAST (Fundamental Astronomy by Space Techniques) Consortium's Baseline and GLOBUS procedures. The Baseline is already operating on satellite data; GLOBUS was invented as an alternative and more rigorous approach to thè adjustment of thè HIPPARCOS astrometric net- work. At present, FAST is supporting a possible introduction of GLOBUS in its data reduction chain. In order to make a true comparison, thè two methods were applied to thè same 1-year simulation of HIPPARCOS mission data. The comparison shows that thè overall precision of thè solution given by thè GLOBUS approach is ~ 20% better than thè Baseline one. INTRODUCTION Schematically, each HIPPARCOS observation is described by an equation relating thè star po- sition measured onto a mono-dimensional grid in thè telescope focal piane (precise to about 10 mas), thè 3 Euler angles specifying thè instantaneous satellite attitude, and thè 5 astrometric parameters (position, parallax, and proper motion) of thè star to be estimated. For sake of sim- plicity, but with no loss of generality, we will neglect in thè following thè complications coming from additional parameters needed to describe thè real instrument. The necessary redundancy is obtained by observing thè same star several (about 1600) times during thè satellite lifetime and by observing different stars with thè "same" attitude. Given some 100,000 program stars and a satellite lifetime of 2.5-3 years, it would build up a set of about 2 • IO8 condition equations and about IO8 unknowns. These numbers and ali thè computations in this paper refer to nominai (pre-launch) mission features. According to recent estimates accessible to thè authors, thè revised mission parameters should not affect most of thè considerations below. In thè following we briefly recali thè main features of thè two approaches adopted by thè FAST Consortium for thè HIPPARCOS data reduction [a complete description can be found in reviews by Kovalevsky /!/, and Bernacca /2/, in thè report edited by Bernacca and Kovalevsky /3/, and thè recent ESA volume edited by Perryman, Lindegren, and Murray /4/]. Then thè results of thè first experiment devoted to compare thè two methods on thè same simulated data are summarized. TWO POSSIBLE REDUCTION SCHEMES The Baseline The first feature of thè Baseline (Lindegren /?/) is a step during which a data compression is performed on a Reference Great Circle (RGC). With reference to Figure 1, thè process of data compression consists essentially in thè projection of thè Instantaneous Great Circle (IGC) star coordinates (77, 6), where only 77 is observed because of thè mono-dimensional grid in thè telescope focal piane, into thè corresponding quantities (ip and £) on thè RGC, TJ) being thè longitude (from thè intersection of thè RGC with thè ecliptic) and £ thè latitude. A RGC can be sjien as thè JASR 11:2-0 (2)79