In order to reach the level of precision targetted by Gaia, many months of observational data must be incorporated into a global, complex data reduction process in order to perform a self-calibration of the satellite and subsequently to determine the astrometric and global parameters. Subtle malfunctions in the spacecraft or payload may impact on the measurement precision and instrument stability and hence degrade the final results. The `First-Look' activity is designed to allow early identification of these types of effects, so that they can be corrected.
One of the major components of the First Look is the ODAS, the One-Day Astrometric Solution. Its goal is very similar to that of the AGIS (see Picture of the week for 2006-01-23 ) but limited to the observations of one single day. A first development step towards an ODAS was the `ODIS' method, the `One-day Iterative Solution', which like AGIS uses a block-iterative scheme to solve for the astrometric parameters (see Picture of the week 2005-05-09).
When the ODIS was used on noisy simulated observations and with realistic errors in the initial values for the unknown astrometric source, attitude, and calibration parameters, convergence was reached only after more than ten thousand iterations. For the longitude coordinates along the Reference Great Circle of the day (given in radians) the first diagram shows the final deviations of the source positions from their `true' (i.e. those assumed in the simulations) values. Units for the differences are milliarcseconds.
The Ring Solution is a direct, non-iterative ODAS method. In contrast to the ODIS, the final solution is reached in one single step involving complex matrix algebra. The second diagram is the result of the Ring Solution. It is practically indistinguishable from the first one!
This means that two completely different numerical methods for the ODAS reached the same result - a very strong verification of both methods.
Besides being a direct method, the Ring Solution also allows a more in-depth investigation of the error budget. Therefore, only the Ring Solution method will be used for the ODAS in the future.
People involved at ARI/ZAH, Heidelberg: Hans Bernstein, Sonja Hirte, Helmut Lenhardt, Ulrich Bastian, Stefan Jordan
[A larger version of each of the two figures is available: top image; lower image]