5.2 Properties of the input data

Author(s): Francesca De Angeli

The PhotPipe system running at DPCI is currently registered as consumer of 248 different data types. Even though not all types are currently in use, this figure gives a fair idea of the complexity of the data handling and of combining all the various inputs to produce the PhotPipe final products.

Here follows a list of the main inputs, without these no processing could be performed:

• •

  Raw observations (Astro and PhotoObservations, AO and PO respectively), containing the decoded telemetry. These are produced on a daily basis in IDT for all FoV transits.

• •

  Object logs, containing a minimal record of each observation. These are needed for the application of the bias calibration which is done in PhotPipe for the BP/RP data.

• •

  CCD PEM NU library, produced off-line in Edinburgh, required for the application of the bias calibration. Full mitigation is applied in PhotPipe.

• •

  IDT/IDU data products, in particular AstroElementaries (AE), containing the results of the Image Parameter Determination, IPD. At each cyclic processing (N), only IDT AEs will be available covering the latest Data Segment (N), while for all previous Data Segments (1…N-1) IDU AEs will be available. Cycle 01 is an exception to this plan, i.e. only IDT AEs were processed in cycle 01 covering both Data Segments 00 and 01. IPD performs a fit to the samples for each CCD transit to produce an estimate of the centroid position (AL only for 1D windows) and of the integrated flux. For this it uses a library of LSF/PSF that is computed off-line (Section 2.3.2). The library used in IDT during Data Segments 00 and 01 did not include dependencies in AC motion or colour.

• •

  Cross-match information, linking each transit to a source either existing in the current source catalogue or to a new source. There are two sources of cross-match data: IDT, which runs a cross-match on a daily basis, and IDU, which runs at the end of each Data Segment covering all date since the start of nominal operations. IDU has the advantage of being able to use multi-epoch information to improve the results and to filter more efficiently spurious observations. So far PhotPipe has always relied on the most recent IDU cross-match results.

• •

  Satellite attitude, required to reconstruct the satellite position and pointing direction at any time. There are three systems producing attitude: IDT does a daily attitude reconstruction (OGA1, Section 2.4.5), ODAS (Section 2.4.5) improves this on a daily basis while AGIS runs in the cyclic processing to provide the best attitude reconstruction (OGA3, Section 3.3.4). PhotPipe requires the accuracy reached by OGA3 for the prediction of centring errors (for the flux loss calibration) and predicted positions (for the geometric calibration of the BP/RP instruments and for the special treatment of observations in crowded areas).

• •

  Source astrophysical coordinates, required to compute predicted positions of the sources onto the focal plane (for measuring centring errors and calibrating the geometry of the BP/RP instruments). ODAS produces these on a daily basis for a fraction of the sources. The initial calibration plan was to use these for the geometry calibration of the BP/RP data, but then it was considered that it was worth adopting an alternative approach to be able to calibrate a larger fraction of the data. The alternative approach relies on the AF centroid contained in the AstroElementaries together with the AF geometric calibration and the attitude to extrapolate the position of sources to the BP/RP CCDs.

• •

  AF geometric calibration, required for the computation of extrapolated positions of sources on the BP/RP CCDs (see the description of the previous item in this list).

The careful reader will notice that the above list does not include a source catalogue. This is because for Cycle 01, PhotPipe did not rely on any bootstrap information for the observed sources. No external catalogue is ever used for the processing of the data. External catalogues are currently only used for validation purposes and are only listed in the validation section 5.5.

For a detailed report on the number of input data for Cycle 01, please refer to the DPCI Processing Configuration section at 1.3.4.

Author(s): Giorgia Busso, Elena Pancino

The external calibration model requires the use of as many calibrators as possible (compatibly with the feasibility of the corresponding on-ground observing campaign), including all spectral types from blue to red (to account for colour dependencies), with smooth spectral energy distribution (SED) but also absorption features, both narrow (atomic lines) and wide (molecular bands). The Gaia end-of-mission requirement for the Spectro-Photometric Standard Star (SPSS) flux precision is $\simeq$1%, and their flux calibration should be tied to Vega (Bohlin and Gilliland 2004; Bohlin 2007, 2014) to within $\simeq$3%. This sets an approximate number of required calibrators of around 200, to ensure an homogeneous sky coverage all year round both in the Northern and Southern hemispheres, and with a suitable magnitude range ($V\sim 9$–$15$ mag) to be observed by both Gaia and several 2–4 m class ground-based telescopes with a good signal-to-noise ratio.

Because no existing set of SPSS in the literature simultaneously meets all these requirements, while at the same time covering the whole Gaia spectral range (330–1050 nm), an initial selection of approximately 300 SPSS candidates was made. These candidates cover all spectral types from hot WD and O/B to cold M stars, i.e., with a temperature range $T_{\mathrm{eff}}$ $\sim 3500$–$\mathrm{80\hspace{0.25em}000}$ K. A substantial observational effort (Pancino et al. 2012; Altavilla et al. 2015) to collect the required data and monitor for constancy started in 2006 and was completed in 2015. The campaign was awarded more than 5000 hours of observing time, mostly in visitor mode, at six different facilities: DOLORES@TNG in La Palma, EFOSC2@NTT and ROSS@REM in La Silla, CAFOS@2.2 m in Calar Alto, BFOSC@Cassini in Loiano, and LaRuca@1.5 m in San Pedro Mártir. Some additional data were also acquired with Meia@TJO in Catalonia. The survey produced more than 100,000 imaging and spectroscopic frames, that are presently being analysed (Altavilla et al. 2015). The raw data, flux tables, and intermediate data products are collected at the ASI Science Data Center in a database that will be opened to the public along with the first official release of SPSS flux tables. The first public release of SPSS flux tables should occur before GDR2.

Two internal releases of SPSS flux tables were prepared so far, a pre-launch version (V0) to test the instrument performance and the pipelines, and a first post-launch version (V1) to actually calibrate the photometry for Gaia DR1. Both V0 and V1 are stored in MDB and contain the best 94 SPSS, i.e. about 50% of the final SPSS sample, observed in strictly photometric conditions and monitored for constancy on timescales of 1–2 hours to exclude stars with magnitude variations larger than $\pm$10 mmag. The quality of the flux tables in V1 already meets the requirements (precision 1% and accuracy 3%). Future releases will increase the number of validated SPSS up to completion of the entire sample, improve the data quality of the flux tables, and complete the data products available for each SPSS (including magnitudes and variability assessment). Fig. 5.5 shows a one-sight view of the current sample where fluxes are normalized at 475 nm, while the colour ranks with the sources spectral type.

For an overview of IDT and IDU data processing see Section 2.4.