Modelling the Gaia instrument Daniele Gardiol D.Bonino, D.Busonero, L.Corcione, M.Gai, M.Lattanzi, D.Loreggia, A.Riva, F.Russo, J.C.Terrazas Vargas INAF - Osservatorio Astronomico di Torino Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 1 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 2 Gaia expected performances Hipparcos Gaia Magnitude limit Completeness Bright limit Number of objects 12 7.3 – 9.0 0 120 000 Effective distance limit Quasars Galaxies Accuracy 1 kpc None None 1 milliarcsec Photometry photometry Radial velocity Observing programme 2-colour (B and V) None Pre-selected 20 mag 20 mag 6 mag 26 million to V = 15 250 million to V = 18 1000 million to V = 20 50 kpc 5 x 105 106 – 107 7 µarcsec at V = 10 10-25 µarcsec at V = 15 300 µarcsec at V = 20 Low-res. spectra to V = 20 15 km/s to V = 16-17 Complete and unbiased Gaia: complete, faint, accurate (from www.rssd.esa.int) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 3 Gaia expected performances Hipparcos Gaia Magnitude limit Completeness Bright limit Number of objects 12 7.3 – 9.0 0 120 000 Effective distance limit Quasars Galaxies Accuracy 1 kpc None None 1 milliarcsec Photometry photometry Radial velocity Observing programme 2-colour (B and V) None Pre-selected 20 mag 20 mag 6 mag 26 million to V = 15 250 million to V = 18 1000 million to V = 20 50 kpc 5 x 105 106 – 107 7 µarcsec at V = 10 10-25 µarcsec at V = 15 300 µarcsec at V = 20 Low-res. spectra to V = 20 15 km/s to V = 16-17 Complete and unbiased → Stringent requirements also in Instrument Modelling performances Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 4 Instrument modelling approach Design parameters In-flight instrument behaviour forward Analysis backward Daniele Gardiol – Modelling the Gaia instrument Instrument predicted performances Real observations LIII congresso SAIt – Pisa 4-8 maggio 2009 5 Instrument modelling approach SIMULATIONS Design parameters In-flight instrument behaviour forward Analysis backward Instrument predicted performances Real observations CALIBRATIONS Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 6 DU4 - Instrument model DU1 Management: Management SW system: X.Luri Orbit/Attitude: Un.Barcelona Optics: DU3 Dispersers: Universe CCD: model PSF/LSF: A.Robin Obs.Besançon BAM: On-Board processing: CU2 - Simulations D.Gardiol Oss.Torino F.Russo Oss.Torino DU2 SW Engineering J.M. Wallut Data Generators R.Keil ZARM Bremen CNRS Tolosa DU5 GASS → telemetry D.Loreggia Oss.Torino E.Masana - Un. Barcelona J.Rebordao INETI Lisbon DU6 GIBIS → pixel level L.Corcione Oss.Torino C.Babusiaux - Obs. Meudon D.Gardiol, D.Busonero OATo DU7 GOG → MDB objects Y.Isasi – Un. Barcelona D.Gardiol Oss. Torino J.Portell Daniele Gardiol – Modelling the Gaia instrument DU4 Instrument model D.Gardiol Oss.Torino Univ. Barcelona LIII congresso SAIt – Pisa 4-8 maggio 2009 7 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 8 Gaia – Disegno ottico LOS2 LOS2 Image credit: www.rssd.esa.int Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 9 Gaia – Piano focale Image credit: www.rssd.esa.int Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 10 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 11 PSF/LSF model for simulations Model described in GAIA-CU2-TN-INAF-DG-011 (ESA-Gaia livelink) The model is based on a dual representation: Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 12 PSF/LSF model for simulations Model described in GAIA-CU2-TN-INAF-DG-011 (ESA-Gaia livelink) The model is based on a dual representation: • numerical library for GIBIS. The starting point is a numerical library (discrete sampling) where the elements are generated from the optical design of the instrument (CodeV generated WFEs) plus some ad-hoc effects. Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 13 PSF/LSF model for simulations Model described in GAIA-CU2-TN-INAF-DG-011 (ESA-Gaia livelink) The model is based on a dual representation: • numerical library for GIBIS. The starting point is a numerical library (discrete sampling) where the elements are generated from the optical design of the instrument (CodeV generated WFEs) plus some ad-hoc effects. • analytical library for GASS/GOG. The elements of the library are generated from fittings of suitable functions to the elements of the numerical library. Interpolation may be used when appropriate. Detailed in GAIA-CU3-TN-INAF-DB-007 (ESA-Gaia livelink) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 14 Advantages of such a model • many effects can be introduced at the level of the numerical library and they will automatically be present in the analytical representation (no need to develop specific models for GASS and GOG) → homogeneity of simulations • the analytical representation requires a minimised number of computations in GASS/GOG → good compromise between realism and efficiency • nonetheless, many effects are not usefully described by means of precomputed libraries (CTI, noise, magnitude, non-linearity/ saturation, …) and have to be treated separately Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 15 Analytical model – example (AF) •S1R1T1 vs. S1R7T1 (V-I = 0.0) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 16 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 17 CCD – modelling radiation damage and CTI • high energy solar radiation creates traps into the semiconductor lattice that capture photoelectrons and release them after some time • this increases the CTI of the CCD • as a result, the charge packet is displaced (retarded) wrt the source Image credit: www.rssd.esa.int Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 18 Radiation damage - test campaign • Three CCD zones: No irradiated, transition, Irradiated (end-ofmission dose) • Three irradiated sessions with different diffuse optical background (DOB): 0, 5, and 10 e-/pixel • Different brightness levels corresponding to nominal 60000, 7000, 2000, 650, 400, and 200 integrated eJ.C.Terrazas Vargas, L.Corcione, M.Gai, M.Lattanzi (OATo) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 19 Photometry – charge loss Not irradiated zone Transition zone Irradiated zone Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 20 Astrometry – centroid bias Centroids relative to the linear best fit from stars A to C in the non-irradiate zone Non irradiated zone Transition zone Irradiated zone not compensated for the mask bias Daniele Gardiol – Modelling the Gaia instrument mask bias subtracted LIII congresso SAIt – Pisa 4-8 maggio 2009 21 21 Radiation test campaign and Simulations Test campaign Microscopic model Raw test data Montecarlo Model simulations verification Test data analysis Models verification Daniele Gardiol – Modelling the Gaia instrument Macroscopic models GASS (CDM1) GIBIS (CDM1&2) Simulated datasets LIII congresso SAIt – Pisa 4-8 maggio 2009 22 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 23 Basic Angle Variation and Monitoring LOS 1 LOS 2 Image credit: Meijer et al., SPIE 7010 Daniele Gardiol – Modelling the Gaia instrument D.Gardiol, A.Riva, F.Russo (OATo) LIII congresso SAIt – Pisa 4-8 maggio 2009 24 Basic Angle Variation Sinusoidal behaviour expected (P = 6 hours) Thermal perturbation (M1#1 – M2#2) ~ 200 µK peak to peak BA response ~ 1200 µas Daniele Gardiol – Modelling the Gaia instrument dBA/dT ~ 6 µas / µK (Gardiol et al., SPIE 5497) LIII congresso SAIt – Pisa 4-8 maggio 2009 25 BAM fringes simulation BAM detailed optical design (Zemax) available to us only since last week. Analytical model implemented, based on ideal BAM instrument For each telescope the fringe pattern is given by: D 2B xi x0 fp(i, j ) k Airy rij 1 V cos f f where xi R(i i0 ) 2 LOS f y j R( j j0 ) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 26 Physical window on the CCD: Size is 360 x 120 pixels = 3.6 x 3.6 mm² Fringe image Logical window size is 360 pixels x 60 samples (binning x 2 AC scan) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 27 BAM optical layout Image credit: Meijer et al., SPIE 7010 Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 28 Overview • Gaia expected performances • Instrument modelling approach • Gaia: the instrument • PSF/LSF model for simulation • CCD Charge Transfer Inefficiency modelling • Basic Angle Variation and Monitoring • Astrometric error prediction for General Relativity Experiment Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 29 Astrometric error prediction near Jupiter for the GAia Relativity Experiment Crosta & Mignard, 2006 GAREX aims at testing General Relativity with Eddington-like differential measurements. Right: light deflection due to the quadrupole of Jupiter predicted by GR, but never actually measured (240 µas at limb) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 30 Astrometric error prediction near Jupiter for the GAia Relativity Experiment • • • • • Analytical law derived from Montecarlo simulations Dependence on star magnitude and distance from Jupiter limb partial-to-full saturation of CCD pixels taken into account background level due to Jupiter's stray-light Error refers to a single CCD transit (differential measurement) Note that the actual values of the errors depend on several assumptions, e.g.: • a specific location algorithm (least-square); • a specific measurement process, depending on the read-out procedure actually foreseen for Gaia • nominal CCD performances (e.g. no CTI) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 31 Astrometric error prediction near Jupiter for the GAia Relativity Experiment CCD 10 g ( m,r ) g (m, r ) A(r ) m B(r ) m 2 C (r ) m D(r ) E (r ) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 32 GRAZIE PER L’ATTENZIONE Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 33 Backup slides Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 34 Current version of the model Numerical library: The starting point is the current library of effective PSFs generated by PSFmaker 2.1 http://gibispc.obspm.fr/~gibis/data-V4.0/psf/Gaia3/AF/effective/ Analytical library: • Fitting functs: Bi-quartic B-Spline (L.Lindegren, GAIA-C2-TN-LU-LL-066) • 31 knots equally spaced (spacing = 0.5 pixels AL) • Result: Analytical function giving pixel readout for any (continuous) AL position: LSF ( x) Ak p1 ,, pn Bk x k • Details of the analytical library coeffs calculation in GAIA-CU3-TN-INAFDB007 (ESA-Gaia livelink) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 35 Parameter space domain Parameter SM AF BP RP RVS Telescopes 2 2 2 2 2 FoV 7 62 7 7 12 12 12 16 18 15 Source spectrum 168 + 1488 + 224 + 252 + 360 = 2492 Size (numerical, GB) 0.672 5.952 0.244 0.252 0.360 = 7.5 GB Size (analytical, MB) 0.148 1.333 0.196 0.316 = 2.2 MB Daniele Gardiol – Modelling the Gaia instrument 0.211 LIII congresso SAIt – Pisa 4-8 maggio 2009 36 Current version of the model SM/AF Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 37 Current version of the model BP/RP/RVS Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 38 Analytical model – example (AF) •FoV domain (coeff n. 15, V-I=0): Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 39 Analytical model – example (AF) Source colour domain (S1R1T1): Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 40 Analytical model - AF (example) Sub-pixel AC-shift (S1R1T1, coeff. N. 15, V-I = 0.0): Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 41 Analytical model for AF – Relative weights of effects PSF shape variation wrt given domain estimated using BQ coeff dispersion (coeff N15): RWeight 100 Max Min Max Min BQ-coeff values dispersion as a function of • FoV position: 30% (0.36 – 0.67 peak to peak) • Colour variation: 13% (0.45 – 0.58 peak to peak) • sub-pixel AC position 0.1% (0.5647-0.5663 peak to peak) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 42 Radiation campaign data analysis results Photometry Astrometry •The charge loss due to trap presence affects the star signal mostly at the first transit •The average amount of charge loss is highly correlated to the signal intensity •Charge loss measurements are in good agreement with a power law. •The adopted levels of DOB contribute in partially mitigating the radiation damage (particularly evident at the faintest signals) •Mechanical uncertainties (mask position jitter, mask yaw) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 • At high signal regimes (S>=2000), centroid bias is correlated with the signal level and DOB apparently mitigates the CTI effects •At low signal levels (S<= 600), the centroid bias is largely affected by mask positioning uncertainties for definitive conclusions 43 Charge loss - summary Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 44 Cose che vado a dire • gaia expected performances-> consequences on instrument modelling (for- and back-wards) • accenno a CU2/DU4 (forward analysis) vs. CU3/AVU (backward analysis) • Struttura di CU2: IM,UM + GIBIS/GASS/GOG e Struttura di DU4 • Disegno ottico / Piano focale di Gaia (2) • PSF/LSF library (5) • CCDs - Radiation damage (5) • BAM simulation (2) • BAV (spie 2004?) • Prediction of astrometric accuracy Near very bright objects (GAREX -> A&A) Daniele Gardiol – Modelling the Gaia instrument LIII congresso SAIt – Pisa 4-8 maggio 2009 45