Precision of GAIA photometry, technical details =============================================== E. Hoeg, C. Fabricius 18 Feb 1999 SAG_CUO_59 ABSTRACT: The standard errors are plotted in CUO_60 for the G band from one field crossing of 16 CCDs (G epoch photometry) and for average photometry of 100 field crossings of the BBP and NBP. The average number of crossings in a 4 year observation period is 67 for Astro-1 and for Astro-2, and 100 for the Spectro telescope. The photometric system described in CUO_58 is assumed, including 5 bands in the BBP and 10 bands in the NBP, i.e. the NBP with dichroic filter. Figures with plots of precision are given separately, in CUO_60. Technical details. ================== Standard errors are calculated based on photon statistics and readnoise according to Eq.(11) of CUO_53, and these errors are multiplied by a factor 1.2 as a rough account for other error sources. Any compression used is assumed to be loss-free. Count rates are adapted from a message by L. Lindegren to E.H. of 8 Feb. 1999. The widths of P to S are taken from Table 3 of CUO_58, ie. the count rates by Lennart are multiplied by a correction factor: width in Table 3 / width in Lennarts table. Astro telescopes, AF and BBP ---------------------------- Not quite the same assumptions as in Table 2 of CUO_53: fgriz has replaced Sloan g'r'i'z', another CCD : Figs 1&2 Fig 1 Fig 2 AF BBP Astro-1 BBP Astro-2 CCD#1B (with UV response) Aperture 1.19 m^2 G fgriz fgriz 1 pixel = 37*111 mas 1 sample: 1*8 pixels 1*8 pixels 6*8 pixels read out noise, r: 5.7 e- 6.7 e- 5.8 e- samples/star, n_s: 6 6 3 samples for background, n_b: 6 6 3 The differences from those used for the precision plots in Fig.3 of CUO_53 appears by comparison with CUO_37 of June 1998, included below. n_b=12 was then assumed for G, r=4.5 e- everywhere, 3*8 pixels/sample for the Astro instruments. The background has a G2V spectrum and an intensity of V=21 mag/sq_arcsec. This gives the background for one sample of 8 or 48 pixels : Astro-1: Photometric bands: G f g r i z Background 1 smpl: 4.59 0.91 0.94 0.89 0.65 0.35 Integration time: 0.9 Correction factor: 1.000 Number of observ.: 16. 100. 100. 100. 100. 100. Astro-2: Photometric bands: G f g r i z Background 1 smpl: 4.59 5.46 5.62 5.34 3.93 2.11 Integration time: 0.9 Correction factor: 1.000 Number of observ.: 16. 100. 100. 100. 100. 100. Fig. 1. Precision versus V magnitude of photometry with Astro-1. Standard errors are given for the average of n_obs observations, for 1*8 pixels/sample in the G band and the BBP. Fig. 2. Precision with Astro-2. 1*8 pixels/sample for the G band and 6*8 pixels/sample in the BBP. Spectro telescope, NBP ---------------------- The same assumptions as in Table 2 of CUO_53 for r, n_s, n_b : CCD#1B (with UV response) Aperture 0.525 m^2 1 pixel = 500*500 mas 1 sample = 1*5 pixels read out noise, r: 3.0 e- samples/star, n_s: 3 samples for background, n_b: 6 The option with dichroic filter was assumed. The differences from those used for the precision plots in Fig.4 of CUO_53 appears by comparison with CUO_38 of June 1998, included below. r=4.5 e- was then assumed. The background has a G2V spectrum and an intensity of V=21 mag/sq_arcsec. This gives the following background for one sample. The correction factor takes into account that the adopted bandwidths differ from those in the message of 8 Feb from L. Lindegren : Spectro: Photometric bands: u P v b Z y S p1 p2 p3 Background 1 smpl: 7.94 8.84 9.84 13.32 13.22 13.47 12.78 14.23 6.18 1.96 Integration time: 4.8 Correction factor: 1.000 1.154 1.053 1.111 0.952 0.870 1.000 1.000 1.000 1.000 Number of observ.: 200. 200. 100. 100. 100. 100. 100. 100. 100. 200. Fig. 3. Precision versus V magnitude for the average of n_obs observations in the NBP for A_V=0 mag. Fig. 4. Precision versus V magnitude for A_V=4 mag. ----------------------------------------------------------------- Plotting in CUO_60 similar to CUO_53 Figs. 3 and 4. Fig. 1. Astro-1 Telescope, A_V=0 mag, plotting vs V Fig. 2. Astro-2 Telescope, A_V=0 mag, plotting vs V plot of standard errors for : G, 1 epoch, 16 obs. f, average of 100 obs. g, average of 100 obs. r, average of 100 obs. i, average of 100 obs. z, average of 100 obs. Fig. 3. Spectro Telescope, A_V=0 mag, plotting vs V Fig. 4. Spectro Telescope, A_V=4 mag, plotting vs V plot of standard errors for : u, average of 200 obs. P, average of 200 obs. v, average of 100 obs. b, average of 100 obs. Z, average of 100 obs. y, average of 100 obs. S, average of 100 obs. p1, average of 100 obs. p2, average of 100 obs. p3, average of 200 obs. Fig. 5. Astro-1 Telescope, A_V=0 mag, plotting vs Ic Fig. 6. Astro-1 Telescope, A_V=0 mag, plotting vs G --------------------------------------------------------------- Annex of CUO_60: It contains 7x3=21 plots from which the above six are taken. #A0-2: Astro-1 vs Ic for the 3 values A_V=0, 2, 4 #A3-5: Astro-2 vs Ic for the 3 values A_V=0, 2, 4 #A6-8: Astro-1 vs G for the 3 values A_V=0, 2, 4 #A9-11: Astro-2 vs G for the 3 values A_V=0, 2, 4 #A12-14: Astro-1 vs V for the 3 values A_V=0, 2, 4 #A15-17: Astro-2 vs V for the 3 values A_V=0, 2, 4 #A18-20: Spectro vs V for the 3 values A_V=0, 2, 4 *********************************************************************** Copies of CUO_37 and 38 from June 1998 describing the previous plots given, e.g., in CUO_53. Precision of FSM photometry II =========================== C. Fabricius, E. Hoeg 18 June 1998 SAG_CUO_37 ABSTRACT: Plots of precision versus magnitude for the average of 100 observations are given. This is a typical number of observations for a 4 year observation period. More realistic values of readnoise have now been assumed (cf. CUO_36) and it is shown that the readnoise is rather negligible even at V=20.0 mag. Fig.5 gives precision only for the four Sloan bands g'r'i'z' for A_V=0.0 mag, suited to be put on the GAIA webpage. A vertical line in a plot indicates that brighter stars give saturation of the CCD. This limit is about Ic=10 mag for most spectral types. The precision of GAIA main field photometry has been estimated using fluxes computed by Lennart Lindegren based on the spectra by Gunn & Stryker (ApJS 52, 121, 1983). The computed fluxes take quantum efficiency, telescope transmittance and filter transmittance into account. Also different values of the interstellar absorption were considered, indicated by A_V on each of the figures. The Equation 9 in LL_017, Appendix B has been used to calculate photometric standard errors. A factor 1.2 has been applied to all standard errors calculated from photon noise as a safety margin. A maximum charge handling capacity of 500 000 e- has been assumed for the serial register. If this limit is exceeded a vertical line appears in the plots. The figures show precisions for 10 different filters and five different stars for the main telescopes. The values are plotted against the infrared magnitude, Ic, which is close to the PSM1 magnitude used in the detection and which is also understandable. The ten filters include the four Sloan bands g', r', i' and z' as well as the PSM2 magnitude G (without filter). For comparison also z", Rc, Ic, B and V are included. The main difference from the figures in CUO_34 is the lower readnoise (RON), for the sample of 1*8 pixels thus decreasing the standard error of y by a factor 2.2 at Ic=20 mag (see the Table 1.) The larger sample of 3*8 pixels gives a better precision for faint stars because fewer samples and therefore less readnoise contributes (cf. Table 1). ------------------------------------------------------------------ Table 1. Precision at Ic=20.0 mag of FSM photometry for Sloan i' for a G2V star. The last two lines show that the influence of RON is rather negligible at this magnitude which is the faintest to be observed by the FSM. Sample RON sg_b sg_i' Note pixels e- mag 2*7 3 =0 0.022 CUO_33, Fig.1, applying a factor 1.2 2*7 5 >0 31 Fig.1 of CUO_34 1*7 10 >0 78 Fig.3 of CUO_34 1*8 4.5 >0 36 Fig.1 3*8 4.5 >0 29 Fig.3 and 5 1*8 3.0 >0 28 not shown 3*8 3.0 >0 25 not shown ------------------------------------------------------------------ Technical details. ================== Main telescopes, FSM and PSM2. ------------------------------ Figs 1&2 Figs 3&4 Fig.5 FSM1 FSM2 as Fig.3 CCD#3 but only Aperture 1.19 m^2 g'r'i'z' 1 pixel = 37*111 mas 1 sample: 1*8 pixels 3*8 pixels read out noise, r: 4.5 e- 4.5 e- samples/star, Ns: 6 3 samples for background, Nb: 12 6 For G (PSM2), the sample is 2*2 pixels, Ns=9, Nb=18 in all figures. The background has a G2V spectrum and an intensity of V=21 mag/sq_arcsec. This gives for one sample of 8 or 24 pixels (4 pixels for G): GAIA FSM: Photometric bands: g' r' i' z' z" G Rc Ic B V Background 8 pix: 1.08 1.75 1.43 0.75 0.57 2.69 1.33 1.25 0.55 1.06 [e-] Background 24 pix: 3.23 5.24 4.29 2.25 1.71 2.69 3.99 3.76 1.66 3.19 [e-] Annex sent separately: Five figures of precision versus magnitude for the average of 100 observations. -----------------------------------*************** Precision of APT photometry II =========================== C. Fabricius, E. Hoeg 18 June 1998 SAG_CUO_38 ABSTRACT: Plots of precision versus magnitude for the average of 100 observations are given. This is a typical number of observations for a 4 year observation period. More realistic values of readnoise have now been assumed (cf. CUO_36) and it is shown that the readnoise is rather negligible even at V=18.0 mag. Slit photometry instead of aperture photometry has been assumed. Fig.3 gives precision only for the seven Stromvil bands for A_V=0.0 mag, suited to be put on the GAIA webpage. A vertical line in a plot indicates that brighter stars give saturation of the CCD. This limit is about V=9.0 mag for most spectral types. The precision of APT photometry has been estimated using fluxes computed by Lennart Lindegren based on the spectra by Gunn & Stryker (ApJS 52, 121, 1983). The computed fluxes take quantum efficiency, telescope transmittance and filter transmittance into account. Also different values of the interstellar absorption were considered, indicated by A_V on each of the figures. The Equation 9 in LL_017, Appendix B has been used to calculate photometric standard errors. A factor 1.2 has been applied to all standard errors calculated from photon noise as a safety margin. A maximum charge handling capacity of 500 000 e- has been assumed for the serial register. Values are not plotted if this limit is exceeded. The Figs. 1 and 2 show the seven Stromvil bands (u, P, v, b, Z, y, S) together with the Sloan u' and the two H-beta bands (Bn and Bw) used for the Stromgren H-beta index. The error of the H-beta index will be only slightly higher than the error for Bn. It is not foreseen to include all bands in the APT, but only the seven Stromvil bands, shown separately in Fig.3. The values are plotted as function of the Johnson V magnitude which is close to the APT SM magnitude used in the detection. A main difference from the figures in CUO_33 is the higher readnoise (RON) for the sample of 1*4 or 1*5 pixels (see the Table 1) which however appears to be rather negligible. ------------------------------------------------------------------ Table 1. Precision at V=18.0 mag of APT photometry for Stromvil y for a G2V star. The last line shows that the influence of RON is negligible at this magnitude which is the faintest to be observed by the APT. Sample RON sg_b sg_y Note pixels e- mag 1*4 2 =0 0.018 CUO_33, Fig.3, applying a factor 1.2 1*4 4.5 >0 18 Fig.1 and 3 1*4 3.0 >0 16 not shown ------------------------------------------------------------------ Technical details. ================== APT --- CCD#1 Aperture 0.525 m^2 1 pixel = 500*500 mas 1 sample = 1*4 pixels or 1*5 pixels (u,P,u',Bn) read out noise, r: 4.5 e- samples/star, Ns: 3 samples for background, Nb: 6 The background has a G2V spektrum and an intensity of V=21 mag/sq_arcsec. This gives for one sample: Photometric bands: u P v b Z y S u' Bn Bw Background: 3.51 4.72 2.87 4.58 6.09 6.98 5.80 14.12 2.73 6.58 [e-] Integration times:6.5 6.5 2.2 2.2 2.2 2.2 2.2 6.5 6.5 2.2 [s] Annex sent separately: 3 figures of precision versus magnitude for the average of 100 observations. -----------------------------------================