A New Method for Color Calibration to a Few mmag Accuracy, Recalibration of Stripe 82, and Implications on Binary Studies Haibo Yuan (苑海波) KIAA-PKU; LAMOST Fellow [email protected] Collaborators: Xiaowei Liu, Maosheng Xiang, Yang Huang, Bingqiu Chen Tuesday, November 4, 2014 Outline • Stellar Color Regression (SCR) method • Re-calibration of stripe 82 • [Fe/H]-dependent stellar locus • Binary fraction for field stars • Stellar-mass BHs in compact binaries • Summary Tuesday, November 4, 2014 Photometric calibration • Uniform calibration plays a key role in modern surveys • Challenges of 1% precision from ground: varying atmosphere extinction & flat-field (Stubbs & Tonry 2006) • Colors more important than brightness in most scientific cases Tuesday, November 4, 2014 • Methods • Based on standard stars (e.g., Landolt 1992 ) • Ubercalibration (Padmanabhan+08) • • • require over-lapping obs. 1-2 % precision in SDSS Stellar locus regression (High+09) • • • degeneracy problem: require a blue filter (e.g, u-band) restricted to low-extinction regions a few per cent accuracy Stellar locus regression (High+09) Tuesday, November 4, 2014 Stellar Color Regression (SCR) method Yuan+14a, submitted • Requirements: 1) a calibrated field 2) extinction of stars in target fields • • Zero-point of color by linear regression • Advantages: straightforward, model-free and can apply to low/high- extinction regions By-product: reddening coefficient Tuesday, November 4, 2014 Obsed. - Intrinsic color LAMOST DR1 y = a*x + b a: reddening coefficient b: color zero point Reddening • The SCR method is able to achieve an accuracy of ~ 1 mmag under good situations Dec (deg) SDSS/Stripe82 Ivezic+07 achieve ~ 1% accuracy due to 1) repeated obs. 2) color corrections from stellar-locus method [Fe/H] RA (deg) errors in flat-field vectors • 574 reference stars: S/N > 20, 4300 < Teff < 7000 K, E(B-V) < 0.04 (red dots) • 23,759 stars: S/N > 20, 4300 < Teff < 7000 K (black dots) Teff (K) E(B-V)SFD fast variations of atmosphere extin. RA (deg) Tuesday, November 4, 2014 Reddening law in stripe 82 ∆g-r ∆u-g Using the star pair technique (Yuan+13) E(B-V)SFD E(B-V)SFD ∆r-i ∆i-z Intrinsic colors of individual stars are accurate to 1- 4% E(B-V)SFD E(B-V)SFD The SFD dust map over-estimate E(B-V) by about 7 % in stripe 82 Tuesday, November 4, 2014 ∆ (u-g) Calibration errors of I07 ∆u-g, Res.=0.0078 ∆r-i, Res.=0.0025 ∆i-z, Res.=0.0026 ∆ (g-r) The tight correlations arise from usage of principle colors (s/w/x/y) to do color corrections g-r ∆ (r-i) ∆u-g, Res.=0.0071 ∆r-i, Res.=0.0025 ∆i-z, Res.=0.0038 s = 0.55(g-r)−0.25(u-g) --> δu-g/δg-r=0.55/0.25 Lines are the s = 0.55(g-r) − 0.25(u-g) expected relations --> δu-g/δg-r = 0.55/0.25 Dec (deg) ∆ (i-z) The small scatterings suggest a calibration accuracy in this work of a few mmag RA (deg) Tuesday, November 4, 2014 g-r External check σ: 0.0082 σ: 0.0030 i-z This work: using stellar colors I07: using red galaxy colors ∆u-g (RA) (This work - I07) σ: 0.0074 ∆g-r (RA） (This work - I07) Errors in I07 from σ=.003 red galaxies are: 5 -10 mmag for u-g 3 - 6 mmag for g-r 2 - 3 mmag for r-i 2 - 3 mmag for i-z ∆r-i (RA) (This work - I07) σ: 0.0026 ∆i-z (RA) (This work - I07) We achieve a precision of ~ 5 mmag for u-g 3 mmag for g-r 2 mmag for r-i 2 mmag for i-z Red galaxies also serve as excellent color standards for calibration Tuesday, November 4, 2014 Color and mag. dependences for 6 CCDs ∆u-g/2 ∆g-r ∆r-i ∆i-z ∆u-g/2 ∆g-r ∆r-i ∆i-z CCD 6 CCD 5 CCD 4 CCD 3 CCD 2 CCD 1 g-i No obvious dependences on color g r i Significant dependences on magnitude in z-band are found for 4 CCDs Exist in all the SDSS imaging data Tuesday, November 4, 2014 z [Fe/H]-dependent stellar locus • Stellar locus is widely used in • • • selecting interesting outliers • calibrations (Izevic+07, High+09) reddening determinations (Schalfly+10; Majewski+11; Berry+12; Green+14; Chen+14) However, no dependences on [Fe/H] / Log g have been considered and the intrinsic widths are unclear Tuesday, November 4, 2014 Main-sequence stars Yuan+14b, submitted 2D polynomial fitting: color (e.g., g-r) = f(g-i, [Fe/H]) using data of Stripe 82 and SDSS DR9 Fe/H: 0 −0.5 −1 Covey+07 .029 ∆g-r ∆u-g err(u)<0.01 −1.5 −2 .026 err(z)<0.01 .0076 .0106 .0067 .0067 .01 0.2 < g-i < 0.6, −1.2 < [Fe/H] < −0.7 ∆i-z ∆r-i g-i .025 g-i One dex decrease in [Fe/H] causes 0.20 & 0.02 mag decrease in u−g & g−r 0.02 & 0.02 mag increase in r−i & i−z Tuesday, November 4, 2014 .0076 All 0.2 < g-i < 0.6, −0.6 < [Fe/H] < 0 g-i g-i All ∆u-g .0064 ∆g-r .0064 ∆r-i .01 ∆i-z Residuals are fully accounted by the phot. errors & [Fe/H] uncertainties, .0067 --> intrinsic widths of the loci are ~zero Non-gaussian residuals --> binary stars Selected Implications of metallicity-dependent stellar loci • Measure photometric [Fe/H] to 0.1 dex (dwarfs) and 0.2 dex (giants) in low-extinction regions, comparable to low-resolution spectroscopy • Measure [Fe/H] (0.20 dex) and E(B-V) (0.09 mag) of stars in highextinction regions • Select halo red giants (80% efficiency, 70% completeness) With data from LSST, Chinese space station • Follow-up observations with TMT • • Tuesday, November 4, 2014 Extremely metal-poor stars Most distant giant stars in/out of the Galactic halo (> 200 kpc) • Significances Binary stars • • • star formation • • • • Visual: biased to wide, equal-mass binaries stellar evolution (BSs, XBs, type Ia SNe, BHs) population synthesis • Biases of previous methods Photometric: biased to equal-mass binaries Limited to the solar neighborhood • Stellar Locus OuTiler (SLOT) method: • • • Model-free: independent of orbital period, insensitive to mass ratio distribution assumed Applicable to a large sample of different populations Requires accurate colors and metallicities Tuesday, November 4, 2014 color Spectroscopic: biased to close, equal-mass binaries binary offset secondary primary color Maximum offset: 0.15 mag in u-g 0.035 mag in others Method ∆i-z ∆u-g ∆g-r Observed data g-i f(q) Simulated data ∝ q0.3 taking into account ∆i-z g-i g-i Fitting result Black: obs. Red: sim. Tuesday, November 4, 2014 g-i phot. errors, calib. errors & [Fe/H] errors ∆g-r ∆u-g black: all single stars grey: all binary stars g-i g-i Results from SDSS/LAMOST Stripe 82 samples SDSS (dots): Stripe 82 photo. + SDSS DR9 LAMOST (stars): Stripe 82 photo. + LAMOST DR1 Binary fraction for field FGK stars is 41±2%, increasing toward bluer colors and lower metallicities Yuan+, 2014c, submitted Tuesday, November 4, 2014 A diagram for binary classifications Yuan+14d, in prep. Vr variation Hot binaries Compact binaries Binaries from Vr variations Hot binaries Main-sequence binaries from color offset max color offset of MS binaries color offset Combining time-domain spectroscopy from LAMOST/SDSS & accurate colors, one can obtain: 1) Candidates of compact binaries; 2) Orbital period distribution and fraction of binaries Tuesday, November 4, 2014 Apply to duplicate targets of Stripe 82 White-dwarf-M-dwarf binary f(M) = (m2*sini)^3/(m1+m2)^2 = 0.4 m1 = 0.8 Msun, m2 >1.15 Msun, NS/BH? Using proper motions from TMT to constrain sin(i) i=0º Pourbaix+ 2005 Tuesday, November 4, 2014 i=45º ~100 μas at 1 kpc i=90º P~10 days Summary • We propose a SCR method to calibrate colors to a few mmag precision with the help from spectroscopy • • • Stripe 82 is re-calibrated to 2-5 mmag accuracy • Combining accurate colors, time-domain spectroscopy, and TMT opens up a new way to identify stellar-mass black holes in binaries [Fe/H]-dependent stellar loci have zero intrinsic widths Binary fraction in field stars is 41±2%, increasing towards blue colors and low metallicities Tuesday, November 4, 2014
© Copyright 2024 ExpyDoc
