ANDROIDS: Calibrating the Optical-NIR SED of M31 and Mapping Stellar Populations

Presented at AAS 225 in Seattle WA. Results are still tentative; please email me before using them in other investigations. (see contact info at http://jonathansick.ca)

Transcript

1. Jonathan Sick @jonathansick ANDROIDS is: Stéphane Courteau (Queen’s) Jean-Charles Cuillandre (CEA) Julianne Dalcanton (UW) Roelof de Jong (AIP) Michael McDonald (MIT) R. Brent Tully (IfA) github.com/jonathansick Thanks: Canada-France Hawaii Telescope Corp. Canadian Advanced Network for Astronomical Research National Sciences & Engineering Research Council David Hogg (NYU); Joel Roediger (DAO) Calibrating the Optical-NIR SED of M31 & Mapping Stellar Populations
2. Why M31? Missing: the global optical-NIR picture of M31 Massive Disk Galaxy Proximity Complex Star Formation History & Structural Evolution Resolved Stars Resolved Stellar Pops. Resolved Stellar Kinematics e.g., Dorman/SPLASH e.g., Lewis/PHAT Halo Maps e.g., PAndAS / /
3. ANDROIDS: All-Disk M31 Science Identify biases/degeneracies of modelling SEDs. Identify tensions in stellar population synthesis (e.g., RGB and AGB phases contributing to the Near-IR light) Global analysis of stellar populations and structure Comparison of Spectral Energy Distribution (SED) modelling to resolved star formation histories (e.g., PHAT) Decomposition of stellar bulge, disk and halo components Bridge M31 observations to more distant galaxies Maps of: star formation history, stellar metallicity, stellar mass, dust attenuation
4. Taylor et al (2011): NIR SED is inconsistent with optical SED. Taylor et al 2011. GAMA Survey u g r i z Y J H K u g r i z Y J H K Solution: determine exactly how resolved stars contribute to NIR SED. e.g., Melbourne et al 2012 Errors in Near-IR isochrones (TP-AGB)? Poor star formation history models? Calibration of Near-IR observations?
5. Photo: Tom Buckley-Houston / Stephen Rahn / NASA GALEX
6. ANDROMEDA OPTICAL & INFRARED DISK SURVEY @ CANADA-FRANCE-HAWAII TELESCOPE
7. ANDROMEDA OPTICAL & INFRARED DISK SURVEY 0h30m 35m 40m 45m 50m RA (J2000) +39 +40 +41 +42 +43 Dec (J2000) PHAT CFHT/MegaCam CFHT/WIRCam WIRCam extension 40 kpc 30 kpc 120 hours on CFHT ugri bands 1˚ x 1˚ 0.18″ pix-1 R = 40 kpc JKs bands 20ʹ x 20ʹ 0.3″ pix-1 R = 28 kpc Resolved Stars & Spectral Energy Distributions
8. 5 gigapixel map µg ≲26 mag arcsec-2 optical (ugr) colour composite
9. optical (ugr) colour composite
10. Near-IR (iJKs) colour composite
11. optical (ugr) colour composite
12. Near-IR (iJKs) colour composite
13. Nodding in Optical! Elixir-LSB 0h30m 40m 50m 1h00m +38 +40 +42 +44 Dec (J2000) Flat-fielded MegaCam frame Real-time background Background subtracted Sky Fields Disk cycled in 1 hour
14. Sky offset optimization Solve ‘sky offsets’ for each image to minimize image-to-image differences - Solved hierarchically (~4K image optimization) - Offsets within sky uncertainty − 1.0 − 0.5 0.0 0.5 1.0 η (degrees) − 1.0 − 0.5 0.0 0.5 ξ (degrees) − 1.0 − 0.5 0.0 0.5 ξ (degrees) 18 .0 19 .5 21 .0 22 .5 16 18 20 22 J K s mag arcsec-2 1% sky residuals Sky Nodding − 1.0 − 0.5 0.0 0.5 1.0 η (degrees) − 1.0 − 0.5 0.0 0.5 ξ (degrees) − 1.0 − 0.5 0.0 0.5 ξ (degrees) 18 .0 19 .5 21 .0 22 .5 16 18 20 22 J K s mag arcsec-2 0.03% sky residuals Sky Nodding+Optimization 1% sky residuals <0.09% sky residuals 1.0 0.5 0.0 0.5 (degrees) 1.0 0.5 0.0 0.5 1.0 (degrees) J 1.0 0.5 0.0 0.5 (degrees) Ks 0.0 0.5 1.0 1.5 RMS (mag É 2) 0.5 1.0 1.5 RMS (mag É 2) 0.10 0.05 0.20 Need absolute ZP calibration Zeropoint uncertainty of sky offset normalization is significant in NIR! Sick et al (2014) AJ 147 109
15. Modelling an SED Model SED (Pop Synthesis: FSPS) Model Parameters {Mass, SFR(t), Metallicity, Dust, …} Observed SED e.g., ugriJKs Observed SED Uncertainties n Fn n ƒn ln p(✓|F) / X X ✓ FX fX(✓) X ◆2 FX BX u, g, r, i, J, Ks (spectral energy distribution) if only!
16. Estimate Backgrounds by Modelling Many Pixels Solution: Hierarchical Bayesian Modelling Pixel n B Fn ƒn n n Pixel 0 Pixel N {✓}i 1 0 ! {✓}i 0 {✓}i 1 N ! {✓}i N Linearly update background estimates: hBin,X = Fn,X f(✓n,i) Bi X N 0 @ P n hBin 2 n P n 1 2 n , 1 P n 2 n 1 A Estimate with Gibbs MCMC Sampler and repeat! … github.com/jonathansick/sedbot Sample stellar pops. of pixels independently:
17. Multi-pixel Background Modelling 0 5 10 15 20 25 30 Rm j (kpc) 16 18 20 22 24 26 28 AB (mag arcsec 2) g0 r0 0 J Ks 0.10 0.05 0.00 0.05 0.10 Jy É 2 , B = 0.0 ± 0.0 Jy rcsec 2 0.10 0.05 0.00 0.05 0.10 Jy É 2 g0, B = 0.0 ± 0.0 Jy rcsec 2 0.15 0.00 0.15 0.30 Jy É 2 r0, B = 0.1 ± 0.1 Jy rcsec 2 0.6 0.4 0.2 0.0 Jy É 2 0, B = 0.3 ± 0.1 Jy rcsec 2 2 1 0 1 2 Jy É 2 J, B = 0.1 ± 0.5 Jy rcsec 2 0 1000 2000 3000 4000 Step 12 10 8 6 4 Jy É 2 Ks, B = 8.0 ± 0.9 Jy rcsec 2
18. Multi-pixel Background Modelling 0.10 0.05 0.00 0.05 0.10 Jy É 2 , B = 0.0 ± 0.0 Jy rcsec 2 0.10 0.05 0.00 0.05 0.10 Jy É 2 g0, B = 0.0 ± 0.0 Jy rcsec 2 0.15 0.00 0.15 0.30 Jy É 2 r0, B = 0.1 ± 0.1 Jy rcsec 2 0.6 0.4 0.2 0.0 Jy É 2 0, B = 0.3 ± 0.1 Jy rcsec 2 2 1 0 1 2 Jy É 2 J, B = 0.1 ± 0.5 Jy rcsec 2 0 1000 2000 3000 4000 Step 12 10 8 6 4 Jy É 2 Ks, B = 8.0 ± 0.9 Jy rcsec 2 0 5 10 15 20 25 30 Rm j (kpc) 16 18 20 22 24 26 28 AB (mag arcsec 2) g0 r0 0 J Ks
19. Bayesian SED Modelling MAGPHYS: da Cunha & Charlot (2008) (Future: MCMC SED modelling with FSPS — github.com/jonathansick/sedbot) Charlot & Bruzual 2007 Population Synthesis Star Formation History Parameters: Exponential+Constant+Stochastic Bursts Birth Cloud + ISM Attenuation Charlot & Fall 2000 Stochastic Library of Parameters (drawn from priors) Stochastic Library of Model SEDs Estimate parameters by marginalization over model probabilities p ( ✓|F ) = exp ( 1 2 X X ✓ FX MfX X ◆2 )
20. Stellar Metallicity Profiles 0 10 20 30 40 Rm j (kpc) 1.0 0.8 0.6 0.4 0.2 0.0 0.2 log Z/Z gr JKs gr gr JKs 0h36m 39m 42m 45m 48m J2000 +40 +41 +42 J2000 0.4 0.2 0.0 0.2 log Z/Z Next: direct comparison of SED and resolved (PHAT) metallicities, etc.. Saglia+ 2010: R=0 kpc: [Z/H] ~ 0.5 R=1 kpc: [Z/H]~ 0.1 NIR coverage drop 10 kpc 20 kpc
21. Mass-weighted Age Profiles 0 10 20 30 40 Rm j (kpc) 0 2 4 6 8 10 12 14 Mass-weighted Age (Gyr) gr JKs gr gr JKs 0h36m 39m 42m 45m 48m J2000 +40 +41 +42 J2000 2 4 6 8 10 Age (Gyr) Saglia+ 2010: R < 1.3 kpc: 12 Gyr R > 1.3 kpc: < 8 Gyr 10 kpc 20 kpc Brown06 Disk
22. Mass-to-Light Profiles 0 10 20 30 40 Rm j (kpc) 0.4 0.2 0.0 0.2 0.4 0.6 log M /L 0 MAGPHYS gr JKs MAGPHYS gr Zibetti+ 09 Taylor+ 11 Tamm+ 2011 (SDSS) 0h36m 39m 42m 45m 48m J2000 +40 +41 +42 J2000 0.4 0.2 0.0 0.2 0.4 log M /L 0 Chabrier (2003) IMF Taylor 2011: 23% lighter disk than ugriJKs SED fit.
23. Summary Andromeda Optical and Infrared Disk Survey M31 Stellar Populations from ugriJKs SED Modelling On-going projects: CFHT Map of M31 • Optical-IR: ugriJKs • Coverage to R=40 kpc • 5 gigapixels per u, g, r, i • Resolved stars & SEDs Background Subtraction! 1. Sky-target Nodding 2. Mosaic sky optimization 3. Hierarchical modelling 4. Next: wide-field Dragonfly map Direct comparison of SED and resolved star formation histories. TiO/CN + JKs Carbon Star Survey. SED modelling shows M31 as if seen in unresolved galaxy light. Clear age, metallicity & M/L gradients. Photometric+Kinematic Stellar and Dark Matter Mass Decomposition Next: