Mapping the Milky Way with SDSS, Pan-STARRS and LSST
Transcript of Mapping the Milky Way with SDSS, Pan-STARRS and LSST
Mapping the Milky Way with SDSS,Pan-STARRS and LSST
Zeljko Ivezic
Mario Juric, Nick Bond, Alyson Brooks, Robert Lupton, et al.
University of Washington, Princeton University
Institute for Astronomy, Honolulu, Sep 31, 2005
1
Outline
1. Motivation: a detailed description of the Milky Way
2. Overview of SDSS and LSST
3. Photometric Parallax Method
4. The properties of thin and thick disks
5. Large overdensity towards l = 300, b = 60 at ∼10-15 kpc
6. The Milky Way Kinematics
2
The Milky Way Maps• The top left panel is not really
the Milky Way :) but it shows
the distance range probed by
SDSS-detected main sequence
stars (out to ∼15 kpc)
• SDSS RR Lyrae and other lumi-
nous tracers, and 2MASS M gi-
ants, demonstrate that the Milky
Way halo extends to ∼100 kpc
and has a lot of substructure
• What is the structure of the disk
component out to a few tens of
kpc?
• SDSS has obtained excellent
photometric data for close to 100
million stars. How can one utilize
these data for studying the disk
component?
3
Constraining Thin/Thick Disk+Halo Models• Observationally, ρ(z|R = R�) is well fit by a sum of
double exponential (thin and thick disk) and power-
law profiles.
• But, very different models (top: thin and thick disk
without halo; middle: single disk and halo, bottom:
the difference) can produce the same ρ(z|R = R�)
• A large sky area is needed to break model degenera-
cies (pencil beam surveys are not conclusive)
• SDSS is the first survey with the required data
4
Overview of SDSS
• Imaging and Spectroscopic Survey
– ∼10,000 deg2 (1/4 of the full sky)
– 5 bands (ugriz: UV-IR), 0.02 mag photometric accuracy
– < 0.1 arcsec astrometric accuracy
– Over 100,000,000, mostly main sequence, stars
– Spectra for >200,000 stars (radial v to ∼10 km/s)
• Advantages for studying the Milky Way structure
– Accurate photometry: photometric distance estimates
– Numerous stars: small random errors for number density
– Large area and faint limit: good volume coverage
5
Smolcic et al. (2004)
SDSS Color-color diagrams• Wide wavelength coverage of
SDSS bandpasses, together with
accurate and robust photometry,
encodes a large amount of infor-
mation
• Stars on the main stellar locus
are dominated (∼98%) by main
sequence stars
• The position of main sequence
stars on the locus is controlled by
their spectral type/effective tem-
perature/luminosity, and thus
can be used to estimate distance:
photometric parallax method
• A preview of results from Juric et
al. (2005)
6
Large Synoptic Survey Telescope
• LSST = d(SDSS)/dt: an 8.4m telescope with single expo-
sures reaching V∼24.5 over a 9.6 deg2 FOV: the whole (observ-
able) sky in two bands every three nights
• LSST = Super-SDSS: an optical/near-IR survey of the ob-
servable sky in multiple bands (grizY) to V∼26.5 (coadded)
Science Drivers• Dark Energy and Dark Matter (through weak lensing, SNe Ia,
clusters)
• The Milky Way Map (main sequence to 150 kpc, RR Lyrae to
400 kpc, parallaxes for all stars within 500 pc)
• The Solar System Map (over a million main-belt asteroids,
∼100,000 KBOs, Sedna-like objects to beyond 150 AU)
• The Transient Universe (a variety of time scales ranging from
∼10 sec, to the whole sky every 3 nights)7
• Immediate public data distribution (transients within 30 sec)
• 30 TB of data per night (3.2 Gpix camera ∼27 SDSS cameras)
• 60 PB of data over ten years
• A collaboration of numerous (∼20) US institutions (NOAO, Re-
search Corporation, UA, UW, . . . JHU, Harvard, . . . DoE Labs,
. . . Google, Microsoft, . . . )
• A combination of government (NSF and DoE) and private fund-
ing
• Already underway with significant private and NSF funding
• The first light around 2013
8
9
Photometric ParallaxMethod
• Adopted a single relation that
agrees with geometric paral-
lax measurements for nearby
M dwarfs, and with globular
cluster CMDs.
• To increase signal-to-noise at
the faint end, stars are ML
projected on the stellar locus
• Applied to 50 million stars in
6500 deg2 and 100 pc to 15
kpc distance range
• Pitfalls: systematic errors in
adopted relation (e.g. metal-
licity effects), contamination
by giants, smaller distance
range than for e.g. red giants
and RR Lyrae
10
Dissecting Milky Way with SDSS
• Traditional approach: assume initial mass function, fold with
models for stellar evolution; assume mass-luminosity relation;
assume some parametrization for the number density distri-
bution; vary (numerous) free parameters until the observed
and model counts agree. Uniqueness? Validity of all assump-
tions?
• SDSS photometric parallax approach: adopt color-luminosity
relation, estimate distance to each star, bin the stars in XYZ
space and directly compute the stellar number density (for
each narrow color bin). There is no need to a priori assume,
the number of, and analytic form for Galactic components
11
Local maps: thin disk• Red(ish) stars have small lumi-
nosity: sampled to a few kpc
• The maps are roughly consistent
with an exponential disk out to
∼1 kpc: the lines of constant
density are straight lines
• The slope of these lines is given
by the ratio of exponential scale
height and scale length
12
13
Thin to thick disk transition• Yellow(ish) stars have interme-
diate luminosities: sample the
transition from thin to thick disk
at a few kpc
14
Is the Thick Disk ReallyNeeded?
• It is not needed to fit ρ(Z|R =
R�: an appropriate halo, with a
thin disk, can explain the counts
• However, when stars are sepa-
rated by metallicity (using u − g
color as a proxy), low-metalicity
stars follow the halo component,
and the density profile for high-
metalicity stars requires the thick
disk component (or, at least, a
profile different from a single ex-
ponential disk)
• A more robust estimate of the
required number of components
could be obtained by the full
2D analysis of the density maps,
but. . .
15
16
17
18
19
20
Large Virgo Overdensity
21
22
23
Summary
• 3D stellar number density maps of the Milky Way from SDSSphotometric observations of ∼50 million stars
• A two-component exponential disk model is in fair agreementwith the data
• Halo properties poorly constrained due to rich substructureand limited sky coverage; however, an oblate halo is alwayspreferred (no strong evidence for triaxial halo)
• A remarkable localized overdensity in the direction of Virgoover ∼1000 deg2 of the sky
• Clumps/overdensities/streams are an integral part of MilkyWay structure, both of halo and the disk(s)
24
The SDSS Kinematics Data
• SDSS has already obtained over 200,000 stellar spectra, withradial velocities accurate to ∼10 km/s
• SDSS-POSS proper motions (50 yrs baseline), limited by thePOSS astrometric accuracy (0.15 arcsec, recalibrated POSSastrometry by Sesar et al. and Munn et al.), resulting inproper motion accuracy of ∼3 mas/yr; usable to g ∼ 20
• SDSS-SDSS proper motions (∼5 yrs baseline) accurate to∼6 mas/yr (using only 2 epochs); usable to g ∼ 21
• SDSS-LSST proper motions (∼15 yrs baseline, limited by theSDSS astrometry) accurate to <1 mas/yr; usable to g ∼ 22
SDSS is revolutionizing kinematic studies of the Galactic struc-ture (3 mas/yr corresponds to 15 km/s at 1 kpc, and radialvelocities are measured out to 10 kpc)
25
The Milky Way Kinematics Studies Enabled by
SDSS
• For a large number of stars, spread with large baseline over
the sky, SDSS has measured all three velocity components
• By selecting different directions on the sky, systematics can
be reliably controlled (in addition to e.g. using quasars to
determine proper motion errors)
• The dependence of velocity on position, vφ(X, Y, Z), vR(X, Y, Z),
and vZ(X, Y, Z), can be studied directly
• One can also tag the stars by e.g. metallicity (using the u−g
color as a proxy) and study the metallicity-kinematics corre-
lations (with 10-100 times larger sample than previously)
26
0 1 2 321
20
19
18
17
16
15
14
0 1 2 321
20
19
18
17
16
15
14
0 1 2 3
0
1
2
0 1 2 3
0
1
2
0 1 2 321
20
19
18
17
16
15
14
0 1 2 321
20
19
18
17
16
15
14
0 1 2 3
0
1
2
0 1 2 3
0
1
2
Photometrically and Astrometrically Variable Ob-
jects
27
360180
0
4,859 (halo) stars with 0.75 < u-g < 0.9
180
-400 -200 0 200 400
radial velocity (km/s)
Radial velocities for starswith u− g excess
• u − g < 1 selects low-
metallicity stars, which are
presumably halo stars
• stars with u− g < 1 show a
strong dipole in the radial
velocity distribution: net
motion relative to the Sun
28
360180
0
Halo stars corrected for solar motion (209.5 km/s towards l=78.2 b=-4.4)
180
-400 -200 0 200 400
radial velocity (km/s)
Radial velocities for starswith u− g excess
• u − g < 1 selects low-
metallicity stars, which are
presumably halo stars
• stars with u− g < 1 show a
strong dipole in the radial
velocity distribution: net
motion relative to the Sun
• when solar motion is ac-
counted for, stars with u −g < 1 show no overall mo-
tion: halo is not rotating
(vrot < 10±X km/s, X∼10)
• the velocity dispersion is
large (∼100 km/s)
29
Halo vs. Disk(s) Kinematics• Kurucz models indicate that
stars with u − g < 1 have low
metallicity (Z∼-2)
• Stars with u − g < 1 have
markedly different kinematics
than stars with u− g > 1
• Low-metallicity stars have fairly
constant kinematics behavior
within a few kpc
• High-metallicity stars have
smoothly increasing rotational
lag and velocity dispersion
with Z for all three velocity
components
• The dependence of the rota-
tional lag on the height above the
plane dominates over the radial
gradient
30
Halo vs. Disk(s) Kinematics• When looked in detail, kinematic
structure is more complex than,
e.g., Schwarzschild (Gaussian)
velocity distribution, and the nor-
malization of individual compo-
nents is inconsistent with rela-
tive normalization inferred from
counts
• Standard kinematic models can-
not explain the data.
31
Grand Summary
• Clumps/overdensities/streams are an integral part of Milky
Way structure, both of halo and the disk(s)
• Analogous complexity is seen in the Milky Way kinematics
SDSS is revolutionizing studies of the Galactic structure, and
Pan-STARRS and LSST will do even better!
32