HST HIGH-PRECISION PROPER MOTIONS OF GLOBULAR CLUSTERS ANDREA BELLINI - COLLABORATORS: JAY ANDERSON (STSCI), ROELAND VAN DER MAREL (STSCI), LAURA ...
←
→
Page content transcription
If your browser does not render page correctly, please read the page content below
HST high-precision proper motions of globular clusters
Andrea Bellini
Collaborators: Jay Anderson (STScI), Roeland van der Marel (STScI),
Laura Watkins (STScI)HST high-precision proper motions of globular clusters
• Part I!
(intro)! Part I: Why we need high-precision HST PMs
! • HST vs. Ground
Part II
(techniques) • Science with PMs
!
Part III !
(the project)
! Part II: Astrometry with the HST
Part IV
(results)
• Undersampling
• CTE / Geometric distortion
• Differential nature
!
Part III: Our project
• Overview
• PM measuring techniques
• The catalogs
!
Part IV: Preliminary results
• Internal motion
• Rotation
• Multiple-population kinematics
• Cluster dynamics
• Absolute motion
Andrea Bellini • Equipartition / (an)isotropyHST high-precision proper motions of globular clusters
• Part I!
(intro)!
!
Part II
HST PMs vs. GROUND PMs
(techniques)
!
FOV:
~ 1′
~ 1º
Part III
(the project)
FWHM:
0.1′′
1′′
!
Sampling:
undersampled
oversampled
Part IV
(results)
Stars:
V > 17
V < 18
Crowding:
d ~ 0.3′′
d ~ 3′′
Distortion:
large, but static
differential chromatic
! atmosphere, optics refraction, seeing,
breathing (small)
gravity flexure
Reference:
differential
can be absolute
Detector:
better pixels
more pixels
Baseline:
a few years
a few decades
Absolute PMs: faint, plentiful galaxies bright, sparse galaxies
… Very different and complementary niches
Andrea BelliniHST high-precision proper motions of globular clusters
• Part I!
!
(intro)!
Scientific applications
Part II
(techniques)
! 1. Cluster-field separation:
Part III • cluster members
(the project) • Stellar exotica
! • luminosity & mass functions
Part IV • targets for spectroscopic follow-ups
(results)
2. Internal motions:
• detailed kinematics and dynamics
3. Absolute motions
• Galactic GC orbits
4. Geometric distances
• distance scale independent from stellar-evolution models and RR Lyrae
5. Clusters rotation
6. Energy equipartition
7. Mass segregation
8. (An)isotropy
9. Full 3D cluster dynamics
• when LoS velocities for the same stars are available
10. Constraints on IMBHs
• “shooting stars”
• sudden increase in the velocity-dispersion profile in the core
11. …
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
!
• Part II!
(techniques)!
!
Part III Part II: Astrometry with HST!
(the project)
!
Part IV
(results)
!
- undersampling (PSF)
- geometric distortion
- CTE defects
- differential nature
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#1: Undersampling
• Part II!
(techniques)!
!
Part III
(the project) Illustration of undersampling conditions
!
Part IV
(results) Where is the center?
Easy
Easy
?
Harder
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#1: Undersampling
• Part II!
(techniques)!
!
Part III
(the project) Illustration of undersampling conditions
!
Part IV
(results) Where is the center?
Anderson & King 2000, PASP 112, 1360
Easy
Easy
?
Harder
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#1: Undersampling and Astrometry
• Part II!
(techniques)!
! • Impossible?!
Part III
(the project) – A point source has “no hair”
!
Part IV
(results)
• 3 parameters (x*,y*,f), ~9 pixels
– Minimal requirements: “slosh”
What is possible?!
– ≲0.01 pixel possible ~ (S/N)−1
- Need good PSF model
- Need good dithering
• Limitations!
– Individual images; no stacks
– Hard in crowded fields
• Neighbor finding/subtraction
– Ideal in “semi-crowded” regime
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#1: what do we mean by the PSF?
• Part II!
(techniques)!
! • ψINST(Δx,Δy): the “Instrumental” PSF:
Part III
(the project)
! – The PSF as it hits the detector
Part IV
(results) – Good theoretical motivations: Gaussians, Moffat
– See ψINST only indirectly in images
– To solve for: must deconvolve the PSF from the
pixels
!
• ψEFF(Δx,Δy): the “Effective” PSF:
– The PSF after pixelization: ψEFF = ψINST ⊗ Π
– Empirical: no natural basis function to describe
– We never deal with anything BUT the effective PSF
• See ψEFF directly in images
• Can measure ψEFF directly from images
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#1: Finding vs. Using the ePSF
• Part II!
(techniques)!
!
Part III
(the project)
!
Part IV
(results) • Degeneracy:
– Finding ψEFF requires (x*,y*,f)
– Finding (x*,y*,f) requires ψEFF
• Iteration
– Dithers break the degeneracy!
Andrea BelliniHST high-precision proper motions of globular clusters
ISSUE#1: High-level PSF issues
•Spatial variability:
- Core intensity varies up to ±10%
over scales of ~500 pixels.
•Time variability (breathing)
- Core intensity varies up to ±10%
from one exposure to the next
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#2: Geometric distortion
• Part II!
(techniques)!
!
Part III
(the project) Why? Fewer reflections, better throughput
!
Part IV
(results)
• Linear “skew”: 500 pixels over 2000
→ Parallelogram pixels
• Non-linear: 50 pixels over 2000
• Filters introduce distortion (~0.1 pixel)
• Detector “stitching” defects
- WFPC2: every 34.1333th row 3% shorter
- ACS/WFC: pattern every 68.2666th column
- WFC3/UVIS: 2-D zones
!
Need empirical approach: plot everything against everything else…
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#2: Geometric distortion
• Part II!
(techniques)!
!
Part III WFC3/UVIS observed distortion
(the project)
!
Part IV
(results)
~10 pix rms
Andrea Bellini Bellini & Bedin 2009, PASP, 121, 1419HST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#2: Geometric distortion
• Part II!
(techniques)!
!
Part III auto-calibration & polynomial solution
(the project)
!
Part IV
(results)
~10 pix rms ~0.03 pix rms
Andrea Bellini Bellini & Bedin 2009, PASP, 121, 1419 Bellini, Anderson & Bedin 2010, PASP, 123, 622HST high-precision proper motions of globular clusters Part I (intro) ! ISSUE#2: Geometric distortion • Part II! (techniques)! ! Table-of-residuals correction Part III (the project) Astrometric flat-field to the < 0.01-pixel level ! Part IV (results)
HST high-precision proper motions of globular clusters Part I (intro) ISSUE#3: CTE defects ! • Part II! Affect ALL CCD detectors (ground- and space-based) (techniques)! ! Increase with time Part III (the project) Are a function of stellar brightness, chip position, and background level ! Part IV (results) Andrea Bellini
HST high-precision proper motions of globular clusters
Part I
(intro) ISSUE#3: CTE defects
!
• Part II! Affect ALL CCD detectors (ground- and space-based)
(techniques)!
! Increase with time
Part III
(the project) Are a function of stellar brightness, chip position, and background level
!
Part IV
(results)
Anderson & Bedin (2010)
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#4: Transformations
• Part II!
(techniques)!
!
Part III All HST astrometry is differential astrometry
(the project)
! • Guide-star precision ~ 0.5″ (improved from 1.5″!)
Part IV
(results) • No reference stars in typical fields
• We never know the true pointing
Always need to define a local reference frame
• Pixels/positions have only relative meaning
• Choosing a frame
* Base it on a population of objects (3+) in the frame
* Must know a priori something about the population
→ absolute μ = 0 (galaxies)
→ average μ = same (clusters)
→ average μ = unchanging (field)
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! ISSUE#3: Transformations
• Part II!
(techniques)!
!
Part III
(the project)
!
Part IV
(results) • Errors in the transformations:
• “Point” associations are not perfect:
(Xn,Yn ; Un, Vn)
- Stars’ measurement error
- Proper motions (dispersion)
- “Fuzzy handles” for galaxies/faint stars
• Distortion not perfectly removed
Make transformations more local
VSYST = σ/√N
Andrea BelliniHST high-precision proper motions of globular clusters
!
!
!
!
!
• Bonus!
!
!
!
!
Courtesy by Jay Anderson
WFI ACS WFC3 PM
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! The project
Part II
(techniques) 23 GCs:
! 4 (PI: Chandar)
• Part III!
(the project)! 4 (PI: Brown)
! 3 (PI: Ford)
Part IV
(results)
1 (PI: van der Marel)
1 (PI: Chanamé)
10 from the archive
(PI: Bellini)
!
!
!
Heterogeneous
datasets:
!
- different epoch
coverage
-different cameras
-different filters
-different S/N
!
!
Homogeneous
Andrea Bellini reductionHST high-precision proper motions of globular clusters
Part I
(intro)
!
Part II Homogeneous reduction:
(techniques) !
! 1- Need of reference frames with similar properties
• Part III!
(the project)! 2- Single-exposure catalogs obtained with the same software and procedures
! 3- Each star position must define a stand-alone epoch
Part IV
(results)
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
!
Part II Homogeneous reduction:
(techniques) !
! 1- Need of reference frames with similar properties
• Part III!
(the project)! 2- Single-exposure catalogs obtained with the same software and procedures
! 3- Each star position must define a stand-alone epoch
Part IV
(results)
GO-10775 GC Treasury Program (PI: Sarajedini)
• 65 MW GCs cores (central 3’x3’)
• Properly-dithered exposures
• Homogeneous and deep F606W
and F814W photometry 100’’
• All taken in 2006 50’’
25’’
NGC 6752 F606W Stack
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
!
Part II Homogeneus reduction:
(techniques) !
! 1- Need of a common reference frame
• Part III!
(the project)! 2- Single-exposure catalogs obtained with the same software and procedures
! 3- Each star position must define a stand-alone epoch
Part IV
(results)
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
!
Part II Homogeneus reduction:
(techniques) !
! 1- Need of a common reference frame
• Part III!
(the project)! 2- Single-exposure catalogs obtained with the same software and procedures
! 3- Each star position must define a stand-alone epoch
Part IV
(results)
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! Mining down systematics effects
Part II !
(techniques)
! 1- The Master frame is not perfect
• Part III! 2- Our corrections are not perfect
(the project)! 15
!
Part IV
(results)
Figure 9. Impact of uncorrected CTE effects on the GO-10775 NGC 7078 master frame. Panel (a) shows the mF606W vs.
stars into 4 magnitude regions, labeled R1, R2, R3 and R4. For each region we computed the locally-averaged differen
X and Y positions and those predicted by our PM fits at the epoch of the master frame. Panels (b1) to (b4) illustrate the
as a function of the stellar location on the master frame, for the magnitude regions R1 to R4. Panels (c1) to (c4) similarl
the Y axis (∆Y). Points are color coded according to the size of the differences. A footprint of the typical location of the
is also shown in black, with individual amplifiers separated by a red line. A strong correlation between the pattern of po
evident. Panels (d1) to (d4) illustrate the position differences on a rotated reference system, so that the rotated Y′ axis
GO-10775 ′ residuals are highlighted by a red line. The fact that these residuals are stro
Figure 9. Impact of uncorrected CTE effects on theexposures. The averaged
GO-10775 NGC ∆Yframe.
7078 master Panel (a) shows the mF606W vs. mF606W − mF814W CMD. We divided the
stars into 4 magnitude regions, labeled fainter
R1, R2,magnitudes
R3 and R4.isFor
a clear
each signature
region weof unaccounted
computed for CTE losses.difference between the GO-10775 master-frame
the locally-averaged
X and Y positions and those predicted by our PM fits at the epoch of the master frame. Panels (b1) to (b4) illustrate these differences for the X positions (∆X)
as a function of the stellar location on the master frame, for the magnitude regions R1 to R4. Panels (c1) to (c4) similarly show theIndifferences
an absolute sense,
in position along the
ally ACS/WFC
the Y axis (∆Y). Points are color coded according to the size of the differences. A footprint of the typical location of the GO-10775 very small. In fact, 5
chip placements
is also shown in black, with individual amplifiers separated by a red line. A strong correlation between the pattern of position have
differences and the chip
locally-averaged layoutPMs
is s
′
evident. Panels (d1) to (d4) illustrate the position differences on a rotated reference system, so that the rotated Y axis is parallel to the−1raw Y direction of the
pixelyrwithfor
GO-10775 exposures. The averaged ∆Y′ residuals are highlighted by a red line. The fact that these residuals are strongly correlated Y′ the X andatthe
and increase
fainter magnitudes is a clear signature of unaccounted for CTE losses. a reference, we recall that w
bright, unsaturated
In an absolute sense, the systematic trends are stars in ea
gener-
ally very small. In fact, 50% ofcision of ∼in0.01
the stars our pixel.
catalogNev
have locally-averaged PMs smaller thethan
master frame
0.0011 andwhere
0.0004the s
Figure 10. Rotated ∆Y′ position offsetspixelyr theXY′and −1
Andrea Bellini as a function
−1
forofthe position ∼ 0.008 pixelyr
the Y component, . TheAs
respectively. availa
using the original GO-10775 positions as the master frame (black, same asHST high-precision proper motions of globular clusters
Part I
(intro)
! Mining down systematics effects
Part II !
(techniques)
! 1- The Master frame is not perfect
• Part III! 2- Our corrections are not perfect
16
(the project)!
!
Part IV
(results)
Andrea Bellini
Figure 11. The top panels show two-dimensional maps of the locally-averaged µα cos δ (a) and µδ (b) components of the PM, as a function of positions withHST high-precision proper motions of globular clusters
Part I
(intro)
! NGC 7078 (M 15) PM Catalog overview
Part II
(techniques)
!
• Part III!
(the project)!
!
Part IV
(results)
Andrea Bellini
Bellini et al. 2014, submitted to ApJHST high-precision proper motions of globular clusters
Part I
(intro)
!
Part II
(techniques)
!
Part III
(the project)
M 15
!
• Part IV!
(results)
Part IV: Scientific results
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! Internal motions
Part II
(techniques)
!
Part III (a) (b) (c)
(the project)
M 15
!
• Part IV!
(results)
(d) (e) (f)
Andrea Bellini
Bellini et al. 2014, submitted to ApJHST high-precision proper motions of globular clusters
Part I
(intro)
Cluster rotation (NGC 104)
!
Part II
(techniques)
!
Part III
(the project)
!
• Part IV!
(results)
Andrea Bellini Bellini et al., in preparationHST high-precision proper motions of globular clusters Part I (intro) ! Multiple-population kinematics (NGC 2808) Part II (techniques) ! Part III (the project) ! • Part IV! (results) Andrea Bellini Bellini et al., in preparation
HST high-precision proper motions of globular clusters
Part I
(intro)
! Absolute motion (NGC 6681)
Part II
(techniques)
!
Part III
(the project)
!
• Part IV!
(results)
Fig. 7.— VPD of the absolute PMs. The red dots indicate the selected background galaxies (see also Fig. 5), who
mean motion corresponds to the zero point of the VPD. The blue ellipses are centered on the measured absolute PM
of the three populations (marked with a blue cross) and their size corresponds to the calculated 68.3% confiden
region. The black arrows indicate their absolute PM vectors. In the proximity of the Sgr dSph estimate, the PM va
predicted by Law & Majewski (2010) is shown as a light green dot, while the Pryor et al. (2010) and Dinescu et
(2005) measurements and their 68.3% confidence regions are shown as magenta and dark green ellipses, respective
Finally, the cyan ellipse describes the prediction on the PM of the field population by the Besançon model, wh
differs from our estimate obtained using the stars in the same magnitude and color range (marked with black cross
see the text for the selection criteria).
Fig. 3.— The upper panels show the Vector Point Diagrams (VPDs) of the relative PMs. In the lower panels the
Massari, Bellini, et al. 2013, ApJ, 779, 81M
CMDs corresponding to the selections applied in the VPDs are displayed. First column: in the VPD the different
is: We compared this value with previous
populations are indicated with different colors (a sample of cluster members in blue, of Sgr dSph stars in red, of the
−1
field in green), but no selection is applied. The corresponding CMD shows the entire PM catalog. Second column: (µ inα cos δ, µδ ) = (−2.54±0.18, −1.19±0.16) mas yr .
(3)
the VPD cluster members are selected within the blue circle and the corresponding CMD displays only well-defined 10
cluster evolutionary sequences. Third column: Sgr dSph selection within the red circle and corresponding CMD.
Andrea Bellini Fourth column: the selection in the VPD (in green) of the bulk-motion of field stars and their location on the CMD.HST high-precision proper motions of globular clusters
Part I
(intro) Dispersion profiles / (An)isotropy x0 [arcsec]
! R [arcsec] R [arcsec]
Part II fit
1
/10 Rhalf Rcore Rhalf fit
1
/10 Rhalf Rcore Rhalf
0.568
(techniques) 0.70
1.6 50
! 0.65
1.4
y 0 [arcsec]
0.60
[km/s]
[mas/yr]
Part III 0.55
1.2
0
t/ r
(the project) 0.50 1.0
! 0.45 0.8
50
• Part IV! 0.40
0.289
0.6
0.35 NGC 6266 NGC 6266 NGC 6266
(results) 0.4
100 101 102 100 101 102 100 50 0
10 R [arcsec] L. L. WATKINS ET AL .
R [arcsec] 0
x [arcsec]
Figure 8. The same as Figure17 for NGC 5139, NGC 5904, NGC 5927 and NGC 6266.
1
fit /10 Rhalf Rcore fit /10 Rhalf Rcore
0.674
100
we consider
0.65 all three predictions here. 1.6 to km s-1 . Once again, the dispersions are generally in good
We show the results of our comparisons in the bottom
1.4 panel agreement, as most points50fall along the 1:1 correspondence
y 0 [arcsec]
[km/s]
of Figure
0.60 13; the King model predictions are shown as red line with little scatter. However, we note that now we tend to
[mas/yr]
1.2
points, the Wilson model predictions as green points and underestimate the central dispersions
0 more than we overesti-
t/ r
the power-law
0.55
model predictions as blue points.1.0The dotted mate.
line highlights where our central dispersion estimates
0.8 and the We consider briefly the 50 two outliers from the Harris (2010)
McLaughlin & van der Marel (2005) model predictions are comparison. Unlike before, our estimate for NGC 6715 is in
0.420
0.6
0.50
equal. As NGC
before,
0104we use the distance estimates0.4 from
NGCHar-
0104 very good agreement with 100theNGC
McLaughlin
0104 & van der Marel
ris (2010) to convert 1our dispersions estimate
2
from mas yr-1 (2005) predictions; this lends100
further
50
weight
0
to our
50 100
assertion
10 10 101 2 10
R [arcsec] HST R
GC S
[arcsec] x0 [arcsec] 11
1 1
fit /10 Rhalf Rcore Rhalf fit /10 Rhalf Rcore Rhalf
0.869
150
1.0 1.6 100
1.4 50
y 0 [arcsec]
0.9
[km/s]
[mas/yr]
1.2 0
t/ r
0.8 1.0 50
0.8 100
0.7
0.491
0.6
150
0.6 NGC 5139 NGC 5139 NGC 5139
0.4
200
100 101 102 100 101 102 100 0 100 200 300
0
R [arcsec] R [arcsec] x [arcsec]
Watkins L. L., Bellini A., et al., in preparation
Andrea BelliniHST high-precision proper motions of globular clusters
Part I
(intro)
! Figure 12. The same as Figure 7 for NGC 7099. Vanilla distances
Part II
25 25
(techniques) 6388
King
! Wilson
6441
Part III
20 6388 20 power-law
(the project) 6441 2808 6266
! 6715
[km/s]
104
[km/s]
• Part IV! 5139
15 7078 15
(results) 6266
7078
362 5139
McLaughlin
2808
Harris
1851
10 104 5904 1851
6715 10 6341
6656
362 6752 6656
6624 6341 6681
5 6397 5904 6397
6681 6752 7099 5
6362 7099
288
6362 288
6535 vs. Harris (2010) vs. Mclaughlin & van der Marel (2005)
6535
0 0
0 5 10 15 20 25 0 5 10 15 20 25
centre [km/s] centre [km/s]
Watkins L. L., Bellini A., et al., in preparation Figure
Figure 14. The ratioofofour
13. Comparisons thecentral
dispersion measured
dispersions withatliterature
the core radius
estimates. core t
that measured at the half-light radius as a function of cluster
Top: central dispersion from Harris (2010) vshalfour central dispersion. The line concentra
tion, the
for which as listed in Harris
estimates (2010).
are equal The red
is marked as line shows
a dotted a straight-line
line. fit to th
Bottom: central
data. from
dispersion half is a proxy
core /McLaughlin for der
& van the Marel
slope of the dispersion
(2005) profile. There is
vs our dispersion.
clear correlation such that more concentrated clusters have steeper dispersio
profiles, as expected.
where ✏ = 0 indicates a perfectly round cluster and increasing
✏ indicates increasingly
culate the flattening
one-dimensional clusters. dispersion
major-axis The minor-majormajor an
anisotropy reflects
minor-axis the difference
dispersion in the velocity dispersions
minor profiles in the same bins tha
alongwere
the used
majorforandtheminor axes,
analysis of where minor / major = kinema
the one-dimensional 1 in-
dicates isotropy,
ics in Section 4.1. / major
minorThen 1 indicates anisotropy
the 1 indicatesvalu
/ majorrepresentative a
preference
againstfor motion
which along theellipticities,
to compare minor axis.we took the weighte
Tomean
determine
of thethe major-axisusing
anisotropies, and minor-axis
the inverse directions, we
square uncertain
Andrea Bellini use position angles from White & Shawl (1987). We then cal-
ties as weights.HST high-precision proper motions of globular clusters
Part I
(intro)
! Equipartition (?)
Part II
(techniques)
!
Part III
(the project)
!
• Part IV!
(results)
vs. Harris (2010)
Andrea BelliniHST high-precision proper motions of globular clusters
Conclusions
High-precision astrometry with HST is challenging but DOABLE
- undersampling (PSF)
- geometric distortion
- differential nature (local transformations)
!
!
Scientific projects with HST’s proper motions of GCs
- Internal kinematics
- dispersion profiles
- anisotropy
- rotation
- IMBHs
- …
!
Our project
- high-precision (~1 km/s) proper motions in the cores of 22+ GCs
- preliminary results encouraging and exciting
- proper-motion catalogs will be made available to the astronomical community
- Future extension to 60+ GCs thanks to new observations
Andrea BelliniYou can also read
