High Precision Astrometry and Parallax from Spatial Scanning

Part 2

Adam Riess and Stefano Casertano

Distortion at the milli-pixel levelSpatial Scanning Provides measurement 1D* precision of < 1 millipixwhich demands comparable control of distortions across frames

Starin

g

Scan

ning (Sta

tistics)

Paralla

x at 2

kpc

Intra-sc

an Velo

city A

berrati

on

Variab

le Jitt

er

Intra-sc

an Rotati

on (FGS)

Inter-sc

an Dist

ortion ("

Breathing")

Total

Inter-sc

an Rotati

on

Static D

istorti

on0

2

4

6

8

10

12

Final PrecisionScanning SystematicScanning or Staring Systematic

mill

ipix

els

*measurement direction always ΔX

Scale change and rotation of scan orientation• Rotation apparent even in back-to-back scans using same guide stars• Rotation of scan orient by a few arcsec, solved for in frame-to-frame registration• Scale change mostly time-dependent velocity aberration, we interpolate va(t) x,y• Each accounts for up to ~20 mpix (800 as) in measured trail separation

Sequential scans— correlation star pairs ΔY vs ΔX due to global scan axis rotations

Combining Frames: Scale and Rotation

Combining Frames: Low Order Distortions Residual geometric distortion

• Amplitude up to 5 mpix (200 as)• Typically well-fitted by registration with low (2nd / 3rd order) polynomial in ΔX (x,y)• After correction, residuals match to < 1 mpix (40 as) for long, high S/N trails• Polynomial corrections show correlation w/ model focus/breathing• Call this time-dependent or breathing dependent distortion, expected

Combining Frames: Fine StructureAlso see static, fine structure residuals along scan (Y) relative to X

• residuals of about 5 mpix scale correlated on 10^2 pixel scale• orbit phase, breathing independent as seen scans separated by min, hrs, up to 10 days• Only apparent between scans with large shifts or local to a scan line, i.e. Y vs. X-<X> • Averages down for long lines but important loss of precision for short lines

Residuals That Persist for Years

Comparing these static residuals in ΔX in X,Y spatial bins (100x100 pixels) across a year for M35 and Cepheid field stars shows strongcorrelation. Such residuals expected from PSF map (see Murphy’s Law)

In Cycle 22 we will map these residuals on 50 pix scale using 20 dithers of scans in M67,M48 to make a look-up table. Also useful for staring mode.

Deep scan (F606W)Cepheid saturated~40 usable trails

Shallow scan (F673W)Cepheid not saturated

~8 usable trails

A challenge for (bright) Cepheids: Dynamic range

Pair shallow and deep scans asline separation filter independent, anchor Cepheid to all reference stars

Current Progress: 40 epochs for 19 Cepheids… full residuals

Reduction to absolute from reference star distances using 12 bands+spectroscopy, e.g., σD=0.3 mag for red giant at 8 kpc => 18 mas, typical field ~ 10-12 mas

2.3 kpc(Riess et al 2014 ApJ, arXiv1401.0484R)

Proper Motion subtracted, Parallax measurements field stars & Cepheid

8 kpc( =p 125 μas15% error)

250 pc( =p 4 mas1.8% error)

Cepheid

First Results from Pilot Program…Sample of 19 underway

Simulatedresults for 20 CepheidsAfter 5 epochs,Cycle 20-22,twice number,precisionof FGS set +HST photometry+ log P > 1

Expect to reach s 0H ~1.8%when done, double current

Given field, catalog, MW model, can fully simulate multi-epoch scans, final uncertainty

Cepheid Parallax Measurements

The End

Reduction to absolute parallax

Each field contains 30-100 reference stars• Will obtain multiband photometry (UVIS broad + medium filters, WFC3 IR, 2MASS) +

spectra + stellar models to estimate individual distances• Typical uncertainty < 0.3 mag (15% in distance)

• One red giant at 8 kpc => 18 mas• All stars contribute; distant stars give best constraints• Estimated final uncertainty ~ 10-12 mas / field

Incorrect reduction to absolute produces systematic bias in the estimated distance modulus

Incorrect luminosity scale produces systematicbias in parallaxes of nearby objects