Foregrounds and Redshifted 21 cm Observations

1
Foregrounds and Redshifted 21 cm Observations

• Judd D. Bowman (Caltech)
• July 14, 2008

2

• Charge To cover the entire arena of EoR/IGM
  and how it relates to foregrounds

• 21 cm reionization science
• Foregrounds and mitigation
• Experiments
• Latest results from simulations

3
History of the IGM
Image Scientific American 2006
4
1. 21 cm reionization science
5
21 cm reionization science

• Goal Use the 21 cm hyperfine transition line of
  neutral hydrogen to probe the high-redshift IGM

CMB Backlight z 1100
z 0
HI Screen z 10
6
21 cm Overview

• Spin-flip hyperfine transition of neutral
  hydrogen
• ?21 cm (rest-frame), ?1420 MHz
• At z 6 ?1.5 m, ? 203 MHz
• At z 15 ?3.4 m, ? 89 MHz
• Use CMB as a backlight, then brightness
  temperature is

7
Pritchard Loeb 2008
8
?Tb Perturbations
?x 400 cMpc ?? ? 16 degrees ?? ? 60
MHz ?z ? 3.5
z 8, xi 0.3 Cube provided by A. Mesinger
S. Furlanetto
9
?Tb Evolution
Early times (z gt 15)
Primarily density fluctuations
5 arcmin
Ionized regions
Late times (z lt 6)
Furlanetto et al. 2004
10
?Tb Power Spectrum
(50 ionized)
(70 ionized)
parallel
perpendicular
(fully neutral)
(10 ionized)
McQuinn et al. 2006
11
?Tb Amplitude vs. Redshift
Pritchard Loeb 2008
12
Inflationary/fundamental physics
z 100
13
Low-z 21 cm (zlt6)
HIPASS
z 0
6dFGS
Pen et al. 2008 (submitted)
14
21 cm landscape - science

• Inflationary physics and cosmology 30 lt z lt 200
  46 gt ? gt 7 MHz
• Probe of the matter power spectrum at very small
  scales l gt 104 to 106
• Perturbations to primordial power spectrum and
  spatial curvature ns , ?
• Neutrino masses, non-Gaussianity, etc.
• Baryon collapse

• Reionization and the Dark Ages 6 lt z lt 30
  203 gt ? gt 46 MHz
• Spin and kinetic temperature history of the IGM
• Reionization history, Stromgren spheres
• Star formation history / models for ionizing
  sources
• Abundance of mini-halos
• Magnetic fields in IGM
• Cosmology

• Large scale structure/galaxy evolution z lt 6
  1420 gt ? gt 203 MHz
• Dark Energy through BAOs, cosmology, neutrino
  masses, etc.
• HI in galaxies/halos, masses of DLAs at z 3
• Indirectly see helium reionization

15
21 cm landscape - techniques

• Inflationary physics and cosmology 30 lt z lt 200
  46 gt ? gt 7 MHz
• Power spectra
• (3D, more modes than CMB, linear
  perturbations)

• Reionization and the Dark Ages 6 lt z lt 30
  203 gt ? gt 46 MHz
• Mean (global) brightness temperature evolution
• Power spectra (3D, intensity, aspherical
  components, polarization)
• Tomography / imaging / Stromgren spheres
• Cross-correlations (CMB, galaxies)
• Absorption toward discrete sources (AGN,
  star-forming galaxies, GRBs)
• Weak lensing (of order 1 effect in power
  spectrum)

• Large scale structure/galaxy evolution z lt 6
  1420 gt ? gt 203 MHz
• Power spectra from unresolved emission
• Cross-correlations (galaxies)
• Absorption toward discrete sources
• HI galaxy-redshift surveys (1 billion galaxies)
• Weak lensing

16
2. Foregrounds and Mitigation
17
Astrophysical foregroundsat 150 MHz

• Galactic continuum emission 200 to 10,000 K
  (70)
• Synchrotron (99)
• Free-free (1)
• Galactic radio recombination lines lt 1 K
• 10 kHz wide and every 1 MHz
• Extragalactic point sources 30 to 70 K (25)
• Radio galaxies, normal galaxies, AGN, radio
  haloes, radio relics
• Extragalactic free-free emission
• (21 cm 25 mK ? 1 part in 104)
• Very few applicable direct observations, must
  rely on models, simulations, and extrapolations
  from other frequencies

Shaver et al. 1999
18
Signal vs. foregrounds (angular)
135 MHz
Annotated by D. Backer
19
Signal vs. foregrounds (freq)
Bennett et al. 2003
Pritchard Loeb 2008
20
Foreground separation

• Intrinsic sky properties
• All foreground components have smooth spectral
  structure (large spectral coherence)
• Extragalactic recombination lines not significant
  (Oh Mack 2003)
• Removal of sources to Scut 0.1 mJy reduces
  extragalactic foregrounds to 21 cm level (Di
  Matteo et al. 2004)
• Extragalactic free-free dominated by zlt3, only
  correlates with redshifted 21 cm at 1 level
  (Cooray Furlanetto 2004)
• Instrumental response
• PSF changes with frequency
• Calibration errors
• Polarization leakage

21
Spectral coherence/subtraction
Foregrounds
21 cm
(??1 MHz)
Santos et al. 2005
Wang et al. 2006
22
Foreground mitigation strategy

• Bright sources Deconvolution of the bright
  sources in the sky with calibration solutions
• Out-of-beam sky Subtraction of sources from
  outside the primary beam, both point and diffuse
  without calibration solutions
• Confusion level sources Subtraction of faint
  sources/diffuse emission which are at or near the
  confusion limit within the primary beam.
• Polarized foreground Subtraction of the
  polarized foreground, which leaks into the Stokes
  I map via small mis-calibrations. Exploit
  rotation measure synthesis to separate leakage
• Sub-confusion level contaminants Template
  fitting of residual power spectrum components

Morales, Bowman Hewitt 2006
23
3. Experiments
24
Experiments

• Foregrounds are the reason for new instruments
• Focus on limiting systematics to allow foreground
  subtraction
• Sky noise dominates system temperature
• Pathfinder experiments under construction
• Global Fluctuations
• EDGES (Rogers Bowman) MWA
  (MIT/CfA/RRI/Australia)
• CoRE (Ron Ekers) LOFAR (ASTRON)
• GMRT (Ue-li Pen)
• PAPER (Don Backer)
• General approach Start from scratch with new
  instruments that exploit modern digital signal
  processing technology to address challenges

25
Radio frequency interference
Annotated by F. Briggs
26
Ionosphere
10 arcmin distortions 10 second
coherence time-scale
D. Mitchell/MWA-RTS Team
27
Global Experiments
EDGES
CoRE
ADC
Four-point antenna
Ground screen
28
EDGES all-sky spectrum
1.5 sky hours 75 mK residuals after 7th-order
polynomial fit
lt 450 mK redshifted 21 cm instantaneous
reionization contribution
Bowman, Rogers Hewitt 2008
29
Giant Metre-wave Radio Telescope (GMRT) EoR
Project

• Started in 2005 (CITA, Ue-li Pen)
• 30 x 45 meter antennas, 14 in central 1 km core
• 150 MHz feeds, 16 MHz bandwidth
• New GMRT Software Backend
• 16 commercial ADC boards (4 inputs/board)
• 16 computer nodes - dual quad core 2.33 GHz
• 7.8 kHz spectral resolution

Pen et al. 2008
30
GMRT-EOR Status

• 100 hours of test observations acquired
• Significant RFI mitigation improvements
• Polarization calibration using pulsar
• Recent results
• Diffuse polarized emission at 150 MHz lt 0.2 K rms
  (Pen et al. 2008, submitted)

31
Precision Array to Probe the Epoch of
Reionization (PAPER)

• Start in 2004 (Don Backer et al.)
• 100 to 200 MHz
• Individual, dual polarization dipoles
• Aeff 7 m2 ? Atot 1500 m2
• PAPER in Green Bank PGB
• 2-antenna interferometer in 2004 August
• 8-antenna array in 2006
• 16-antenna deployment this month half with dual
  polarization, test new balun design, longer
  baselines (current max at 300m)
• PAPER in Western Australia PWA
• 4-dipole array deployed 2007 July
• 32-antenna deployment 2008 October
• Annual campaigns growing array up to 256-antenna

D. Backer
32
D. Backer
33
PAPER/CASPER Packetized Correlator
Berkeley Wireless Research Center's BEE2
Two Dual 500 Msps ADCs IBOB
A. Parsons, J. Manley
34
PAPER GB7 all sky map
Aaron Parsons
35
Murchison Widefield Array (MWA)

• Collecting area 8000 m2 (1 SKA)
• Spectral coverage 80 to 300 MHz
• Instantaneous bandwidth 32 MHz (?z 2)
• Spectral resolution 10 kHz (40 kHz)
• 512 antenna tiles within 1.5 km diameter
• Field of view 100 to 1000 deg2
• Angular resolution 3 to 10 arcmin
• Sky noise dominated
• Key science
• EOR, sun/heliosphere/ionosphere, transients

36
MWA antenna tile
1
2
3
37
MWA system diagram
38
MWA status

• First fringes w/ 8 tiles
• 32 antenna tiles installed at site
• Receivers and correlator in progress
• Sub-system tests through June 2009
• Build to full array 2009 H2
• EoR observing 2010

39
MWA antenna distribution
Antenna layout
Baseline distribution
Rotation synthesis

• 125000 baselines
• 10 in tightly packed core
• Completely sample uv-plane within 500
  wavelengths
• Short baselines probe both large and small
  spatial scales

40
MWA thermal uncertainty
z 6
z 8
z 10
z 12
B 8 MHz t 360 hours
Bowman, Morales Hewitt 2006
41
Low Frequency Array (LOFAR)

• Being built in Netherlands/Europe
• General purpose cosmic rays, solar, magnetic,
  surveys, transients, EoR
• Low band 30 80 MHz
• High band 110 240 MHz
• EoR will use high band antennas in central core
• 18-24 stations x 2 substations x 24 antenna tiles
  per substation (substation diameter 35 meter)
  960 tiles
• 1250 baselines, 40 m to 2 km
• 32 MHz bandwidth, 0.7kHz resolution
• Target 115 to 180 MHz
• FOV 5 degrees

M. Brentjens / Jelic et al. 2008
42
LOFAR status

• Tiles covered due to weather (rodents)
• Start building stations in September 2008
• 13 core stations by April 2009
• Calibration survey in July 2009 (million
  sources, shallow)
• Full core by December 2009
• Science 2010

43
3. Latest results from simulationsIncluding
instrumental responses
44
LOFAR foreground subtraction
Jelic et al. 2008
45
MWA dirty cubes (Ap, As)
dirty image cube
residuals in image cube, 2nd-order Polynomial fit
residuals uv cube
Bowman, Morales Hewitt 2008 (in prep)
46
MWA foreground subtraction
B 8 MHz t 360 hours
Foregrounds Noise
Noise only
Recovered 21 cm
Bowman, Morales Hewitt 2008 (in prep)
47
Subtraction parameter space
2D (angular) power spectrum

• Source peeling cutoff
• Rotation synthesis
• Polynomial fit order
• Tile arrangement
• Beam manipulations

Liu, Tegmark Zaldarriaga (in prep.)
48
Summary

• 21 cm landscape full of rewarding astrophysics
  and cosmology, but serious foreground challenges
  exist
• Intrinsic properties of low-frequency radio sky
  are known qualitatively and suggest foreground
  subtraction possible, but quantitative details
  remain largely unknown
• Pathfinder experiments for both global and
  anisotropy 21 cm signals are in progress to
  demonstrate foreground mitigation and to detect
  signal at z gt 6
• Simulation efforts have shown minimal impact of
  frequency-dependent PSF polarization leakage is
  next simulation goal