Low frequency sky surveys with the Murchison Widefield Array (MWA) Gianni Bernardi Harvard-Smithsonian Center for Astrophysics SKA SA project/MeerKAT observatory IRA-INAF, February 4 th 2013

The MWA is an SKA low frequency precursor: low frequency ( MHz: 32 MHz bandwidth), large-N (correlation rich) array, 3km maximum baselines; Science cases: search for the 21cm emission from the Epoch of Reionization; search for the 21cm emission from the Epoch of Reionization; Solar and heliospheric science; Solar and heliospheric science; radio transients; radio transients; Galactic and extragalactic science (this talk today); Galactic and extragalactic science (this talk today); Aperture array antenna elements, 4x4 arrays of dual polarization dipole – “tiles”; Initially 128 tiles, expandable to 256

350m : 32 element (32T) prototype

Survey concept and calibration The drift scan maintains the tile primary beams constant with time and all equal to each other, because all the dipoles have zero delay (do not underestimate this simplification!); J ( MHz, α = -0.81, a few arcmin in size at 1.4 GHz) is used as flux, phase and passband calibrator; Each 5 min uv data set (“snapshot”) is calibrated from the J solutions and imaged. All the snapshots are mosaiced together and deconvolved jointly; Polarization leakage is less than 4% over 20º in average less than 1.8% (Calibration: Mitchell et al. 2008, Imaging: Ord et al. 2010, Deconvolution: Bernardi et al. 2011) The primary beam model fitted to the data is good at the 2% level over MHz;

α δ 0h0h 23 h 22 h 1h1h 2h2h 3h3h 4h4h 5h5h -30 º -15 º 6h6h A MHz drift scan survey with the 32T array (Bernardi et al. 2013, ApJ, submitted) MHz, 15.6 arcmin resolution, 8 hour integration, 2400 square degrees confusion limited at 200 mJy/beam (polarization noise ~ 15 mJy/beam)

α α α δ δ δ 0h0h 23 h 22 h 1h1h 2h2h 3h3h 4h4h 5h5h 0h0h 23 h 22 h 1h1h 2h2h 3h3h 4h4h 5h5h 0h0h 23 h 22 h 1h1h 2h2h 3h3h 4h4h 5h5h -30 º -15 º -30 º -15 º -30 º -15 º Stokes I Stokes Q Stokes U

The bright source sample (calibration accuracy) A catalogue of 137 unresolved sources brighter than 4 Jy (29 sources measured for the first time below 200 MHz) Comparison with 160 MHz measurement (Slee 1977) from the literature: 98% of sources matched, 19% rms difference above 4 Jy. The average spectral index between and 160 MHz is α = ± 0.17

Polarization: RM synthesis 1) Linear polarization vector: P = Q + iU = p I e 2i  where I, Q, U (V) are the Stokes parameters, p = % polarization, and  = 0.5 atan(U/Q) is the polarization angle 2) When observing the polarized power P at a range of 2 we can define: where F(  ) is the complex polarized power per unit Faraday depth  first defined by Burn (1966), and W( 2 ) is the window function 3) This relation can be Fourier inverted to yield F(  ) The quantity F(  ) is convolved with a response function, called the RMSF, which is the Fourier transform of the window function W( 2 ) in 2 space. The output of the RM synthesis is a cube of images in Faraday depth space with 4.3 rad m 2 resolution

ϕ = 0 rad m 2 ϕ = +2 rad m 2 ϕ = +4 rad m 2 ϕ = +6 rad m 2 ϕ = +8 rad m 2 ϕ = +34 rad m 2 PMN J % polarized

Where are polarized AGNs? Radio sources at 1.4 GHz have an average polarization fraction of ~ 7%, with peaks up to 20% (Taylor et al. 2009); A 4 Jy source, 7% polarized  ~18σ detection… ionospheric Faraday rotation  15% depolarization why do we see only one polarized source? 20% 1.4 GHz, MHz GHz (polarized intensity) MHz Stokes I significant RM variations on scales smaller than the sources size (within the synthesized beam) significant RM variations on scales smaller than the sources size (within the synthesized beam) lead to beam depolarization: J : p’ ~ 18  σ RM ~ 0.5 rad m 2 very plausible also for smaller depolarization fractions

Where do RM variations occur? “Small scale variation in the Galactic Faraday rotation”, Leahy,1987, MNRAS, 226, samples of 3C sources variations due to faint HII regions along the line of sight (Galactic foreground)

What is the origin of the diffuse polarization? Most of the diffuse polarization at low frequencies has no counterpart in total intensity, originated by small scale structure in the ISM which rotates a fairly smooth polarized synchrotron background n e, B uniform Stokes Q background n e, B ISM clouds emerging Stokes Q with structure on the cloud size Stokes Q structures detected against a uniform background which remains resolved out

What is the origin of the diffuse polarization? Most of the diffuse polarization at low frequencies has no counterpart in total intensity, originating by small scale structure in the ISM n e, B uniform Stokes Q background n e, B ISM clouds emerging Stokes Q with structure on the cloud size Stokes Q structures detected against a uniform background which remains resolved out WSRT observations at 350 MHz, 5 arcmin resolution (Haverkorn, Katgert & de Bruyn, 2003) Total intensity Polarized intensity

What is the origin of the diffuse polarization? WSRT observations at 150 MHz, 4 arcmin resolution (Bernardi et al. 2009) Total intensity Polarized intensity

What is the origin of diffuse polarization? Polarized 188 MHz Polarized 1.4 GHz (Gaensler et al. 2011) Low observed RM values ( < 15 rad m 2 ) indicate that the emission should be more local than pc. Confirmed by the comparison with pulsars of known RM. LOCAL ISM Magnetized, subsonic turbulence in the local, diffuse ionized gas is able to generate a complex filamentary web of discontinuities in gas density and magnetic field (Gaensler et al. 2011)

The only resolved source: Fornax A Lanz et al two X-ray cavities Either the 1.4 GHz radio image does not account for the full radio emission or the central SMBH generated at least two outbursts. This question can be answered by high brightness sensitivity observations, especially at low frequencies

An MWA 32T image of Fornax A integrated MHz: ~ 519 ± 26 Jy Fornax 1.4 GHz with the VLA 1.4 GHz (VLA) Fornax A is four beams across  not enough resolution to claim the existence of a bridge at low frequencies, need to wait for the 128T

Polarization in Fornax A? Fornax 1.4 GHz with the VLA Fornax 1.4 GHz (VLA) Polarization fraction at 1.5 GHz, 22” resolution (Fomalont et al. 1989): dark is 40-65%, average is 20% Depolarization: region 5 is due to a foreground elliptical galaxy which belongs to the cluster; unknown the rest No polarized emission detected at MHz Polarization fraction must be less than 1% (set by the polarization calibration) Easily explained by beam depolarization

MWA and all sky survey program development Deployment of the full array started in August 2012; Deployment of the full array started in August 2012; New receivers, new hybrid CPU-GPU correlator; New receivers, new hybrid CPU-GPU correlator; Commissioning of the full array started in August 2012; Commissioning of the full array started in August 2012; For practical reasons, the array was divided in four sub-arrays of 32T each (separate deployment and commissioning); For practical reasons, the array was divided in four sub-arrays of 32T each (separate deployment and commissioning); Deployment completed in December 2012; Deployment completed in December 2012; Full array operations (all the 128 tiles simultaneously) expected to start in February 2013; Full array operations (all the 128 tiles simultaneously) expected to start in February 2013; First call for proposal (including open sky) later in the year (see First call for proposal (including open sky) later in the year (see

beta array all-sky survey (114 MHz, 40 kHz bandwidth, max baselines ~380m). Courtesy N. Hurley-Walker

gamma array drift-scan survey (114 MHz, 40 kHz bandwidth, max baselines ~2.9 km). Courtesy N. Hurley-Walker

Conclusions We have conducted a 2400  survey with the MWA 32 element array with 16 arcmin resolution at 189 MHz: Total intensity images are limited by confusion at ~200 mJy/beam. A source catalogue only marginally improves over previous measurements Total intensity images are limited by confusion at ~200 mJy/beam. A source catalogue only marginally improves over previous measurements Drift scans have been demonstrated to be a very effective observational strategy: beam stability, gain (fairly) stability, relatively easy to calibrate and to obtain full polarization images  to be employed for the 128 all-sky survey; Drift scans have been demonstrated to be a very effective observational strategy: beam stability, gain (fairly) stability, relatively easy to calibrate and to obtain full polarization images  to be employed for the 128 all-sky survey; We detected polarization from only one catalogue source: a comparison with their cm- wavelength polarization fraction indicate that they are likely to be beam depolarized We detected polarization from only one catalogue source: a comparison with their cm- wavelength polarization fraction indicate that they are likely to be beam depolarized A wealth of diffuse polarization across almost the whole area at low RM values with peaks up to ~20 K/RMSF  tracing turbulence in the local (< 150 pc) magnetized, diffuse, ionized gas A wealth of diffuse polarization across almost the whole area at low RM values with peaks up to ~20 K/RMSF  tracing turbulence in the local (< 150 pc) magnetized, diffuse, ionized gas The all sky survey program has continued during commissioning and it will receive dedicated observing time in 2013 with the full 128T array (~200 h, full MHz coverage, a survey team and theme, lead by Dr. R. Wayth) The all sky survey program has continued during commissioning and it will receive dedicated observing time in 2013 with the full 128T array (~200 h, full MHz coverage, a survey team and theme, lead by Dr. R. Wayth) THANK YOU