GALFACTS Team Ready to Go

Russ Taylor (U. Calgary) & Chris Salter (NAIC)

[No Frames Version]

(This article also appeared in the December 2007 NAIC Newsletter.)

Over the past year the GALFACTS data facility has been put in place and the code for the processing pipeline software written and exercised on test observations taken in essentially full GALFACTS mode. GALFACTS observed commensally with the GALFA - H I observing program A2174. That project, led by Lewis Knee and James Di Francesco of the Herzberg Institute of Astrophysics, was a 21-cm H I line observation using the GALSPECT spectrometer to image an 18 x 12 degree region of the Perseus molecular cloud complex. Those observations supplement the Spitzer ``Cores to Disks'' (C2D) legacy survey, and the related ``Co-ordinated Molecular Probes Line Extinction and Thermal Emission'' (COMPLETE) survey of this region. The goal of A2174 was to address the formation, evolution and destruction of molecular clouds, connecting stars to their atomic gas origins. The GALFACTS commensal observations, in addition to testing the GALFACTS processing pipeline, provide full-Stokes polarimetry of the region offering the potential of complementary information on the relationship between magnetic fields and the molecular and atomic gas in this region.

For the A2174 continuum observations, the WAPP spectrometer was used as backend, providing 100 MHz of bandwidth in 256 channels. For actual GALFACTS observations we will use the new spectrometer to cover a 300 MHz bandwidth. A switched (25 Hz) low-intensity noise-diode signal added to calibrate the electronic gains of the fourteen channels (two polarizations per beam). Our GALFA - H I partners graciously accepted the resulting 0.5 K increase in their system temperature. The observations were taken in fast meridian scanning mode as will be done for GALFACTS. The commensal drive program for the observations was specified by the GALFACTS team, and implemented in short order by Mikael Lerner (NAIC).

Following arrival in Calgary, the data are first run through an RFI detection and excision routine. The complex gains of the electronic signal chain are then measured and corrected for using the switched noise diode signal. This signal is also used to correct the band shape based on measurements of the frequency dependence of the noise diode amplitude using continuum calibration sources. The data, now on a Kelvin scale, are ``basket-weaved'' using the sky-crossing points of the ``UP'' and ``DOWN'' sets of meridian scans. This process achieves an internally-consistent zero level for each beam data set over all observing days, employing an extension and refinement of traditional single-pixel basket-weaving (Haslam, Quigley & Salter, 1970, MNRAS, 147, 405).

One of the final stages in the data reduction is to apply a ``Multi-beam CLEAN'' deconvolution algorithm to remove the effects on the imaging of the near-in side lobes of the ALFA beams. This is particularly critical for the large coma-lobes that exist for the six outer beams of the ALFA array. The technique has been developed by S. Guram and R. Taylor and is the subject of Guram's M.Sc. Thesis, recently completed at the University of Calgary. Because the images are constructed from scans made with all seven beams, the point source response function is dependent upon its position in the map -- varying depending on which of the seven ALFA beam tracks traverses the vicinity of the source. Multi-beam Clean is a variation on the Högbom (1974) CLEAN algorithm, which was developed for interferometers. In Multi-beam Clean the deconvolution iterates between the image and the time-domain data, removing the sidelobes in the time-series data based on a knowledge of the time-domain response function of each beam in the scan data, and the location of the sources (Clean components) in the map.

Multi-beam Clean requires an accurate mathematical model for each of the ALFA beams. These beam models must allow the response function in the time domain to be derived along a track that has arbitrary distance and orientation relative to the beam center. We have derived mathematical representations of the beams using an iterative technique that fits the PSF of each beam with linear combinations of Zernike polynomials. The Zernike polynomials model the PSF in terms of error terms in the telescope optics (amplitude and phase wavefront errors), which allow the aperture illumination and phase distribution to be derived.

Figure 1 shows the observed beam PSF of Beam 5, from an image of a calibration source, along with the model PSF from the Zernike polynomial fits. Figure 2 shows the corresponding Beam 5 aperture illumination function and aperture phase distribution. An example of Multi-beam Clean for Stokes-I is shown in Figure 3. The algorithm removes the complex near-in side-lobe structures surrounding strong compact sources and accurately reproduces the sky structure around the sources as can be seen by comparison with the higher resolution NVSS image. The Clean-component model of the Stokes-I image derived via Multi-beam clean will also be used for removing the Stokes-I leakage response patterns from the Stokes Q, U and V image cubes.

The preliminary continuum images of the Perseus molecular cloud region for Stokes I and U are shown in Figure 4, along with images of the same area at 6cm from the GB6 survey and the VTSS+WHAM H-alpha emission (Finkbeiner 2003, ApJS, 146, 407). The Stokes U image shows features over a range of spatial scales that are primarily Faraday rotation structures caused by the magneto-ionic medium. Many of the features in U are reproduced in high-resolution (arcminute) VTSS H-alpha emission maps. Combined analysis of H-alpha intensity (dependent on the ne and Te) and of polarization position angle variations (depending on B and ne) can be used to derived the magnetic fields strengths in the ISM. The comparison between the GALFACTS and GB6 images demonstrates the depth of the GALFACTS survey compared to the similar resolution Green Bank survey, even over the lower bandwidth (~ 80 MHz) of the WAPPs. Also clear is the much better response to low-spatial frequency structures (e.g. the diffuse emission from the California Nebula (upper left)).

The basic scientific data products from GALFACTS will be spectro-polarimetric image cubes with Stokes I, Q, U and V as a function of frequency. The observing band includes the atomic hydrogen line. Figure 5 shows a wide-area view of the ISM in the Perseus molecular cloud region as a composite of the GALFACTS radio continuum and H I emission as well as the IRAS 60 micron emission from dust. The ISM is seen to be complex and highly structured. The study of the relationship of magnetic fields to the gas and dust, and their role in the evolution of ISM, are among the key science goals of GALFACTS.

Figure 1. Observed and model beam shapes for ALFA Beam 5. The right hand panel shows the observed image-plane beam shape of Beam 5 as an image of a calibration source. The left hand side is an image of the model beam shape derived by iteratively fitting a linear combination of Zernike polynomials to the wavefront error function. The lowest contours are 2%, 4%, 6% and 10% of peak intensity, thereafter increasing by 10% intervals.

Figure 2. The aperture plane illumination function (left) and the aperture phase distribution (right) for Beam 5 derived from the Zernike polynomial fits to the beam 5 response function. For the aperture illumination, dark regions have higher weight. The lowest contour is 10% of maximum and the interval is 10%. For the phase distribution, green contours are negative angles.

Figure 3. Sample results of Multi-beam Clean. The top panel shows the initial ``dirty'' image of a section of the data. The middle panel shows the effect of Multi-beam Clean. The near-in sidelobe responses around compact sources are largely removed. The bottom panel shows the same region from the NVSS. The faint sources surrounding the bright compact source at the right were previously masked by side-lobe structure but become visible in the `Cleaned' image.

Figure 4. Top panels are GALFACTS images of A2174. Left is Stokes I and right is Stokes U. These images are band-averaged over the central 80 MHz of the WAPPs' band. Bottom panels are the same region from the GB6 survey (left: Gregory et al. 1996, ApJS, 103, 427) and H-alpha from WHAM and VTSS (right).

Figure 5. Composite image of the ISM in the Perseus molecular cloud region. The image shows radio continuum (gray-blue) and atomic hydrogen (diffuse yellow-brown) gas from GALFACTS mode observations of A2174, as well as IRAS 60 micron emission from dust (orange). The H II region IC 348 is seen near the centre and the California nebula at upper left. Complex regions of structure in gas and dust surround both nebulae.