The VLA E-array, a compact configuration planned as part of the VLA expansion (EVLA) phase-two proposal, offers a unique combination of brightness sensitivity and dynamic range which will make possible observations in a completely new regime.  Its potential applications are exciting for many areas of astronomy, including cosmology, evolution and interaction of galaxies, Galactic structure and interstellar medium studies, and solar system observations.  Some examples of areas where the E-array could make fundamental and unique contributions include :

• studies of fluctuations in the cosmic microwave background, including CMB linear polarization searches, the Sunyaev-Zeldovitch effect, and Thompson scattering halos around high redshift, radio-bright quasars

• HI mapping of extended disks of galaxies, including tidal tails and bridges like the Magellanic Stream, and of dwarf galaxies and intergalactic clouds similar to those seen as "Lyman Edge" systems in high redshift optical absorption studies

• mapping star formation regions in NH₃ and other molecules to trace the onset of gravitational collapse and fragmentation

• mapping large shells and chimneys in the diffuse interstellar medium using the HI 21-cm line in emission and mapping cold atomic clouds using 21-cm self-absorption (HISA)

• studying the density and temperature distribution in PDR's by mapping diffuse, partially ionized gas through hydrogen and carbon recombination lines

• quantitative mapping of the interstellar magnetic field through Zeeman aplitting of the  HI line,  recombination lines, and molecular lines

• mapping the structure of the Galactic magnetic field through diffuse linear polarized emission and its Faraday rotation

• tracking eruptive prominences and coronal mass ejections from solar active regions

The E-array concept considered here is specified in the VLA Development Plan  (Bastian and Bridle, 1995) as the E2 Array, which optimally concentrates the existing 25 m antennas with roughly hexagonal packing inside an area about 200m square.  The E1 Array, which is easier to build, accomplishes some of the objectives described here, but it is less sensitive by about a factor of 1.5 in the critical range of 50m to 150m baselines.

A simple characterization of the brightness sensitivity of a telescope is given by the covering factor, f, of the antennas, i.e. the combined area of the antennas divided by the area delimited by the longest baselines of the array.  Ignoring taper and aperture efficiency, the antennas have physical area of 491m² so the full array has physical area 1.33 x 10⁴ m².  This gives f = 0.33 for the E2 array, compared with the D array's f = 0.03 .  For the more extended configurations of the VLA, the covering factor decreases by a factor of 9 each time, so the C array has f=3 x 10^-3 , the B array has f=4 x 10^-4 , and the A array has f=4 x 10^-5.  The VLBA has f = 6 x 10^-11.

Single dish telescopes all have f=1, so that their brightness sensitivity is given by the radiometer equation, with appropriate factors for clipping, calibration, etc.  An aperture synthesis telescope has in principle zero brightness sensitivity, since without the zero uv spacing the dirty map has no sensitivity to smoothly distributed emission.  However, cleaning or other deconvolution techniques replace the dirty beam (beam solid angle zero) with a clean beam whose solid angle gives the conversion between brightness units of Jy/beam and K.   The result of this is the radiometer equation with rms noise in brightness temperature (sigma_T) increased by 1/f, so

where T_sys is an effective system temperature, B is the appropriate bandwidth, and tau the integration time.

An example of this equation is the mosaic survey mode, such as used in wide area HI surveys of the Magellanic Clouds  (Kim et al. 1998,   Staveley-Smith et al. 1995) ,  the Canadian Galactic Plane Survey (Taylor et al. 1997) , and the Southern Galactic Plane Survey (McClure-Griffiths et al. 2000, 2001) .  The relative speed of covering a given area to a given sensitivity (1 K rms at 0.8 km/s spectral resolution) is shown for several telescopes on figure 1 as a function of angular resolution.

The crosses indicate existing single dish telescopes, the curves represent aperture synthesis arrays with varying amounts of taper, which improves the brightness sensitivity by widening the beam, until a significant fraction of the baselines are weighted down too heavily.

The point of figure 1 is that the E array fills a niche in the 3' resolution area which is unique.  No other telescope comes close to it for speed of mapping and resolution.  If Arecibo were fitted with a multibeam like the one at Parkes this would be competitive.  However, Arecibo will never come close to the VLA in terms of dynamic range.  This is a critical point for many of the science applications of the E array.

Results from mosaic surveys with the ATCA 375m and 210m arrays show that dynamic range of better than 1000:1 can be achieved routinely with this technique  (example, Stokes I map of the SGPS test region) .  The VLA often achieves much higher dynamic range.  Single dish telescopes like Arecibo are generally limited to 10:1 dynamic range.  The GBT offers the possibility of much better performance, but at lower resolution.  For any application which requires detection of a faint object in the vicinity of one or more bright sources, dynamic range is critical.  An example is the detection of continuum emission (in Stokes I or Q and U) scattered from the ionized intergalactic medium at high redshifts around bright QSO's.  This would be a valuable probe of conditions during the epoch of galaxy formation, but it requires good surface brightness sensitivity and very high dynamic range (perhaps 10⁵ : 1).  Such an experiment would be worth trying with the VLA E array, but it is hopeless with a single dish, even one with sufficient resolution.

The dynamic range issue applies to any experiment which pushes the confusion limit.  Low and moderate resolution Stokes I continuum mapping at centimeter waves generally is limited not by radiometer noise but by the combined effects of all the unrelated continuum sources in the vicinity.  Combining observations from multiple arrays can help to beat this limit, as the high resolution maps can be used to subtract the effect of discrete sources from the low resolution data, leaving the low surface brightness, distributed emission unpolluted.  This is also a useful technique for linear polarization mapping in some cases.  This technique would be straightforeward for cleaning up maps made with the E array, but it is not so easy for single dish maps because of the imperfect calibration of the high spatial frequencies of the single dish data.
 

In many applications it is necessary to combine high resolution data taken with an aperture synthesis telescope with a single dish map to fill in the zero spacing and large scale structure.  An example is shown in figure 2, which is three images of the same structure (a large HI chimney detected in the SGPS,  McClure-Griffiths et al. 1999 ).  In this case the combination was simple, as the single dish data is from

Figure 2.  Images of the galactic chimney GSH277+0+36 made with Parkes (top), the ATCA interferometer (middle), and the combined data (bottom).  These show the necessity for combining data from single dish and aperture synthesis telescopes to adequately map large structures in the Milky Way.

Figure 3.  The gap between the D array and the GBT on the uv plane.  This histogram shows the sampling of a typical D array snapshot at 21-cm, with superposed in green an estimate for the GBT's uv plane sensitivity function.

GBT uv sensitivity (green curve).  There is a huge gap between these, with not much overlap.  This lack of overlap is problematic for combining observations taken with the two telescopes, both because of poor sensitivity to structures in the gap region, and because of the difficulty in calibrating the two datasets.  It is very important to have a lot of overlap in order to match the calibration factors of the two observations by comparing the amplitude of features in the overlapping region of the uv plane.
 

A large fraction of the scientific applications of the E array involve observations of spectral lines from atomic and molecular transitions with excitation temperatures of a few tens of K up to a few hundred K.  The 21-cm line from the Galactic ISM has an absolute upper limit of about 130 K in brightness (based on single dish surveys).  Non-maser molecular lines typically have much lower excitation temperatures.  So even if the optical depth is high, the brightness temperatures are low for this type of radiation.  Observations of objects with brightness temperature of 10 K is possible but very difficult with the D array.  The sensitivity table  shows that for a 1 km/s channel width at 21-cm the rms after a full 12 hour integration with the D array is about 0.2 K, after 5 minutes it is 2.4 K.  Thus large scale mosaics of thermalized lines in emission are very time consuming.  At higher frequencies the situation is not much better.  Channel widths in kHz increase as the rest frequency, but system temperatures are worse, and typical excitation temperatures are lower.  So for ammonia at 24 GHz the brightness sensitivity of the D array is about 0.6 K for 1 km/s channels after 5 minutes.  An example of what is possible now is shown in figure 4, from  Wiseman and Ho, 1998,  which shows the ridge-line of the Orion molecular cloud as traced by NH₃ lines.  This probably represents only the "tip of the iceberg", as single dish maps indicate that there is much more ammonia emission in this region, but most of it is resolved away by the D-array.

Figure 4.  A first moment map showing the velocity field in the high density ridges of the Orion nebula as mapped in ammonia by Weisman and Ho (1998) using the VLA D array.  The E array would show perhaps 5 to 10 times more emission, which is resolved out by the D array.

more

In preparing this I was struck by several things in the literature.  One is the very few papers which include combination of VLA data with single dish data.  This has been done a few times for HI data on nearby galaxies, but otherwise I have not found any references to it.   Another thing that strikes me is how few published papers present VLA maps of Galactic HI emission.  There are more of these, but still not many.  There are no papers on the diffuse Galactic polarized emission, which has been a topic of intense study by other telescopes in the last few years ( Gaensler et al, 2000  see figure 5).  Overall, people seem to have stayed away from projects which involve low surface brightness lines and polarized continuum.  This must be an effect of the difficulty of doing such observations because they "fall through the crack" between single dish telescopes and the D array.  There may be many more phenomena waiting for the E array to open up this neglected regime.

Figure 5.  Diffuse linear polarization from the Galactic non-thermal background, mapped with the SGPS by Gaensler et al. (2000).  This is an example of low surface brightness emission which has been neglected for many years because of its relative inaccessibility in the "E array gap".