1: \documentclass[12pt,preprint]{aastex}

2: %\received{2003 September~15}

3: %\revised{2001 April~26}

4: %\accepted{2001 May~17}

5: %\journalid{136}{2001 October}

6: %\articleid{START PAGE}{END PAGE}

7: \paperid{58977}

8: %\cpright{PD}{2000}

9: %\ccc{}

10: %

11: \shorttitle{74~MHz VLA}

12: \shortauthors{Kassim et al.}

13: 

14: \newcommand{\mjybm}{\mbox{mJy~beam${}^{-1}$}}

15: \newcommand{\jybm}{\mbox{Jy~beam${}^{-1}$}}

16: \newcommand{\task}[1]{\texttt{#1}}

17: \newcommand{\aips}{\textsc{aips}}

18: \renewcommand\anchor[2]{{#2}\footnote{#1}}%

19: \renewcommand\url[1]{\mbox{#1}}%

20: 

21: \begin{document}

22: \title{The 74~MHz System on the Very Large Array}

23: \author{N. E. Kassim, T.~Joseph~W.~Lazio}

24: \affil{Naval Research Laboratory, Remote Sensing Division, Code~7213,

25: 	Washington, DC 20375-5351}

26: \email{Namir.Kassim@nrl.navy.mil}

27: \email{Joseph.Lazio@nrl.navy.mil}

28: 

29: \author{W. C. Erickson}

30: \affil{U.~of Tasmania, School of Math.\ \& Physics, G.P.O.~Box 252-21,

31: 	Hobart, Tasmania 7001, Australia}

32: 

33: \author{R. A. Perley}

34: \affil{National Radio Astronomy Observatory, P.O.~Box O, Socorro, NM 

35: 	87801}

36: \email{rperley@nrao.edu}

37: 

38: \author{W.~D.~Cotton}

39: \affil{National Radio Astronomy Observatory, 520 Edgemont Road,

40: 	Charlottesville, VA  22903}

41: \email{bcotton@nrao.edu}

42: 

43: \author{E.~W.~Greisen}

44: \affil{National Radio Astronomy Observatory, P.O.~Box O, Socorro, NM 

45: 	87801}

46: \email{egreisen@nrao.edu}

47: 

48: \author{A.~S.~Cohen, B.~Hicks}

49: \affil{Naval Research Laboratory, Remote Sensing Division, Code~7213,

50: 	Washington, DC 20375-5351}

51: \email{Aaron.Cohen@nrl.navy.mil}

52: \email{Brian.Hicks@nrl.navy.mil}

53: 

54: \author{H.~R.~Schmitt}

55: \affil{Naval Research Laboratory, Remote Sensing Division, Code~7215,

56:        Washington, DC 20375-5351\\

57:        and\\

58:        Interferometrics, Inc., 13454 Sunrise Valley Drive, Suite 240, Herndon,

59: VA\,20171}

60: \email{Henrique.Schmitt@nrl.navy.mil}

61: 

62: \and

63: 

64: \author{D.~Katz}

65: \affil{U.S. Naval Academy, Physics 9C, Annapolis, MD21402-5026}

66: \email{dkatz@usna.edu}

67: 

68: 

69: \begin{abstract}

70: The Naval Research Laboratory and the National Radio Astronomy

71: Observatory completed implementation of a low frequency

72: capability on the Very Large Array at~73.8~MHz in 1998.  This frequency

73: band offers unprecedented sensitivity ($\sim 25$~\mjybm) and

74: resolution for low-frequency observations.  The longest baselines in

75: the VLA itself provide 25\arcsec\ resolution; the system has recently

76: been extended to the nearby Pie Town antenna of the Very Long Baseline

77: Array, which provides resolutions as high as 12\arcsec.  This paper

78: reviews the hardware, the calibration and imaging strategies of this

79: relatively new system.  Ionospheric phase fluctuations pose the major difficulty

80: in calibrating the array, and they influence the choice of calibration

81: strategy.  Over restricted fields of view (e.g., when imaging a strong

82: source) or at times of extremely quiescent ionospheric ``weather''

83: (when the ionospheric isoplanatic patch size is larger than the field

84: of view), an \emph{angle-invariant} calibration strategy can be used.

85: In this approach a single phase correction is devised for each

86: antenna, typically via self-calibration; this approach is similar to

87: that used at higher frequencies.  Over larger fields of view or at

88: times of more normal ionospheric ``weather'' when the ionospheric

89: isoplanatic patch size is smaller than the field of view, we adopt a

90: \emph{field-based} strategy in which the phase correction depends upon

91: location within the field of view.  In practice we have implemented

92: this second calibration strategy by modeling the ionosphere above the

93: array using Zernike polynomials.  Images of 3C sources of moderate

94: strength are provided as examples of routine, angle-invariant

95: calibration and imaging. Flux density measurements of a sub-sample

96: of these sources with previously well determined low frequency spectra

97: indicate that the 74 MHz flux scale at the Very Large Array is

98: stable to a few percent, and that flux densities tied to the Baars

99: et al. value of Cygnus A are reliable to at least 5 percent.

100: We also present an example of a wide-field

101: image, devoid of bright objects and containing hundreds of weaker

102: sources, constructed from the field-based calibration.  The paper also

103: reviews other practical aspects of low frequency observations, in so

104: far as they differ from those encountered at higher frequencies,

105: including aspects of interference excision and wide-field imaging. We

106: close with a summary of lessons the 74~MHz system offers as a model for

107: new and developing low-frequency telescopes.

108: \end{abstract}

109: 

110: \keywords{instrumentation: interferometers --- techniques: high

111: angular resolution --- techniques: image processing --- techniques:

112: interferometric --- astrometry}

113: 

114: \section{Introduction}\label{sec:intro}

115: 

116: Radio astronomy began with the discovery of celestial radio emission

117: by K.~Jansky at~20.5~MHz \citep{j35}.  Throughout the 1950s and 1960s,

118: key discoveries and technological advances in radio astronomy---at

119: low-frequencies ($\nu \lesssim 100$~MHz) in particular---helped form

120: the basis of modern astronomy, including:

121: \begin{itemize}

122: \item The introduction of non-thermal processes as an astrophysical

123: source of emission \citep{ah50,k50,s52}, motivated by early

124: observations of the diffuse Galactic radio emission at~160~MHz

125: \citep{r40}, though it took some time for the importance of

126: non-thermal processes to be accepted widely;

127: 

128: \item The discovery of pulsars through observations at~81~MHz

129: \citep{hbpsc68}; and

130: 

131: \item The  development of aperture synthesis interferometry at~38~MHz

132: \citep{rv46,rse50,m52,r52}.  (See also the work by \cite{pp-sm46} and

133: \cite{mpp-s47} using a ``sea interferometer'' at frequencies

134: between~75 and~3000~MHz.)

135: \end{itemize}

136: By the late 1960s and early 1970s, however, interest turned to

137: obtaining high-resolution images.  Powerful centimeter-wavelength

138: interferometers began to provide sub-arcminute angular resolution (or

139: sub-arcsecond resolution in the case of the VLA by the late 1970s)

140: with dynamic ranges of several hundred or better.

141: 

142: Even so, a number of low-frequency interferometers continued to be

143: constructed, some with truly impressive collecting areas ($\gtrsim

144: 10^5$~m${}^{2}$), including the UTR-2 \citep{bms78}, the Culgoora

145: Radioheliograph,

146: the Cambridge Synthesis Telescope, and the Clark Lake TPT Radio

147: Telescope \citep{eme82}.  However, with the exception of the MERLIN

148: \citep{thom86,leah89} and GMRT 151~MHz

149: systems \citep{swarup90} and for all telescopes operating below 100 MHz, the

150: maximum baselines were relatively limited ($\lesssim 5$~km). The corresponding

151: angular resolution was relatively poor ($\gtrsim 10\arcmin$), and the

152: resulting high confusion levels meant poor sensitivities ($\gtrsim

153: 1$~Jy).  The primary constraint on baseline length were the phase

154: distortions imposed by the Earth's ionosphere over the intrinsically

155: wide fields of view and the lack of suitable algorithms to compensate

156: for the distortions.  On baselines longer than

157: a few kilometers, ionospheric phase distortions are severe enough to cause

158: decorrelation, making higher-resolution imaging difficult to

159: impossible, especially at the lowest frequencies ($\leq 100$ MHz).

160: 

161: This paucity of large aperture, high-sensitivity, synthesis

162: instruments operating below~100~MHz has left this portion of the radio

163: spectrum poorly explored.  Yet, there are a variety of topics that

164: could be addressed by a sensitive, high-angular-resolution,

165: low-frequency telescope including:

166: \begin{itemize}

167: \item Continuum spectra over much larger frequency dynamic ranges for

168: studies of shock acceleration and spectral aging in Galactic

169: (supernova remnants) and extragalactic (radio galaxies and galaxy

170: cluster relics) sources;

171: 

172: \item Efficient detection of large numbers of steep spectrum sources,

173: which can be imaged in some cases, including high-redshift radio

174: galaxies, shocks driven by infalling matter in clusters of galaxies, and

175: pulsars in the Milky Way and possibly in external galaxies;

176: 

177: \item Probing the ionized interstellar medium (ISM) via measurements

178: of radio-wave scattering and absorption, the distribution of low-density ionized gas

179: toward nonthermal sources, and hydrogen and carbon recombination line

180: observations of very high Rydberg state atoms;

181: 

182: \item The large opacity of \ion{H}{2} regions below~100~MHz can enable

183: distance determinations to various foreground objects, in both the

184: Galaxy and external galaxies, from which the three-dimensional

185: distribution and spectrum of cosmic-ray emissivity can be determined

186: as well as being used to measure their emission measures,

187: temperatures, pressures, and ionization states

188: 

189: \item Detection of coherent emission from sources such as the Sun,

190: Jupiter, pulsars, and possibly radio bursts from nearby stars and

191: extrasolar planets.

192: \end{itemize}

193: 

194: While earlier experience with MERLIN and the VLA (and more recently

195: with the GMRT) demonstrated that large ($>$ 5~km) interferometers could

196: compensate for ionospheric effects below $\sim$330~MHz, a prototype

197: 74~MHz system was the first to demonstrate that self-calibration

198: techniques can correct for the large ionospheric-induced phase errors

199: below 100 MHz. The prototype 74~MHz system consisted of eight of the

200: VLA's 28 antennas equipped with 74~MHz receivers, and \cite{kped93}

201: were able to produce images with sub-arcminute resolutions and

202: sub-Jansky sensitivities, thereby demonstrating that these

203: self-calibration techniques were able to correct for ionospheric phase

204: errors on baselines at least as long as the longest VLA baselines

205: (35~km) and, in principle, much longer.  Because of the limited number

206: of antennas, the prototype 74~MHz system had relatively poor $u$-$v$

207: coverage, so only the strongest ($\geq 500$~Jy) sources, numbering a

208: dozen or so, were imaged.

209: 

210: In~1998 January all 28 VLA antennas were equipped with~74~MHz

211: antennas.  This improved the capability of the instrument greatly, and

212: it is now possible to detect hundreds of sources in single fields at

213: high angular resolution and sensitivity.  At the same time, a number

214: of innovative procedures and software solutions have been developed to

215: handle the data, some of which are significantly different than at

216: centimeter wavelengths and others of which are applicable to, and new

217: to, centimeter wavelengths.  Figures~\ref{fig:resolution}

218: and~\ref{fig:sensitivity} illustrate the levels of resolution and

219: sensitivity now possible with this new system, improvements that are

220: all the more impressive given that the relatively modest collecting

221: area and low efficiency of the VLA ($\simeq 2 \times 10^3$~m${}^2$

222: effective collecting area, $\lesssim 10$\% of many of the other

223: telescopes shown).

224: 

225: This paper describes the fully operational 74~MHz system (hardware and

226: software) on the \hbox{VLA}.  In \S\ref{sec:history} we summarize

227: briefly the characteristics of the prototype array as they relate to

228: the current system.  In \S\ref{sec:general} we review general

229: characteristics of the current low frequency system, summarize its

230: general performance, and highlight those aspects that are

231: significantly different than at centimeter wavelengths.  We describe

232: the calibration of 74~MHz observations in \S\ref{sec:calibrate} and

233: their imaging in \S\ref{sec:imaging}.  In \S\ref{sec:dynamic} we suggest

234: how the current system could be improved via dynamic scheduling.  In

235: \S\ref{sec:future} we discuss possible future expansions of

236: low-frequency synthesis instruments using the lessons from the 74~MHz

237: system, and we present our conclusions in \S\ref{sec:conclude}.

238: In Appendix~\ref{sec:example} we present selected examples of

239: imaging of moderately strong 3C sources.

240: 

241: Throughout the paper we shall illustrate various effects with images

242: or other figures produced from 74~MHz observations.  The examples we

243: show are a heterogeneous lot, resulting from a number of different

244: observations of different sources acquired for different purposes.

245: Our objective is to present a representative sample of various effects,

246: but not all effects will necessarily be present in every observation.

247: 

248: \section{Low-Frequency Systems on the VLA}\label{sec:history}

249: 

250: The original design of the Very Large Array included only four

251: frequency bands, centered near wavelengths of~21, 6, 2, and~1.3~cm

252: \citep{nte83}.  However, there is no fundamental reason a

253: low-frequency system cannot operate on the array---the principles

254: of aperture synthesis are as applicable to~50~MHz as they are

255: to~5~GHz.  More importantly, the key components of the array---the

256: signal collection (antennas), signal transmission (waveguide), and

257: signal processing (correlator and post-processing)---are essentially

258: frequency independent within the radio part of the spectrum.  As soon

259: as the construction phase of the VLA ended, discussions on

260: implementing a low-frequency capability began.

261: 

262: \cite{pe84} advocated a free-standing array that would make use of the

263: VLA's infrastructure (most importantly, the waveguide transmission

264: system) to achieve approximately 25\arcsec\ resolution.  However, no

265: source of funding was obvious, and it was decided subsequently that

266: trial systems could be implemented on the VLA itself to address key

267: questions regarding the calibration and imaging of low frequency,

268: long-baseline data.

269: 

270: The initial low-frequency system, operating at~90~cm (300 to~340~MHz),

271: was installed between~1983 and~1989.  It is a prime-focus system, as

272: it is impractical to implement a secondary focus system at such a low

273: frequency.  The feed is a crossed dipole, situated in front of the

274: Cassegrain subreflector, which thus acts as a (rather imperfect)

275: ground plane.  Because of this, and because the phase center is

276: located approximately 50~cm ($\approx \lambda/2$) in front of the true

277: focus, this system has both a low efficiency (less than 40\%) and a

278: very broad shoulder of width approximately 12\arcdeg\ in the antenna

279: power pattern.  Nevertheless, it has been a very successful and widely

280: used frequency band at the \hbox{VLA}.  Most importantly, it

281: encouraged the development of the multi-faceted imaging algorithms

282: \citep{cp92} needed for wide-field, low-frequency observing, as

283: described later in this paper. Its operation also demonstrated the

284: robustness of angle-independent self-calibration (\S\ref{sec:pcal}) 

285: for removing ionospheric distortions across the large ($\sim$2.5 \arcdeg FWHP) 

286: ~90~cm field of view.

287: 

288: The success of the 330~MHz system soon led to consideration of a lower

289: frequency facility.  A protected radio astronomy frequency allocation

290: exists between~73 and~74.6~MHz.  Again, funding constraints led to the

291: decision to deploy a trial system, comprising a simple feed system on

292: a few of the VLA's antennas.

293: 

294: The feed system chosen is essentially the same as that used

295: at~330~MHz---crossed dipoles in front of the subreflector.  Because of

296: the long wavelength, the defocussing errors that affect 330~MHz

297: performance severely are not serious at~74~MHz.  However, because the

298: antenna itself is only approximately $6\lambda$ in diameter, the

299: subreflector is an imperfect ground plane, and the profound effect of

300: the antenna quadrupod structure, it was anticipated that the forward

301: gain and sidelobe structure would be fairly poor---as subsequent

302: measurements have borne out.  \cite{kped93} describe the prototype

303: 74~MHz system in more detail and describe the initial data calibration

304: and imaging methodologies.

305: 

306: 

307: \section{The 74~MHz System on the VLA}\label{sec:general}

308: 

309: 

310: Amplifiers and feeds for the complete 74~MHz system were built during

311: the summer of~1997 by two of us (WCE and BH) at the NRL and deployed

312: during the fall of that year.  All antennas were equipped with dipoles

313: by~1998 January.

314: 

315: Because of concerns about blockage at higher frequencies, a deployable

316: crossed-dipole feed was designed.  The half-wavelength dipoles

317: contribute to blockage and a higher system temperature, resulting in a

318: total sensitivity loss of about~6\% at~1.4~GHz and smaller losses at

319: higher frequencies, so they are deployed only during a fraction of the

320: time in each configuration.  A simple mounting system is used---two

321: ropes, each of which supports one dipole, are threaded through

322: eyebolts located on opposite quadrupod legs at the appropriate height.

323: The ends of these ropes are tied to cleats located at a convenient

324: height on the quadrupod legs.  The signal cables drop about~7~m to the

325: antenna surface, where they pass through the roof of the vertex room

326: to the amplifiers.  Figure~\ref{fig:mount} shows the dipoles and

327: mounting system developed.

328: 

329: The receiver units combine the linearly polarized signals of the

330: dipoles to produce circular polarized signals, then amplify and

331: bandpass filter these signals, and pass them to the VLA intermediate

332: frequency (IF) system.  They also contain an integral noise

333: calibration source.  In order to produce serviceable receivers on a

334: short timescale and at low cost, they were constructed almost entirely

335: from commercial components.

336: 

337: The VLA signal transmission system allows for two pairs of two

338: parallel-hand signals to be transmitted.  The receiver system is

339: designed so that the two senses of circular polarization from the

340: 74~MHz receivers occupy one pair of signal transmission channels while

341: the two senses of circular polarization at~330~MHz occupy the other

342: pair of signal transmission channels.  Thus, 74 and~330~MHz

343: observations can be acquired simultaneously.  \cite{kped93} used this

344: simultaneous, dual-frequency capability for ionospheric calibration

345: via phase transfer from~330 to~74~MHz (see \S\ref{sec:calibrate}).  An

346: alternate signal transmission approach is to use one pair of signals

347: for the upper half of the 1.5~MHz bandpass and the other pair for the

348: lower half, thereby obtaining higher spectral resolution, primarily

349: for radio frequency interference (RFI) excision purposes

350: (\S\ref{sec:rfi}). Measurements of circular polarization are normally

351: available and have been used for both astronomical 

352: (solar, T. Bastian, private communication, 1998) and RFI excision

353: purposes.

354: \footnote{VLA correlator modes 'PA' or 'PB' allow obtaining full 

355: polarization information at 74 MHz alone, at the expense of halving

356: the number of channels, and thus reducing the ability to purge 

357: narrowband RFI. Tests observing a strong, unpolarized source indicate

358: cross-polarization leakage of at least $\sim$30 \%, and since 

359: initial attempts at polarization calibration using AIPS task POLCAL

360: failed, users have not been encouraged to utilize this mode. It is

361: possible that a dedicated effort with the EVLA correlator might 

362: permit full polarization astronomical measurements in the future,

363: although it is suspected that many, except relatively nearby sources,

364: might be depolarized at this frequency due to Faraday rotation.}

365: 

366: Figure~\ref{fig:receiver} presents a complete block diagram of a

367: receiver.  In detail, the receiver units are comprised of the

368: following components:

369: 

370: \begin{enumerate}

371: \item The two orthogonal linear feeds are converted by a full quad

372: hybrid 

373: into right and left circular polarizations (RCP, LCP).

374: 

375: \item An onboard source 

376: injects a noise calibration signal into both the RCP and LCP chains

377: via directional couplers.

378: This source is directly powered by a ``Cal'' signal provided by the site.

379: 

380: \item Out-of-band rejection filtering is provided by high-Q cavity

381: filters 

382: with a center frequency of~73.9~MHz and a 1.7~MHz bandwidth.  Such a

383: narrow bandwidth is necessitated by the close proximity of local

384: television stations.  (Tests have been conducted with a 3~MHz

385: bandwidth; the TV signals saturated the receivers.)

386: 

387: \item For the VLA, the signal is transferred directly to the IF

388: system while for the PT receiver, the signal is upcoverted in order to

389: pass into the VLBA IF system, by mixing the RCP and LCP channels with

390: a reference local oscillator (LO) signal.

391: 

392: \item Two power combiners consolidate the 74 and 330~MHz channels for

393: transport to the correlator.

394: \end{enumerate}

395: 

396: The initial work on the 74~MHz system focused on the VLA alone.

397: Concurrently NRAO was in the process of testing fiber optic

398: transmission techniques by tying in the nearby Pie Town antenna (PT)

399: of the Very Long Baseline Array (VLBA).  First fringes between a VLA

400: subarray and the PT antenna were observed in~1998 December, and a

401: successful test with the PT antenna and the full VLA was conducted

402: in~1999 September.  Routine observations on this facility began

403: in~2000 October.  Consequences of this fiber-optic link to Pie Town

404: (the PT link) are that the longest VLA baselines are extended to

405: approximately 70~km, but the number of antennas in the VLA decreases

406: to~26 because the PT antenna replaces one of the antennas in the VLA

407: and the VLA signal electronics requires removing another antenna from

408: the array to accommodate the PT signal.

409: 

410: Initial efforts on the PT link focused on frequencies above~1000~MHz.

411: The VLBA does have a standard observing frequency at~330~MHz, but it

412: does not have a 74~MHz operating capability.  In~2000, NRL and NRAO

413: initiated a program to add a 74~MHz capability to the PT antenna and

414: to operate the PT link at both 74 and~330~MHz.  Initial tests were

415: conducted with the 74~MHz receiver replacing the 330~MHz receiver, but

416: the electronics path has been modified subsequently to operate at both

417: 74 and~330~MHz simultaneously.

418: Figure~\ref{fig:ptlink} shows the first successful fringes at~74~MHz,

419: obtained on the quasar \objectname[3C]{3C~123} in the fall of~2001.

420: In the remainder of this paper we shall focus on the VLA system alone.

421: Many of the techniques we describe are equally applicable to

422: observations with the PT link, though experience with that system is

423: considerably more limited and has only been used to observe relatively 

424: bright, isolated objects \citep{Gizani2005, lazio06, Lane2006, Delaney2004}.

425: See Appendix~A for an example of the full synthesis VLA+PT image of Cas~A

426: (Figure~\ref{fig:casapt}).

427: 

428: Table~\ref{tab:perform} summarizes the performance characteristics of

429: the VLA's 74~MHz system.  We quote a sky-noise dominated system 

430: temperature that is appropriate for the Galactic polar caps. While

431: the sky-noise for fields on the Galactic plane can be up to ten times

432: higher, in practice the low forward gain of the primary beam smooths

433: out and lowers the variations in T$_{sys}$ to typically a factor

434: of two or less (\S\ref{sec:ampcal}) The sensitivity

435: listed is typical for the A and~B configurations, for regions away

436: from the Galactic plane.  Sensitivities for the smaller configurations

437: are considerably poorer, because of confusion (\S\ref{sec:confuse})

438: and presumably because of low-level, broad-band RFI from the antennas

439: and equipment located at the VLA site.

440: 

441: Figure~\ref{fig:beam} shows the primary beam power pattern measured

442: for one of the antennas.  Other antennas show similar power patterns.

443: Table~\ref{tab:perform} cites the primary beamwidth as 11\fdg7, but it

444: is clear that the beam has only a modest forward gain and a broad

445: plateau with a poor sidelobe structure.  These result from the

446: aforementioned aspects of the system that the antenna itself is only

447: approximately $6\lambda$ in diameter and the antenna quadrupod

448: structure. Notice that the sharp edge seen on the right side of the

449: beams presented in Figure~\ref{fig:beam} is due to the fact that this

450: portion of the beam was not sampled. This was due to motion limitations

451: of the VLA antennas during the holography measurements used to determine

452: the beam power pattern. Nevertheless, this sampling issue did not affect

453: the mapping of the most important part of the beam, down to 20-25~dB

454: from the peak. The poor primary beam definition gives rise to significant

455: sidelobe confusion, as discussed below and further in \S\ref{sec:confuse}.

456: 

457: Figures~\ref{fig:bandwidth} and~\ref{fig:time} show the system sensitivity,

458: as measured by the rms noise level in an image, as a function of increasing

459: receiver bandwidth and integration time. This behaviour is typical of

460: observations obtained in the A and B configurations. These images were

461: excised of narrow-band RFI, calibrated and imaged following the procedures

462: described in Sections 4 and 5. The deviation from a $\Delta\nu^{-1/2}$

463: dependence with increasing receiver bandwidth in Figure~\ref{fig:bandwidth}

464: indicates that the system is not thermal noise limited. On the other hand,

465: Figure~\ref{fig:time} shows an approximate $t^{-1/2}$ dependence normally

466: indicative of a thermal noise limited response. Taken together we conclude

467: that we are 

468: mainly sidelobe confusion limited since its effects should be independent of

469: bandwidth to first order, but scale roughly as $t^{-1/2}$ because sources

470: moving through the sidelobes contribute noise in a random walk fashion that

471: averages out with time \footnote{One might expect this

472: dependence to disappear once the $u$-$v$ coverage repeats.  In practice,

473: however, position shifts caused by the ionosphere (\S\ref{sec:pdelay}) act

474: to ``thermalize'' the sidelobe confusion contribution.  Thus, we believe

475: that the noise due to sidelobe confusion will continue to decrease, at least

476: initially, as $t^{-1/2}$ even after the $u$-$v$ coverage repeats.}

477: For integrations of $\geq$1~hr, the

478: deviation from thermal noise is typically a factor of 2-4.

479: 

480: The residual effects of incompletely removed RFI must also play a role in

481: the reduced sensitivity. In fact the sensitivity in the C and D

482: configurations is much poorer than expected from confusion alone, and our

483: hypothesis is that the effects of low-level, broad-band RFI are responsible

484: for that. In practice the 74 MHz system is rarely used in either the C or D

485: configurations.

486: 

487: We note that while Table~\ref{tab:perform} quotes an 8 hour sensitivity

488: limit of ~25 mJy, in practice the achievable sensitivity may vary

489: considerably due to the positionally dependent sky noise dominated system

490: temperature, and more importantly from the relative proximity to the handful

491: of extremely bright sources that often dominate the sidelobe confusion

492: (e.g. \objectname[]{Cyg~A}, \objectname[]{Cas~A}, etc.)

493: In \S\ref{sec:confuse} we discuss both sidelobe

494: and classical confusion further, and show that in the more compact

495: configurations the latter effect can become significant.

496: 

497: \section{Calibration of 74~MHz VLA Data}\label{sec:calibrate}

498: 

499: Observations of the brightest sources in the sky (e.g.,

500: \objectname[]{Cyg~A}, \objectname[]{Tau~A}) with the prototype 74~MHz

501: system demonstrated that the methodologies and algorithms that had

502: been developed for calibration at the standard VLA frequencies were

503: generally sufficient for 74~MHz \citep{kped93}, with self-calibration

504: being particularly important.  The larger number of antennas now

505: available makes self-calibration even more robust, but it has also

506: revealed its limits more clearly.

507: 

508: In this section we motivate and describe the procedures we have

509: developed for the calibration and imaging of~74~MHz data.  The

510: procedure can be summarized as follows:

511: \begin{description}

512: \item[Bandpass calibration (\S\ref{sec:bpass})] Observations at~74~MHz

513: are acquired in a spectral line mode, both to enable RFI excision and

514: to avoid bandwidth smearing over the relatively large fields of view.

515: Therefore we need to apply a baseline correction to the data. 

516: 

517: \item[RFI excision (\S\ref{sec:rfi})] Largely because of

518: self-interference, 74~MHz data always must be edited to remove

519: \hbox{RFI}.

520: 

521: \item[Amplitude calibration (\S\ref{sec:ampcal})] As at higher

522: frequencies, a source whose flux density is presumed to be known must

523: be observed to set the flux density scale.  The primary flux density

524: calibrator for the VLA at~74~MHz is \objectname[]{Cyg~A}. Other

525: sources that can be used when \objectname[]{Cyg~A}

526: is not available are  \objectname[]{Cas~A}, \objectname[]{Vir~A},

527: \objectname[]{3C~123} and \objectname[]{Tau~A}.

528: 

529: \item[Phase calibration] The dominant source of phase corruption at

530: low frequencies is due to the Earth's ionosphere.  Unlike at higher

531: frequencies, one cannot employ as a phase calibrator a source nearby

532: in the sky to one's target source or field.  Instead we have developed

533: two strategies:

534: \begin{itemize}

535: \item When the isoplanatic patch (scale over which the rms phase

536: difference between two lines of sight is approximately 1~radian)

537: size is larger than the field of view of interest

538: (\S\ref{sec:pcal}), a single phase calibration can be

539: applied to the entire field of view.  It is most useful in the more

540: compact configurations (C and~D) or in the larger configurations (A

541: and~B) when imaging a strong source.  This strategy relies heavily on

542: current implementations of self-calibration, and it is similar to the

543: calibration strategy used by \cite{kped93}. As such, this strategy is

544: a confirmation of their prediction that self-calibration can be used

545: to remove ionospheric phase fluctuations.

546: 

547: \item When the isoplanatic patch size is smaller than the field of

548: view of interest (\S\ref{sec:pdelay}), an angular dependence

549: \emph{within the field of view} of interest must be used in the phase

550: calibration. We have used a method called ``field-based'' calibration

551: to do this, which  models the ionosphere as a phase-delay screen and

552: uses a grid of background sources to solve for the ionospheric

553: refraction, both globally and differentially within the field of view.

554: \end{itemize}

555: \end{description}

556: Following sections describe imaging requirements (\S\ref{sec:imaging})

557: and present examples designed to illustrate the efficacy of both phase

558: calibration strategies (\S\ref{sec:example}).  This section parallels

559: closely the discussion in our online

560: \anchor{http://rsd-www.nrl.navy.mil/7210/7213/LWA/tutorial}{tutorial},

561: which contains more detailed descriptions of the procedures, as well

562: as sample inputs for reducing data within \aips.

563: 

564: \subsection{Bandpass Calibration}\label{sec:bpass}

565: 

566: As described below (\S\ref{sec:rfi}), the VLA generates considerable

567: internal radio frequency interference.  Consequently, observations are

568: performed in a spectral-line mode, which also avoids bandwidth

569: smearing over the large regions (typically) imaged

570: (\S\ref{sec:imaging}).  Characteristic spectral channel bandwidths are

571: 12~kHz in a 1.5~MHz total bandpass.  As with any spectral line

572: observation, the amplitude variations across the band (and phase gradients

573: due to delay errors) must be removed by bandpass calibration. Fortunately,

574: the flux density of \objectname[]{Cyg~A} ($\simeq 17$~kJy) is nearly

575: always much greater than the equivalent flux density of any RFI, even

576: in the narrowest channels that the VLA correlator can produce

577: (12~kHz), meaning that one can use observations of it to calibrate the

578: bandpass prior to excising \hbox{RFI}.  Other sources---such as

579: \objectname[]{Cas~A}, \objectname[]{Vir~A}, \objectname[]{Tau~A}---can

580: also be used; their lower signal-to-noise ratios on longer baselines

581: can require judicious choices of time or frequency ranges or both in

582: order to calibrate the bandpass.

583: 

584: Bandpass calibration is traditionally performed with the assumption

585: that the flux density of the calibrator can be represented by a single

586: value across the bandpass.  In the case of 74~MHz observations with

587: the VLA, however, the fractional bandwidth of the system is large

588: enough that the intrinsic visibility of \objectname[]{Cyg~A} can

589: change across it.  A rough estimate shows that the effect of

590: resolution can be ignored if

591: \begin{equation}

592: \pi\frac{\Delta\nu}{\nu}\frac{\theta_{\mathrm{src}}}{\theta_{\mathrm{HPBW}}}

593:  \ll 1,

594: \label{eqn:bpass}

595: \end{equation}

596: where $\Delta\nu/\nu$ is the fractional bandwidth and

597: $\theta_{\mathrm{src}}/\theta_{\mathrm{HPBW}}$ is the source size

598: expressed in units of the synthesized beam.  In the A configuration,

599: the left-hand side of equation~(\ref{eqn:bpass}) is 0.3, so there is a

600: 30\% change (worst case) in visibility across the bandpass on the

601: longest baseline; in the smaller configurations, the error introduced

602: by the fractional bandwidth is proportionally less.

603: 

604: Two options exist for dealing with the effects of the intrinsic

605: visibility change across the bandpass.  First, one can set an upper

606: limit to the baseline length to be considered, effectively decreasing

607: $\theta_{\mathrm{src}}/\theta_{\mathrm{HPBW}}$.  Second, one can

608: divide the observed visibilities at each frequency by a model of the

609: source, thereby transforming the visibilities into what would have

610: been obtained had a point source been observed.  In practice, the

611: second option is used most often in order to employ the maximum amount

612: of the data possible, and

613: \anchor{http://rsd-www.nrl.navy.mil/7210/7213/LWA/tutorial/}{models} of

614: \objectname[]{Cyg~A}, \objectname[]{Cas~A}, \objectname[]{Vir~A},

615: \objectname[]{3C~123} and \objectname[]{Tau~A} are available online.

616: 

617: This second method, dividing the visibilities by a model of the

618: source, is done before the visibility amplitudes are calibrated.  At

619: higher frequencies, the standard bandpass calibration procedure at the

620: VLA is to divide by the ``Channel~0'' continuum data, formed by

621: averaging the inner 75\% of the bandpass.  Rather than using a

622: ``Channel~0'' continuum, which would be contaminated by the presence

623: of the 100~kHz comb (\S\ref{sec:rfi}), one first solves for the individual channel

624: corrections (both amplitude and phase).  These corrections are then

625: normalized so that the mean correction across the spectrum is unity.

626: Various weighting schemes can be used during the normalization

627: process.  In practice, the weights used in normalization are often

628: taken to be independent of channel; the converse, scaling the weights

629: by some function of the amplitude in each channel, can have the effect

630: of giving more weight to baselines with more RFI (higher amplitudes)

631: and the short baselines (which are most often affected by RFI).

632: 

633: In general a single bandpass correction is determined from all

634: observations of the bandpass calibrator, and this correction is

635: applied to all other observations within that ``observing run.''  For

636: bright sources, residual bandpass calibration errors can introduce

637: systematic errors in the image (appearing in the shape of the beam,

638: but which cannot be \textsc{clean}ed) and thereby limit the dynamic

639: range of the image.  In these cases, one of two strategies can be

640: adopted.  If one has multiple observations of the bandpass calibrator,

641: one can attempt to form the bandpass correction as a function of time.

642: Alternately, during self-calibration one can produce a different phase

643: correction for each channel (or small number of channels).

644: 

645: \subsection{RFI Excision}\label{sec:rfi}

646: 

647: Radio frequency interference (RFI) is a significant problem for low

648: frequency VLA observations, though we note that there is little

649: \emph{external} RFI within the passband of 73.0--74.6~MHz. RFI is the

650: major limiting factor for the dynamic range of the data observed with

651: the more compact VLA configurations (C and D), due to the interference

652: between the individual antennas of the array. It also constitutes an

653: important issue for the more extended configurations (A and B),

654: and needs to be dealt with appropriately.

655: Essentially all non-astronomical signals are generated internally by

656: the electronics of the VLA's antennas and (to a lesser degree) signals

657: emanating from other sources on the VLA site.  By far the most common

658: is a 100~kHz ``comb'' generated by the VLA's monitor and control

659: system.  The oscillators responsible for this are located in every

660: antenna.  The oscillators can be coherent, with low phase rates such

661: that phase differentials of less than 1~radian are obtained in an

662: integration time, so

663: that the coherence is maintained over long periods of time.  The

664: result is spurious correlation, especially between certain pairs of

665: antennas whose oscillators appear to maintain coherence regardless of

666: where they are located in the array.  The net (averaged over frequency)

667: spurious visibility is not great ($\lesssim 100$~Jy) and can be

668: effectively removed in the spectral domain.  (By contrast, RFI

669: at~330~MHz is mainly externally generated but can be excised using

670: the same procedures as described below.)  Figure~\ref{fig:rfi} (left

671: panel) shows an example of the 100~kHz ``comb'' for a single baseline.

672: 

673: Various procedures exist to accomplish RFI excision, depending upon

674: the dynamic range requirements. Notice that the final image dynamic range

675: will also depend on the position of the source in the field, usually being

676: lower closer to brighter sources, and how well ionospheric effects can be

677: solved (Sections 4.4 and 4.5). RFI excision appropriate for moderate

678: dynamic range images (DR$\sim10^3$) can be accomplished with the following

679: procedure.\footnote{% 

680: Within \aips, this procedure is implemented using \task{FLGIT} or by

681: \task{VLAFM}, a special-purpose task added to \aips.

682: }  One begins by identifying a ``baseline'' region within the spectrum to

683: which only the astronomical source(s) and the noise within the system

684: (i.e., $T_{\mathrm{sys}}$) contribute.  Usually, this ``baseline''

685: region is non-contiguous and does not include the ``comb.''  Next,

686: visibilities with excessive amplitudes (e.g., $100\sigma$) are

687: flagged, regardless of where they appear in the spectrum (inside the

688: baseline region or not).  A linear fit to the ``baseline'' region is

689: made and subtracted.  The objective of this linear fit is to avoid (or

690: reduce) the extent to which the flagging is biased by the astronomical

691: source(s) and the system.  Finally, visibilities whose residual values

692: exceed a user-defined threshold (typically 6--$8\sigma$) are flagged

693: in channels both inside and outside of the ``baseline'' region.

694: The results of this procedure are presented in the right panel

695: of Figure~\ref{fig:rfi}, where we show the example of the visibility

696: spectrum on a baseline after RFI editing. An inspection of this image

697: shows that this procedure has eliminated the RFI emission present in the

698: left panel.

699: 

700: A similar procedure, also appropriate for moderate dynamic range

701: imaging, is to flag all of the channels comprising the 100~kHz

702: comb\footnote{%

703: Within \aips, the task \task{SPFLG} can be used.

704: } and then clip any remaining data with excessively large amplitudes.

705: 

706: While elementary, these approaches have proven to be reasonably effective

707: in removing RFI at~74 (and~330) MHz.  Additional tests also have been

708: found to be useful.  For instance, channels in which the ratio of the

709: real part of the visibility to the imaginary part exceeds a given

710: value can be flagged.  Various observers have noted that RFI can often

711: be strongly circularly polarized so a test on the amplitude of the

712: circular polarization (Stokes V parameter) in each channel is also

713: useful. For the highest dynamic range images (DR$\sim10^4$), it is

714: often necessary to inspect the data by hand and perform additional

715: editing.\footnote{ Within \aips, the tasks \task{SPFLG}, \task{TVFLG},

716: and \task{WIPER} can be used either singly or in combination.} \cite{Lane2005}

717: describe in more detail post correlation RFI excision with the VLA and VLBA

718: (see also \cite{Golap2006}).

719: 

720: The limited spectral resolution of the VLA's original correlator means that

721: even the narrowest channels provided are 12~kHz wide.  As a result, the

722: procedure we describe can remove a significant fraction of the

723: data---typically 10 to~25\%.  A key design goal of any future correlator

724: would be to provide considerably more channels with higher frequency

725: resolution, allowing much more precise removal of these artificial signals.

726: Such improvements are forthcoming for the 74 MHz system with the transition

727: to the EVLA and its WIDAR correlator. Furthermore, techniques of

728: pre-correlation RFI excision would clearly offer key advances over current

729: approaches.

730: 

731: A final note is that the discussion above focuses on narrow-band RFI,

732: much of whose source is relatively well understood. Since the sensitivity

733: and dynamic range

734: in the rarely used C and D configurations appears worse than attributable

735: to confusion or poorly excised narrow-band RFI, we hypothesize that 

736: an additional form of low-level, broad-band RFI is limiting the system

737: performance in those configurations. That suggests that 

738: the source of interference is the VLA itself, or equipment

739: at the site. Clearly another lesson for future instruments is design them

740: with a focus on eliminating or at least shielding against self-generated RFI.

741: 

742: \subsection{Amplitude Calibration}\label{sec:ampcal}

743: 

744: \subsubsection{Low-Frequency Calibration Problems and Advantages of the VLA System}\label{sec:ampcal1}

745: 

746: Despite relatively limited sensitivity compared to that typical of

747: higher frequency observations, the sensitivity of the 74~MHz VLA is

748: unprecedented for frequencies below~100~MHz, and so it has now become

749: possible to obtain flux density measurements for thousands of radio

750: sources that have never been measured before at these frequencies.

751: However, it is important to note that flux density scales are rather

752: uncertain below~100~MHz for several reasons. 

753: 

754: First, the Galactic background dominates system noise

755: temperatures, so system temperatures vary as an antenna

756: is scanned across the sky and system gain and temperature must be

757: measured continuously.  Antenna gains vary at less than the few \%

758: level as a function of time and are consistent with elevation dependent

759: deformation of the feed or dish structure being negligible at 4-m wavelength.

760: This trend is consistent with elevation-dependent gain variations at

761: L-band (20-cm) measured at less than 1\%, while variations at P band

762: (90-cm) have never been seen. $T_{\mathrm{sys}}$ certainly changes with

763: time, as it depends on $T_{\mathrm{sky}}$ which is a strong function of

764: sky position, being much greater on the Galactic plane than off. Early

765: single antenna tests showed variations of up to a factor of 2 between

766: ``on'' and ``off'' regions against the Galactic plane - being greatest

767: towards the Galactic center. This is much less than the known contrast

768: in brightness temperature between inner regions of the Galaxy and cooler

769: parts of the extragalactic sky as determined by early, lower resolution

770: measurements (e.g. \citealt{y68}). However that real variation is diluted

771: by the poor directivity of the 25-m dishes at 74 MHz, with $\leq$20\%

772: of the power in the main beam and the rest scattered in sidelobes across

773: the sky. Hence while $T_{\mathrm{sys}}$ does change as a function of time,

774: the variation is greatly smoothed out and is normally less than a factor

775: of two over the course of any pointed, full synthesis observations. A

776: related consideration is that calibrator observations of calibrators like

777: \objectname[]{Cyg~A} not drive the system response into a regime in which

778: a non-linear correction

779: of its correlation coefficient is required. Fortunately the increase on

780: $T_{\mathrm{sys}}$ when observing \objectname[]{Cyg~A} is modest. A rough

781: estimate from existing low resolution sky maps suggests an increase in

782: $T_{\mathrm{sys}}$ of $\leq$40\% in the main beam alone, again consistent

783: with early single antenna tests. Thus the contribution from the distributed

784: ``hot sky'' still dominates the measured power, and are consistent with the

785: measurements described below indicating that variations in system gain and

786: temperature are being tracked accurately (see also Sections 3 and 5.1).

787: 

788: Second, many systems utilize fixed antennas

789: whose power patterns are not very well determined.  The relative flux

790: densities of sources at adjacent declinations can be determined

791: reliably because they traverse the same parts of the fixed antenna's

792: pattern, but measurements of sources at widely separated declinations

793: are difficult.  Also, many measurements must be made far from the

794: maxima of the primary response patterns of the antennas, where the

795: patterns are less stable than at the maxima.  Finally, ionospheric

796: amplitude scintillations often disturb measurements below 50~MHz and

797: ionospheric absorption can affect them below 20~MHz.

798: 

799: The VLA has features that avoid most of these problems.  First, the

800: gain and noise temperature of each antenna are monitored continuously

801: by the injection of noise calibration signals into every preamplifier.

802: At higher frequencies, experience has shown that the gain properties

803: of the VLA electronics system are stable at the 0.1\% level, and that

804: linearity is preserved even if the T$_{\mathrm{sys}}$ changes by a

805: factor of 3 or more.  Second, the VLA antennas are pointed at each

806: source under observation, and the antenna power pattern is stable,

807: at least for measurements made near its center.  Sources at different

808: declinations should be directly comparable.  Third, 74~MHz is a high

809: enough frequency that ionospheric amplitude scintillations and

810: absorption are minimal, though occasional  episodes of scintillations

811: have been seen during observations of strong sources.

812: 

813: \subsubsection{VLA Calibration Method}

814: 

815: Because of the high system temperature of the 74~MHz system, the best

816: source to use for calibrating the visibility amplitudes is

817: \objectname[]{Cyg~A}.  However, as it is partially resolved, a good model

818: is required for accurate amplitude and phase calibration.  We have

819: used multiple observations in multiple VLA configurations to generate

820: such a model, and it is available

821: \anchor{http://rsd-www.nrl.navy.mil/7210/7213/lazio/tutorial/}{online}.

822: We base our flux density scale on that of \cite{bgp-tw77}, in which

823: \objectname[]{Cyg~A} has a flux density of~17\,086~Jy at~74~MHz. Although

824: \objectname[]{Cyg~A} is by far the best amplitude calibrator, other

825: sources can be employed, if necessary.  \objectname[]{Vir~A} is

826: acceptable in all VLA configurations, while \objectname[]{Tau~A} and

827: \objectname[]{Cas~A} can be used in the smaller configurations.  As

828: with \objectname[]{Cyg~A}, all of these sources are resolved

829: significantly, so model images (available online) are necessary for

830: calibration. The amplitude calibration is done in the usual way in AIPS,

831: with the task CALIB used to derive the antenna complex gains. Residual

832: amplitude errors are corrected by self calibration in the final stages

833: of the reduction. As described below, measurements demonstrate that

834: reliable fluxes, at the 5\% level, can be obtained down to the half power

835: beam point. In the rare case of severe ionospheric weather conditions

836: such as ionospheric scintillations (\S\ref{sec:pdelay_motivate}),

837: amplitude correction is not possible and the data are discarded.

838: 

839: 

840: \subsubsection{Reliability of the VLA flux-scale at 74~MHz}

841: 

842: In order to quantify our ability to cope in practice with the problems

843: described above (\S\ref{sec:ampcal1}), it is necessary to examine the robustness

844: of the flux density scale for the VLA. As part of a snapshot survey

845: (\S\ref{sec:example}), we observed a number of strong sources, which

846: we have used to constrain the gain stability of the 74~MHz system.

847: 

848: \cite{kwp-tn81} presented the spectra of~518 extragalactic radio

849: sources that have flux densities above~1~Jy at~5~GHz, using

850: data compiled from many catalogs.  The absolute flux density

851: scale was based on that of \cite{bgp-tw77}.  Although the data

852: below~100~MHz are rather sparse, the spectra of \cite{kwp-tn81} have

853: proven to be quite reliable in most cases.  They do not include any

854: sources near the Galactic plane and give analytic expressions for

855: spectra only when they are fairly simple (straight or moderately

856: curved).  Sources with complex spectra are included in the catalog,

857: but no attempt was made to fit such spectra with analytic expressions.

858: 

859: Of the 29 sources in our snapshot survey, eleven have spectra for

860: which \cite{kwp-tn81} give analytic expressions.

861: Table~\ref{tab:fluxscale} compares our measured 74~MHz flux densities

862: with the value predicted at~73.8~MHz from those expressions.  (See

863: also Table~\ref{tab:list}.)

864: 

865: Because of their relatively simple structure even at the highest angular

866: resolution, the four most reliable sources for comparison appear to be 

867: \objectname[3C]{3C~98}, \objectname[3C]{3C~123},

868: \objectname[3C]{3C~219}, and~\objectname[3C]{3C~274}.  These yield an

869: average flux density ratio (VLA/\citealt*{kwp-tn81}) of~0.99 $\pm$

870: 0.06.  A simple average of all eleven ratios yields 0.98 $\pm$ 0.13.

871: In either case the agreement between the \cite{kwp-tn81} and VLA flux

872: density scales appears to be as good as 2\%. We conclude that 74 MHz

873: flux density measurements are stable at the few percent level, and

874: that the absolute flux density scale as tied to the \citet{bgp-tw77}

875: value for Cygnus A is accurate to at least 5\%.  These results also

876: confirm that antenna system temperatures and gains are being tracked

877: correctly, and that power is being detected linearly. Hence the 74~MHz

878: flux scale is reliable for those sources never observed before at this

879: frequency. However, \cite{kwp-tn81} give correction factors that should

880: be used to adjust the data from the various catalogs to their flux

881: density scale. When making comparisons, these corrections should be used.

882: 

883: \subsection{Angle-Independent Phase Calibration}\label{sec:pcal}

884: 

885: The phase calibration procedure employed commonly at centimeter

886: wavelengths makes use of a secondary calibrator, a modestly strong

887: source much closer in the sky to the target source than the amplitude

888: calibrator.  It is chosen to be strong enough to dominate the total

889: flux within its field of view, so that a unique phase can be derived

890: for the contribution of the array's electronics.

891: 

892: In contrast, at~74~MHz the field of view is so large that few sources

893: are strong enough to dominate the field.  Thus, an \emph{initial}

894: phase calibration is determined from \objectname[]{Cyg~A} (or another

895: strong source such as those listed above, \S\ref{sec:ampcal}), which

896: usually provides sufficient coherence on enough short ($\leq 10$~km)

897: baselines to form an initial image even if the target source is in a

898: completely different portion of the sky. Oftentimes, coherence is

899: retained even after the phases are smoothed over long time scales,

900: even longer than the duration of the observation. 

901: 

902: \emph{This technique works only for bright sources. Implicit in this

903: procedure is the assumption that a single phase suffices for the entire

904: field of view---hence its \emph{angle-invariant} designation.}

905: 

906: 

907: This initial phase calibration works because relatively larger scales

908: ($\gtrsim 50$~km, i.e., larger than the VLA) present in the ionosphere dominate

909: the phase fluctuations. It is also the reason previous low frequency

910: ($<100$~MHz) systems with baselines less than about~5~km were able to function

911: without ionospheric correction techniques---except for a constant refractive

912: shift arising from the large-scale structures, short baselines ($\lesssim

913: 5$~km) were, to first order, relatively unaffected by the ionosphere

914: \citep{e84}\footnote{Short baselines are undoubtedly affected by small scale

915: ionospheric structure beyond simple refractive shifts, but those effects

916: were negligible for past arrays where the intrinsic sensitivity was already

917: confusion limited to Jy-levels. As a new generation of low frequency arrays

918: aims to achieve much greater levels of sensitivity, the phase variations due

919: to small scale ionospheric structure may well be the limiting factor.}.

920: The initial image obtained from this calibration must be subsequently

921: astrometrically corrected and self-calibrated.

922: 

923: We write the observed phase on a given baseline as

924: \begin{equation}

925: \phi(t) = \phi_{\mathrm{src}}(t) + \phi_{\mathrm{VLA}} + \phi_{\mathrm{ion}}(t)

926: \label{eqn:visphase}

927: \end{equation}

928: where $\phi_{\mathrm{src}}$ is the phase contributed by the intrinsic

929: structure of the source, $\phi_{\mathrm{VLA}}$ is the phase

930: contributed by the VLA electronics, and $\phi_{\mathrm{ion}}$ is the

931: ionospheric contribution.  We write $\phi(t)$ to emphasize that the

932: phase is a function of time, not only because the Earth-rotation

933: synthesis means that the phase can change as the array samples

934: different portions of the $u$-$v$ plane ($\phi_{\mathrm{src}}$) but 

935: also because of the temporal variations imposed by the ionosphere

936: ($\phi_{\mathrm{ion}}$).\footnote{%

937: In practice, $\phi_{\mathrm{VLA}}$ can also be a function of time

938: because of phase jumps in the electronics.  These are sufficiently

939: rare that we shall ignore them.%

940: }  During calibration observations of a strong source, the first

941: term~$\phi_{\mathrm{src}}$ can be calculated and removed using a model

942: of the source, so we shall ignore it henceforth.

943: 

944: The smallest scales in the ionosphere over which the rms phase varies

945: by more than 1~radian at~74~MHz are typically no smaller than 5~km.  As a

946: result, the ionospheric phase term, $\phi_{\mathrm{ion}}$, can be

947: considered to be relatively constant across at least the short

948: baselines in the array ($\lesssim 5$~km).  Transferring the phases

949: determined toward the calibrator source (i.e., \objectname[]{Cyg~A})

950: to another direction in the sky introduces a term,

951: $\phi_{\mathrm{ion,cal}}$, the ionospheric phase distortion toward the

952: calibrator.  However, for the short baselines, this term is

953: effectively constant.  It introduces little more than a refractive

954: shift of the apparent source positions \citep{e84}.  This refractive

955: shift can be removed by registering the image with an existing all-sky

956: survey at a higher frequency (i.e., the NVSS, \citealt*{ccgyptb98};

957: WENSS, \citealt*{rtdmbrb97}; and SUMSS, \citealt{bls99}).

958: 

959: This initial, crude phase calibration is sufficient to produce an

960: image.  The initial image then serves as the initial model for hybrid

961: mapping, which consists of iterative loops between self-calibration

962: and imaging.  With the large number of antennas in the full VLA,

963: convergence occurs rapidly.  In this respect, the process is similar

964: to that employed often in very long baseline interferometry (VLBI) in

965: which a crude initial model is combined with several iterations of

966: hybrid mapping to produce the final image \citep{w95}.

967: 

968: Phase self-calibration is warranted only if a sufficient

969: signal-to-noise level can be obtained \citep{cf99}.  In practice there

970: are two signal-to-noise level thresholds that must be met, but both

971: can be exceeded quite easily.  The first signal-to-noise threshold

972: that must be met is that there must be a source or sources that are

973: strong enough to be detected in the approximate ionospheric coherence

974: time ($\approx 1$~min.).  Both an extrapolation of higher frequency

975: source counts \citep{b99} and source counts derived from 74~MHz

976: observations indicate that there should be roughly 150~Jy of flux

977: from different sources within the primary beam, originating from

978: sources stronger than about~5~Jy, which is about~5 times the rms

979: noise level obtained in a 1~min.\ integration time.

980: 

981: The second signal-to-noise threshold is that the phase derived from

982: the calibrator source must dominate over weaker sources in the field.

983: This criterion can be understood by considering the calibrator, with a

984: flux density~$S_{\mathrm{cal}}$, to be immersed in a ``sea'' of

985: randomly-located background sources, with a typical flux

986: density~$S_b$.  Treating the visibilities of the sources as phasors,

987: the background sources will contribute to a jitter of the phase

988: determined for the calibrator.  A rough estimate of the magnitude of

989: this jitter is

990: \begin{equation}

991: \delta\phi \sim \sqrt{N}\frac{S_b}{S_{\mathrm{cal}}},

992: \label{eqn:phasestable}

993: \end{equation}

994: where $N$ is the number of background sources in the field of view.

995: For the purposes of a rough estimate, we take $\delta\phi \sim 0.2$

996: (implying a phase jitter signal-to-noise threshold of~5).  With $S_b

997: \sim 5$~Jy and $N \sim 40 (= 200\,\mathrm{Jy}/5\,\mathrm{Jy})$, we

998: find $S_{\mathrm{cal}} \sim 150$~Jy.  Though clearly a rough estimate,

999: experience has shown that it is possible to achieve successful phase

1000: self-calibration with as little as 50~Jy from bright sources in the

1001: initial model.  More generally, nearly every randomly-picked field of

1002: view will contain at least one 3C object (or equivalent at southern

1003: declinations) whose flux density alone is close to this minimum value,

1004: and existing all-sky surveys at higher frequencies can be used to

1005: identify the (few) strongest objects totaling at least 100~Jy within

1006: the field of view. (The phase fluctuations in the top panel of

1007: Figure~\ref{fig:ptlink} are consistent with the estimate derived here

1008: for the phase jitter due to background sources.)

1009: 

1010: A potential weakness of this method is its limited utility to the

1011: larger configurations (A and~B).  Depending upon the state of the

1012: ionosphere, it may be difficult or impossible to obtain sufficient

1013: signal-to-noise on the more limited number of short baselines.  In

1014: turn, this may impair one's ability to produce an initial model for

1015: self-calibration.

1016: 

1017: An alternate phase calibration strategy does not rely on the phases

1018: transferred from a distant source.  Instead, a strong source, like

1019: \objectname[]{Cyg~A}, is used to establish the amplitude scale.  An

1020: all-sky survey (particularly the NVSS or WENSS) is used to construct a

1021: sky model\footnote{%

1022: Within \aips, the sky model is constructed using \task{FACES}.

1023: } within the primary beam; for sources north of

1024: declination~$+30\arcdeg$, the lower frequency WENSS is preferable to

1025: \hbox{NVSS}.  This sky model then serves as an initial model for

1026: self-calibration.  In effect, no attempt is made to use an external

1027: calibrator to calibrate the phases.  A primary benefit of this

1028: approach is to produce a map with astrometrically correct positions,

1029: as the coordinate system of the map is locked to the \textit{a priori} 

1030: known position of sources in the sky, while self-calibration with a

1031: model produced from raw data locks the final position to an arbitrary

1032: sky position determined by the position of the model or ionospheric

1033: refraction. This alternate strategy

1034: is a simplified application of the field based phase calibration technique,

1035: described in the next section.

1036: 

1037: \subsection{Field-Based Phase Calibration}\label{sec:pdelay}

1038: 

1039: The prior calibration strategy is most useful at~74~MHz within a

1040: restricted field of view containing a strong source or strong sources

1041: located relatively close together.  In this section we first motivate

1042: why phase angle independent self-calibration is insufficient for all

1043: observations, then describe the method we have developed to handle

1044: more general observations.

1045: 

1046: \subsubsection{Motivation}\label{sec:pdelay_motivate}

1047: 

1048: The isoplanatic patch is the characteristic scale over which the rms

1049: phase difference between two lines of sight is approximately 1~radian.

1050: At low frequencies the size of the isoplanatic patch is determined

1051: primarily by the ionosphere.  The phase contributed by a cold plasma

1052: is

1053: \begin{eqnarray}

1054: \phi &=& r_e\lambda\int_0^D ds\,n_e(s),\nonumber\\

1055:      &=& r_e\lambda N_e,

1056: \label{eqn:plasma}

1057: \end{eqnarray}

1058: where $r_e$ is the classical electron radius, $\lambda$ is the

1059: free-space wavelength, $n_e$ is the electron number density, and $N_e$

1060: is the electron column density (also known as the total electron

1061: column, TEC).  For reference, two lines of sight for which $\Delta\phi

1062: \sim 1$~rad at~74~MHz would differ in electron column density (TEC) by $\Delta

1063: N_e \sim 10^{14}$~m${}^{-2}$.

1064: 

1065: Two lines of sight separated by the diameter of the primary beam

1066: pierce the F-layer of the ionosphere (the most dense region of the ionisphere

1067: at an altitude of $\sim$400~km) at a linear separation as large

1068: as 80~km.  This is larger than the size of the array itself, even with

1069: the PT link, and the ionosphere can have significant column density

1070: variations at much smaller scales ($\lesssim 10$~km), so the primary

1071: beam typically includes multiple isoplanatic patches.

1072: Figure~\ref{fig:isoplane}, which itself is a portion of a larger

1073: image, shows the effect of constructing an image larger than the

1074: isoplanatic patch.  Clearly evident is a systematic distortion of the

1075: sources attributable to incorrect phase calibration.

1076: 

1077: Figures~\ref{fig:ionshift}--\ref{fig:ionsmall} illustrate the

1078: successively higher order ionospheric effects that lead to the

1079: break-down of a simple angle-invariant self-calibration.

1080: Figures~\ref{fig:ionshift} and~\ref{fig:ionwedge} shows the effects of

1081: the largest scale ($> 1000$~km) ionospheric structure, a ``wedge''

1082: that acts to shift the entire field of view, without source

1083: distortion, on time scales of minutes. This structure dominates the

1084: total electron content (TEC) and also causes Faraday rotation of

1085: linear polarization. A GPS-based method to correct this effect has

1086: been demonstrated at the VLA \citep{epfk01}, but self-calibration

1087: alone can compensate for it if the time scale on which the phase

1088: corrections are calculated is sufficiently short to track this gross

1089: refraction.  It will, however, leave the field with a gross

1090: astrometric offset from the correct source positions.

1091: 

1092: Figure~\ref{fig:iontid} shows the phase effects imposed by ionospheric

1093: mesoscale structure.  These mesoscale structures are due typically to

1094: traveling ionospheric disturbances (TIDs), with scale sizes on the order of

1095: hundreds of kilometers, column densities of order $10^{15.5}$~m${}^{-2}$

1096: and periods less than 90 minutes.  While the size of the VLA---even in

1097: its A configuration---is smaller than that of a TID, it is still

1098: sufficiently large that TIDs can impose a (mainly) linear phase

1099: gradient down the arms of the \hbox{VLA}.  The effect of this linear

1100: phase gradient can be seen as a (nearly) linear increase in the phase

1101: with increasing distance from the array center; the amplitude of the

1102: offset also increases with baseline, as the antenna separation becomes

1103: a larger fraction of the \hbox{TID}.  (In the example shown in

1104: Figure~\ref{fig:iontid}, the traveling nature of the disturbance is

1105: also present as a lag between the times when the maximum phase offset

1106: is obtained at the various antennas.)  Self-calibration can remove the

1107: effects of TIDs only for a limited region in the field of view.  At

1108: any given instant, sources outside this region are ``differentially

1109: refracted'' or shifted by varying amounts in proportion to their

1110: distance from nominal direction of the self-calibration solution.

1111: Because a typical observation lasts much longer than the time it takes

1112: a TID to pass over the array, the differential refraction changes with

1113: time.  Thus, sources in a map made using all of the data {are} blurred

1114: as well.

1115: 

1116: \cite{je92a,je92b} conducted an extensive study of acoustic gravity wave

1117: generated TIDs having phase speeds less than 200~m~s${}^{-1}$ and

1118: periods greater than $10^3$~s using the VLA at~330~MHz.  They found

1119: them to have a quasi-isotropic distribution in azimuth.  On shorter

1120: time scales ($< 300$~s) they found the ionospheric phase effects to be

1121: dominated by faster ($> 200$~m~s${}^{-1}$) magnetic eastward-directed

1122: disturbances (MEDs).  Over the baseline relevant to the extended VLA

1123: configurations and the PT link ($\sim 50$~km), both TIDs and MEDs

1124: contribute to~74~MHz phase effects.

1125: 

1126: Figure~\ref{fig:ionsmall}a-f were obtained from an 8 hour observation

1127: towards a strong source (Virgo A), and illustrate a variety of effects

1128: due to ionospheric phenomena typically observed at the VLA on scales

1129: sizes both larger and smaller than TIDs. Figures~\ref{fig:ionsmall}~a

1130: and b illustrate first order effects on the visibility phase

1131: towards Virgo A. Figures~\ref{fig:ionsmall}~c and d track the flux density and 

1132: apparent position of Virgo A, while Figures~\ref{fig:ionsmall}~e and f

1133: reflect differential refraction towards five field objects after the

1134: first order term has been

1135: removed. These differential phase effects due to smaller scale ionospheric

1136: structures, on scales of tens of kilometers, are the most intractable and

1137: pose A MAJOR challenge for future instruments. (The high frequency

1138: ``jitter'' superposed on the TID-produced structure in

1139: Figure~\ref{fig:iontid} is also contribution from smaller scale structure.)

1140: These phase effects are generated by turbulent structures comparable to OR

1141: SMALLER than the size of the VLA, and lead to source distortion.

1142: There is no current

1143: means of correcting for these effects over a wide field of view and when

1144: they are severe, the data must be discarded.  If these small-scale 

1145: structures generate phase distortions larger than 1~radian on scales

1146: smaller than the size of the VLA antennas, the result is ionospheric

1147: scintillations (also illustrated in Figure 14).  We believe that there

1148: will never be a means for compensating for this effect, at least so far

1149: as imaging applications are concerned, as the data are effectively smeared

1150: in phase before they arrive at the antennas.  (Ionospheric scintillations

1151: can be useful for studying the ionosphere itself, though.) Fortunately 

1152: ionospheric scintillations are rare at the VLA at~74~MHz, but when they

1153: do occur those portions of the data must be removed.

1154: 

1155: In principle, a data-adaptive calibration scheme, based on

1156: self-calibration, could be used to remove the phase errors that result

1157: from imaging a region larger than the isoplanatic patch.  Current

1158: implementations of self-calibration\footnote{%

1159: This description includes \texttt{CALIB} within \aips.%

1160: } model the phase error on the $i^{\mathrm{th}}$ antenna at time~$t$

1161: as a single quantity, $\delta\phi_i(t)$, equivalent to assuming that

1162: the isoplanatic patch has an infinite extent above the array.  The key

1163: assumption for self-calibration---that errors can be modeled as being

1164: antenna-dependent---remains true for a data-adaptive scheme.  Thus, a

1165: joint multi-source self-calibration, in which the calibration

1166: correction would become a function of position on the sky,

1167: $\delta\phi(t; \alpha, \delta)$, seems possible.  Limited testing

1168: suggests that such a scheme could work, though it probably requires

1169: considerably better signal-to-noise than can be achieved with the VLA;

1170: no comprehensive attempt to implement and assess such a scheme has

1171: been performed yet. In principle such a technique might allow 

1172: wide-field imaging across the full extent of the primary beam

1173: on the longest baselines allowed by natural limits of brightness temperature

1174: sensitivity (for a related discussion, see also \cite{Erickson2006}).

1175: 

1176: When it is desired to image an entire field of view, the assumption of

1177: an infinite isoplanatic patch is no longer valid. The effects of such

1178: an assumption can be seen as a tendency to detect a larger number of

1179: sources toward the direction of the strongest source in the field

1180: \citep{cohenetal03}. Figure~\ref{fig:SCbias} illustrates this effect

1181: in a field containing \objectname[3C]{3C~63} ($\approx 35$~Jy).

1182: The density of sources across the field is clearly non-uniform.

1183: Besides the effects of anisoplanaticity, which shows up at scales of

1184: 5-10$^{\circ}$, the non-uniform density can also arise from two other

1185: related causes. First, if sufficiently high-order ionospheric distortions

1186: are present, they contribute to increasing phase errors at increasing

1187: distances from the effective phase center.  Sources at large distances

1188: will be blurred, thereby decreasing their brightnesses. Second,

1189: Figure~\ref{fig:wander} demonstrates that the apparent source positions

1190: also wander over time. Because the amount and direction of the wander can

1191: vary both over the field of view and with time, sources are smeared further

1192: and their brightnesses decrease further.  Only in the neighborhood of a

1193: strong source are the self-calibration solutions able to track the

1194: ionospheric phase distortions accurately.  If one is interested in

1195: imaging only a strong source, the non-uniform distribution of sources

1196: is usually unimportant.  If a field does not contain a strong source,

1197: the self-calibration solutions represent some ``average'' ionospheric

1198: phase distortion across the field.  Again, depending upon the scientific

1199: problem being attacked, using angle-independent self-calibration may be

1200: sufficient.

1201: 

1202: \subsubsection{Implementation}\label{sec:pdelay_implement}

1203: 

1204: Compensation for higher order ionospheric phase distortions is

1205: required \citep{cc02,cotton04}, when the entire field of view is of interest

1206: (e.g., in the VLA Low-frequency Sky Survey \cite{cohen07, cohen06})

1207: or when the object of interest is ``close'' to a strong source.  

1208: 

1209: 

1210: The latter case happens

1211: fairly frequently given that the typical separation of 3C sources on

1212: the sky is approximately 8\arcdeg, comparable to the size of the VLA

1213: field of view.  Moreover, the dynamic range required to detect or

1214: image a weak source close to a strong source may impose much more

1215: stringent constraints on the need to determine the ionospheric phase

1216: than the approximate 1~radian criterion that we gave at the beginning

1217: of the previous section.

1218: 

1219: In such cases, we model the ionosphere as a phase-delay

1220: screen.\footnote{%

1221: Within \aips, this procedure is implemented using \task{VLAFM}, a

1222: special-purpose task added to \aips.

1223: }  We return to equation~(\ref{eqn:visphase}), assume that

1224: $\phi_{\mathrm{src}}$ is calculable and can be removed, and expand the

1225: ionosphere term as

1226: \begin{equation}

1227: \phi

1228:  = \phi_{\mathrm{VLA}} + \phi_{\mathrm{ion,lo}} 

1229:  + \phi_{\mathrm{ion,hi}}.

1230: \label{eqn:ioncal}

1231: \end{equation}

1232: Here $\phi_{\mathrm{ion,lo}}$ refers to the phase distortion

1233: introduced by low spatial frequency structures in the ionosphere while

1234: $\phi_{\mathrm{ion,hi}}$ refers to high spatial frequency structures.

1235: In order to make a division between ``high'' and ``low'' spatial

1236: frequencies (or ``long'' and ``short'' wavelengths) we make a

1237: ``frozen-flow'' approximation in which the ionospheric structures are

1238: assumed not to change internally over the time it takes for them to be

1239: transported across the array.  Under this assumption, spatial and

1240: temporal scales become equivalent.  The smallest structures ($\lesssim

1241: 10$~km) are transported at speeds of order 100~m~s${}^{-1}$ on time

1242: scales of order 100~s.  We consider these to be high spatial

1243: frequencies.  Structures of~100~km or larger are transported across in

1244: time scales of order 15~min.; these are low spatial frequencies.

1245: 

1246: The large-scale structures are larger than even the maximum baselines

1247: of the A configuration.  We therefore decompose the low-frequency

1248: ionosphere term into low-order Zernike polynomials

1249: \begin{equation}

1250: \phi_{\mathrm{ion,lo}} = \sum_{n=1}^2\sum_l A^l_n Z^l_n

1251: \label{eqn:zernike}

1252: \end{equation}

1253: where $Z^l_n$ is the Zernike polynomial, $A^l_n$ is the coefficient of

1254: that polynomial, and the standard conditions apply that $n \ge |l|$

1255: and $n - |l|$ is even.  The $n = 1$ terms account for the large-scale

1256: refractive shift of the field of view while the $n = 2$ terms describe

1257: astigmatism or differential refraction within the field of view.  The

1258: $n = 0$ term is not used because it represents an overall phase

1259: advance or delay (``piston'') to which the interferometer is

1260: insensitive.  We use Zernike polynomials because they represent a

1261: class of polynomials orthogonal on a circle.  Thus, they are useful

1262: for representing distortions across the aperture of the

1263: array.\footnote{%

1264: The phase-delay screen representation and Zernike modeling of the

1265: ionosphere is not strictly correct as the ionosphere is not in the

1266: far-field of the array.  Nonetheless, we view the Zernike

1267: polynomials as a useful first step in modeling the ionosphere.

1268: }

1269: Table~\ref{tab:zernike} summarizes the polynomials used.  The

1270: methodology is quite similar to that used in adaptive optics systems

1271: in optical astronomy.  One important difference between our use of the

1272: Zernike polynomials and that in adaptive optics, however, is that

1273: these polynomials are used to describe wavefront errors in or near the

1274: aperture for the case of adaptive optics systems in optical astronomy

1275: whereas here they represent errors quite far from the aperture plane.

1276: 

1277: In order to derive the required corrections, snapshot images of

1278: sources in an ``astrometric grid'' are produced, and the offsets

1279: between the apparent and expected locations of sources in the

1280: astrometric grid are determined.  The snapshots must be formed on

1281: short enough timescales so as to track the ionospheric phase

1282: variations, typically 1~min.\ or shorter.  Both the NVSS and WENSS can

1283: be used to produce this astrometric grid as both are constructed at a

1284: high enough frequency and from a sufficiently large number of sources

1285: that the positions of sources within these catalogs is known to an

1286: accuracy much better than the synthesized beam at~74~MHz, even in the

1287: A configuration.  From the source offsets, the coefficients of the

1288: Zernike polynomials can be found in a least-squares minimization.

1289: Figure~\ref{fig:zernike} shows an example of a (subset) of an

1290: astrometric grid and the resulting $\phi_{\mathrm{ion,lo}}$ over the

1291: VLA for a particular observation.

1292: 

1293: The snapshot images of the astrometric grid sources are useful only if

1294: two conditions are met.  First, the ionosphere must be stable enough

1295: that sources are not defocused seriously but merely shifted from their

1296: expected positions by refraction.  Second, one must be able to

1297: determine $\phi_{\mathrm{VLA}}$ prior to forming the snapshot images.

1298: As in \S\ref{sec:pcal}, the initial phase calibration is determined

1299: from observations of \objectname[]{Cyg~A}, or another primary calibrator.

1300: For a single source at a

1301: well-known position, one need solve for only the $n = 1$ terms

1302: describing an overall refractive shift.  Correcting for this global

1303: refractive shift should yield $\phi_{\mathrm{VLA}}$.  If

1304: $\phi_{\mathrm{ion,hi}}$ is non-negligible (meaning that sources may

1305: be defocused), it will corrupt estimates of $\phi_{\mathrm{VLA}}$.

1306: Because no phase calibration is performed during this strategy (but

1307: see below), any errors in determining $\phi_{\mathrm{VLA}}$ remain in

1308: the data during all subsequent processing.

1309: 

1310: Figure~\ref{fig:SCunbias} presents the same field as in

1311: Figure~\ref{fig:SCbias}, only this time calibrated using this

1312: field-calibration strategy.  The density of sources across the field

1313: is seen to be far more uniform.  Figure~\ref{fig:sf} quantifies the

1314: improvement that the field-based calibration; it shows the rms jitter

1315: in the apparent separations of pairs of sources of various

1316: separations.  This rms jitter is a fairly direct measure of the

1317: refraction differences induced by the low spatial frequency

1318: ionospheric irregularities.  For separations greater than 2$^\circ$,

1319: the phase screen corrections dramatically reduce the jitter.  Also

1320: important is that the rms jitter shows no trends as distance from the

1321: phase center increases.

1322: 

1323: One weakness in our current implementation of this strategy is that

1324: the ionosphere is modeled in a piece-wise fashion in time.  No

1325: ``smoothness'' constraint is applied to Zernike models from adjoining

1326: snapshots.  Work is ongoing to rectify this weakness.

1327: 

1328: In order to assess if (or to what extent) $\phi_{\mathrm{ion,hi}}$ is

1329: corrupting our estimate of $\phi_{\mathrm{VLA}}$, we observe

1330: \objectname[]{Cyg~A} multiple times during the course of an observing

1331: run.  By comparing or, more often, averaging the various estimates of

1332: $\phi_{\mathrm{VLA}} + \phi_{\mathrm{ion,hi}}$, we seek to minimize

1333: the contribution of $\phi_{\mathrm{ion,hi}}$ to our estimate of

1334: $\phi_{\mathrm{VLA}}.$\footnote{%

1335: This process is implemented in the task \texttt{SNFLT}, a

1336: custom-designed task designed to work within \aips\ and available from

1337: the authors upon request.

1338: }

1339: 

1340: We close this section with a few general comments on our choice of

1341: Zernike polynomials and our implementation.  From the work of

1342: \cite{je92a,je92b} showing that much of the ionospheric structure

1343: above the VLA is in the form of waves, one might wonder if a Fourier

1344: representation would not be more appropriate. We have chosen to use

1345: Zernike polynomials to describe the ionospheric structures precisely

1346: because they were invented for the purpose of describing phase errors

1347: across a circular aperture.  Moreover, compared to a rectangular

1348: Fourier transform, many fewer terms of Zernike polynomials are

1349: required to describe the ionospheric phase fluctuations.  Given the

1350: limited sensitivity of the VLA, this criterion is quite important.

1351: Finally, although it is not yet contained within our implementation, a

1352: natural extension of our method would involve requiring the

1353: ionospheric phase corrections to be smooth in time.  In order to

1354: impose this requirement, one requires an orthogonal basis for the

1355: modeling since the interpolation in time is by interpolating the

1356: coefficients and this only works if the terms being interpolated are

1357: orthogonal.

1358: 

1359: We emphasize that our division between $\phi_{\mathrm{ion,lo}}$

1360: and~$\phi_{\mathrm{ion,hi}}$ is phenomenological and has the effect of

1361: making the division based on the order of the Zernike polynomials used

1362: rather than on physical properties of the ionosphere.  The range of

1363: spatial scales in the ionosphere implies that one could use

1364: higher-order Zernike polynomials to decompose the phase distortions.

1365: In principle, by incorporating a sufficient number of Zernike

1366: polynomials one could reduce $\phi_{\mathrm{ion,hi}}$ to a

1367: sufficiently low level so as to be unimportant; orders as high as 80

1368: are not unusual for correcting large optical telescopes.  However, the

1369: number of Zernike coefficients that can be determined is limited by

1370: both the sensitivity of the system and the \emph{aperture}

1371: distribution of the \hbox{VLA}.  The sensitivity of the VLA is an

1372: issue because the field must be imaged on less than the ionospheric

1373: coherence time, which in turn depends upon the array configuration as

1374: the amount of phase contributed by the ionosphere can depend upon the

1375: baseline length (Figure~\ref{fig:iontid}).  In 1~min., a coherence

1376: time appropriate for the B configuration, a typical field of view

1377: contains no more than 15--20 sources strong enough to be detected; for

1378: A configuration, a more typical coherence time is approximately 20~s.

1379: Even during times of ionospheric quiescence, fewer than 10 sources may

1380: actually be detected; during poor ionospheric ``weather'' conditions

1381: or when sidelobes from strong sources obscure weaker sources, the

1382: actual number detected with any confidence may be no more than 5

1383: sources.  Hence, in order to avoid (or minimize the occurrence of)

1384: spurious Zernike coefficients, we have restricted the modeling to only

1385: the $n < 3$ terms.  This model of the ionosphere is constructed anew

1386: every 1--2~min.

1387: 

1388: The aperture distribution (as opposed to the more traditional concerns

1389: in interferometry regarding the $u$-$v$ distribution) is related to

1390: the range of spatial scales sampled in the ionosphere.  

1391: In order to characterize large-scale ionospheric structures (e.g.,

1392: TIDs) that are of concern under typical ``weather'' conditions, only a

1393: modest number of pierce points through the ionosphere are required.

1394: In this respect the aperture distribution of the compact

1395: configurations (C and~D) provides adequate sampling of the relevant

1396: spatial scales, though such sampling is also often not needed because

1397: the array remains nearly coherent in these configurations.  In

1398: contrast, in order to characterize smaller scale ionospheric

1399: structures, a high density of closely spaced pierce points is

1400: required, which in turn requires high spatial frequencies in the

1401: aperture plane, rather than in the $u$-$v$ plane.  In the extended

1402: configuration (A and~B), the sparseness of the aperture distribution

1403: means that small spatial scales are hardly sampled at all. The design of

1404: future low frequency instruments may require a compromise between the

1405: good uv coverage important for imaging and the good aperture plane 

1406: coverage that might be required to allow sufficient modelling of small

1407: scale ionospheric structure.

1408: 

1409: 

1410: \subsection{Phase Transfer}\label{sec:ptransfer}

1411: 

1412: Under especially severe ionospheric weather conditions (with

1413: ionospheric phase rates on long baselines in excess

1414: of~1~deg~s${}^{-1}$), it may become necessary to scale the 330~MHz

1415: ionospheric-induced phase rates, transfer, and remove them from the

1416: 74~MHz data stream \citep{kped93}.  In practice, this dual-frequency

1417: ionospheric phase transfer technique has been required only rarely, as

1418: even in the A configuration phases transferred from a strong source

1419: anywhere in the sky retain sufficient coherence on enough short

1420: spacings to provide an initial model for self-calibration.  Subsequent

1421: iterations of self-calibration then improve the phases on the longer

1422: baselines. As with straight-forward self-calibration, phase transfer

1423: does not compensate for the main failing of self-calibration, which is

1424: the lack of an angular dependence on the antenna based phase

1425: solutions. However phase transfer may well be of great benefit to future

1426: low frequency instruments, especially those planned to operate at lower

1427: frequencies and longer baselines than the VLA, such as the LWA 

1428: \citep{Kassim2004,Kassim2006} and LOFAR (Kassim et al. 2004). Therefore it is

1429: important that their design does not preclude the possibility of

1430: simultaneous observations at multiple frequencies.

1431: 

1432: \subsection{Evolving Techniques of Phase Calibration}

1433: 

1434: The previous sections illustrate successive schemes of ionospheric

1435: phase calibration that continue to evolve with real observational

1436: experience. A technique such as joint, multi-source self-calibration

1437: is required to realize the full potential of emerging larger

1438: low-frequency instruments such as the LWA and LOFAR but has yet to

1439: be developed and tested. Other related appoaches to the calibration

1440: of large, low frequency arrays continue to be proposed and discussed

1441: \citep{Erickson2006,Nijboer2006,Cotton2006,Brentjens2005,Noordam2004,Lane2004}.

1442: 

1443: 

1444: Table~\ref{tab:ioncalib} presents a simple overview of these evolving

1445: schemes; the illustrations in Cotton et al. (2004) may help

1446: orient the reader with respect to the applicable geometries.

1447: 

1448: \section{Imaging at Low Frequencies}\label{sec:imaging}

1449: 

1450: In this section we summarize the essential procedures needed for

1451: effective imaging in the presence of effects that are particularly

1452: significant at these low frequencies.

1453: 

1454: \subsection{Confusion and Sidelobes}\label{sec:confuse}

1455: 

1456: A key limitation to previous low frequency interferometers operating below

1457: 100 MHz has been the poor and confusion-limited sensitivity that arose from

1458: their low angular resolution.  The high angular resolution of the larger (A-

1459: and B-) configurations of the VLA provides some mitigation, but the poor

1460: forward gain of the antennas increases the confusion.  In fact, two effects

1461: are at work, and we distinguish between \emph{classical} confusion and

1462: \emph{sidelobe} confusion.  The former occurs when the density of sources

1463: within the synthesized beam becomes so large that they cannot be separated

1464: (e.g., 1 source per every 10~beams is a common criterion for the onset of

1465: classical confusion).  The latter results from the incompletely removed

1466: sidelobe response to bright sources.

1467: 

1468: Table~\ref{tab:confuse} summarizes the classical confusion limits for

1469: the four VLA configurations, where we have taken classical confusion

1470: to occur when there is one source per 10 synthesized beams.  In order

1471: to estimate these classical confusion limits, we have adopted a $\log

1472: N$-$\log S$ relation derived from existing 74~MHz observations,

1473: \begin{equation}

1474: N(>S)

1475:  = 1.25\,\mathrm{deg}^{-2}\,\left(\frac{S}{1\,\mathrm{Jy}}\right)^{-1.25}

1476: \label{eqn:counts}

1477: \end{equation}

1478: where $N(>S)$ is the areal density of sources, in units of

1479: deg${}^{-2}$, stronger than $S$~Jy.  Strictly, this $\log N$-$\log S$

1480: relation is valid only for flux densities $S \gtrsim 0.25$~Jy, based

1481: on the flux densities of the sources from which it was determined.

1482: Expectations of source densities, based on higher frequency source

1483: counts, suggest that this $\log N$-$\log S$ relation will tend to

1484: overestimate the confusion flux density levels, though, and estimates

1485: using source counts scaled from higher frequencies give similar

1486: results.  Even with the uncertainty in the low-frequency $\log

1487: N$-$\log S$ relation, the results presented in  Table~\ref{tab:confuse}

1488: show that classical confusion is unlikely to limit the sensitivity

1489: of~74~MHz observations in the A configuration and probably not in

1490: the B configuration (possibly only in crowded regions like the Galactic

1491: center), but it is a serious factor for C- and D-configuration observations.

1492: 

1493: Sidelobe confusion results from the improperly subtracted response to bright

1494: sources both inside and outside the main field of view, and is exacerbated

1495: when those sources are unresolved by the synthesized beam. The dominant

1496: effect is often from the pathlogically brightest sources (e.g. \objectname[]{Cyg~A}, \objectname[]{Cas~A},

1497: etc) outside the main field of view whose emission ``rumble'' through the

1498: primary beam sidelobes.  This problem is particularly severe for the 74~MHz

1499: VLA because of the relatively small aperture ($\approx 6\lambda$) and the

1500: sidelobes caused by the feed support structure. Thus the primary antenna gain is low, the primary beam is large (FWHM

1501: $\approx 11\arcdeg$), and the sidelobe levels are high and asymmetric (peak

1502: sidelobes are typically only 20~dB down from the main lobe), with a large

1503: area of sky in the ``close-in sidelobe'' region (Figure~\ref{fig:beam}).

1504: Moreover, the receiver noise is a small fraction of the total system

1505: temperature, so objects all over the sky, including those seen through the

1506: sidelobes, produce measurable coherence.  Because of this ``signal-rich''

1507: environment at~74~MHz, it is often necessary to image the entire primary

1508: beam and a small number of sources outside it.

1509: 

1510: Competing factors mitigate the effects of sidelobe confusion, including 

1511: delay beam and "ionospheric" smearing of the residuals of subtracted 

1512: sources. Nevertheless remaining artifacts from the few, strongest 

1513: sources outside the main field of view often limit the sensitivity and 

1514: dynamic range. Sidelobe confusion can still dominate over classical 

1515: confusion even in the more compact configurations, since in those 

1516: cases the arc-minute size scale of the normally offending sources 

1517: (most notably \objectname[]{Cyg~A}, \objectname[]{Cas~A}, 

1518: \objectname[]{Vir~A}, \objectname[]{Her~A}, \objectname[]{Hyd~A}, and 

1519: \objectname[]{Tau~A}) prevents them

1520: from being resolved out, and hence their residual effects are worse. (The

1521: dominance of sidelobe confusion over thermal noise was indicated earlier in

1522: Figures 7 and 8 and discussed In \S\ref{sec:general}. As noted earlier, the

1523: sensitivity in the compact configurations is worse than attributable to

1524: either form of confusion alone, and our hypothesis is that low-level,

1525: broad-band RFI, possibly self-generated, are limiting the performance.

1526: 

1527: 

1528: This situation is in direct contrast to that at centimeter wavelengths

1529: \citep[viz.\ Figure~15]{ccgyptb98} where the receiver temperatures

1530: dominate the sky temperature, a more uniform aperture illumination

1531: produces lower sidelobes, and the non-thermal spectra result in most

1532: sources being fainter than at low frequencies, so that only rarely

1533: does one have to contend with sources outside the primary beam.

1534: 

1535: \subsection{Wide-field Imaging}\label{sec:widefield}

1536: 

1537: The standard two-dimensional Fourier inversion of visibility data

1538: requires that $w\theta^2 \ll 1$ where $w$ is component of the

1539: interferometric baseline in the direction of the source and $\theta$

1540: is the field of view.  This assumption is not valid for 74~MHz

1541: observations over any significant hour angle.\footnote{%

1542: In general, snapshot observations are not viable with the 74~MHz

1543: \hbox{VLA}.  The large primary beam and crowded fields means that a

1544: snapshot observation produces too few visibilities for an adequate

1545: representation of the field of view.  The exception is for sources

1546: that are so strong, e.g., \objectname[]{Cyg~A}, \objectname[]{Vir~A},

1547: \objectname[]{Cas~A}, that their flux densities dominate the flux

1548: densities of other sources in the field of view.%

1549: }  Consequently, inverting the visibility data requires either a

1550: three-dimensional Fourier inversion or ``polyhedral imaging,'' in

1551: which the field of view is tessellated into facets over which the

1552: assumption of a two-dimensional inversion is valid (e.g.,

1553: Figure~\ref{fig:SCunbias}).  For further details and examples, see

1554: \cite{cp92} and \cite{p99}.  In general, polyhedral imaging is the

1555: more common technique.\footnote{%

1556: The \aips\ task \texttt{IMAGR} and the special-purpose, low-frequency

1557: imaging task \texttt{VLAFM}  implement polyhedral imaging.

1558: }  Table~\ref{tab:facet} gives general guidelines as to the size and

1559: number of facets required for acceptable polyhedral imaging over the

1560: entire primary beam.  In constructing Table~\ref{tab:facet} we have

1561: made use of the criterion \citep{tms86,p99} that the radius of a facet (in

1562: radians) should be

1563: \begin{equation}

1564: \theta_{\mathrm{facet}} \approx \frac{1}{3}\sqrt{\theta}

1565: \label{eqn:facet}

1566: \end{equation}

1567: where $\theta$ is the (FWHM) diameter of the synthesized beam ($\sim

1568: \lambda/b$ for a baseline~$b$).  Larger facets can be used to reduce

1569: the computational burden at the expense of increased phase errors at

1570: the edges of the facets.

1571: 

1572: An alternate strategy, ``targeted faceting,'' exploits two aspects of

1573: the radio astronomical sky.  First, the majority of the sidelobe

1574: confusion results from the strongest sources either inside or outside the

1575: primary beam.  The variance in the flux density is given by the

1576: expectation value of $S^2 N(S)$, which because $N(S) \propto S^x$

1577: with $x < 2$ (equation~\ref{eqn:counts}) means that the variance

1578: is dominated by the strongest sources. Second, most of the the sky

1579: remains largely dark (empty of sources), even at these frequencies.

1580: Based on these facts, instead of tessellating the entire primary beam,

1581: which would take a large amount of time and computer power, one

1582: makes use of \emph{a priori} knowledge of the field of

1583: view, e.g., from the NVSS or WENSS, to place small facets only where

1584: there are sources. We usually have a few hundred small facets per pointing.

1585: \footnote{ Determining the location of the facets can be done by the \aips\

1586: task \texttt{SETFC} or the special-purpose, low-frequency imaging task

1587: \texttt{VLAFM}.} Depending upon the sensitivity of one's image and the

1588: desired image, this strategy can produce a useful computational savings.

1589: Recently, a new technique for wide-field imaging know as the "w-projection"

1590: has been proposed. It projects the intrinsically 3-D data onto a 2-D plane 

1591: with an appropriate Fernel-like convolution, and avoid the teidum involved

1592: with polyhedron imaging and targetted facetting. This technique is currently

1593: being tested at the VLA \citep{Cornwell2006}.

1594: 

1595: 

1596: \subsection{Astrometry}\label{sec:astrometry}

1597: 

1598: Even with the relatively large synthesized beam, compared to that

1599: attainable at higher frequencies, accurate astrometry at~74~MHz is

1600: desired.  The astronomical motivations are varied but include spectral

1601: index studies, for which one wants to align images obtained at

1602: different frequencies (to the extent allowed by any

1603: frequency-dependent shifts within the source), and followup at other

1604: wavelengths, for which positional accuracies of 1\arcsec\ or better

1605: can be required.

1606: 

1607: If self-calibration is used during phase calibration

1608: (\S\ref{sec:pcal}), self-calibration will ``freeze'' the large-scale

1609: refraction but leave the image with an arbitrary absolute position (as

1610: in Figure~\ref{fig:ionshift}).  The NVSS or WENSS sources that appear

1611: in the image can then be utilized to re-register the astrometry to an

1612: accuracy of approximately 5\arcsec.  If the ionosphere is treated as a

1613: phase-delay screen (\S\ref{sec:pdelay}), the astrometric source grid

1614: used in the procedure results in source positions comparable to that

1615: of the survey from which the astrometric grid was constructed (e.g., a

1616: few arcseconds for NVSS), provided that there are no other systematic

1617: effects.

1618: 

1619: \section{Solar Effects, Ionospheric Weather, and Dynamic

1620: 	Scheduling}\label{sec:dynamic}

1621: 

1622: A potential operational improvement in conducting 74~MHz observations 

1623: (and low-frequency observations in general) would be to determine when 

1624: ionospheric conditions are sufficiently quiescent to make useful 

1625: observations.  Currently, observations are scheduled on the telescope 

1626: well in advance and proceed at the scheduled time.  This procedure 

1627: carries the risk that the ionosphere could be in a sufficiently 

1628: disturbed state so as to preclude useful observations.  A better 

1629: approach would be to schedule the telescope in a ``dynamic'' fashion. 

1630: In order to do this, one would have to conduct a test or identify a 

1631: proxy observable that could establish the state of the ionosphere 

1632: rapidly.  If the ionosphere was disturbed so that 74~MHz observations 

1633: would be unlikely to be successful, observations at a higher frequency 

1634: could proceed with the 74~MHz observations being deferred until such 

1635: time as the ionosphere is more amenable to correction via methods 

1636: described here.\footnote{A similar observing strategy is in place for 

1637: high frequency observations on the VLA (22 and 43 GHz) for which the 

1638: dominant source of phase fluctuations is the troposphere.  In this 

1639: case, if the tropospheric conditions appear poor, the high frequency 

1640: observations can be deferred with lower frequency observations being 

1641: observed instead. }  Our experience suggests a number of ways in which 

1642: such dynamic scheduling could be implemented.

1643: 

1644: In general, the ionospheric phase effects that dominate calibration 

1645: issues at low frequencies are a manifestation of highly unpredictable 

1646: ionospheric weather conditions. The key consideration for low frequency 

1647: observations is not the total thickness of the ionosphere, as measured 

1648: by the TEC, but variations in TEC such as TIDs and smaller scale 

1649: ionospheric structures (\S\ref{sec:pdelay_motivate}).  Stable 

1650: ionospheric conditions can occur at mid-day, when the TEC is highest. 

1651: At the same time, highly disruptive ionospheric scintillations can 

1652: occur during the middle of the night. Periods of particular ionospheric 

1653: instability are [after] sunset and

1654: especially

1655: sunrise (Figure~\ref{fig:ionsmall}), but it is often easy to avoid 

1656: observations during these times.

1657: 

1658: Solar activity can affect 74 MHz observations both directly, due to 

1659: radio

1660: emission

1661: from the sun that occur during the daytime,

1662: and indirectly, from ionospheric turbulence generated by manifestations 

1663: of space weather linked to solar activity that can occur any time of 

1664: the day or night.

1665: 

1666: At~74~MHz the quiet \objectname[]{Sun} is a benign

1667: 2~kJy disk (compared to \objectname[]{Cyg~A} at~17~kJy), and, at 

1668: 30\arcmin\ or larger in size, is significantly resolved out with the A 

1669: and~B configurations.  However, nonthermal solar radio noise [storms] 

1670: can have flux densities in excess of~1~MJy, and by entering through the 

1671: far-out sidelibes of the primary beam render observations completely 

1672: incapable of being calibrated. However they are usually short-lived, on 

1673: the order of 30 minutes or less, and  can be excised from the data

1674: in way similar to the treatment of narrow-band RFI.

1675: 

1676: The second direct effect is due to the scattering effects of the solar 

1677: wind. These lead to asymetric angular broadening and distortion of the 

1678: brightness distribution of observed sources, and the effects are worst 

1679: in the A and B configuration when the angular resolution is highest. 

1680: Experience with both of these direct effects indicates that a useful 

1681: rule of thumb is to allow at least 60 degree stand-off betweeen the 

1682: target source and the sun.

1683: 

1684: The indirect effects are more important since bad ionospheric weather 

1685: conditions related to a geomagnetic storm precipitated by solar 

1686: activity can persist for days and render observations useless 

1687: throughout. One well known manifestation are the massive solar 

1688: ejections of plasma and magnetic field known as Coronal Mass Ejections 

1689: (CMEs). Energetic Earth-ward directed CMEs

1690: take

1691: 20 to 48 hours to arrive and, given a suitable magnetic field 

1692: orientation,

1693: their impact

1694: can transfer massive amounts of energy into the upper atmosphere. The 

1695: resulting geomagnetic storms can damage satellites, knock out power 

1696: grids, and disrupt communications. Ionospheric disturbances of 

1697: terrestrial origin, for example acoustic gravity waves can also 

1698: generate ionospheric

1699: variations and poor low frequency

1700: observing conditions. However the most severe ionospheric disturbances 

1701: are usually associated with solar activity.

1702: 

1703: Various spacecraft and ground-based observatories monitor the Sun, and 

1704: indices exist to quantify the level of solar and geomagnetic activity.

1705: Hence proxies exist to

1706: predict both direct and indirect effects of solar and geomagnetic 

1707: activity, and in principle these could be used to provide an 

1708: "ionospheric weather" forecast to guide low frequency observations. 

1709: However no systematic effort has been conducted yet to assess the 

1710: correlation between various indices and satisfactory ionospheric 

1711: imaging conditions. In the mean time, the self-calibration solution, or 

1712: short time scale stability of the phase on a few baselines from a short 

1713: scan on a strong 74~MHz source (\objectname[]{Cyg~A}, 

1714: \objectname[]{Vir~A}) is the most useful means of determining whether 

1715: ionospheric weather conditions are suitable for observations. If they 

1716: are deemed too poor to observe, the observations should be postponed, 

1717: and conditions revisited on the timescale of several hours, or sooner 

1718: if monitoring of strong sources can be efficiently built in to a 

1719: dynamic scheduling system.  Only limited attempts have been made to 

1720: quantify the usefulness of this latter approach; while this procedure 

1721: appears useful, it may not be able to distinguish excellent ionospheric 

1722: imaging conditions from only fair conditions.

1723: 

1724: 

1725: 

1726: 

1727: 

1728: \section{The Future of the 74~MHz System}\label{sec:future}

1729: 

1730: The 74~MHz system expanded from an initial trial system to an

1731: operational component of the \hbox{VLA}.  Even so, its sensitivity is

1732: limited fundamentally by the VLA's modest collecting area ($\sim 10^4$~m${}^2$,

1733: which translates to $\sim2\times10^3$~m$^2$ given the $\sim$15\% aperture

1734: efficiency at this frequency) and poor sensitivity. Even if the VLA

1735: telescopes could be made 50\% efficient, and the bandwidths increased

1736: significantly (\S\ref{sec:general}), it is not possible for the VLA

1737: (or the EVLA or the GMRT) to be more sensitive in plausible integration 

1738: time ($\lesssim$ 10 hrs) than roughly 1~\mjybm\ at 74~MHz---100 times less

1739: sensitive than at~1400~MHz to normal-spectrum objects (i.e., those with

1740: spectral indices $\alpha < -0.7$, $S_\nu \propto \nu^\alpha$).

1741: 

1742: The only way to achieve a sensitivity at these wavelengths comparable

1743: to that available at centimeter wavelengths is to build a system with

1744: much greater collecting area. The Long Wavelength Array (LWA\footnote{lwa.unm.edu},

1745: \citealt{kassimetal06, Kassim2006, Taylor06, KE98}), Low-Frequency Array

1746: (LOFAR\footnote{www.lofar.org}, \citealt{kassimetal00, Kassim2004, Falcke2005})

1747: and Mileura Widefield Array (MWA\footnote{www.haystack.mit.edu/ast/arrays/mwa/},

1748: \citealt{Morales2006, bowmanetal07}) seek to accomplish this at different

1749: levels, with different collecting areas, spatial resolutions and frequency

1750: ranges. The objective of these projects is to produce

1751: aperture synthesis instruments, operating at frequencies between~10

1752: and~300~MHz, with collecting areas of up to $\sim10^6$~m${}^2$ at~20~MHz with

1753: maximum baselines of up to 500~km or longer. As such, they would have

1754: considerably more collecting area than the VLA (and more than any

1755: other previous low-frequency telescope as well) and angular

1756: resolutions comparable to that of the VLA at frequencies near~1~GHz.

1757: 

1758: 

1759: These projects must take into account the lessons learned from previous

1760: low frequency instruments. Here we emphasize some of these lessons, which

1761: became apparent during the deployment of the 74~MHz VLA system:

1762: \begin{itemize}

1763: \item The system should be designed with careful attention to the

1764: signals it emits so that the instrument does not pollute itself with

1765: \hbox{RFI}.

1766: 

1767: \item The complex gain and noise temperature of each antenna should be

1768: monitored continuously.  Measurements of the round-trip phase through

1769: the system should also be available.

1770: 

1771: \item Channel bandwidths should be kept small ($\Delta\nu/\nu \ll 1$)

1772: so as to avoid strong terrestrial transmitters as often occur at low

1773: frequencies and to enable wide-field imaging.

1774: 

1775: \item Wide-band receiving systems should be employed so that more than

1776: a single frequency can be observed.  Compare the VLA to the CLRO in

1777: Figures~\ref{fig:resolution} and~\ref{fig:sensitivity}.

1778: 

1779: \item Large, well-filled primary collectors (presumably composed of

1780: numerous individual dipole antennas phased together) should be used so

1781: as to produce a primary beam with high forward gain.

1782: 

1783: \item Ionospheric calibration should be considered during the design

1784: of the array.  In particular, adequate ionospheric calibration

1785: may require good \emph{aperture}-plane coverage, rather than good

1786: $u$-$v$ plane coverage usually sought in synthesis imaging. Also, the

1787: ability to make simultaneous measurements at widely spaced frequencies

1788: (for phase transfer) and pointing positions (for modelling the ionosphere over

1789: the array) should be available. Rapid (< 10sec) switching may suffice over

1790: true simultaneous measurements, given sufficient sensitivity to track

1791: ionospheric changes on short time scales.

1792: \end{itemize}

1793: 

1794: \section{Conclusions}\label{sec:conclude}

1795: 

1796: We have described the 74~MHz system that is installed on the NRAO Very

1797: Large Array and the Pie Town antenna of the Very Long Baseline Array.

1798: All 29 antennas have been outfitted with 74~MHz receivers.  With these

1799: low-frequency receivers and in the VLA's A configuration, the VLA is

1800: the world's highest resolution, highest dynamic range interferometer

1801: operating below~150~MHz.  Working in conjunction with the PT antenna,

1802: the VLA-PT interferometer, with baselines approaching 70~km,

1803: represents the longest baselines ever used for connected element 

1804: synthesis imaging below ~150~ MHz.

1805: 

1806: The calibration strategy for the 74~MHz VLA incorporates a number of

1807: features not common at higher frequencies.  Chief among these is the

1808: importance of the isoplanatic patch size relative to the primary beam

1809: diameter.  At~74~MHz, the isoplanatic patch size is determined by the

1810: ionosphere and the size of the array, particularly in its larger

1811: configurations, and can be smaller than the primary beam.  We have

1812: developed new \emph{field-based} methods of calibrating 74~MHz VLA

1813: data that are not restricted to assuming that a single phase

1814: correction must apply to the entire field of view.  For certain types

1815: of observations, e.g., a field dominated by a single strong source

1816: that itself is the object of interest, normal, position-independent

1817: self-calibration (such as is used at centimeter-wavelengths) can be

1818: employed to obtain images with reasonable dynamic ranges. 

1819: We have presented snapshot images of 3C

1820: sources of moderate strength as examples of routine, angle-invariant

1821: calibration and imaging, and a sub-sample of these sources with previously

1822: well determined low frequency spectra indicate that the 74 MHz flux scale at

1823: the Very Large Array is stable and reliable to at least 5 percent. The

1824: absolute flux density scale is tied to a model of Cygnus A with a flux

1825: density fixed to the \citet{bgp-tw77} value.

1826: 

1827: Other aspects of calibration and imaging are not unique to 74~MHz but

1828: assume greater importance relative to higher frequencies.  For

1829: instance, at~1400~MHz one can observe in a pseudo-continuum mode

1830: (i.e., low spectral resolution mode) in order to maintain a large

1831: field of view, to identify RFI, or both.  At~74~MHz, the internal RFI

1832: is sufficiently bad that a pseudo-continuum mode (with somewhat higher

1833: spectral dynamic range than used typically at~1400~MHz) is essential.

1834: Similarly, at~1400~MHz there are typically other sources in the field

1835: of view, and it is possible for there to be strong sources outside the

1836: field of view that must be \textsc{clean}ed in order that their

1837: sidelobes do not reduce the dynamic range in the image.  At~74~MHz,

1838: the presence of many sources in the field of view and of strong

1839: sources outside the field of view is guaranteed.

1840: 

1841: Although we expect that the 74~MHz system will continue to be a

1842: productive, ``facility-level'' system of the VLA (and EVLA) for many more years,

1843: we also anticipate that the 74~MHz system will be superseded

1844: eventually by LOFAR and the LWA and possibly the

1845: low-frequency end of the Square Kilometer Array (SKA).  Nonetheless,

1846: we believe that it offers many lessons for these future high

1847: resolution telescopes, as

1848: well as the apparatus to explore new science in its own right.

1849: 

1850: \acknowledgements

1851: 

1852: We have benefited from discussions with many individuals. A partial

1853: listing includes M.~Bietenholz, K.~Blundell, C.~Brogan, T.~Clarke,

1854: J.~Condon, K.~Dyer, T.~Ensslin, C.~Lacey, W.~Tschager, and K.~Weiler.

1855: The National Radio Astronomy Observatory is a facility of the National

1856: Science Foundation operated under cooperative agreement by Associated

1857: Universities, Inc.  ASC and WML were supported by National Research

1858: Council-NRL Research Associateships.  Basic research in radio

1859: astronomy at the NRL is supported by 6.1 base funding.

1860: 

1861: \appendix

1862: 

1863: \section{Snapshot Observations of Bright Sources}\label{sec:example}

1864: 

1865: The prototype 8-antenna 74~MHz system \citep{kped93} was capable of

1866: imaging only the dozen or so strongest sources in the sky whose flux

1867: densities were hundreds to thousands of Janskys. After the full deployment

1868: of the VLA 74~MHz system, a multi-configuration snapshot survey of

1869: sources from the 3C catalog was started. Observations were obtained in

1870: a succession of three VLA configurations in 1998, A configuration

1871: 7-8 March, B configuration 4-5 October, and C configuration 21 November

1872: and 4-5 December. Data in both circular polarizations was obtained

1873: simultaneously in 1 IF each at 74 and 330~MHz. The data were obtained

1874: in spectral line mode with 32 channels at 74~MHz and 64 channels at

1875: 330~MHz after online Hanning smoothing. The total available bandwidth

1876: was $\sim$1.5 and $\sim$3~MHz at 74 and 330~MHz, respectively.

1877: 

1878: All sources were observed numerous times in cycling snapshot fashion to

1879: maximize the hour angle coverage. Our typical scan lengths were 5-10

1880: minutes. Many of the sources were sufficiently small in angular

1881: extent that the A configuration run was sufficient to generate a

1882: good image. More extended sources required B and sometimes C array

1883: data, especially at 330~MHz, because of the higher intrinsic angular

1884: resolution. Table~\ref{tblobs} summarizes the observations and the 

1885: final image beams at 74 and 330~MHz. Notice that in most cases we

1886: convolved the final images with a gaussian, to produce a circular

1887: beam.

1888: 

1889: The data reduction followed the prescriptions described in the previous

1890: sections of this paper. All of the images were produced from multiple

1891: snapshot observations using an angle-invariant calibration (self-calibration)

1892: strategy, which is sufficient for these strong sources because they dominate

1893: the self-calibration solution and sidelobe confusion can be ignored.

1894: In some cases, work is underway to produce even higher resolution, higher

1895: dynamic range images. Superior images are readily obtained with full synthesis

1896: observations as compared to the snapshot images presented here.

1897: 

1898: Because of the calibration method used, the locations of the 330 and

1899: 74~MHz images are uncertain due to ionospheric wander. We used scaled

1900: subtractions ($330{\rm~MHz-\alpha_{ave}}\times73{\rm~MHz}$) to test for shifts

1901: between the maps and adjusted the 74~MHz images to agree with those at

1902: 330~MHz. Note that 330~MHz astrometry could be off by as much as 5\arcsec\

1903: from the true radio reference frame. Alternatively, by using positions

1904: from the NVSS survey one can achieve positional accuracies of $\sim$1\arcsec\

1905: at this frequency.

1906: 

1907: Figure~\ref{fig:3c} shows 74 and 330~MHz images of a variety of moderately

1908: resolved 3C sources with flux densities on the scales of tens of Janskys or

1909: higher.  Images such as these can now be made routinely with snapshot

1910: observations of tens of minutes or less. In most cases these are the first

1911: sub-arcminute images of these sources below~100~MHz. Table~\ref{tab:list}

1912: reports the peak brightness and flux density of the source, but, for the

1913: resolved sources for which we present images, these values may be lower

1914: limits, as some of the flux may have been resolved out. We discuss the

1915: resolved sources briefly but make no quantitative analyzes, as our

1916: discussion here is intended only to be illustrative.

1917: 

1918: 

1919: \begin{description}

1920: 

1921: \item[{3C~10}]

1922: Figure~\ref{fig:3c}\textit{a} shows the images of Tycho's supernova remnant

1923: at 74 and 330~MHz. We can see in these images a limg-brightened spherical

1924: shell with a diameter of $\sim$8\arcmin, and enhanced emission toward the NE

1925: half of the shell. This structure is similar to that observed at 330 MHz 

1926: and 1.4~GHz by \cite{katz00}, and at 610~MHz by \cite{duim75}. 

1927: 

1928: 

1929: \item[{3C~33}]

1930: Figure~\ref{fig:3c}\textit{b} presents the 330 and 74~MHz images of this

1931: radio galaxy. The 74~MHz image shows the lobes and fainter emission

1932: between these two structures. The 330~MHz image resolves some of the

1933: details of this source, separating each lobe into 2 components, as well as

1934: showing diffuse emission extending perpendicular to the jet axis in regions

1935: between the lobes and the nucleus. The structures seen in our images are

1936: similar to the ones seen at 1.5~GHz by \cite{leahy91}. A low resolution

1937: 160~MHz image of this source \citep{s77} was able to resolve it only

1938: into two components.

1939: 

1940: 

1941: \item[{3C~84}]

1942: Figure~\ref{fig:3c}\textit{c} shows the 74~MHz image of the central

1943: regions of the Perseus cluster, which contains a number of strong radio

1944: sources including \objectname[3C]{3C~84}. An enlarged plot of the 330~MHz

1945: Figure~\ref{fig:3c}\textit{d}

1946: image of this radio galaxy shows a strong core, two lobes in the N-S direction

1947: and some extended emission beyond the lobes. These structures are similar to

1948: the ones detected by \cite{pedlar90} at 1.4~GHz, 330~MHz and 151~MHz.

1949: The 74~MHz image does not resolve the nucleus and lobes, but shows some

1950: diffuse emission extending to the NW and SW, corresponding to previous

1951: outbursts (see \cite{fcbkp02} for a more detailed discussion of this

1952: image and the relation between these structures and X-ray holes).

1953: 

1954: \item[{3C~98}]

1955: Figure~\ref{fig:3c}\textit{e}

1956: The 74~MHz image shows a double

1957: lobe structure with diffuse emission in between. The 330~MHz image

1958: shows a similar structure, although with higher resolution, allowing

1959: the detection of the hotspots and part of the jet. These structures were

1960: previously imaged at higher frequencies (1.4 and 4.8 GHz) by

1961: \cite{Leahy1997} and \cite{Young2005}. A 160~MHz image of this galaxy

1962: was presented by \cite{s77}.

1963: 

1964: \item[{3C~129}]

1965: Figure~\ref{fig:3c}\textit{f} shows \objectname[3C]{3C~129} at~74~MHz.

1966: A study of the 330~MHz image \citep{lkehp02} have confirmed

1967: the existence of a steep-spectrum ``crosspiece'' at the head of

1968: \objectname[3C]{3C~129}, along the NE-SW direction, perpendicular to

1969: its main tail. This structure was previously detected at 600~MHz

1970: \citep{jag83} and have recently been observed by \cite{lal04} with the

1971: GMRT at 240 and 610~MHz.  We see no indication of this structure at~74~MHz,

1972: but our resolution is considerably lower, so it is probably blended

1973: with the head of \objectname[3C]{3C~129}.

1974: 

1975: \item[{3C~144}]

1976: Figure~\ref{fig:3c}\textit{g} shows \objectname[3C]{3C~144} (the

1977: \objectname[]{Crab Nebula}, \objectname[]{Tau~A}) at~74~MHz

1978: and 330~MHz. This source can be used as an

1979: amplitude calibrator in more compact configurations (C and D).

1980: The 74~MHz compact source in the center of the nebula is the

1981: \objectname[]{Crab pulsar} (\objectname[PSR]{PSR~B0531$+$21}).

1982: A higher sensitivity 330~MHz image of this source is presented

1983: by  \cite{Frail1995}, while \cite{Bietenholz1997} present

1984: a study of the 74/330 MHz radio spectral index and find evidence

1985: for intrinsic thermal absorption.

1986: 

1987: \item[{3C~218} (Hydra~A)]

1988: Figure~\ref{fig:3c}\textit{h} present the 330 and 74~MHz images of Hydra A,

1989: where we can see that this radio galaxy has a complex structure, consisting

1990: of several outbursts. \cite{laneetal2004} present a detailed study of this source,

1991: the spectral indices of the different components and their correlation of the

1992: X-ray emission from the cluster of galaxy where it resides.

1993: 

1994: \item[{3C~219}]

1995: The 74~MHz emission from this radio galaxy (Figure~\ref{fig:3c}\textit{i})

1996: shows two lobes and diffuse emission between them. This emission is resolved

1997: into better detail by the 330~MHz image, where we can see the N and S

1998: hotspots, the jet and some diffuse emission. These structures are similar to

1999: the ones seen at 1.4~GHz by \cite{Clarke1992}.

2000: 

2001: \item[{3C~274} (Vir~A)]

2002: Figure~\ref{fig:3c}\textit{j} shows \objectname[3C]{3C~274}

2003: (\objectname[]{Vir~A}), one of the sources that can be used as a

2004: primary bandpass and flux density calibrator, in addition to or

2005: instead of \objectname[]{Cyg~A} (particularly if \objectname[]{Cyg~A}

2006: is not above the horizon). Both 330~MHz and 74~MHz images show 

2007: a complex structure, indicative of the interaction between the radio

2008: plasma and the intra cluster medium. A more detailed discussion about

2009: the 330~MHz image of this source is presented by  \cite{oek00}.

2010: 

2011: 

2012: \item[{3C~327}]

2013: The 74~MHz image of this galaxy (Figure~\ref{fig:3c}\textit{k})

2014: shows a double structure along the E-W direction. The E component

2015: is elongated, indicating the presence of multiple components. A

2016: low resolution 160~MHz image from \cite{s77} showed only two blobs

2017: along the E-W direction. The structure seen at 74~MHz are confirmed

2018: by the 330~MHz image, where we can clearly see the hotspots and some

2019: diffuse emission, similar to the structure observed at 8.4~GHz

2020: by \cite{Leahy1997}. 

2021: 

2022: \item[{3C~353}]

2023: The 74~MHz image of this radio galaxy (Figure~\ref{fig:3c}\textit{l})

2024: is broken into 3 components aligned along the E-W direction. Previous

2025: low frequency (160~MHz) images by \cite{s77} were not able to detect

2026: multiple components, detecting only an extended source in the E-W

2027: direction. The 330~MHz image shows the hotspots and diffuse emission in

2028: the lobes, similar to the structure detected by \cite{Baum1988} at 5~GHz.

2029: 

2030: \item[{3C~390.3}]

2031: In Figure~\ref{fig:3c}\textit{m} we can see that the 74~MHz image of this

2032: radio galaxy is resolved into 2 lobes and diffuse emission associated

2033: with them. This structure is similar to the one detected at 610~MHz 

2034: by radio \cite{jag87}, using WSRT observations. The 330~MHz image shows

2035: the hotspots and diffuse emission in better detail, as well as the nucleus.

2036: The structures seen at this frequency are similar to the ones detected

2037: with 1.4~GHz VLA observations \citep{Leahy1995}.

2038: 

2039: \item[3C~392]

2040: Figure~\ref{fig:3c}\textit{n} presents the images of this supernova

2041: remnant, which has similar structure at 74 and 330~MHz. The 330~MHz

2042: image shows details similar to the ones seen at 1.4 GHz

2043: \citep{Jones1993,Giacani1997}. Higher resolution, full synthesis images

2044: at both frequencies of this SNR, also known as W44,

2045: are presented in \cite{castellettietal07}.

2046: 

2047: 

2048: \item[{3C~405} (Cyg~A)]

2049: Figure~\ref{fig:3c}\textit{o} shows our primary bandpass and flux density

2050: calibrator, \objectname[3C]{3C~405} or \objectname[]{Cyg~A}.  This

2051: image is dynamic range limited. 3C405 is almost unresolved at 74~MHz,

2052: however, higher resolution (VLA $+$ Pie Town) images at both

2053: 74 and 330~MHz are presented by \cite{lazio06}. The VLA 330~MHz image

2054: is able to resolve the emission into hotspots and associated diffuse

2055: emission, similar to the structure detected by \cite{pe84} at

2056: 1.4~GHz, although without the detection of the nucleus and the jets.

2057: 

2058: \item[{3C~445}]

2059: Figure~\ref{fig:3c}\textit{p} shows that 74~MHz image of this

2060: radio galaxy is composed of two bright lobes and some faint extended

2061: emission around the nucleus. A lower resolution 160~MHz image of

2062: this galaxy \citep{s77} shows only three blobs. The VLA 330~MHz

2063: image resolves the lobes into hotspots and some associated 

2064: diffuse emission, similar to  that observed at higher frequencies

2065: \citep{Kronberg1986,Leahy1997}.

2066: 

2067: 

2068: \item[{3C~452}]

2069: The 74~MHz image of this radio galaxy is presented in

2070: Figure~\ref{fig:3c}\textit{q}, which shows a double lobed structure,

2071: similar to the one detected at 610~MHz by \cite{jag87}, using WSRT.

2072: The 330~MHz image resolves the structure of this galaxy into better

2073: detail, hotspots and diffuse emission along the jet, similar to the

2074: structure detected at 1.4~GHz by \cite{dennett99}.

2075: 

2076: 

2077: \item[{3C~461}]

2078: Figure~\ref{fig:3c}\textit{r} shows the 74 and 330~MHz of 

2079: \objectname[]{Cas~A}, another source that can be used as a

2080: primary bandpass and flux density calibrator in more compact

2081: configurations (C and D). The 330 and 74~MHz images look similar,

2082: although of lower resolution compared to the 1.4 GHz images

2083: presented in \citet{Anderson1991}. \cite{Kassim1995} found

2084: evidence of internal thermal absorption using the prototype

2085: 8-antenna system, that was subsequently confirmed by \cite{Delaney2004}

2086: using the full 74 MHz VLA + PT Link (Figure~\ref{fig:casapt}).

2087: These observations were obtained with a maximum baseline is ~72 km,

2088: corresponding to an angular resolution of 8.5\arcsec.

2089: 

2090: 

2091: \end{description}

2092: 

2093: 

2094: 

2095: \begin{thebibliography}{}

2096: \bibitem[\protect\citeauthoryear{Alfv\`en \& Herlofson}{1950}]{ah50}

2097: 	Alfv\`en, H.\ \& Herlofson, N.  1950, Phys.\ Rev., 78, 616

2098: 

2099: \bibitem[\protect\citeauthoryear{Anderson et al.}{1991}]{Anderson1991}

2100:         Anderson, M., Rudnick, L., Leppik, P., Perley, R., \&

2101:         Braun, R.\ 1991, \apj, 373, 146

2102: 

2103: \bibitem[\protect\citeauthoryear{Baars et al.}{1977}]{bgp-tw77} Baars,

2104: 	J.~W.~M., Genzel, R., Pauliny-Toth, I.~I.~K., \& Witzel, A.

2105: 	1977, \aap, 61, 99

2106: 

2107: \bibitem[\protect\citeauthoryear{Baum et al.}{1988}]{Baum1988} Baum, S. A.,

2108:         Heckman, T. M., Bridle, A., van Breugel, W. J. M., \& Miley, G. K.

2109:         1988, \apjs, 68, 643

2110: 

2111: \bibitem[\protect\citeauthoryear{Bietenholz et al.}{1997}]{Bietenholz1997}

2112:         Bietenholz, M.~F., Kassim, N., Frail, D.~A., Perley, R.~A.,

2113:         Erickson, W.~C., \& Hajian, A.~R.\ 1997, \apj, 490, 291

2114: 

2115: 

2116: \bibitem[\protect\citeauthoryear{Bowman et al.}{2007}]{bowmanetal07}

2117:         Bowman, J. D. et al. 2007, \aj, in press

2118: 

2119: \bibitem[\protect\citeauthoryear{Braude, Men, \& Sodin}{Braude et

2120: 	al.}{1978}]{bms78} Braude, S.~Ia., Men, A.~V., \& Sodin,

2121: 	L.~G.  1978, Antenny, 26, 3

2122: 

2123: \bibitem[\protect\citeauthoryear{Brentjens}{2005}]{Brentjens2005}

2124:         Brentjens, M. A., 2005, Astron. Nachr., 326, 609-609

2125: 

2126: \bibitem[\protect\citeauthoryear{Bridle}{1999}]{b99} Bridle, A.~H.

2127: 	1999, in Synthesis Imaging in Radio Astronomy, eds.\

2128: 	R.~A.~Perley, F.~R.~Schwab, \& A.~H.~Bridle (ASP: San

2129: 	Francisco) p.~443

2130: 

2131: \bibitem[\protect\citeauthoryear{Bridle \& Purton}{1968}]{bp68}

2132: 	Bridle, A.~H.\ \& Purton, C.~R.  1968, \aj, 73, 717

2133: 

2134: \bibitem[\protect\citeauthoryear{Bock, Large, \& Sadler}{Bock et

2135: 	al.}{1999}]{bls99} Bock, D.~C.-J., Large, M.~I., \& Sadler,

2136: 	E.~M.  1999, \aj, 117, 1578

2137: 

2138: \bibitem[\protect\citeauthoryear{Castelletti et al.}{2007}]{castellettietal07}

2139: 	Castelletti, G., Dubner, G., Brogan, C., Kassim, N.~E. \aap, in press

2140:         (astro-ph/0702746)

2141: 

2142: \bibitem[\protect\citeauthoryear{Clarke et al.}{1992}]{Clarke1992}

2143:         Clarke, D.~A., Bridle, A.~H., Burns, J.~O., Perley, R.~A., \&

2144:         Norman, M.~L.\ 1992, \apj, 385, 173

2145: 

2146: \bibitem[\protect\citeauthoryear{Cohen et al.}{2006}]{cohen06}

2147:         Cohen, A.S., Lane, W.M., Kassim, N.E., Lazio, T.J.W., Cotton,

2148:         W.D., Condon, J.J., Perley, R.A., Erickson, W.C. 2006, in

2149:         "From Clark Lake to the Long Wavelength Array: Bill Erickson's

2150:         Radio Science", ASP Conference Series, Vol. 345, Proceedings of

2151:         the Conference held 8-11 September, 2004 in Santa Fe, New Mexico,

2152:         USA. Edited by N. Kassim, M. Perez, M. Junor, and P. Henning, p.299.

2153: 

2154: \bibitem[\protect\citeauthoryear{Cohen et al.}{2003}]{cohenetal03}

2155: 	Cohen, A.~S., R{\"o}ttgering, H.~J.~A., Kassim, N.~E., et al.

2156: 	2003, \apj, 591, 640

2157: 

2158: \bibitem[\protect\citeauthoryear{Cohen et al.}{2003}]{cohen07}

2159:         Cohen, A.~S., et al. 2007, in preparation

2160: 

2161: \bibitem[\protect\citeauthoryear{Condon et al.}{1998}]{ccgyptb98}

2162: 	Condon, J.~J., Cotton, W.~D., Greisen, E.~W., Yin, Q.~F.,

2163: 	Perley, R.~A., Taylor, G.~B., \& Broderick, J.~J.

2164: 	1998, \aj, 115, 1693

2165: 

2166: \bibitem[\protect\citeauthoryear{Cornwell, Golap, \& Bhatnagar}{2006}]{Cornwell2006}

2167:         Cornwell, T.J., Golap, K., Bhatnagar, S. in "From Clark Lake to

2168:         the Long Wavelength Array: Bill Erickson's Radio Science", ASP

2169:         Conference Series, Vol. 345, Proceedings of the Conference held 8-11

2170:         September, 2004 in Santa Fe, New Mexico, USA.Edited by N. Kassim, M.

2171:         Perez, M. Junor, and P. Henning, p.350.

2172: 

2173: 

2174: \bibitem[\protect\citeauthoryear{Cornwell \& Fomalont}{1999}]{cf99}

2175: 	Cornwell, T.~J.\ \& Fomalont, E.~B.  1999, in Synthesis

2176: 	Imaging in Radio Astronomy~II, eds.\ G.~B.~Taylor,

2177: 	C.~L.~Carilli, \& R.~A.~Perley (ASP:San Francisco) p.~187

2178: 

2179: \bibitem[\protect\citeauthoryear{Cornwell \& Perley}{1992}]{cp92}

2180: 	Cornwell, T.~J.\ \& Perley, R.~A. 1992, \aap, 261, 353

2181: 

2182: \bibitem[\protect\citeauthoryear{Cotton}{2006}]{Cotton2006}

2183:         Cotton, W.D. in "From Clark Lake to the Long Wavelength Array: Bill

2184:         Erickson's Radio Science" ASP Conference Series, Vol. 345, Proceedings

2185:         of the Conference held 8-11 September, 2004 in Santa Fe, New Mexico,

2186:         USA. Edited by N. Kassim, M. Perez, M. Junor, and P. Henning, p.337

2187: 

2188: \bibitem[\protect\citeauthoryear{Cotton \& Condon}{2002}]{cc02}

2189: 	Cotton, W.~D.\ \& Condon, J.~J.  2002, Proc.\ URSI General

2190: 	Assembly, J3.0.2

2191: 

2192: \bibitem[\protect\citeauthoryear{Cotton et al.}{2004}]{cotton04}

2193:         Cotton, W.~D., Condon, J.~J., Perley, R.~A., Kassim, N.,

2194:         Lazio, J., Cohen, A., Lane, W., \& Erickson, W.~C.\ 2004,

2195:         \procspie, 5489, 180 

2196: 

2197: \bibitem[\protect\citeauthoryear{Delaney et al.}{2004}]{Delaney2004}

2198:         Delaney, T., Rudnick, L., Jones, T., Fesen, R., Hwang, U., Petre, R., \&

2199:         Morse, J. 2004, in X-Ray and Radio Connections Conference Proceedings,

2200:         eds. L.O. Sjouwerman and K.K. Dyer. Published electronically by NRAO,

2201:         http://www.aoc.nrao.edu/events/xraydio. Held 3/6 February 2004 in

2202:         Santa Fe, New Mexico, USA (E4.05)

2203: 

2204: \bibitem[\protect\citeauthoryear{Dennett-Thorpe, et al.}{1999}]{dennett99}

2205:         Dennett-Thorpe, J., Bridle, A. H., Laing, R. A., \& Scheuer, P. A. G.

2206:         1999, \mnras, 304, 271

2207: 

2208: \bibitem[\protect\citeauthoryear{Duim \& Strom}{1975}]{duim75} Duim, R. M., \&

2209:         Strom, R. G. 1975, \aap, 39, 33

2210: 

2211: \bibitem[\protect\citeauthoryear{Erickson}{1984}]{e84} Erickson, W.~C.  1984, J.\ Astrophys.\

2212: 	Astron., 5, 55

2213: 

2214: \bibitem[\protect\citeauthoryear{Erickson et al.}{2001}]{epfk01}

2215: 	Erickson, W.~C., Perley, R.~A., Flatters, C., \& Kassim, N.~E.

2216: 	2001, \aap, 366, 1071

2217: 

2218: \bibitem[\protect\citeauthoryear{Erickson, Mahoney, \& Erb}{Erickson

2219: 	et al.}{1982}]{eme82} Erickson, W.~C., Mahoney, M.~J., \& Erb, K.

2220: 	1982, \apjs, 50, 403

2221: 

2222: \bibitem[\protect\citeauthoryear{Erickson}{2006}]{Erickson2006}

2223:         Erickson, W.C. in "From Clark Lake to the Long Wavelength

2224:         Array: Bill Erickson's Radio Science", ASP Conference Series,

2225:         Vol. 345, Proceedings of the Conference held 8-11 September,

2226:         2004 in Santa Fe, New Mexico, USA. Edited by N. Kassim, M.

2227:         Perez, M. Junor, and P. Henning, p.317

2228: 

2229: 

2230: \bibitem[\protect\citeauthoryear{Fabian et al.}{2002}]{fcbkp02}

2231: 	Fabian, A.~C., Celotti, A., Blundell, K.~M., Kassim, N.~E., \&

2232: 	Perley, R.~A.  2002, \mnras, 331, 369

2233: 

2234: \bibitem[\protect\citeauthoryear{Falcke}{2005}]{Falcke2005}

2235:         Falcke, H. 2005, Astron. Nachr., 326, 612

2236: 

2237: \bibitem[\protect\citeauthoryear{Frail et al.}{1995}]{Frail1995}

2238:         Frail, D.~A., Kassim, N.~E., Cornwell, T.~J., \& Goss,

2239:         W.~M.\ 1995, \apj, 454, L129

2240: 

2241: \bibitem[\protect\citeauthoryear{Giacani et al.}{1997}]{Giacani1997}

2242:         Giacani, E.~B., Dubner, G..~M., Kassim, N.~E., Frail, D.~A.,

2243:         Goss, W.~M., Winkler, P.~F., \& Williams, B.~F.\ 1997, \aj, 113, 1379

2244: 

2245: \bibitem[\protect\citeauthoryear{Gizani et al.}{2005}]{Gizani2005}

2246:         Gizani, Nectaria A. B., Cohen, A., Kassim, N. E., 2005 \mnras, 358, 1061

2247: 

2248: \bibitem[\protect\citeauthoryear{Golap et al.}{2006}]{Golap2006}

2249:         Golap, K., Cornwell, T.J., Perley, R.A., Bhatnagar, S. in

2250:         "From Clark Lake to the Long Wavelength Array: Bill Erickson's

2251:         Radio Science", ASP Conference Series, Vol. 345, Proceedings of

2252:         the Conference held 8-11 September, 2004 in Santa Fe, New Mexico,

2253:         USA. Edited by N. Kassim, M. Perez, M. Junor, and P. Henning, p.366.

2254:                                                                                 

2255: \bibitem[\protect\citeauthoryear{Hewish et al.}{1968}]{hbpsc68}

2256: 	Hewish, A., Bell, S.~J., Pilkington, J.~D.~H., Scott, P.~F., \&

2257: 	Collins, R.~A. 1968, \nat, 217, 709

2258: 

2259: \bibitem[\protect\citeauthoryear{Jansky}{1935}]{j35} Jansky, K.~G.  1935, 

2260: 	Proc.\ I.\ R.\ E., 23, 1158 

2261: 

2262: \bibitem[\protect\citeauthoryear{Jacobson \& Erickson}{1992a}]{je92a} 

2263: 	Jacobson, A.~R.\ \& Erickson, W.~C.  1992a,

2264: 	Planetary and Space Sci., 40, 447

2265: 

2266: \bibitem[\protect\citeauthoryear{Jacobson \& Erickson}{1992b}]{je92b} 

2267: 	Jacobson, A.~R.\ \& Erickson, W.~C.  1992b,

2268: 	\aap, 257, 401

2269: 

2270: \bibitem[\protect\citeauthoryear{Ja\"gers}{1987}]{jag87} Ja\"gers, W.~J.

2271:         1987, \aaps, 67, 395 

2272: 

2273: \bibitem[\protect\citeauthoryear{Ja\"gers \& de Grijp}{1983}]{jag83}

2274:         Ja\"gers, W.~J., \& de Grijp, M.~H.~K.\ 1983, \aap, 127, 235

2275: 

2276: \bibitem[\protect\citeauthoryear{Jones, Smith \& Angellini}{1993}]{Jones1993}

2277:         Jones, L.~R., Smith, A., \& Angellini, L.\ 1993, \mnras, 265, 631

2278: 

2279: \bibitem[\protect\citeauthoryear{Kassim \& Erickson}{1998}]{KE98}

2280:         Kassim, N.~E., \& Erickson, W.~C.\ 1998, \procspie, 3357, 740 

2281: 

2282: \bibitem[\protect\citeauthoryear{Kassim et al.}{2006a}]{kassimetal06}

2283:         Kassim, N.~E., et al. 2006, Long Wavelength Astrophysics, 26th

2284:         meeting of the IAU, Joint Discussion 12, 21 August 2006, Prague,

2285:         Czech Republic, JD12, \#56, 12

2286: 

2287: \bibitem[\protect\citeauthoryear{Kassim et al.}{2001}]{kassimetal00}

2288: 	Kassim, N.~E., Lazio, T.~J.~W., Erickson, W.~C., et al.  2000,

2289: 	in Radio Telescopes, Proc.\ SPIE, ed.\ H.~R.~Butcher,

2290: 	vol.~4015, p.~328

2291: 

2292: \bibitem[\protect\citeauthoryear{Kassim et al.}{1995}]{Kassim1995}

2293:         Kassim, N. E., Perley, R. A., Dwarakanath, K. S., Erickson, W. C. 1995,

2294:         ApJ, 455, L59

2295: 

2296: \bibitem[\protect\citeauthoryear{Kassim et al.}{1993}]{kped93}

2297: 	Kassim, N.~E., Perley, R.~A., Erickson, W.~C., \& Dwarakanath,

2298: 	K.~S. 1993, \aj, 106, 2218

2299: 

2300: \bibitem[\protect\citeauthoryear{Kassim et al.}{2004}]{Kassim2004}

2301:         Kassim, N. E., Lazio, T. J. W., Ray, P. S., Crane, P. C., Hicks, B. C.,

2302:         Stewart, K. P., Cohen, A. S., Lane, W. M., 2004, Planetary

2303:         and Space Science, Volume 52, Issue 15, p. 1343-1349

2304: 

2305: 

2306: \bibitem[\protect\citeauthoryear{Kassim}{1988}]{k88} Kassim, N.~E.

2307: 	1988, \apjs, 68, 715

2308: 

2309: \bibitem[\protect\citeauthoryear{Kassim et al.}{2006b}]{Kassim2006}

2310:         Kassim, N. E., Polisensky, E. J., Clarke, T. E., Hicks, B. C.,

2311:         Crane, P. C., Stewart, K. P., Ray, P. S., Weiler, K. W.,

2312:         Rickard, L. J., Lazio, T. J. W., Lane, W. M., Cohen, A. S.,

2313:         Nord, M. E., Erickson, W. C., Perley, R. A. in "From Clark

2314:         Lake to the Long Wavelength Array: Bill Erickson's Radio

2315:         Science", ASP Conference Series, Vol. 345, Proceedings of

2316:         the Conference held 8-11 September, 2004 in Santa Fe, New

2317:         Mexico, USA. Edited by N. Kassim, M. Perez, M. Junor, and

2318:         P. Henning, p.392

2319:                                                                                 

2320: \bibitem[\protect\citeauthoryear{Katz-Stone et al.}{2000}]{katz00}

2321:         Katz-Stone, D. M., Kassim, N. E., Lazio, T. J. W., O'Donnell, R.

2322:         2000, \apj, 529, 453

2323: 

2324: \bibitem[\protect\citeauthoryear{Kiepenheuer}{1950}]{k50} Kiepenheuer, 

2325: 	K.~O.  1950, Phys.\ Rev., 79, 738

2326: 

2327: \bibitem[\protect\citeauthoryear{Kronberg, Wielebinski \& Graham}{1986}]

2328:         {Kronberg1986}Kronberg, P. P., Wielebinski, R., \& Graham, D. A.

2329:         1986, \aap, 169, 63

2330:         

2331: \bibitem[\protect\citeauthoryear{Krymkin \& Sidorchuk}{1994}]{ks94}

2332: 	Krymkin, V.~V.\ \& Sidorchuk, M.~A.  1994, \apss, 213, 1

2333: 

2334: \bibitem[\protect\citeauthoryear{K\"uhr et al.}{1981}]{kwp-tn81}

2335: 	K\"uhr, H., Witzel, A., Pauliny-Toth, I.~I.~K., \& Nauber, U.

2336: 	1981, \aaps, 45, 367

2337: 

2338: \bibitem[\protect\citeauthoryear{Lane et al.}{2004a}]{laneetal2004}

2339:         Lane, W. M., Clarke, T. E., Taylor, G. B., Perley, R. A.,

2340:         Kassim, N. E. 2004, \aj, 127, 48

2341: 

2342: \bibitem[\protect\citeauthoryear{Lane et al.}{2002}]{lkehp02} Lane,

2343: 	W.~M., Kassim, N.~E., En{\ss}lin, T.~A., Harris, D.~E., \&

2344: 	Perley, R.~A.  2002, \aj, 123, 2985

2345: 

2346: \bibitem[\protect\citeauthoryear{Lane et al.}{2004b}]{Lane2004}

2347:         Lane, W., Cohen, A., Cotton, W. D., Condon, J. J., Perley, R. A.,

2348:         Lazio, J., Kassim, N., Erickson, W. C. in "Ground-based Telescopes".

2349:         Edited by Oschmann, Jacobus M., Jr. Proceedings of the SPIE, Volume

2350:         5489, pp. 354-361 (2004)

2351: 

2352: \bibitem[\protect\citeauthoryear{Lane et al.}{2005}]{Lane2005}

2353:         Lane, W. M., Cohen, A. S., Kassim, N. E., Lazio, T. J. W.,

2354:         Perley, R. A., Cotton, W. D., Greisen, E. W., Radio Science,

2355:         40, RS5S05, 2005. (See also Golap et al. 2006.)

2356:                                                                                 

2357: \bibitem[\protect\citeauthoryear{Lane et al.}{2006}]{Lane2006}

2358:         Lane, W.M., Cohen, A.S., Kassim, N.E., Lazio, T.J.W., in "From Clark

2359:         Lake to the Long Wavelength Array: Bill Erickson's Radio Science", ASP

2360:         Conference Series, Vol. 345, Proceedings of the Conference held 8-11

2361:         September, 2004 in Santa Fe, New Mexico, USA. Edited by N. Kassim, M.

2362:         Perez, M. Junor, and P. Henning, p.203.

2363:                                                                                 

2364: \bibitem[\protect\citeauthoryear{Lal \& Rao}{2004}]{lal04} Lal, D.V., \&

2365:         Rao, A.~P.\ 2004, \aap, 420, 491 

2366: 

2367: \bibitem[\protect\citeauthoryear{Lazio et al.}{2006}]{lazio06}

2368:         Lazio, T.~J.~W., Cohen, A.~S., Kassim, N.~E., Perley, R.~A., Erickson,

2369:         W.~C., Carilli, C.~L., \& Crane, P.~C.\ 2006, \apjl, 642, L33 

2370: 

2371: \bibitem[\protect\citeauthoryear{Leahy et al.}{1997}]{Leahy1997} Leahy, J. P.,

2372:         Black,A.~R.~S., Dennett-Thorpe, J., Hardcastle, M.~J., Komissarov, S.,

2373:         Perley, R.~A., Riley, J.~M., \& Scheuer, P.~A.~G. 1997, \mnras, 291, 20

2374: 

2375: \bibitem[\protect\citeauthoryear{Leahy, Muxlow \& Stephens}{1989}]{leah89}

2376:         Leahy, J.~P., Muxlow, T.~W.~B., \& Stephens, P.~W.\ 1989, \mnras,

2377:         239, 401 

2378: 

2379: \bibitem[\protect\citeauthoryear{Leahy, \& Perley}{1991}]{leahy91}

2380:         Leahy, J. P. \& Perley, R. A. 1991, \aj, 102, 537

2381: 

2382: 

2383: \bibitem[\protect\citeauthoryear{Leahy \& Perley}{1995}]{Leahy1995}

2384:          Leahy, J.~P., \& Perley, R.~A.\ 1995, \mnras, 277, 1097 

2385: 

2386: \bibitem[\protect\citeauthoryear{Mills}{1952}]{m52} Mills, B.~Y.

2387: 	1952, Aust.\ J.\ Sci.\ Res., A5, 456

2388: 

2389: \bibitem[\protect\citeauthoryear{McCready, Pawsey, \&

2390: 	Payne-Scott}{McCready et al.}{1947}]{mpp-s47} McCready,

2391: 	L.~L,. Pawsey, J.~L., \& Payne-Scott, R.  1947, Proc.\ Royal

2392: 	Soc.~A, 190, 357

2393: 

2394: \bibitem[\protect\citeauthoryear{Morales et al.}{2006}]{Morales2006}

2395:         Morales, M.F., Lonsdale, C.J., Cappallo, R.J., Hewitt, J.N.,

2396:         Doeleman, S., in "From Clark Lake to the Long Wavelength Array:

2397:         Bill Erickson's Radio Science", ASP Conference Series, Vol. 345,

2398:         Proceedings of the Conference held 8-11 September, 2004 in Santa

2399:         Fe, New Mexico, USA. Edited by N. Kassim, M. Perez, M. Junor, and

2400:         P. Henning, p. 452.

2401: 

2402: \bibitem[\protect\citeauthoryear{Nijboer, Noordam \& Yatawatta}{2006}]{Nijboer2006}

2403:        Nijboer, R. J., Noordam, J. E., Yatawatta, S. B. in "Astronomical Data

2404:        Analysis Software and Systems XV" ASP Conference Series, Vol. 351,

2405:        Proceedings of the Conference Held 2-5 October 2005 in San Lorenzo de

2406:        El Escorial, Spain. Edited by Carlos Gabriel, Christophe Arviset, Daniel

2407:        Ponz, and Enrique Solano. San Francisco: Astronomical Society of the

2408:        Pacific, 2006., p.291

2409: 

2410: \bibitem[\protect\citeauthoryear{Noordam}{2004}]{Noordam2004}

2411:         Noordam, J.E. 2004, in "Ground-based Telescopes". Edited by Oschmann,

2412:         Jacobus M., Jr. Proceedings of the SPIE, Volume 5489, pp. 817-825

2413: 

2414: \bibitem[\protect\citeauthoryear{Pedlar, A. et al.}{1990}]{pedlar90}

2415:         Pedlar, A., Ghataure, H.~S., Davies, R.~D., Harrison, B.~A.,

2416:         Perley, R., Crane, P.~C., \& Unger, S.~W.\ 1990, \mnras, 246, 477 

2417: 

2418: \bibitem[\protect\citeauthoryear{Napier, Thompson, \& Ekers}{Napier et

2419: 	al.}{1983}]{nte83} Napier, P.~J., Thompson, A.~R., \& Ekers,

2420: 	R.~D.  1983, Proc.\ IEEE, 71, 1295

2421: 

2422: \bibitem[\protect\citeauthoryear{Owen, Eilek, \& Kassim}{Owen et

2423: 	al.}{2000}]{oek00}  Owen, F.~N., Eilek, J.~A., \& Kassim,

2424: 	N.~E.  2000, \apj, {543}, 611

2425: 

2426: \bibitem[\protect\citeauthoryear{Pawsey, Payne-Scott, \&

2427: 	McCready}{Pawsey et al.}{1946}]{pp-sm46} Pawsey, J.~L.,

2428: 	Payne-Scott, R., \& McCready, L.~L.  1946, \nat, 157, 158

2429: 

2430: \bibitem[\protect\citeauthoryear{Perley}{2002}]{p02}

2431: 	Perley, R.~A.  2002, Proc.\ URSI General Assembly, J4.0.2

2432: 

2433: \bibitem[\protect\citeauthoryear{Perley}{1999}]{p99} 

2434: 	Perley, R.~A. 1999, in Synthesis Imaging in Radio Astronomy

2435: 	II, eds.\ G.~B.~Taylor, C.~L.~Carilli, \& R.~A.~Perley (San

2436: 	Francisco: ASP) p.~383

2437: 

2438: \bibitem[\protect\citeauthoryear{Perley \& Erickson}{1984}]{pe84}

2439: 	Perley, R.~A.\ \& Erickson, W.~C. 1984, VLA Scientific

2440: 	Memorandum \#146

2441: 

2442: \bibitem[\protect\citeauthoryear{Reber}{1940}]{r40} Reber,

2443: 	G. 1940, \apj, 91, 621

2444: 

2445: \bibitem[\protect\citeauthoryear{Rees}{1990}]{r90} Rees, N.

2446: 	1990, \mnras, 244, 233

2447: 

2448: \bibitem[\protect\citeauthoryear{Rengelink et al.}{1997}]{rtdmbrb97}

2449: 	Rengelink, R.~B., Tang, Y., de~Bruyn, A.~G., Miley, G.~K.,

2450: 	Bremer, M.~N., R\"ottgering, H.~J.~A., \& Bremer, M.~A.~R.

2451: 	1997, \aaps, 124, 259

2452: 

2453: \bibitem[\protect\citeauthoryear{Roger, Costain, \& Stewart}{Roger et

2454: 	al.}{1986}]{rcs86}  Roger, R.~S., Costain, C.~H., \& Stewart,

2455: 	D.~I.  1986, \aaps, 65, 485

2456: 

2457: \bibitem[\protect\citeauthoryear{Ryle}{1952}]{r52} Ryle, M.  1952,

2458: 	Proc.\ Royal Soc.~A, 211, 351

2459: 

2460: \bibitem[\protect\citeauthoryear{Ryle, Smith, \& Elsmore}{Ryle et

2461: 	al.}{1950}]{rse50} Ryle, M., Smith, F.~G., \& Elsemore, B.

2462: 	1950, \mnras, 110, 508

2463: 

2464: \bibitem[\protect\citeauthoryear{Ryle \& Vonberg}{1946}]{rv46} Ryle,

2465: 	M.\ \& Vonberg, D.~D.  1946, \nat, 158, 339

2466: 

2467: \bibitem[\protect\citeauthoryear{Shepherd, Claussen, \&

2468: 	Kurtz}{Shepherd et al.}{2001}]{sck01} Shepherd, D.~S.,

2469: 	Claussen, M.~J., \& Kurtz, S.~E.  2001, Science, 292, 1513

2470: 

2471: \bibitem[\protect\citeauthoryear{Shkolvsky}{1952}]{s52} Shkolvsky,

2472: 	I.~S.  1952, \azh, 29, 418

2473: 

2474: \bibitem[\protect\citeauthoryear{Slee}{1977}]{s77} Slee, O.~B.

2475: 	1977, Au. J.\ Phys.\ Astron., 43, 1

2476: 

2477: \bibitem[\protect\citeauthoryear{Slee \& Higgins}{1975}]{sh75}  Slee,

2478: 	O.~B.\ \& Higgins, C.~S.  1975, Au.\ J.\ Phys.\ Astron., 36, 1

2479: 

2480: \bibitem[\protect\citeauthoryear{Slee \& Higgins}{1973}]{sh73}  Slee,

2481: 	O.~B.\ \& Higgins, C.~S.  1973, Au.\ J.\ Phys.\ Astron., 27, 1

2482: 

2483: \bibitem[\protect\citeauthoryear{Subramanian et al.}{1986}]{sghs86}

2484: 	Subramanian, K.~R., Gowda, C.~N., Hameed, A.~T.~A., \& Sastry,

2485: 	C.~V.  1986, Astron.\ Soc. India. Bulletin, 14, 236

2486: 

2487: \bibitem[\protect\citeauthoryear{Swarup}{1990}]{swarup90}

2488:         Swarup, G.\ 1990, Indian Journal of Radio and Space Physics, 19, 493

2489: 

2490: \bibitem[\protect\citeauthoryear{Taylor}{2006}]{Taylor06}

2491:         Taylor, G.~B.\ 2006, Long Wavelength Astrophysics, 26th meeting

2492:         of the IAU, Joint Discussion 12, 21 August 2006, Prague, Czech

2493:         Republic, JD12, \#17, 12,  

2494: 

2495: \bibitem[\protect\citeauthoryear{Thomasson}{1986}]{thom86}

2496:          Thomasson, P.\ 1986, \qjras, 27, 413 

2497: 

2498: \bibitem[\protect\citeauthoryear{Thompson, Moran, \& Swenson}{Thompson 

2499: 	et al.}{1986}]{tms86} Thompson, A.~R., Moran, J.~M., \&

2500: 	Swenson, G.~W., Jr.  1986, Interferometry and Synthesis in

2501: 	Radio Astronomy (Wiley: New York)

2502: 

2503: \bibitem[\protect\citeauthoryear{Walker}{1995}]{w95} Walker, R.~C.

2504: 	1995, in Very Long Baseline Interferometry and the VLBA, eds.\ 

2505: 	J.~A.~Zensus, P.~J.~Diamond, \& P.~J.~Napier (ASP:San

2506: 	Francisco) p.~247

2507: 

2508: \bibitem[\protect\citeauthoryear{Yates}{1968}]{y68}  Yates, K.~W.

2509: 	1968, Au.\ J.\ Phys., 21, 167

2510: 

2511: \bibitem[\protect\citeauthoryear{Young et al.}{2005}]{Young2005}

2512:         Young, A., Rudnick, L., Katz, D., DeLaney, T., Kassim,

2513:         N.~E., \& Makishima, K.\ 2005, \apj, 626, 748 

2514: 

2515: \end{thebibliography}

2516: 

2517: \clearpage

2518: 

2519: \begin{figure}

2520: \epsscale{0.7}

2521: {\plotone{f1.eps}}

2522: \caption[]{Angular resolution (arcseconds) as a

2523: function of frequency (MHz) for past and present imaging

2524: instruments in the 10 to~200~MHz range. The various 

2525: telescopes include the UTR-2 \citep{ks94}, Gauribidanur

2526: \citep{sghs86}, 8C \citep{r90}, Clark Lake Radio Observatory

2527: \citep[CLRO,][]{k88}, Culgoora \citep{sh73,sh75,s77}, Dominion Radio

2528: Astrophysical Observatory \citep[DRAO,][]{bp68,rcs86}, MERLIN

2529: \citep{thom86,leah89} and the Mauritius radio telescope.

2530: The 74~MHz VLA is respresented by the filled triangle.}

2531: \label{fig:resolution}

2532: \end{figure}

2533: 

2534: \begin{figure}

2535: \epsscale{0.7}

2536: {\plotone{f2.eps}}

2537: \caption[]{Sensitivity (mJy) to a point source as a function

2538: of frequency for the same instruments shown in

2539: Figure~\ref{fig:resolution}.  The sensitivities are estimates of the

2540: minimum detectable flux density provided by past and present

2541: telescopes.  The sensitivity of most of the telescopes shown here were

2542: or are confusion limited; for the VLA and the GMRT, an integration

2543: time of~8~hr was assumed in calculating their sensitivities.}

2544: \label{fig:sensitivity}

2545: \end{figure}

2546: 

2547: 

2548: \begin{figure}

2549: %\rotatebox{90}{\plotone{f3.eps}}

2550: \caption[]{A picture of a 74~MHz dipole mounted on a VLA

2551: antenna.  In the center of the picture is the subreflector, supported

2552: by the quadrupod legs.  The 74~MHz (crossed) dipoles are in the lower

2553: center of the picture.  The cable that carries the signals from the

2554: dipoles to the receivers drops from the intersection of the dipoles to

2555: the bottom of the antenna's surface.  Also visible just below the

2556: subreflector are the 330~MHz dipoles.}

2557: \label{fig:mount}

2558: \end{figure}

2559: 

2560: 

2561: \begin{figure}

2562: \epsscale{0.6}

2563: %\plotone{f4.eps}

2564: \caption[]{The block diagram of the 74~MHz receiver on the Pie Town

2565: VLBA antenna.  The receivers on the VLA antennas are similar, with the

2566: main difference being that the VLA receivers are not hetrodyne

2567: receivers.  Rather the 74 and~330~MHz signals are transferred directly

2568: to the intermediate frequency (IF) transmission system.}

2569: \label{fig:receiver}

2570: \end{figure}

2571: 

2572: 

2573: \begin{figure}

2574: \rotatebox{-90}{\plotone{f5.eps}}

2575: \caption[]{First fringes using the PT Link at~74~MHz.  The source

2576: observed was \protect\objectname[3C]{3C~123}.  

2577: \textit{Top} Phase as a function of time for VLA antenna~\#7, located

2578: relatively close to the center of the array.

2579: \textit{Bottom} Phase as a function of time for the PT antenna.  In

2580: both cases, the phases are measured relative to an antenna near the

2581: center of the array.  The scales on the ordinates differ.}

2582: \label{fig:ptlink}

2583: \end{figure}

2584: 

2585: 

2586: \begin{figure}

2587: \epsscale{0.85}

2588: %\plottwo{f6a.eps}{f6b.eps}

2589: \caption[]{The primary beam power pattern at~74~MHz from one of the

2590: antennas.  Other antennas have similar power patterns.  The left panel

2591: shows the left circular polarization, and the right panel shows the

2592: right circular polarization.  The axes are the sines of the offset

2593: angle.  The contours show the decrease, in dB, from the peak of the

2594: pattern (drawn at 3, 6, 10, 15, 20, 25, and 30 dB levels).

2595: The sharp edge on the right hand side of the plots results

2596: from the motion limits on pointing for the VLA antennas.}

2597: \label{fig:beam}

2598: \end{figure}

2599: 

2600: 

2601: \begin{figure}

2602: \epsscale{0.7}

2603: \rotatebox{-90}{\plotone{f7.eps}}

2604: \caption[]{The rms image noise level as a function of receiver

2605: bandwidth.  The dots show the measured values from a 90~min.\

2606: integration calibrated and imaged in the fashion described in

2607: \S\S\ref{sec:calibrate} and~\ref{sec:imaging}.  The solid line shows

2608: the behavior expected, $\Delta\nu^{-1/2}$, if the system performance

2609: is limited by thermal noise. This behaviour is typical of observations

2610: obtained in the A and B configurations.}

2611: \label{fig:bandwidth}

2612: \end{figure}

2613: 

2614: \begin{figure}

2615: \epsscale{0.7}

2616: \rotatebox{-90}{\plotone{f8.eps}}

2617: \caption[]{The rms image noise level as a function of integration

2618: time.  Multiple points at the same integration time indicate

2619: integration times for which multiple, independent images were

2620: constructed from the approximately 1-hr total integration time.  The

2621: solid line shows the behavior expected, $t^{-1/2}$, if the system

2622: performance is limited by thermal noise. This behaviour is typical of observations

2623: obtained in the A and B configurations.}

2624: \label{fig:time}

2625: \end{figure}

2626: 

2627: \clearpage

2628: 

2629: \begin{figure}

2630: \epsscale{0.65}

2631: %\plotone{f9.eps}

2632: \caption[]{A demonstration of the excision of 74~MHz \hbox{RFI}.

2633: \textit{Left:} The visibility amplitudes on a single baseline are

2634: displayed in a (linear) gray scale format with frequency on the

2635: abscissa and time on the ordinate.  The time axis is not linear, and

2636: the thick horizontal black stripes indicate times when other sources were

2637: being observed.  The bright vertical stripes represent the 100~kHz

2638: ``comb''.  The strength of the comb varies from baseline to baseline

2639: and as a function of time within individual baselines, depending upon

2640: the orientation of the antennas and how strongly they couple.  For

2641: this illustration, we have chosen a particularly severe example; there

2642: were other baselines in the same data set for which the comb was

2643: barely visible.

2644: \textit{Right:} The same data after RFI have been flagged with \task{FLGIT}.

2645: This panel shows a much more homogeneous brightness distribution than the

2646: one to the left, indicating that most of the RFI (strongest interference) was

2647: removed.}

2648: \label{fig:rfi}

2649: \end{figure}

2650: 

2651: \begin{figure}

2652: \epsscale{0.8}

2653: \plotone{f10.eps}

2654: \caption[]{An example of the distortions introduced by imaging a

2655: region larger than the isoplanatic patch.  This image is a subimage of

2656: a larger image.  The sources shown are roughly 7\arcdeg\ from the

2657: phase center of the larger image; the two brightest sources should be

2658: point-like or nearly so while the third, fainter source is extended,

2659: with a nearly north-south orientation.  The image was produced from

2660: combined B- and C-configuration observations; the beam is 95\arcsec\

2661: $\times$ 83\arcsec\ and is shown in the lower left.  The rms noise

2662: level in the image is approximately 30~\mjybm, and the contour levels

2663: are 30~\mjybm\ $\times -4$, $-2$, 2, 4, 6, 10, 20, 40, 60, 100,

2664: and~150.  The bandwidth and time averaging used in producing this

2665: image are sufficiently small that both contribute a negligible amount

2666: ($< 1$ beamwidth) of smearing.}

2667: \label{fig:isoplane}

2668: \end{figure}

2669: 

2670: 

2671: \begin{figure}

2672: \epsscale{0.8}

2673: %\plotone{f11.eps}

2674: \caption[]{An illustration of the position offsets arising from phase

2675: distortions caused by the largest spatial structures $(\gtrsim

2676: 1000$~km) in the ionosphere.  The contours show the 74~MHz image; the

2677: gray scale is an overlay of the NVSS in this region.  The NVSS is at a

2678: sufficiently high frequency that the ionospheric refraction is much

2679: less than a beamwidth.  Note that not all NVSS sources have a 74~MHz

2680: counterpart.}

2681: \label{fig:ionshift}

2682: \end{figure}

2683: 

2684: \begin{figure}

2685: \epsscale{0.75}

2686: %\plotone{f12.eps}

2687: \caption[]{The temporal variation in the refractive shift caused by

2688: the largest spatial scales in the ionosphere.  Shown is the offset, in

2689: both right ascension and declination, of \protect\objectname[]{Cyg~A}

2690: from its known position as a function of time, with a 1~min.\ sampling

2691: interval.  These observations were taken in the B configuration.}

2692: \label{fig:ionwedge}

2693: \end{figure}

2694: 

2695: 

2696: \begin{figure}

2697: \epsscale{0.8}

2698: %\plotone{f13.eps}

2699: \caption[]{An illustration of the phase distortions caused by

2700: ionospheric mesoscale structure, in this case a traveling ionospheric

2701: disturbance (TID), with the typical scale of order hundreds of

2702: kilometers.  Shown are the phases, measured relative to an antenna

2703: near the center of the array, as a function of time for three antennas

2704: along the west arm of the array.}

2705: \label{fig:iontid}

2706: \end{figure}

2707: 

2708: \clearpage

2709: 

2710: \begin{figure}

2711: \epsscale{0.8}

2712: \rotatebox{-90}{\plotone{f14a.eps}}

2713: \hspace{-8.0cm}

2714: \caption[]{

2715: a) The phase of three antennas relative to a central antenna during an

2716: approximately 8 hr observations of \protect\objectname[]{Vir~A}

2717: illustrating many of the ionospheric phenomena typically observed at

2718: the VLA. All three antennas are located approximately 8 km from the

2719: reference antenna, but represent different azimuths. For the TID the

2720: observed parameters were a period of 750i~s, phase slope of

2721: 50~deg~km$^{-1}$, and a time lag of $\sim$50 seconds over 20 km allowing

2722: \cite{p02} to derive a TID wavelength of 750 km and velocity of 200 m~s$^{-1}$;

2723: b) Same as in panel a except for two antennas at different distances

2724: along the same azimuth, indicating that to first order the phase effects of

2725: all the phenomena are proportional to baseline length;

2726: c) The instantaneous amplitude or apparent defocusing of

2727: \protect\objectname[]{Vir~A} over the same time scale as in panel a,

2728: (Hour Angle -4 corresponds to IAT Time$\sim$8~hrs), sunrise

2729: occurred at $+3^{\mathrm{h}}$;

2730: d) The refraction (or apparent postion wander in both RA and Dec) of Virgo

2731: A over the same time scale as panel a;

2732: e,f)  The apparent differential refraction in RA (e) and declination (f) as

2733: measured towards 5 objects located within 6 degrees of

2734: \protect\objectname[]{Vir~A} and of sufficient strength to be detected and

2735: tracked over the course of the observations. The time scale is the same as

2736: in panel c.

2737: }

2738: \label{fig:ionsmall}

2739: \end{figure}

2740: 

2741: \begin{figure}

2742: \rotatebox{-90}{\plotone{f14b.eps}}

2743: \end{figure}

2744: 

2745: 

2746: \clearpage

2747: 

2748: \begin{figure}

2749: \plotone{f14c.eps}

2750: \end{figure}

2751: 

2752: \begin{figure}

2753: \rotatebox{-90}{\plotone{f14d.eps}}

2754: \end{figure}

2755: 

2756: \clearpage

2757: 

2758: \begin{figure}

2759: \plotone{f14e.eps}

2760: \end{figure}

2761: 

2762: \begin{figure}

2763: \plotone{f14f.eps}

2764: \end{figure}

2765: 

2766: \clearpage

2767: 

2768: \begin{figure}

2769: \epsscale{0.9}

2770: \rotatebox{-90}{\plotone{f15.eps}}

2771: \caption[]{An example of the bias introduced by self-calibration.

2772: Symbols show the location of sources and are proportional to their

2773: flux densities.  This field contains the (strong) source

2774: \protect\objectname[3C]{3C~63} in the upper right.  A clear

2775: non-uniform distribution of other sources across the field is evident,

2776: in addition to the decrease in number density expected from the

2777: primary beam attenuation.  The jagged edges in the image indicate the

2778: boundaries of facets (\S\ref{sec:imaging}); regions beyond the edges

2779: of the image have not been imaged.}

2780: \label{fig:SCbias}

2781: \end{figure}

2782: 

2783: \begin{figure}

2784: \epsscale{0.8}

2785: \rotatebox{-90}{\plotone{f16.eps}}

2786: \caption[]{Differential, ionospheric-induced source wander within a

2787: field of view.  The expected locations of various moderately strong

2788: sources within a single field of view is shown.  The direction of each

2789: vector indicates direction of the position shift in five 5-min.\

2790: intervals; the length of each vector is 100 times the actual

2791: displacement.  Because the magnitude and direction of the wander is

2792: not the same for all sources, averaging over time results in sources

2793: being smeared out and apparently disappearing from view (viz.\

2794: Figure~\ref{fig:SCbias}).}

2795: \label{fig:wander}

2796: \end{figure}

2797: 

2798: \clearpage

2799: 

2800: \begin{figure}

2801: \epsscale{0.9}

2802:  \plottwo{f17a.eps}{f17b.eps}

2803: \caption[]{\textit{Left}: A mosaic of NVSS sources assembled from a

2804: 1~min.\ snapshot of a full-field image at~74~MHz.  The typical

2805: distance of each source from the phase center is 3\arcdeg.  The

2806: cross in each panel marks the nominal location of the NVSS source.

2807: Offsets from this nominal position are due to ionospheric refraction.

2808: \textit{Right}: The Zernike model for $\phi_{\mathrm{ion,lo}}$ over

2809: the VLA for the same time as the mosaic was constructed.  See

2810: equation~(\ref{eqn:zernike}).  Shown is the phase delay screen above

2811: one VLA antenna; because of the ``small-array'' approximation used,

2812: the phase delay screen over all other antennas is essentially the

2813: same.  At a typical 350~km altitude, the phase delay screen shows

2814: structures on scales of roughly 100~km.  Solid contours represent a

2815: phase advance, relative to a nominal phase, while dashed contours

2816: represent a phase delay.  The most negative contour (lower right)

2817: represents a phase delay of~$-30\arcdeg$, and the most positive

2818: contour (upper left) represents a phase advance of~$+30\arcdeg$.}

2819: \label{fig:zernike}

2820: \end{figure}

2821: 

2822: \begin{figure}

2823: \epsscale{0.9}

2824: %\rotatebox{-90}{\plotone{f18.eps}}

2825: \caption[]{The same field as in Figure~\ref{fig:SCbias}, but

2826: calibrated by treating the ionosphere as a phase-delay screen.  The

2827: distribution of sources across the field is far more uniform than

2828: before, though the primary beam attenuation near the edges of the

2829: image is still apparent.  Abscissa: Right ascension; Ordinate:

2830: Declination.  The jagged edges in the image indicate the boundaries of

2831: facets (\S\ref{sec:imaging}); regions beyond the edges of the image

2832: have not been imaged.  The black circle outside the field of view in

2833: the upper right is the location of an outlier field; this field

2834: contains a strong source outside the field of view that was also

2835: imaged.}

2836: \label{fig:SCunbias}

2837: \end{figure}

2838: 

2839: \clearpage

2840: 

2841: 

2842: \begin{figure}

2843: \begin{center}

2844: \epsscale{0.54}

2845: \rotatebox{-90}{\plotone{f19a.eps}}\\

2846: \rotatebox{-90}{\plotone{f19b.eps}}

2847: \end{center}

2848: \vspace{-0.33cm}

2849: \caption[]{

2850: The rms jitter in apparent separations of pairs of sources with

2851: different separations made from a sequence of one minute snapshots

2852: taken over a period of several hours on a day with a moderately

2853: disturbed ionosphere.

2854: \textit{Top} The rms jitter with no correction to the

2855: position difference.

2856: \textit{Bottom} The rms jitter after applying an ionospheric model

2857: described by a 5-term Zernike phase screen determined from

2858: that snapshot.  The large rms values at angular separations greater

2859: than approximately 15\arcdeg\ result from \protect\objectname[]{Cyg~A}

2860: which was not corrected as part of this Zernike modeling.}

2861: \label{fig:sf}

2862: \end{figure}

2863: 

2864: \clearpage

2865: 

2866: \figcaption{Various (resolved) 3C sources at~74~MHz (left panel) and 330~MHz

2867: (right panel). For each source, we quote the rms noise level and the beam

2868: diameters in Table~\ref{tblobs}. All beams are circular unless otherwise

2869: noted. The contour  levels are given in terms of the  rms noise level in

2870: the image i($-3$, 3, 6, 12, 24, 48, \ldots times that rms noise level)

2871: and the beam is shown in the lower left.

2872: (\textit{a}) \protect\objectname[3C]{3C~10};

2873: (\textit{b}) \protect\objectname[3C]{3C~33};

2874: (\textit{c}) The central strong sources of the \protect\objectname[]{Perseus

2875: cluster} at~74~MHz.  The source in the lower left is

2876: \protect\objectname[3C]{3C~84}, the compact source near the center is

2877: \protect\objectname[3C]{3C~83.1A}, and the extended radio galaxy in

2878: the upper right is \protect\objectname[NGC]{NGC~1265};

2879: (\textit{d}) \protect\objectname[3C]{3C~84};

2880: (\textit{e}) \protect\objectname[3C]{3C~98};

2881: (\textit{f}) \protect\objectname[3C]{3C~129} (western source) and \protect\objectname[3C]{3C~129.1} (eastern source);

2882: (\textit{g}) \protect\objectname[3C]{3C~144};

2883: (\textit{h}) \protect\objectname[3C]{3C~218};

2884: (\textit{i}) \protect\objectname[3C]{3C~219};

2885: (\textit{j}) \protect\objectname[3C]{3C~274} or \protect\objectname[]{Vir~A};

2886: (\textit{k}) \protect\objectname[3C]{3C~327};

2887: (\textit{l}) \protect\objectname[3C]{3C~353};

2888: (\textit{m}) \protect\objectname[3C]{3C~390.3};

2889: (\textit{n}) \protect\objectname[3C]{3C~392};

2890: (\textit{o}) \protect\objectname[3C]{3C~405} or \protect\objectname[]{Cyg~A};

2891: (\textit{p}) \protect\objectname[3C]{3C~445};

2892: (\textit{q}) \protect\objectname[3C]{3C~452};

2893: (\textit{r}) \protect\objectname[3C]{3C~461}.

2894: \label{fig:3c}}

2895: 

2896: 

2897: %\begin{figure}

2898: %\epsscale{1}

2899: %\plottwo{f20a.eps}{f20b.eps}

2900: %\end{figure}

2901: %

2902: %\begin{figure}

2903: %\plottwo{f20c.eps}{f20d.eps}

2904: %\end{figure}

2905: %

2906: %\begin{figure}

2907: %\plottwo{f20e.eps}{f20f.eps}

2908: %\end{figure}

2909: %

2910: %\begin{figure}

2911: %\plottwo{f20g.eps}{f20h.eps}

2912: %\end{figure}

2913: %

2914: %\begin{figure}

2915: %\epsscale{1}

2916: %\plottwo{f20i.eps}{f20j.eps}

2917: %\end{figure}

2918: %

2919: %\begin{figure}

2920: %\epsscale{1}

2921: %\plottwo{f20k.eps}{f20l.eps}

2922: %\end{figure}

2923: %

2924: %\begin{figure}

2925: %\plottwo{f20m.eps}{f20n.eps}

2926: %\end{figure}

2927: %

2928: %\begin{figure}

2929: %\plottwo{f20o.eps}{f20p.eps}

2930: %\end{figure}

2931: %

2932: %\begin{figure}

2933: %\plottwo{f20q.eps}{f20r.eps}

2934: %\end{figure}

2935: %

2936: %\begin{figure}

2937: %\plottwo{f20s.eps}{f20t.eps}

2938: %\end{figure}

2939: %

2940: %\begin{figure}

2941: %\epsscale{1}

2942: %\plottwo{f20u.eps}{f20v.eps}

2943: %\end{figure}

2944: %

2945: %\begin{figure}

2946: %\plottwo{f20w.eps}{f20x.eps}

2947: %\end{figure}

2948: %

2949: %

2950: %\begin{figure}

2951: %\epsscale{1}

2952: %\plottwo{f20y.eps}{f20z.eps}

2953: %\end{figure}

2954: %

2955: %\begin{figure}

2956: %\plottwo{f20aa.eps}{f20ab.eps}

2957: %\end{figure}

2958: %

2959: %\begin{figure}

2960: %\plottwo{f20ac.eps}{f20ad.eps}

2961: %\end{figure}

2962: %

2963: %\begin{figure}

2964: %\epsscale{1}

2965: %\plottwo{f20ae.eps}{f20af.eps}

2966: %\end{figure}

2967: %

2968: %\begin{figure}

2969: %\epsscale{1}

2970: %\plottwo{f20ag.eps}{f20ah.eps}

2971: %\end{figure}

2972: %

2973: %\begin{figure}

2974: %\epsscale{1}

2975: %\plottwo{f20ai.eps}{f20aj.eps}

2976: %\end{figure}

2977: 

2978: \begin{figure}

2979: %\plotone{f21.eps}

2980: \caption[]{VLA + PT Link full synthesis 74~MHz image of Cas A, with a resolution of

2981: $\sim$8\arcsec. The bottom bar shows the color transfer function in Jy.

2982: Image courtesy T. Delaney \cite{Delaney2004}}

2983: \label{fig:casapt}

2984: \end{figure}

2985: 

2986: \clearpage

2987: 

2988: \begin{deluxetable}{lr}

2989: \tablecaption{Performance Characteristics of the VLA 74~MHz

2990: 	System\label{tab:perform}}

2991: \tablewidth{0pc}

2992: \tablehead{}

2993: \startdata

2994: Center Frequency                  & 73.8~MHz \\

2995: Bandwidth                         & 1.6~MHz  \\

2996: Primary Beamwidth (FWHM)          & 11\fdg7  \\

2997: System Temperature (minimum)      & 1500~K   \\

2998: Aperture Efficiency               & $\approx 15$\% \\

2999: Point Source Sensitivity (A or B configs., 8~hr) & 25$-$50~mJy\\

3000: Resolution (A configuration)      & 25\arcsec \\

3001: \phantom{Resolution} (VLA $+$ PT) & 12\arcsec \\

3002: \enddata

3003: \tablecomments{See also Figures~\ref{fig:beam}, \ref{fig:bandwidth},

3004: and~\ref{fig:time}.}

3005: \end{deluxetable}

3006: 

3007: \begin{deluxetable}{lcccl}

3008: \tablewidth{0pc}

3009: \tablecaption{Comparison of VLA and \citeauthor{kwp-tn81} Flux Density

3010: 	Estimates\label{tab:fluxscale}}

3011: \tabletypesize{\footnotesize}

3012: \tablehead{

3013:  \colhead{Name} & \colhead{\citeauthor{kwp-tn81}} & \colhead{VLA} &

3014: 	\colhead{Ratio} & \colhead{Remarks} \\

3015:                 & \colhead{(Jy)}            & \colhead{(Jy)}

3016: }

3017: \startdata

3018: 

3019: \objectname[3C]{3C~48}    & 69.1  & 67.6 $\pm$ 0.4   & 0.978 &

3020: 	\citeauthor{kwp-tn81} spectrum fit below observations \\

3021: 			   	    		       	        

3022: \objectname[3C]{3C~98}    & 98.9  & 98.0 $\pm$ 1.1   & 0.991 & \\

3023: 			   	    		       	        

3024: \objectname[3C]{3C~123}   & 387.6  &  414.7 $\pm$ 1.3 & 1.070 & 

3025: 	Both \citeauthor{kwp-tn81} spectrum and VLA data look good \\

3026: 			   	    		       	        

3027: \objectname[3C]{3C~147}   &  67.1 &  55.3 $\pm$ 0.8  & 0.825 &

3028: 	\citeauthor{kwp-tn81} estimate could be high, only one datum \\

3029:                           &        &                  &       &

3030: 	below~178~MHz and spectrum fit well below it \\

3031: 			   	    		       	        

3032: \objectname[3C]{3C~219}   &  96.6 &  94.8 $\pm$ 0.5  & 0.981 & 

3033: 	\citeauthor{kwp-tn81} spectrum looks good \\

3034: 

3035: \\

3036: 

3037: \objectname[3C]{3C~274}   & 2281.3 & 2084.6 $\pm$ 1.3    & 0.914 & 

3038: 	Both \citeauthor{kwp-tn81} spectrum and VLA data look good \\

3039: 

3040: \objectname[3C]{3C~327}   &  93.1 & 118.4 $\pm$ 0.8  & 1.272 &

3041: 	\citeauthor{kwp-tn81} spectrum could be low, \\

3042:                           &        &                  &       &    

3043: 	falls below~38~MHz datum but fits 80~MHz datum \\

3044: 			   	    		       	        

3045: \objectname[3C]{3C~353}   & 437.0  & 443.8 $\pm$ 0.8  & 1.016 &

3046: 	\citeauthor{kwp-tn81} spectrum is unreliable below~160~MHz \\

3047: 			   	    		       	        

3048: \objectname[3C]{3C~390.3} & 107.2  &  97.1 $\pm$ 0.4  & 0.906 &

3049: 	\citeauthor{kwp-tn81} spectrum may be high, \\

3050:                           &        &                  &       &

3051: 	misses 38~MHz datum \\

3052: 			   	    		       	        

3053: \objectname[3C]{3C~445}   &  62.6 & 49.2 $\pm$ 0.3 & 0.785 & 

3054: 	VLA flux density unestimated?; B-configuration data only \\

3055: 

3056: \\

3057: 

3058: \objectname[3C]{3C~452}   &  142.5 & 142.6 $\pm$ 0.2  & 1.001 & \\

3059: 

3060: \enddata

3061: \tablecomments{The uncertainties given for the VLA measurements are

3062: merely the formal standard deviations from fits to the images.  Little

3063: significance should be attached to them.}

3064: 

3065: \end{deluxetable}

3066: 

3067: \begin{deluxetable}{ccccc}

3068: \tablecaption{Zernike Polynomials $Z^l_n(\rho, \phi)$\label{tab:zernike}}

3069: \tablewidth{0pc}

3070: \tablehead{

3071: 		      & \multicolumn{4}{c}{$l$} \\

3072: 	\colhead{$n$} & 0 & 1 & 2 & 3}

3073: \startdata

3074: 

3075: 1 & \nodata & $\rho e^{i\phi}$ & \nodata & \nodata \\

3076: 2 & $2\rho^2 -1$ & \nodata & $\rho^2 e^{2i\phi}$ & \nodata \\

3077: 3 & \nodata & $(3\rho^3 - 2\rho) e^{i\phi}$ & \nodata & $\rho^3 e^{3i\phi}$ \\

3078: \enddata

3079: \tablecomments{The polynomials are expressed in terms of a radial

3080: distance from the phase center~$\rho$ and a position angle~$\phi$.

3081: The $Z^0_0$ polynomial is not used because it represents a total phase

3082: contribution (``piston'') to which the interferometer is insensitive.}

3083: \end{deluxetable}

3084: 

3085: \begin{deluxetable}{llllll}

3086: \tablecaption{Evolving Techniques for Ionospheric Calibration\label{tab:ioncalib}}

3087: \tablewidth{0pc}

3088: \tablehead{\colhead{Method}&\colhead{Parameterization}&

3089: \colhead{Imaging}&\colhead{FoV}&\colhead{Baseline}&\colhead{Reference}\\

3090: \colhead{}&\colhead{}&\colhead{Capability}&\colhead{}&\colhead{}}

3091: \startdata

3092: Simple &None – re-register &No restrictions&Full field &$\leq$5 km&\cite{e84}\\

3093:  geometric shift&position of known&&($\lambda$/D$_{station}$)&&\\

3094: & sources&&&&\\

3095: &&&&&\\

3096: Classical& $\phi_i$(t) - one term& Bright, isolated& $\leq$15\arcmin&$\leq$400 km&\cite{kped93}\\

3097: Self Calibration&per i station&sources (e.g.& &&\cite{Gizani2005}\\

3098: && 3C objects)& &&\cite{lazio06}\\

3099: &&&&&\\

3100: Field-based &$\phi$(t,$\alpha$,$\delta$) - single & No restrictions& Full field&$\leq$12 km&\cite{cotton04}\\

3101:  calibration&term for entire&(used for VLSS)&($\lambda$/D$_{station}$)&&\\

3102: &array&&&&\\

3103: &&&&&\\

3104: Joint &$\phi$(t,$\alpha$,$\delta$) -  one& No restrictions& Full field&$\leq$400 km& TBD\\

3105: Multi-source&term per&&($\lambda$/D$_{station}$)&&\\

3106: Self-calibration&i stations&&&&\\

3107: \enddata

3108: \end{deluxetable}

3109: 

3110: \begin{deluxetable}{ccc}

3111: \tablecaption{74~MHz VLA Classical Confusion

3112: 	Limits\label{tab:confuse}}

3113: \tablewidth{0pc}

3114: \tablehead{

3115: 	\colhead{Configuration} & \colhead{Resolution}  &

3116: 	\colhead{Confusion Level} \\

3117: 	                        & \colhead{(\arcsec)} &

3118: 	\colhead{(mJy)}}

3119: \startdata

3120: A &  25 &    7 \\

3121: B &  77 &   40 \\

3122: C & 240 &  250 \\

3123: D & 744 & 1500 \\

3124: \enddata

3125: 

3126: \end{deluxetable}

3127: 

3128: \begin{deluxetable}{ccc}

3129: \tablecaption{74~MHz VLA Polyhedral Imaging\label{tab:facet}}

3130: \tablewidth{0pc}

3131: \tablehead{

3132: 	\colhead{Configuration} & \colhead{$\theta_{\mathrm{facet}}$}  &

3133: 	\colhead{$N_{\mathrm{facet}}$} \\

3134: 	                        & \colhead{(\arcmin)}}

3135: \startdata

3136: A & 13 & 720 \\

3137: B & 22 & 250 \\

3138: C & 39 & \phantom{1}80 \\

3139: D & 69 & \phantom{1}25 \\

3140: \enddata

3141: \end{deluxetable}

3142: 

3143: \begin{deluxetable}{llrrrrrrrrr}

3144: \tabletypesize{\scriptsize}

3145: \tablecaption{74~MHz and 330~MHz Observations of Sources\label{tblobs}}

3146: \tablewidth{0pt}

3147: \tablehead{

3148: \colhead{Source}&\colhead{Alternative}&\multicolumn{3}{c}{Integration Time}&\colhead{} &\multicolumn{2}{c}{74~MHz}&\colhead{} &\multicolumn{2}{c}{330~MHz}\\

3149: \cline{3-5} \cline{7-8} \cline{10-11}\\

3150: \colhead{} & \colhead{Name} & \colhead{A-conf.}& \colhead{B-conf.}& \colhead{C-conf.}&\colhead{} &\colhead{beam}&\colhead{rms}&\colhead{} &\colhead{beam}&\colhead{rms}\\

3151: \colhead{}&\colhead{}&\colhead{(min)}&\colhead{(min)}&\colhead{(min)}&\colhead{} &\colhead{(\arcsec)}&\colhead{(Jy)}&\colhead{} &\colhead{(\arcsec)}&\colhead{(mJy)}}

3152: \startdata

3153: 3C010   &Tycho SNR         &71 &80 &35  && 80&0.36 &&7.9$\times$6.7&2.0\\

3154: 3C033   &                  &41 &62 &29  && 25&0.14 &&7.0&7.4\\

3155: Perseus Cluster&           &51 &73 &111 && 94&0.16 &&20.0&6.7\\

3156: 3C084   &NGC1275           &51 &73 &111 && 25&0.08 &&6.0&2.6\\

3157: 3C098   &                  &44 &53 &36  && 25&0.13 &&8.0&3.9\\

3158: 3C129   &                  &13 &93 &87  && 83$\times$75&0.10&&65.0&6.5 \\

3159: 3C144   &M1, Crab SNR      &92 &14 &19  && 25&0.12 &&18.0&47.4\\

3160: 3C218   &Hydra A           &16 &16 &18  && 30&0.17 &&10.0&26.0\\

3161: 3C219   &                  &82 &82 &58  && 25&0.06 &&6.0&1.9\\

3162: 3C274   &M87, Virgo A      &71 &71 &49  && 25&0.08 &&23.0&13.8\\

3163: 3C327   &                  &52 &51 &21  && 25&0.38 &&6.0&4.3\\

3164: 3C353   &                  &51 &51 &27  && 25&0.16 &&7.5&7.3\\

3165: 3C390.3 &                  &72 &81 &27  && 29$\times$25&0.05&&7.5&2.4 \\

3166: 3C392   &W44, SNR G34.7-0.4&51 &51 &27  && 300&1.10 &&25.0$\times$22&15.6 \\

3167: 3C405   &Cyg A             &9  &9  &10  && 31$\times$26&4.63 &&5.0&53.0\\

3168: 3C445   &                  &31 &42 &\nodata&& 30&0.03 &&10.4$\times$9.0&3.0\\

3169: 3C452   &                  &76 &82 &33  && 25&0.12 &&7.0&2.8\\

3170: 3C461   &Cas A             &7  &9  &7   && 30&17.0 &&18.0&227.0\\

3171: \enddata

3172: \end{deluxetable}

3173: 

3174: 

3175: \begin{deluxetable}{lcc}

3176: \tablecaption{Peak Intensities and Flux Densities of 3C Sources

3177: 	at~74~MHz\label{tab:list}}

3178: \tablewidth{0pc}

3179: \tablehead{

3180:  \colhead{Name} & \colhead{$I$} & \colhead{$S$} \\

3181:                 & \colhead{(Jy~beam${}^{-1}$)} & \colhead{(Jy)}}

3182: \startdata

3183: 

3184: \objectname[3C]{3C~10}  &  13.0  &  252.1 $\pm$ 1.94   \\

3185: \objectname[3C]{3C~33}  &  55.3  &  105.3 $\pm$ 0.41   \\

3186: \objectname[3C]{3C~48}  &  66.9  &   67.6 $\pm$ 0.44	   \\

3187: \objectname[3C]{3C~84}  &  82.2  &  171.3 $\pm$ 0.52   \\

3188: \objectname[3C]{3C~98}  &  35.7  &   98.0 $\pm$ 1.12	   \\

3189: 							     

3190: \\							     

3191: 							     

3192: \objectname[3C]{3C~123} &  387.8 &  414.7 $\pm$ 1.31 \\

3193: \objectname[3C]{3C~129} &  27.8  &   70.4 $\pm$ 0.48   \\

3194: \objectname[3C]{3C~144} &  305.6 & 1811.3 $\pm$ 3.07 \\

3195: \objectname[3C]{3C~147} &  57.2  &   55.3 $\pm$ 0.82   \\

3196: \objectname[3C]{3C~161} &  85.6  &   87.5 $\pm$ 0.39   \\

3197: 							     

3198: \\							     

3199: 							     

3200: \objectname[3C]{3C~196} &  133.1 &  129.8 $\pm$ 0.95 \\

3201: \objectname[3C]{3C~218} &  270.2 &  644.2 $\pm$ 1.83 \\

3202: \objectname[3C]{3C~219} &  38.9  &   94.8 $\pm$ 0.50   \\

3203: \objectname[3C]{3C~273} &  142.2 &  140.6 $\pm$ 1.42   \\

3204: \objectname[3C]{3C~274} &  567.2 & 2084.6 $\pm$ 1.29 \\

3205: 							     

3206: \\							     

3207: 							     

3208: \objectname[3C]{3C~286} &  27.6  &   27.2 $\pm$ 0.76   \\

3209: \objectname[3C]{3C~295} &  111.6 &  107.9 $\pm$ 1.62 \\

3210: \objectname[3C]{3C~298} &  95.3  &   92.4 $\pm$ 1.22   \\  

3211: \objectname[3C]{3C~327} &  52.8  &  118.4 $\pm$ 0.77  \\

3212: \objectname[3C]{3C~353} &  152.2 &  443.8 $\pm$ 0.76 \\

3213: 							   

3214: \\							     

3215: 							     

3216: \objectname[3C]{3C~380} &  124.4 &  143.7 $\pm$ 2.57 \\

3217: \objectname[3C]{3C~390.3} & 45.4 &   97.1 $\pm$ 0.42 \\

3218: \objectname[3C]{3C~392} &   9.2  &  715.7 $\pm$ 2.88   \\

3219: \objectname[3C]{3C~405} & 9308.3 & 17205.0 $\pm$ 1.44 \\

3220: \objectname[3C]{3C~445} &  9.2   &  49.2 $\pm$ 0.32	 \\

3221: 							   

3222: \\							     

3223: 							     

3224: \objectname[3C]{3C~452} &  38.1   &  142.6 $\pm$ 0.19   \\

3225: \objectname[3C]{3C~461} &  2362.2 &  17693.9 $\pm$ 12.12 \\

3226: \objectname[3C]{3C~468.1} & 42.0  &  40.7 $\pm$ 0.66 \\

3227: 							  

3228: \enddata

3229: \end{deluxetable}

3230: 

3231: \end{document}

3232: