Please Note: the e-mail address(es) and any external links in this paper were correct when it was written in 1995, but may no longer be valid.
Mullard Radio Astronomy Observatory, Cavendish Laboratory, Madingley Road, Cambridge CB3 0HE, UNITED KINGDOM
Temperature anisotropies in the CMBR are expected to result from inhomogeneities in the distribution of matter at the surface of last scattering. Measurements of these primordial anisotropies provide powerful constraints on theories of structure formation in the early Universe. Observations on different angular scales probe the amplitude of structures on different mass scales. On angular scales greater than , the Sachs-Wolfe effect (resulting in a scale-invariant spectrum) is expected to be the principal cause of anisotropy Sachs & Wolfe (1967). The COBE satellite has already detected anisotropies at a level of on an angular scale of about Smoot et al. (1992). Many current experiments are sensitive to degree scales, with both detections and upper limits similar to that of COBE (see Schuster et al. (1993),Gundersen et al. (1993),Cheng et al. (1994),Ganga et al. (1993)). There have also been attempts to detect anisotropies on arcminute scales (e.g. Subrahmanyan et al. (1993),Fomalont et al. (1993),Myers et al. (1993)). On angular scales in the range Doppler effects at the surface of last scattering are expected to result in large peaks in the amplitude of the fluctuation spectrum. The Cosmic Anisotropy Telescope (CAT) has been designed to detect primordial anisotropies in this angular-scale range.
The CAT Robson et al. (1993) is a three-element interferometer which can operate at any frequency between 13 and 17 GHz with a bandwidth of 500 MHz. This frequency range was chosen as a compromise between the effects of atmospheric emission, which increase with frequency, and Galactic synchrotron and bremsstrahlung emission, which decrease with frequency. The CAT has a system temperature of approximately 50 K. Variations in the system temperature are continuously measured using a modulated 1-K noise signal injected into each antenna. The baselines can be varied from 1 to 5 m, and are scaled to give the same synthesised beam at different frequencies. The antennas have a primary beam FWHM of at 15 GHz. The CAT simultaneously records data from orthogonal linear polarizations. Its alt-az mount causes the plane of polarization to rotate on the sky as the telescope tracks a given field.
The CAT is situated within a 5-m high earth bank which is lined with aluminium. This shielding reduces the effect of spillover and terrestrial radio interference, but limits observations to elevations above . The control hut is located about 100 m away. Each element of the telescope is a corrugated conical horn with a parabolic reflector. The horns are mounted on a single turntable which can track in azimuth. Each antenna has an individual elevation drive. Preliminary test have shown that crosstalk, correlator offsets, and antenna shadowing - particular problems associated with interferometers - do not affect the performance of the CAT at elevations greater than (Robson (1994),O'Sullivan et al. (1995)).
There are several factors to be considered when choosing a field for deep CMBR observations. Most importantly it must lie away from regions of bright radio emission such as the Galactic plane and bright discrete radio sources. The 408-MHz all-sky radio survey of Haslam et al. (1982) was convolved with a beam and used to locate suitable low brightness regions of the sky. Fields at high declinations were most suitable for CAT observations as they were above the screen for longer periods. Fields were discarded near to the north celestial pole, where the natural fringe frequency of the sky was too low to provide useful distinction between wanted and unwanted signals. The Green Bank 4.85-GHz sky maps Condon et al. (1989) were used to locate regions of the sky relatively free from bright discrete radio sources, difficult for such a large field of view as that of the CAT. One field (referred to as CAT1) at RA , DEC (B1950) contained few sources brighter than 5 mJy (at 4.85 GHz) within the half-power points of the primary beam, and observations were carried out here over an extended period.
The CAT1 field was observed for six weeks from 1993 December to 1994 February. During this period the field was above the screen for about 16 hours each night. No observations were made below an elevation of . The observations were made at night to avoid solar interference. Cas A was observed for 1 hour at the beginning and end of each night's run for calibration purposes.
The measured visibility amplitudes were adjusted to take account of the variations in gain as a result of system temperature fluctuations. Flux density and phase calibrations were then carried out using the observations of Cas A. The phases of the recorded signals were reduced in software to a common reference at the pointing centre. Calibrated data were averaged over 100 s before being converted to FITS format and passed to the NRAO AIPS package for further processing. Each night's data was scanned by eye and all sections obviously affected by interference or atmospheric emission were discarded. Even if the interference was only evident on the shorter spacings, data taken during the relevant time period were discarded from all baselines. The bad points were easy to detect (Robson et al. (1994)) and more than 60 per cent of the data were used in making the final map. A 3- clip was also applied to minimize the effect on the final map of single-point excursions. The rms noise level of the remaining data showed no variation with baseline length, as expected if no weather contamination remained at a significant level. Successive data files were then concatenated and maps were made using a total of about 300 hours of good data.
Figure 1: 13.5-GHz map of the CAT1 field showing the difference between orthogonal polarizations. The FWHM of the CAT primary beam is approximately .
Figure 1 shows a map made by subtracting data recorded in linear orthogonal polarizations. The unpolarized components of the astronomical signals cancel. Polarized components would be expected to appear towards the centre of the map, in that region covered by the primary beam of the telescope. Radio sources are typically polarized at, or below, the few per cent level and the contribution from any polarized component of the CMBR is negligible. The rotation of the plane of polarization on the sky further reduces any unpolarized signal appearing on this map. The noise on the map is entirely consistent with the calculated instrumental noise given the system temperature of the telescope. (For a conversion between flux density sensitivity and sensitivity to brightness temperature on the sky see Hobson et al. (1995) and for its application to these data see Scott et al. (in preparation)). Figure 2 shows a map made by adding together data recorded in orthogonal linear polarizations.
Figure 2: 13.5-GHz map of the CAT1 field with orthogonal polarizations added.
The noise variance well away from the centre is the same as that on the previous map, , but now astronomical signals appear in that part of the map covered by the primary beam. The variance, , in this region of the map includes contributions from the instrumental noise , radio sources , the Galaxy and the CMBR . The variance due to radio sources can conveniently be considered as the sum of the variance due to sources with flux densities greater than 10 mJy, , (the Ryle Telescope completeness limit,see below) and that due to sources with a flux density less than 10 mJy, , (which give rise to confusion noise). The variance can be written as
After correction for the primary beam pattern it is found that in the field of view of the antennas.
Each candidate field was scanned using a raster technique with the Ryle Telescope (RT; also at Cambridge) to produce a map of the central region with a noise level of 2 mJy and a resolution of (Pooley, in preparation). Bright peripheral sources identified from the 4.85-GHz survey were also observed with the RT to determine their flux densities at 15.2 GHz. Flux densities at 1.4, 2.7, 5 and 10.7 GHz, where available Kühr et al. (1981), along with the RT data, were used to determine the source flux densities at 13.5 GHz. All sources brighter than 10 mJy in the central , and all other sources likely to contibute 10 mJy or more to the final map (taking into account the attenuation of the primary beam away from the map centre) were subtracted. In all, it was found necessary to subtract the visibilities of 31 sources from the measured visibility. Figure 3 shows a simulated map showing what would have been recorded had the CAT observed only these 31 radio sources in the absence of any noise.
Figure 3: 13.5-GHz simulated noiseless map of the CAT1 field containing only foreground (discrete) sources.
Figure 4: 13.5-GHz map of the CAT1 field with 31 sources subtracted.
Figure 4 shows the map made after source subtraction. Comparison with Figure 1 reveals an excess variance towards the centre of the map, given by
After correction for the primary beam it is found that . A source detection limit of 10 mJy gives , so that
There is clear evidence for residual emission at 13.5 GHz consistent with reported levels of primordial anisotropy, but further measurements at different frequencies will be needed to separate the Galactic and CMBR components.
Preliminary tests have shown that the CAT is reaching its design sensitivity. During typical winter conditions 60 per cent of the data collected can be used. One candidate field (CAT1) has been observed at 13.5 GHz, with approximately 300 hours of data being used to make the final map. The Ryle Telescope has made raster scans at 15.2 GHz with sufficient sensitivity to subtract bright foreground radio sources, and after source subtraction the data show evidence for residual structure. The CMBR component of this anisotropy will be determined using future multi-frequency measurements.