What do we have to observe
to understand the universe?
Saskatoon Home Page
The Saskatoon (SK) telescope was ground-based, operated between 26 to 46
GHz, and probed angular scales between 0.5 and 3 degrees (roughly between
multipoles l=60 and 360). The data was taken with Ka (26-36 GHz) and Q-Band
(36-46 GHz) receivers. Observations were made in the winters of 1993-1995
at an observing site in Saskatoon, Saskatchewan, Canada.
The figure below shows the configuration of the SK experiment. The details of the instrument may be found in Wollack et al. (1993) and Wollack et al. (1996). Observations were made by two independent radiometers, that sequentially shared the same telescope: one operated in the Ka-Band (26-36 GHz) and another in the Q-Band (36-46 GHz). Each of these bands were broken up into 3 frequency sub-bands and 2 polarizations: the sub-bands 26-29, 29-32, 32-36 GHz in the Ka-Band, and 36-39.5, 39.5-43, 43-46 GHz in the Q-Band. Total power radiometers based on HEMT (High Eletron Mobility Transistor) amplifiers were used.
The 6 frequency sub-bands (or channels) in each radiometer viewed the sky in a single, and approximatelly frequency-independent, beam formed by a cryogenic horn and an ambient temperature off-axis parabolic reflector. The FWHM beam of the Ka-Band was (1.42+/-0.02) degrees, while the FWHM beam of the Q-Band was (1.04+/-0.02) degrees. After the parabola, the beam was reflected in a chopping flat plate with dimensions (90 x 150) cm. The chopping flat was swept in azimuth at 3.9 Hz, while the radiometer channels were synchronously sampled at 250 Hz. As the plate moved, the receiver outputs were rapidly sampled. The elevation of the beam was fixed at 52 degrees, the latitude of the town Saskatoon. To point the telescope, the receiver, parabolic reflector, near ground-screen, electronics, and chopping plate were moved in azimuth as a unit. As a result, the beam was scaned back and forth about the vertical axis, giving an effective beam pattern that was sythesized in software by appropriately weighting each sample during the chopper sweep. This allowed us to probe multiple angular scales, to optimize the spatial frequency coverage, to remove atmospheric contamination, as well as to perform a variety internal systematic checks.
All of the beam-forming optics were inside a large fixed aluminum ground screen. A large stationary ground screen shielded the radiometer from terrestrial and solar emission (in the data, contaminating signals from the ground and the Sum were less than 1 micro-Kelvin). The instrument was calibrated using Cassiopeia-A (Cas-A), which was also used to map the beam. Details about the data selection and reduction may be found in Netterfield et al. (1994) and Netterfield et a.l (1996).
In the analysis of this data, there were 3 main goals:
To obtain the angular power spectrum of the CMB;
To make a sky map of the Saskatoon observing region; and
To show that this data was not contaminated by foreground emissions.
Determination of the Angular Power Spectrum:
Comparison of the data with theoretical predictions may be done by comparison with the angular power spectrum. Although the power spectrum was presented in 3 ways in the literature, we decided to present it here in the most compact way: combined into 5 angular scale bins. To produce each of this bins, all the data which is sensitive to a given range of angular scales was combined using the full correlation matrix.
The results are shown in the figure on the left. In this figure, predicted
spectra from 6 representative theories are also shown. They are, from top
to botton at l=160 a flat Lambda+CDM model, a standard CDM, a texture
model, a PPI model, and a model with reionization. Unless the texture model
that was arbitrarily normalized, all other models were COBE normalized.
The angular power spectrum shows a distinct rise from T=49(+8,-5) micro
K at l=87 to T=85(+10,-8) micro K at l=237. Details about
how this power spectrum were obtained, may be found in Netterfield
et a.l (1996).
The Sky Map:
When reducing a CMB data set, one usually wants to produce either an estimate of the angular power spectrum C(l) or a map. Although it is the former that is utimately used to constrain cosmological parameters, there are a number of reasons why one wants to make maps as well
made a Wiener-filtered map of the CMB fluctuations in a cap with 15 degrees
of diameter, centered in the NCP. The map was based on the 1993-1995 data
from the Saskatoon experiment, with an angular resolution around 1 degree
in the frequency range of 30-40 GHz. The signal-to-noise ratio in the map
is of order 2, and some individual hot and cold spots are significant
at the 5 sigma level. The spatial features are found to be consistent from
year to year, which reinforces the conclusion that Saskatoon results are
not dominated by residual atmospheric contamination or other non-celestial
signals. Details about this map-making process can be found in Tegmark
et al. (1996).
One of the major goals of any CMB anisotropy analysis is to determine if the observed signal is due to real CMB fluctuations or due to some foreground contaminant. At the frequency range and angular scale of the SK experiment, there are two major potential sources of foreground contamination: diffuse Galactic emission and unresolved point sources.
The diffuse Galactic contamination includes at least three components: synchrotron and free-free radiation, which are important mainly at frequencies below 60 GHz, and thermal emission from dust particles, which is important mainly at frequencies above 60 GHz. From the theoretical point of view, it is possible to distinguish these three components by observing their different frequency dependence and spatial morphology. In practice, however, there is no emission component for which both the frequency dependence and spatial template are currently well known.
We cross-correlated the SK Ka and Q-Band CMB data with different maps to
quantify possible foreground contamination. We detected a marginal correlation
(around 2 sigma) with the Diffuse Infrared Background Experiment (DIRBE)
240, 140 and 100 microm maps, but we found no significant correlation with
point sources (PS), with the Haslam 408 MHz map (Has) or with the Reich
and Reich 1420 MHz map (RR). The rms amplitude of the component correlated
with DIRBE is about 20% of the CMB signal. Interpreting this component
as free-free emission, this normalization agrees with that of Kogut et
al. (1996) and supports the hypothesis that the spatial correlation between
dust and warm ionized gas observed on large angular scales persists to
smaller angular scales. Subtracting this contribution from the CMB data
reduces the normalization of the Saskatoon power spectrum by only a few
percent. Details about this analysis may be found in de
Oliveira-Costa et al. (1997).
Mark J. Devlin (Info about Saskatoon next-generation: MAT experiment)
C. Barth Netterfield
Edward J. Wollack
Just email email@example.com with the request.
This page is maintained by Angelica de Oliveira-Costa and Lyman Page. It was last modified in August 06, 1998. Comments? Qustions? Just email firstname.lastname@example.org.