Spectral Baselines

By far the greatest loss of centimeter-wave telescope observing efficiency due to poor spectrum quality comes from solar interference. Measurements of weak extragalactic HI is nearly impossible during the daytime without a large fraction of the time spent calibrating off-source baselines in connection with clever techniques such as those described by Frank Briggs elsewhere in these proceedings. The main instrumental problem is that we do not know the frequency response of our telescope to strong sources in far sidelobes. This is the same problem that we have in creating nulls in the directions of man-made interference.

The instrumental model that we most commonly use to calibrate spectral observations is close to the one shown in Figure  2. We assume that the only difference between signal and reference spectra is the desired signal and that all of the spectrometer filters follow the single source of white noise.

  
Figure 2: Simple receiver model used in most spectrometer calibrations.

Figure 3 is closer to reality. There are many sources of white noise, each of which is injected through its own combination of filter paths. The filters represented in Figure 3 are a combination of antenna multi-path and receiver transfer functions. To the extent that the sources and filter functions common to the signal and reference paths can be assumed to be identical, a lot of complexity can be hidden in the guise of the simpler model.

Image temporarily unavailable. Figure 3: Receiver model that is closer to an actual spectrometer noise and filter distribution.

However, the only three noise sources shown in Figure 3 which can be made to be the same between signal and reference phases are the ground, atmosphere, and receiver noise. The other sources will inevitably be different in the reference because, when the antenna is moved off the signal source, the response of our antenna to the other noise sources changes. A more subtle effect is that, when the fraction of total system noise ahead of each receiver filter in the lower part of the diagram changes, the relative contribution of each filter in the chain to the total system spectral shape will change.

Before we can make major improvements to spectral baselines we must understand our telescope systems to at least the level of complexity shown in Figure 3. Which components shown in this diagram are stable enough to be measured once and made part of a routine correction? Can our antenna theory be strengthened to the point of being able to interpolate a sparsely sampled far sidelobe pattern where solar radiation is expected to enter? What calibration procedures are needed to determine the time variable components? Are there useful areas of the antenna sidelobes where the response is low enough to observe with small modifications to current techniques?

© Copyright Astronomical Society of Australia 1997