Telescopes do not only detect the astronomical signal of interest, but everything that enters their beam. And this fact might be a major problem when the aimed signal is several orders of magnitude fainter than everything else. Some examples of “contaminants” are:
- Diffuse galactic foregrounds (emitting dust, synchrotron emission from supernovae, spectral line emission from interstellar medium)
- Extragalactic foregrounds like radio galaxies and starburst galaxies
- Atmospheric foregrounds
In the above sequence the Planck collaboration illustrates the galactic and extragalactic foregrounds in this case for CMB measurements, beginning by the total signal (including the apparently undetectable CMB signal) at frequencies between 24 GHz and 1 THz, and then progressively subtracting out: emission from both point sources (galactic and extragalactic) and interstellar material. After removing the contaminants, CMB is revealed.
As an example, the foreground for HI intensity mapping at 1GHz, for an angular scale of 1⁰, along a 10⁰ sky strip, at a declination of 45⁰ and galactic latitude of |b| > 30⁰ is T ~ 4600 mK with δT ~82 mK, whereas the signal that we want to measure is δT_HI ~ 0.1 mK, see the following table from Battye et al 2013:
The foreground signals are relatively simple to subtract, as even if they are very strong, their spectra is quite smooth (the fluctuations are much less than the signal), which is not the case of the signal of interest, as the fluctuation expected from HI large scale emission is of the order of the signal.
Radio Frequency Interferences (RFI) and systematics
The true “signal killers” are human made signals, broadly called RFI (radio frequency interferences), that include different sources such as: mobile communications, satellite and airplane communications, TV broadcasting, photovoltaic solar plants, windmills. Below an example of signal from satellite:
Simulation showing the RMS of the power variation received from global navigation satellites across the BINGO bandpass over one year (50 MHz channel widths): solid lines show the variations when satellites within 0, 1, 5 and 10 degrees of the BINGO line-of-sight, also shown the foreseen 50 K system temperature noise RMS, and HI RMS. Adapted from Harper&Dickinson 2016
All sum up, the magnitude of foreground signal is huge compared with the scientific signal one aims to measure. In order to try to detect BAO, all sources of unwanted signal must be suppressed or at least minimized. RFI is minimised by choosing the best site, ie a radio quiet site, and through calibration subtract any existent signal. Noise from telescope electronics also adds up, and that may be minimised via correlation receivers and calibration. The existent worldwide radio telescopes, independently of their size, were built under RFI requirements and systematics, but not at the level BAO detection requires, and that is the reason why so far BAO have not been detected in the radio band.