Methods for detection and analysis of weak radio sources with single-dish radio telescopes

The detection of faint radio sources (mJy/sub-mJy events) is an open issue in single-dish astronomy. To date, observations of undetected radio sources in single-dish mode often result in a rough upper limit based on the root mean square (RMS) of the image. Using the Sardinia Radio Telescope (SRT), we compare the sensitivity and robustness of three methods applied to the detection of faint radio sources from raster maps around a known source position: the smart "quick-look" method, the "source extraction" method (typical of high-energy astronomy), and the fit with a 2-D Gaussian. The targets of interest are the faint radio afterglow emission (due to synchrotron radiation) from gamma-ray bursts (GRBs), flashes of high-energy radiation of cosmological origin. To this aim, we developed a Python code specific for the analysis of point-like radio sources applied to the SRT C-band (6.9 GHz) observations of both undetected sources (GRB afterglows of 181201A and 190114C) and the detected Galactic X-ray binary GRS1915+105. Our comparative analysis of the different detection methods made extensive use of simulations as a useful complement to actual radio observations. The best method for the SRT data analysis is the fit with a 2-D Gaussian, as it pushes down the sensitivity limits of single-dish observations -- with respect to more traditional techniques -- to ~ 1.8 mJy, improving by ~ 40 % compared with the initial value. This analysis shows that -- especially for faint sources -- an accurate knowledge of the radio background is essential.