Simulated data-points (black) and test-points (red) to demonstrate the observable quantity Q2 and the effect of the elliptical correction. Panel a: Q2 before the elliptical correction. Panel b: Q2 after the elliptical correction. X-axis: Q2 (normalized high-pass filtered intensities) for the ON-beam. Y -axis: same for the OFF-beam. Points with high values in the ON- and OFF-beam (i.e. close to the main diagonal) are due to either residual RFI or ionospheric fluctuations. One of the main detection criteria is based on the number of points with high values only in the ON-beam or only in the OFF-beam. For this, the regions used in Q3e and Q4e are hatched (orange for the ON-beam, and blue for the OFF-beam; see text for the precise definition) for the case of τ = 3σ (i.e. a threshold of 3σ). This figure also illustrates the effect of the elliptical correction described in Appendix B. The red test data-points allow for the visualization of the displacement of individual points that leads to the circularization of the cloud. Using these red data-points, it can be seen that data-points on the x- and y-axes are the least affected by this procedure; data-points close to the main diagonal are most strongly affected. The black points represent what we expect from an observation, namely sky noise plus a few signal datapoints (visible in panel a at ON ~ 2.0 and OFF ~ 0.0 in this example). Before the elliptical correction, there are no points in the orange and blue hatched regions. After the elliptical correction, the injected data-points are indeed in the orange hatched region.