|Description:|| Winsorization might be a convenient tool for RFI mitigation.The example shows the effectiveness of this statistical procedure. The figure top left shows the input data: LOFAR measurement-set samples, cross-correlation, imaginary part, one sub-band (@131.25MHz), with 256 frequency channels and 1sec integration time. The data are processed in the following way:|
1) The samples (lower left figure, upper panel), along the time axis in each spectral channel are sorted in ascending order (same figure, centre panel).
2) For the chosen value of ¦Ã, 0¡Ü¦Ã¡Ü0.5 and k = [¦Ãn] winsorization of the sorted data consists of setting the samples unchanged if their numbers are between k and N-k. Otherwise they are equal to the values x_k or x_N-k, respectively, (see figure lower left, lower panel).
3) The same procedure is then repeated in the transversal direction, i.e. along the frequency axis.
4) The result of RFI mitigation is shown in the lower right figure, while the top right figure shows the averaged cross-spectra: the input cross-spectrum with RFI and the output cross-spectrum (winsorized mean)
|Copyright:||All rights reserved|