@@ -137,6 +137,8 @@ To illustrate the impact of the imperfect FIR filter response $\left | H \left (
Figure~\ref{fig:observed_aliasing} shows an example of aliasing seen in an actual APERTIF commissioning observation analysed by Tom Oosterloo. Both spectra are created by summing the amplitude of all visibilities per frequency. The large peak in the center is Galactic HI. A bandpass correction was determined by calculating the spectra bandpass for all antennas using an observation on a strong calibrator, i.e., by the standard WSRT bandpass calibration procedure. The left panel shows the observed spectrum before bandpass correction clearly showing the response of the subband filter. The right panel shows the result after bandpass correction. The standard bandpass correction takes care of the amplitude of the subband filter response, but does not correct for aliasing effects. This produces the series of small spikes seen at the edges of the subbands. The spikes around channels 9280 and 9405 are larger due to aliased power from the channels around channel 9340. This example clearly shows that about 7 channels per subband need to be flagged if no anti-aliasing filter is applied. The scientific impact of this will be discussed in more detail in the next section.
{\color{red} action SJW: try to apply AAF to spectra in Fig.~\ref{fig:observed_aliasing}.}
\subsection{Impact on observations}
% source smearing? missed lindes? velocity field reconstruction issues?
...
...
@@ -209,6 +211,10 @@ Action EK: analyse impact of oversampled polyphase filter bank on APERTIF signal
{\color{red}
Action SvdT: describe AAF for both autocorrelations and cross-correlations (before September 19). Information on APERTIF fringe stopping and delay tracking is provided by in the e-mail by EK sent on August 30. The link has been forwarded to SvdT by SJW on September 7.}
{\color{red} Action SvdT: make figure showing where an aliased signal ends up in neighbouring subbands.}
{\color{blue} Update 3-10: SvdT has completed an analysis of phase errors and decorrelation errors in the cross-correlations produced by the delay compensation implementation. This will cause a phase difference of about 1\% (0.01 rad) between the direct signal and the aliased signal in the adjacent subband. Next step is to try and simulated this effect to see to how this affects the effectiveness of the AAF.}
{\color{magenta}
\begin{itemize}
\item action: describe AAF for autocorrelation spectra
