Correlator mapping, removing Tsys
july04
The correlator mapping steps are:
-
input the data, cals
-
scale to kelvins
-
bandpass correct
-
baseline removal
When doing line searches, the baseline removal
might remove Tsys from every spectra. This will get rid of most continuum
sources. The simplest way to do this is to subtract the mean value of the
bandpass corrected spectra. This can lead to non-zero baselines if you
have a strong line source in the band. Forcing the mean to 0 (with the
line source included) will cause the median value to sit a little below
0 (since the line source sticks up above 0). Some options are:
-
Just use the mean value and forget about it.
-
Remove the median value rather than the mean of the bandpass corrected
spectra.
-
Remove a robust average rather than the median. This should have better
statistics that the median value.
The robust average is computed by removing the mean,
computing the rms, throwing out all points above 3 sigma, and then looping
if you have thrown out any points.
The plot
shows a comparison of these 3 methods. The plot\ is a 1 second
spectrum of a data with a methanol maser present. The data was bandpass
corrected and then Tsys was removed over the entire spectra (excluding
8% of the edges). This was repeated using the 3 techniques.
-
Fig 1 top: This is the spectra after bandpass correction and removal of
the mean bandpass. The maser sticks up to about 120 Kelvins. The blue lines
shows the frequency channels that were used to compute the mean values
to compare the 3 methods.
-
Fig 2 bottom: This is a blowup showing the mean (black) and median (red)
baseline removals (the robust average layed on top of the median). The
dashed horizontal lines are the mean values computed between the vertical
lines (channels 600 thru 1800) for the 3 techniques: black:mean, red: median,
green:robust average.
Residual Mean channels 600-1800
| removal method |
mean channels 600-1800
deg K |
cormapBc
Tsys=value |
| mean |
-1.0 |
tsys=2 |
| median |
-.11 |
tsys=4 |
| robustAverage |
-.043 |
tsys=5 |
For this one instance the robust average did a slightly better job than
the median. It was offset by .043K while the median was offset by .11 K.
The final column shows the Tsys keyword value in the cormapbc()
routine that enables each option.
To use thi
processing: x101/obstech/calbpc/callbn.pro
home_~phil