It turns out that the bandpass at the offset frequencies are not necessarily a good match the bandpass at the source frequency. This is unfortunate. When applying the bandpass correction to the phase calibrator, we should see a flat slope. However, we don't. Here are some possible ways of fixing it (if it's a big problem):
- apply to the phase cal and smooth (smooth 7 30 0 -> boxcar smoothing of 30 channels?), or fit a lower-order polynomial to the bandpass (2-3 order)
- can we bootstrap fluxes from B-configuration?
- can we use self-cal?
- see if phase calibrator fluxes have changed over the past year: ie, "proposal" to check fluxes of phase cals
- our line is narrow but so is our band - what is the slope across our line and do we care? A 5% slope or less is probably okay.
- flatten the HI data and see?
To save the smoothed, vector-averaged, calibrated possm spectrum: set outtex.
We will want a saved version of the smoothed bandpass calibrator to estimate this effect.