[mmaimcal] Atmospheric Noise memo

[Darrel Emerson]

Hi Robert,

   I was very interested to read your memo "Reducing Atmospheric Noise
in Single Dish observations with ALMA."  My first guess was that there
would never be enough difference in atmospheric emission across the ALMA
IF band for this to be a workable technique, but you have shown
otherwise; my first guess was wrong.

   I have a comment in further support of your suggestion of using
separate sidebands, rather than just separate sub-bands within the same
sideband, as a way of deriving the component of varying atmospheric
emission.  As you know, to first order it doesn't make much difference
astronomically which precise frequency is used for continuum
observations.  A good observing strategy is obviously to choose that
frequency in the particular atmospheric window where the atmospheric
noise is least.  However, if one does that, then by definition one has
chosen a frequency where the rate of change of atmospheric noise with
frequency is zero - i.e. the worst possible choice for your technique,
because all the IF sub-bands have nearly equal atmospheric emission at
the bottom of the window.  

   I would suggest that, looking at the width of a "typical" atmospheric
window compared to out IF bandwidth and our USB and LSB frequency
separation, while it will usually be possible to fit a single 8-GHz band
(i.e. one sideband) into the optimum part of the atmospheric window, it
will rarely be possible to have both sidebands (i.e. a 24-GHz frequency
range) simultaneously in the optimum part of any given window.  However,
if one sideband is put in the optimum part of the window, in most cases
the other sideband will inevitably be on an edge of the window - making
it probably of little use for useful astronomical data, but of maximum
potential use for deriving the fluctuations in atmospheric emission
within your technique.

   So, my suggestion is that in practice we would rarely try to apply
your technique using adjacent sub-bands of the same IF, but that we may
very often apply your technique using data from the other sideband
that's not near the minimum of atmospheric noise picked for the
astronomical sideband.

   The implications of this are that, in single dish continuum
observing, sideband separation, as opposed to unwanted sideband
suppression, may be more important that we'd been considering so far. 
Also, near simultaneous sideband separation is required - for example,
we need independent sideband data sampled perhaps every 10 ms.  Unlike
the interferometer case of requiring complete sets of (e.g.) Walsh
functions before sidebands can be separated, implying a 1-second time
cycle, for the single dish case I think this shouldn't be a problem -
but we do need to make sure it's possible within the system design.
I think the main impact is in the receiver design.

   Does this make sense?
	
		Cheers,
			Darrel.