Weighting is wrong following opacity correction.

[Max Voronkov]

Hi Malte,

I looked into opacity business in the current code and spotted a potential bug. The opacity correction (STMath::opacity) scales just the spectrum. This is fine, but if all spectra corresponding to different times are then averaged together, the same noise is assumed, although the low elevation scans should have a lower weight. The easiest way to fix it would be to scale TSYS column as well inside STMath::opacity. I can make the appropriate changes, but wanted to discuss it first as this change would alter the meaning of the TSYS column (i.e. it will become TSYS above the atmosphere).

Cheers,

Max

P.S. Theoretically we can also implement the same Bob's scheme of correcting for opacity as used for ATCA data. I look at the code and it is probably a day of work to migrate all the stuff into ASAP and test it. I don't know whether we want to do it for the coming release or sometime later.

[Malte Marquarding]

Hi Max,

I think we should scale the TSYS value as well. We should just document this in the doc string for he function? Is the scaling factor the same for TSYS?

We should leave introducing Bob's scheme for after the release. Maybe just find out what the miriad is doing first.

[Max Voronkov]

OK. I think the scaling factor may be slightly different for TSYS (although very close for our relatively low frequencies). An additional contribution to the noise comes from the atmosphere emission. I can figure out the appropriate formula and make the changes to the code tomorrow if you want me to do it (I have to leave soon for Sydney).

I agree to leave Bob's scheme till later time. It seems to me I understand now quite well what miriad is doing and have extracted a fortran routine to compute opacity. Because it has a number of tabulated empirical factors in the code, one must be careful porting it to C++ (usual indexing issues known to cause problems).

[Malte Marquarding]

scantable.opacity documentation says TSYS will be scaled already. So I fixed this. We can open a new ticket later for addition of Bob's algorithm.