Spectral Model Subtraction [Verified to work OK]

[len067]

Hi All,

I don't yet have definite proof of this as I haven't had a chance to investigate thoroughly but it appears to me that the task uvmodel does not make use of the spectral index plane that is generated in mfclean models (at least this is how it sounds in the help for uvmodel). Effectively, this means that a simple flat spectrum model is subtracted (or whatever operation is set in options) from the u-v data rather than the model with spectral information. This can be problematic if dealing with large bands.

Cheers,

Emil.

[Mark.Wieringa]

Hi Emil,

looking at the code it looks like it does try to handle mfclean models with 2 planes when options=mfs is specified.
E.g., it contains the following:
if(nz.eq.2) Intp(j) = Intp(j) +
* log(real(sfreq(j)/freq0))*Intp(j+1)
(If there are 2 planes in the model and mfs has been specified, calculate the pixel value using 2nd plane and frequency)

It may still be true it doesn't work properly, if that is true please submit a bug report.

Cheers,

Mark

[len067]

Thanks for checking that Mark.

That's good news! I'm glad it was just my misunderstanding of the help file.

BTW, I've been running a few tests on this over the past week or so and it appears that the code in uvmodel does appear to be working ... or at least the model-subtracted uvdata (mfclean -> uvmodel -> invert) results in an image that is pretty much identical to the residual image for that model (mfclean->restor[options=residual]).

Cheers,

Emil.