UVMODEL amplitude question

[Laura]

Hi

I've been trying to create a fake gaussian source with the same u-v coverage as my other data, using:

----------------------------------------------------------------
invert vis=mydata map=mymap beam=mybeam stokes=i line=channel,75,550,1,1 robust=-1.0 cell=0.2,0.2 imsize=5000,9000 options=mfs

imgen in=mymap out=fakegauss object=gaussian factor=0 spar=1.0,0,0,180,180,0

uvmodel vis=mydata model=fakegauss out=fakedata options=replace,mfs

invert vis=fakedata map=fakemap beam=fakebeam stokes=i robust=-1.0 cell=0.2,0.2 imsize=3000,3000 options=mfs
----------------------------------------------------------------

When I use mfs in invert, I get an amplitude I can't understad in the final "fakemap" image.
When I do an imstat on the central 1000 pixels of the fakemap image, I get:

Axis 3 (FREQ-LSR)
plane Frequency Sum Mean rms Maximum Minimum Npoints
1 2.0 5.096E+09 5.085E+03 2.349E+03 9.691E+03 -55.0 1002001

So to check, I re-did the second invert above for a range of channels individually, using
line=channel,75,550,1,1 slop=1,zero

I then did an imstat on the central 1000 pixels of the image, and I get:

plane Frequency Sum Mean rms Maximum Minimum Npoints
1 2.0 1.365E+10 1.363E+04 6.405E+03 2.619E+04 -384. 1002001
2 2.0 1.513E+10 1.510E+04 5.529E+03 2.592E+04 3.291E+03 1002001
3 2.0 1.379E+10 1.376E+04 6.431E+03 2.638E+04 -272. 1002001
4 2.0 1.346E+10 1.343E+04 6.381E+03 2.597E+04 -523. 1002001
5 2.0 1.395E+10 1.392E+04 6.524E+03 2.673E+04 -291. 1002001
6 2.0 1.389E+10 1.386E+04 6.490E+03 2.660E+04 -338. 1002001
7 2.0 1.206E+10 1.203E+04 6.325E+03 2.452E+04-1.087E+03 1002001
8 2.0 1.374E+10 1.371E+04 6.404E+03 2.633E+04 14.6 1002001
9 2.0 1.393E+10 1.390E+04 6.525E+03 2.671E+04 -358. 1002001
10 2.0 1.391E+10 1.389E+04 6.519E+03 2.669E+04 -309. 1002001
11 2.0 1.398E+10 1.395E+04 6.473E+03 2.666E+04 -95.5 1002001
12 2.0 1.392E+10 1.390E+04 6.548E+03 2.676E+04 -341. 1002001
13 2.0 1.395E+10 1.393E+04 6.551E+03 2.680E+04 -336. 1002001
14 2.0 1.397E+10 1.394E+04 6.557E+03 2.682E+04 -405. 1002001
...

So the mfs map has a maximum of 9.691E+03, where in the per-channel map I have a maximum of ~2.6E+04 in each channel.

* How does this relate to the amplitude of 1 that I gave my original Gaussian image?
* Why is the mfs amplitude lower than the per-channel maxima?

Thanks!

[Mark.Wieringa]

Hi Laura,

the image produced by imgen will have a peak flux of 1 as specified, but the total flux will be quite large, due to the small cellsize and large size gaussian. When you use this image as input for uvmodel it assumes the input is in Jy/pixel, so it is the total flux that will appear in the centre of the uvplane. It then depends on the details of your uv coverage how much of this will appear in the image. An mfs image will have a beam that is smoother, with lower sidelobes than a single channel image, but you would need to run a deconvolution algorithm before you could compare the fluxes easily.

In the cube, the beam size would vary significantly from one end of the spectrum to the other. At the low frequency end you'll get a much higher peak flux for the extended gaussian, just because the beam is bigger.

Generally only point sources will appear with easily predictable peak fluxes in the image, for extended sources you need to deconvolve and integrate over the size of the source and divide by the beam area to get a sensible number to compare with.

Cheers,

Mark