Wednesday, January 28, 2009

Clean Boxes from SETFC

Here is a question from an anonymous reader:

I'm new to AIPS, and noticed that using SETFC automatically sets clean boxes for every facet, the same size as every facet. When I've been wide field imaging, I've just been adding smaller clean boxes around sources on top of these. Is this ok, or is it better to delete these automatic boxes, and just manually add ones around obvious sources? Thanks.


I think I know the answer to this, if you click to read on...

You definitely want to delete the big clean box. By and large, you don't want clean boxes to overlap, and you do not want to clean your entire field (which is what you're doing if you leave that big clean box). Someone at NRAO once really stressed to me that you want your clean boxes placed as tightly around the real flux in your sources as possible. So get rid of those big boxes.

5 comments:

Adrienne said...

Eric Greisen once commented that he had tried to very carefully box an HI data cube for cleaning and ended up missing some lower level flux, compared to a different (possibly multi-scale/-resolution clean) method of cleaning the cube.

So just a caveat on using extremely tight CLEAN boxes!

amanda said...

I think the point of relatively tight cleaning boxes is to avoid putting the first sidelobes in the box and accidentally cleaning them instead of real emission. If you have a data set with lots of articles and you set the clean boxes too big, CLEAN thinks the artifacts are real and proceeds to reconstruct them.

amanda said...

articles-> artifacts in the above comment

Michael said...

Sorry but I very much diagree, getting rid of the large CLEAN box systematically biases your image when in general you do not have sufficient a-prioi knowledge of the source you are imaging to accurately place a CLEAN box. As a general rule CLEAN boxes should always be avoided, and the entire field CLEANed. Doing otherwise has little or no scientific validity.

amanda said...

Michael: The CLEAN algorithm is non-linear, so setting clean boxes is important for getting CLEAN to run smoothly. See pages 157-158 in Synthesis Imaging in Radio Astronomy II for details. It is possible to clean without setting boxes, but getting the algorithm to converge to a consistent model is difficult.