Post Reply 
 
Thread Rating:
  • 0 Votes - 0 Average
  • 1
  • 2
  • 3
  • 4
  • 5
the size of a background mesh
04-08-2007, 01:21
Post: #1
Question the size of a background mesh
Hi,

I was using swarp to stack several MegaCam images, each contains 36 CCD chips. And the final stack
image has artificial bright features at the corner of each CCD. And the size of the bright feature seems to depend
on the 'BACK_SIZE' parameter in the swarp configuration file. I attach the jpeg image which shows part
of the stack image. The left panel was generated with 'BACK_SIZE' = 64, while the right panel was
generated with 'BACK_SIZE' = 128. My question is how to choose a proper size of the background mesh
in order to get rid of the bright fake features? What is the minimum value of 'BACK_SIZE' one can adopt?

My second question is regarding the paramter 'obversampling' in swarp configuration file. Accoding to the
uesr's guide, '0' means 'automatic'. What does that actually mean by 'automatic'? Does it use
the oversampling factor = 1 for this case?

thanks for any suggestion in advance,
Catherine

   
Find all posts by this user
Quote this message in a reply
04-09-2007, 20:49
Post: #2
RE: the size of a background mesh
Hi Catherine,

this one has nothing to do with the size of the background mesh actually. What you need to do is to use a weight map as well as the input science frames (my guess is, looking at your images, is that no weight-map was used).

You can create a weight map using the TERAPIX weight-watcher software and a flat-field frame. Weights are necessary to tell swarp where the edges of the CCDs and bad columns are. btw, don't worry about the oversampling option, there is no point in changing it for properly sampled MEGACAM images.

a typical weight-watcher command line might be something like this, where you have a mask and flat being combined to make a weight-map and a flag map:

ww -WEIGHT_NAMES $calib_dir/nKs.flat.fits,$mask -WEIGHT_MIN 0.7,0.9 -WEIGHT_MAX 1.3,1e9 -OUTWEIGHT_NAME $weight

cheers
henry
Visit this user's website Find all posts by this user
Quote this message in a reply
04-10-2007, 23:33
Post: #3
RE: the size of a background mesh
Hi Henry,

Thanks for your suggestion!

I followed your instruction of using the weight map when running swarp, and now it wroks! Appended is a small portion of the stacked image I get.
Now my final stack image no longer has the artificial bright features at the edge of each CCDs (although some boundary features can still be seen).

There is still one minor question I encountered when using the weight map: If I just use the flat image as my weight image for swarp, then the bright features are still present. I get rid of that feature only when the weight image is produced by combining flat + mask files. Is it because that only mask image contains the formation of the CCD boundary while flat image doesn't? I checked the header of the produced weight map, and couldn't identify the keywords which specify the boundary information. Did I miss something?

cheers,
Catherine


Attached File(s) Thumbnail(s)
   
Find all posts by this user
Quote this message in a reply
04-11-2007, 09:18
Post: #4
RE: the size of a background mesh
Dear Catherine,

>I followed your instruction of using the weight map when running swarp, and now it wroks! Appended is a small >portion of the stacked image I get.
>Now my final stack image no longer has the artificial bright features at the edge of each CCDs (although some >boundary features can still be seen).

That's great! Those residual features come from the fact that your dithers are probably too small (the default CFHT dither pattern isn't great).

>There is still one minor question I encountered when using the weight map: If I just use the flat image as my
>weight image for swarp, then the bright features are still present. I get rid of that feature only when the weight >image is produced by combining flat + mask files. Is it because that only mask image contains the formation
>of .the CCD boundary while flat image doesn't? I checked the header of the produced weight map, and couldn't
>identify the keywords which specify the boundary information. Did I miss something?

I can't really comment on this, except that you can see if the weight map is being used or not by the output from swarp during stacking if the you see 'UNWEIGHTED' then there's no weight map (do you see that when using the flat?). I never tried making a weight-map without a mask, however. The boundary information is coming from the pixel values in the mask as the edges of the flat field are probably 'ragged' (for sure this information is not stored anywhere in the image header). Anyway, you certainly have to use the mask in order to remove the bad columns and other stuck pixels.

have fun!

cheers
henry
Visit this user's website Find all posts by this user
Quote this message in a reply
04-11-2007, 19:27
Post: #5
RE: the size of a background mesh
catherine Wrote:There is still one minor question I encountered when using the weight map: If I just use the flat image as my weight image for swarp, then the bright features are still present. I get rid of that feature only when the weight image is produced by combining flat + mask files. Is it because that only mask image contains the formation of the CCD boundary while flat image doesn't? I checked the header of the produced weight map, and couldn't identify the keywords which specify the boundary information. Did I miss something?

cheers,
Catherine

Hi Catherine,
you need mask information because MegaCam images (and flats) are not trimmed. If you do not mask (put to zero weight) the overscan, it is taken into account by swarp in the background estimation and subtraction.
Cheers,
Chiara.
Find all posts by this user
Quote this message in a reply
04-13-2007, 01:09
Post: #6
RE: the size of a background mesh
Dear Henry and Chiara,

Thanks for solving my puzzle, which is very helpful! I'm using the mask '2003A.mask.0.36.01.fits' provided by elixir. And I do see the pixel values = 0 outside the regions where science images cover, which confirms your comment The mask image can be read using fv, but not by ds9. Is the mask file not reguar fits image file or it has special format?

cheers,
Catherine
Find all posts by this user
Quote this message in a reply
04-13-2007, 08:50
Post: #7
RE: the size of a background mesh
catherine Wrote:The mask image can be read using fv, but not by ds9. Is the mask file not reguar fits image file or it has special format?

cheers,
Catherine

Dear Catherine,
you can visualise Elixir masks with ds9 by setting the default scale to 'minmax' before loading the file.
The default scale parameter is in general 'zscale'. When images have many pixels with identical values, ds9 cannot build the histogram and crashes.

Cheers,
Chiara.
Find all posts by this user
Quote this message in a reply
Post Reply 


Forum Jump:


User(s) browsing this thread: 1 Guest(s)