spectro-aras.com

Dear Christian,

Since SMSW-2, we have weekly Spectroscopy Discussion Group online meetings. Last week we concentrated on using the Masters tab in ISIS for the proper preparation of master calibration images. The following questions arose that we could not answer, and we are hoping for your advice and guidance.
1) Are images median combined or averaged on the Master Tab? Specifically Offset, Dark and Flat master images.
2) How best to determine Threshold value to use when creating the Cosmetic file?
3) How is the Cosmetic file data applied? What is being done to each pixel identified in the cosmetic file?
4) On the Instrument Tab with eShel tab selected, what does having Sigma-Clipping selected do? Is it applicable to the LISA, ALPY and LHIRESIII?
5) With the Atik cameras we have tested (Atik314ex, Atik414ex and Atik460ex, taking Darks appears to add more noise to the final result than it removes. What do you think?

Sincerely and many thanks,

Woody Sims

Hi Woody, I am not Christian but I can say how I approached 2).

I took a dark frame and ran it through cosmetic file creation process with different thresholds. I then plotted the number of marked pixels v threshold. Then from this I judged where to make the cut.

I had a sharp rise as the threshold cut into the good pixels. There was a shelf of warm pixels which dropped off leaving just the hot pixels. This was with a Sony chip.I decided to cut the warm pixels as well.

Regards Andrew

Hi again Woody.

On 5) yes dark subtraction adds randon noise which is why you need to combine many of them to minimise it. However, it removes systematic error (from the addition of the dark current to the signal) which can be important in calibration. Obviously, if it is low enough it could be missed out as is sometimes done by the Pros using liquid N2 cooling.
Regards Andrew

Hi Woody,

Concerning point #1 :

For the Offset and Dark master I compute a median sum. But before, I add a constant for equalize the background level (to the latest image background level of the sequence). The constant attached to each image is find by computing the median value of all pixel intensity (all the image is considered).

For the Flat master, I compute a simple mean of sequence images (no scaling applied).

Concerning point #2 : see the excellent Andrew's comment. Typically on a ATIK460EX I select between 500 to 2000 deviant pixels (a very small fraction of total number pixel in fact).

Concerning point #3 : on each identified "hot pixel" ISIS compute the median of surrounding pixel (a one pixel ring) and replace the actual value of the hot pixel by the median.

Concerning point #4 : Yes, it's an asymmetry : the Sigma-Clipping algorithm is not implemented for LISA, ALPY, ... for the moment.

Concerning point #5 : a clean Dark master (and Offset) is necessary (more then 10 images in the sequence and an exposure time equal or upper to the science images for the Dark). Also, see Andrew answer.

Clear sky,

Christian Buil

Hi Andrew and Christian,

Thank you both for your clear answers to our questions. I shared them with our Spectroscopy Discussion Group in our meeting yesterday. And Andrew, your practical method for determining a value for the Cosmetic Threshold sounds very sensible. I currently have 3,299 values in my cosmetic file for my Atik414ex cooled to -10C (used with my Shelyak LISA) using a Threshold value of 500. I will implement your process with my darks to determine a new Threshold value.

Best Regards,

Woody

Hi Woody,

If you are interested, here is the discussion on the distribution of warm pixels seen with the
Sony ICX694/674 CCD
http://www.spectro-aras.com/forum/viewt ... f=8&t=1919

On the subject of darks, It takes time to get enough to reduce the noise contribution to an insignificant level but that is what cloudy nights are for

Cheers
Robin