read noise and gain in APALL
Mao yewei wrote on Apr 15, 2008
Hi, in APALL task, there are two parameters named "readnoise" and "gain". Does anyone know the "readnoise" and "gain" play what role when extracting spectrum using APALL task? Different values of them make shape of extracted spectra quite different. So I want to know how to set their values properly. I need some help.
Thanks very much in advance!
Owen
Thanks very much in advance!
Owen
Francisco Valdes wrote on Apr 15, 2008
These parameters define a model for the noise or pixel uncertainties. The model is
Typically these are the same descirption as for the noise in a CCD array. So the CCD values can be used with raw or data which has just been bias subtracted and divided by a normalized(!) flat field (that is, the counts are not much different than the raw counts). If other things are done to the image, such as subtracting a background or dividing by some value then it is up to the user to determine how the CCD parameters need to be adjusted. If you have no information then you have to determine the values yourself.
The simplest thing is to measure the standard deviation at some point in the image that is purely noise. You can use implot or imexam for this. Then you can ignore the gain part and simple have
If the values are not a good description of the pixel noise then the cleaning and optimal extraction will be wrong. In particular the cleaning will either find most pixels are cosmic rays or none are.
Yours,
Frank Valdes
variance = rdnoise**2 + gain * data
Typically these are the same descirption as for the noise in a CCD array. So the CCD values can be used with raw or data which has just been bias subtracted and divided by a normalized(!) flat field (that is, the counts are not much different than the raw counts). If other things are done to the image, such as subtracting a background or dividing by some value then it is up to the user to determine how the CCD parameters need to be adjusted. If you have no information then you have to determine the values yourself.
The simplest thing is to measure the standard deviation at some point in the image that is purely noise. You can use implot or imexam for this. Then you can ignore the gain part and simple have
variance = (standard deviation)**2
rdnoise = "the standard deviation"
gain = 0
If the values are not a good description of the pixel noise then the cleaning and optimal extraction will be wrong. In particular the cleaning will either find most pixels are cosmic rays or none are.
Yours,
Frank Valdes
Last post on Apr 15, 2008