iraf.net

Hi,I'm working on some spectra both slit and echelle. I've applied all the reduction steps (bias-dark-flat correction, extraction, scattered light correction etc. and wavelength calibration) except for flux calibration which I didn't need to. My problem is that I don't know how exactly I should find signal-to-noise ratio (S/N) from the spectra I have. Is there any special task for it other than splot? I know I can determine S/N for every single point on the spectrum and for a wavelenght range by splot but I do need other ideas. I think it will be wiser to get as much S/N values for continuum points and then get the average of them but is there any better way to do it? Can you rrecommend something for me to read? Any help will be appreciated.

Well, I still don't know how exactly I can determine the S/N. Is it a silly question or there are so many solutions to this problem or is it a practice-dependent issue? If anyone could say "I don't know it either, i think it's a hard task" or "it is so easy, you idiot.." , i'll be glad... Ozgur Basturk
Ankara University
Astronomy and Space Science Department

Ozgur,Frank is probably the best person to answer but is away until next week, I'm sure he regrets missing your earlier question. The S/N computed in SPLOT is just the mean divided by the rms of the values within the specified range of the 'm' keystroke. The answer to your question is that is probably *is* dependent on exactly what you're doing, but there are a number of ways you could do a similar calculation.For example, the BPLOT task lets you run SPLOT in a sort of batch mode, so you can script many 'm' commands at once. Tasks like LISTPIX and WSPECTEXT can dump the pixel values as text and AVERAGE task in the LISTS package can read these values to compute the needed values over the entire spectrum or just the set of ranges you choose (i.e. you would need to select the value beforehand with a script or something and then use AVERAGE). IMREPLACE or IMEXPR could be used to "clip" absorption features in continuum normalized spectra and the listpix/average solution could then be applied to just the continuum regions, and so on.Hope this helps, Frank may have more comments when he returns.Cheers,
-Mike

Mike,Thanks a million for your reply. Frank wrote something on the "S/N per pixel spectrum" topic. I posted my question there too. I tried bplot and got rid of the pain of taking a lot of measurements by splot and average them.Ozgur