spectro-aras.com

Je viens d'écrire un article qui décrit avec des détails ma façon de procéder pour calculer la réponse instrumentale, un sujet très recurant dans les discussions (!). Vous pouvez le consulter ici :

http://www.astrosurf.com/buil/instrument_response/

J'ai essayé de rationaliser un peu la chose.

An article about the computation of instrument response based on artificial and natural light sources (in French but illustrated).

Christian

Hello Christian,

Can you confirm if this summary is correct please

I. You no longer correct the spectral images in pre processing by dividing by the flat. You just correct with dark and offset
(But see also my comment here http://www.spectro-aras.com/forum/viewt ... =10#p12661 )

2. You subtract the background in the flat (an excellent idea) and use the flat to calculate an approximate instrument response, assuming a flat lamp temperature. The important thing is to transfer the high frequency variations in the flat to the instrument response
(The advantage of doing this is that a wide bin zone can be used so the noise is reduced compared with a conventional flat correction, particularly in the blue)

3. You then calculate a true master instrument response using a reference star corrected for extinction using air mass and a model of the atmosphere (Instead of using a reference star close in time and position for each observation)

4. You then apply this master instrument response to any other stars (not doing a flat correction), Correcting for extinction using air mass and the model

Thanks
Robin

Robin,

I confirm very well the summary!

You are right, the problem of dusts, but also variations of the pixel-to-pixel gain is not concerned here.

It is a compromise between efficiency and complexity. My focus concern low spatial frequency variations of the gain - the most important problem.

There must be a big difference between a "flat-field" profile wavelength calibrated and an image in pixel coordinates. Due to the flexure of the instrument and other instabilities, the latter is always difficult to manage in practice. My hypothesis is that the surface of the sensor and the other surfaces are reasonably free from dust!

The pixel to pixel detector variation gain (more perceptible on CMOS cameras) is a specific topic : if necessary, I suggest to extract the high-frequency gain variations by generating a flat-field image by using the tools "Masters" -> "Make a PRNU map", and use this image as a flat on "General" panel.

Christian

Oops, this topic is not in the right place. Sorry. He is transferred here:

http://www.spectro-aras.com/forum/viewt ... f=8&t=2318

For those of us "French" challenged, here is Christian's transled article above with all images and diagrams included:
http://www.astrosurf.com/buil/instrument_response_us



James