ISIS Processing Issue

[Keith Graham]

A couple of nights ago, an unusual occurrence took place when using ISIS 5.3.1 to process my ALPY spectra. I have been successfully processing my Alpy spectra with ISIS for the past 5 months. However the last couple of Nova Del and Z And spectra series processed out to a fairly smooth concave curve upward from lower left to upper right with no emission lines. There was no semblance to how those spectra should appear. After looking at the fits header. I noticed a discrepancy in the NAXIS between the properly processed spectra and the newer improperly processed spectra. The NAXIS readings for the good spectra were between 4000 and 4400, but he bad ones were above 5000. I figured there must be something causing ISIS to not see the spectrum as it was processing the images.

The ISIS manual shows that the only box that is checked on the General tab is Optimal binning, so that is how I had been successfully processing my spectra all along. As an experiment, I thought I would try checking the Fixed Y value for sequence box to see what would happen. Viola, this solved the problem. And the NAXIS readings for the corrected spectra were in the low 4000s as were the all of my previous spectra.

Next, I tried reprocessing a series that was successfully processed without the Fixed Y value box checked, but this time I checked that box just to see the results. When I clicked the GO button, all went well to a point when I got the message that ISIS could not proper;y normalize the spectrum and that I should check the Y position of the spectra. When I unchecked the Fixed Y box, processing went through successfully.

I checked my images and the spectra in all images were at the same Y position. So I have no idea why this sudden change in how ISIS processes my images. I am thinking that there must be some thing about these newer images that is different from the other images that is causing ISIS to not recognize the spectra, but I can not seem to find any irregularities. So, can anyone shed any light on what might be happening here and perhaps offer suggestions as to what I should look for so I can return to processing spectra without checking the Fixed Y box?


Cheers,

Keith Graham

[Christian Buil]

Private answer,

Christian

[Christian Buil]

To be carreful with automatic Y-value evaluation if the
spectrum if faint (low SNR), in presence of cosmic rays, if
the spectrum is nearly only composed of emission lines, if the
spectrum trace is not well horizontal, ...

For difficult situations (andf also extended object of course)
I suggest to unselect the Y position automatic search algorithm.
Attached, the screen copy shows how ISIS compute the Y central
position of the spectum (for each individual image).

The full algorithm is you autorize ISIS to compute vertical
position of the spectrum :

(1) determination of Y position for the individual images of a sequence (from
the given initial value - here give a good first guest value - see Calibration tab !)
(2) vertival registration of all the spectra
(3) addition of all the individual 2D imagse
(4) a new determination of Y pos from the sum image
(5) use this Ypos estimation for the final processing.

Keith I recommand a better orientation of horizontal
trace of the spectrum (tilt < 0.5°). Your actual value is 1.1°.

Christian

[Christian Buil]

The "Auto" button of calibration tab indicate to ISIS to
automatically compute the Y-position then you click on the Next button
of General tab.

For the future I project to mask this option if the Y position
automatic option is not required by the user.

Christian

[Keith Graham]

Hi Christian,

Thanks for the detailed reply and for the explanatory diagram. I will certainly straighten that tilt. As an aside, I did try running this sequence today using various scenarios of checking and unchecking the Fixed Y value as well as checking and unchecking the Auto box. The only one that would work is when the Y box is checked and it made no difference if the Auto box was checked or not. With other sequences taken on other nights, the Y box needed to be unchecked or I would get a normalization error in the final processing. So, bottom line, let me try leveling the tilt to see what effect that has. Once that is done, I will be interested in seeing how playing with the Y and Auto boxes affects the outcome.

Cheers,

Keith