Processing a lot versus a few spectra

David Cejudo
Posts: 45
Joined: Mon Apr 23, 2018 9:01 pm

Processing a lot versus a few spectra

Post by David Cejudo »

Hello.

I am finding a problem with ISIS. I am observing the same star through the night and obtaining around 200 images each season.
ISIS can process all of them at the same time, but whenever i do it i always find that a bunch of spectra look weird, obviously not well extracted. I have observed this behavior several nights. Maybe about 10 or 15 of the spectra are wrong.
So i processed those weird looking spectra again, this time having only 10 images to process at the same time. In this case the extraction is good.

I am attaching an image to show it. In the image, the upper profile was obtained while processing 170 images at the same time. There is something not correct going on between 4600 and 4800 Angstron.
The second spectra is obtained from the same image but this time only 9 images were processed at the same time.
Attachments
ISIS profiles.png
David Cejudo.
Observatorio El gallinero.
El berrueco, Madrid.
Olivier GARDE
Posts: 1243
Joined: Thu Sep 29, 2011 6:35 am
Location: Rhône Alpes FRANCE
Contact:

Re: Processing a lot versus a few spectra

Post by Olivier GARDE »

What's the time exposure for each raw spectrum ?
If time exposure is too short you can have issue about the spectrum if the position of the star is not very well centered in the slit, so some strange profil in the blue part of the spectrum.
LHIRES III #5, LISA, e-Shel, C14, RC400 Astrosib, AP1600
http://o.garde.free.fr/astro/Spectro1/Bienvenue.html
David Cejudo
Posts: 45
Joined: Mon Apr 23, 2018 9:01 pm

Re: Processing a lot versus a few spectra

Post by David Cejudo »

Hello.

The exposure time for each image was two minutes. Not short!

Anyway, i managed to solve the problem. IWhen i observed the problem, i was using a low Rejection Coefficient of 5.
When changed to a value above 25 -i tried many different values-, the process goes well.

What i do not understand is why with a low value, processing of 10 images is correct but processing of 170 is not. Is this a normal behaviour?

I forgot to mention that i am not using the Cosmic Ray filter. With it the results are worse.
Also, my equipment is a LISA with an Atik 460 camera.
David Cejudo.
Observatorio El gallinero.
El berrueco, Madrid.
Olivier GARDE
Posts: 1243
Joined: Thu Sep 29, 2011 6:35 am
Location: Rhône Alpes FRANCE
Contact:

Re: Processing a lot versus a few spectra

Post by Olivier GARDE »

What I can say is that if the autoguiding is not perfect then you can have continuum problems that become a bit incurgent as in the following two examples.

Here I voluntarily positioned the target not correctly in the slit (with a LISA spectrograph)

Image

and here the target is well positioned in the slit.

Image


It's very important to have perfect images at the beginning, otherwise you can have problems during the process.

Have you checked all 170 individual poses? And deleted some of them if they are not correct?
LHIRES III #5, LISA, e-Shel, C14, RC400 Astrosib, AP1600
http://o.garde.free.fr/astro/Spectro1/Bienvenue.html
Tom Love
Posts: 64
Joined: Sun Apr 21, 2019 5:57 am

Re: Processing a lot versus a few spectra

Post by Tom Love »

I'd be interested in hearing what the practice is with other forum users, but if I'm doing a total exposure that long, I'd tend to use much longer times than 2 minutes for individual science frames - for anything that's 1 hour or longer on a faint object I use individual exposures of 1800 seconds (it's a bit time consuming to take the dark frames, but you can do that on a cloudy night). This also assumes, as Olivier says, pretty good guiding.

Out of interest, are you using a high gain setting, with such short exposures? I did some experiments with the gain turned up and shorter individual exposures, but ended up deciding that a gain setting close to unity was best, in order to give decent dynamic range for a single exposure. Also, with the gain turned up high, it was difficult to get a good calibration frame without saturation. But I'd be interested in hearing about other experiences on gain and exposure length.
-------------------------------------------------
Martinborough, New Zealand. Alpy, Lhires RC12
David Cejudo
Posts: 45
Joined: Mon Apr 23, 2018 9:01 pm

Re: Processing a lot versus a few spectra

Post by David Cejudo »

Hello.

I have checked the few extrange looking images, yes. They look fine to me.
Guiding was also checked, it went well, ±1 arcsec the whole night on average. There are a few peaks of 1.5 or 2 arc sec error but none of them coincide with the times those images were taken.

And the strange issue is that processing those spectra individually, or together with ten or twenty, the result is good. Processing them with all the rest 170 together, the result is bad.
I also tried with BASS software and the result is good as well.

I am attcaching an image of three spectra. They are uncalibrated and without flats or darks applied. The central blue one is a spectrum that always results fine after processing. The other two, with an offset applied for clarity, result always wrong after processing all the 170 together, but fine individually or with only a few more.
In the image all three look very similar to me.
Anyway, as i wrote in the previous post, raising the rejection coefficient to 30 the result is always good for all the profiles.
Attachments
chart_.png
David Cejudo.
Observatorio El gallinero.
El berrueco, Madrid.
David Cejudo
Posts: 45
Joined: Mon Apr 23, 2018 9:01 pm

Re: Processing a lot versus a few spectra

Post by David Cejudo »

And answering to Tom.

The 2 minutes exposure was requested for the project.
I was asked to expose between 2 and 5 minutes, being 2 the preferred time. The idea is to observe a flare and those normally last a few minutes.

Regearding the gain, i think you are refering to a CMOS? I am using an Atik 460, still with the traditional CCD chip.
David Cejudo.
Observatorio El gallinero.
El berrueco, Madrid.
Tom Love
Posts: 64
Joined: Sun Apr 21, 2019 5:57 am

Re: Processing a lot versus a few spectra

Post by Tom Love »

Ah yes, if looking for short flares, that makes a lot of sense! And yes, I was talking about CMOS.
-------------------------------------------------
Martinborough, New Zealand. Alpy, Lhires RC12
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: Processing a lot versus a few spectra

Post by Robin Leadbeater »

Hi David,

Is optimal binning (in the general tab) ticked ? (This works with the rejection coefficient and selects the rows with best SNR but is only needed for weak noisy spectra) If so try with it off

Check for saturation in individual spectra

When using the cosmic filter it is important to always check the @map intermediate files to check what is being healed. On bright narrow spectra it can delete complete parts of rows. Adjusting the threshold can be used to find the best level (under the setup tab)

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
David Cejudo
Posts: 45
Joined: Mon Apr 23, 2018 9:01 pm

Re: Processing a lot versus a few spectra

Post by David Cejudo »

Thanks, Robin.

Optimal binning is checked.

Saturation also was checked. The images are very far from saturation.

I am not using the cosmic filter. I did many tests with different levels but results are worse always in many spectra. Besides, in Astrosurf manual says, regarding the cosmic filter, "However, it must be much more careful with LISA spectra or sometimes spectra acquired with simple Star Analyser grating. These spectra are often so thin that the filter can alter the real parts confusing them with cosmic rays. Therefore caution, and with these data, avoiding using such a filter, if possible."
David Cejudo.
Observatorio El gallinero.
El berrueco, Madrid.
Post Reply