ELODIE 3.1 - 1388 high res flux calibrated reference spectra

Design, construction, tuning of spectroscopes
Information and discussion about softwares (telescope remote, autoguiding, acquisition, spectral processing ...)
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

ELODIE 3.1 - 1388 high res flux calibrated reference spectra

Post by Robin Leadbeater »

The ELODIE archive
http://atlas.obs-hp.fr/elodie/
is a useful source of high resolution spectra but the quality varies significantly so is rather unreliable as a source of reference spectra.
I have however just come across the ELODIE 3.1 library of spectra. It is a set of 1388 stars selected from the ELODIE archive and fully flux calibrated to 2.5%
http://arxiv.org/abs/astro-ph/0703658
http://www.obs.u-bordeaux1.fr/m2a/soubi ... brary.html
It looks like a potentially very useful set of high resolution reference star spectra. Has anyone looked at these before? If not I will investigate them further.

Cheers
Robin

EDIT: revised link to ELODIE 3.1 download
http://perso.astrophy.u-bordeaux.fr/CSo ... nload.html
Last edited by Robin Leadbeater on Mon Oct 09, 2017 1:10 pm, edited 1 time in total.
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

Very interesting topic Robin. I hope amateurs can find a spectral library best suitable for intensity calibration of mid and high resolution spectra.

I have a doubt concerning Elodie spectrum of Vega whose profile is enclosed in ISIS database. If I compare it with Calspec spectrum, I can see a large deviation of the continuum slope (I think considering the small interval):

Image

Note it's the same with Pickles A0V. Which one is correct? Is the Elodie Vega profile the same of 3.1 library?

Paolo
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Robin Leadbeater »

Hello Paolo,

I agree. There is a problem with the ELODIE spectrum of Vega in ISIS. This is the danger when using spectra directly from the ELODIE archive. If you look in the ELODIE archive,there are several Vega spectra but they do not all look the same! This is the advantage of the ELODIE 3.1 library. There are 3 Vega spectra in the library which are very similar. They also agree well with the CALSPEC spectrum.
vega_ref_spectra.png
vega_ref_spectra.png (20.82 KiB) Viewed 10908 times
(The zero values in the ELODIE 3.1 spectrum are telluric wavelengths)
Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

Hi Robin, thanks a lot for your analysis. I've downloaded the two archives. There are many stars per A/B classes. I've done a quick statistic (to be verified the sky coverage):

Image

Concerning tellurics, they seem significantly shifted compared to the rest position. In addition they appear different or missing depending on profiles:

Image

I dont know if I was wrong in the loading phase (I loaded the profiles directly in ISIS/VSpec). Tellurics in Vega Elodie spectrum in ISIS library (the one with the strange continuum slope) are however consistent.

Perhaps we should remove (cut) tellurics from Elodie (3.1 library) profiles before division (and remove the tellurics from our spectra too). In your opinion, what is the better way to proceed?

Paolo
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Robin Leadbeater »

Hi Paolo,

I suspect the shift in Telluric position may due to heliocentric correction. I am not not sure why some are missing though. Perhaps they are set to zero based on their magnitude (ie if they are detectable above the noise, which will vary depending on the air mass and SNR of the spectrum) This is just a guess though. There may be some information about this in the ELODIE pipeline description. In any case we will need to remove them, probably best by interpolation. I think it is already good practise to remove the tellurics in our high resolution reference star spectra before calculating the instrument response. (I am undecided about the best way to deal with tellurics at low resolution. Sometimes I try to include the telluric bands in the instrument response so they are removed in the target spectrum, other times I have removed them from the instrument response so they remain in the target spectrum)

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

Thanks Robin. Thinking about spectra for BeSS submission I also would like to mantain telluric lines in the final profile.

With ISIS continuum tool I tried to cut and interpolate the telluric "holes". In blue the resulting Elodie "net" profile:

Image

I did a test with a reference star I observed with Lhires III 1200 (same spectral type but a different star, just an experiment). With tellurics removed on both profiles, they are ready for division to calculate the response curve:

Image

The resulting response curve (blue smoothed). The spectral features on the center-left part of the division seems due to the different observed stars. With the same star I would expected a more regular division curve (maybe some residual due to the slightly different spectral resolution).

Image

The corrected profile of the reference star:

Image

It seems to me that the Elodie 3.1 library profiles could usefully be involved in the spectral reduction. The resolution R=10000 is better compatible than Miles profiles with mid and high res spectra (i.e. Lhires III with 1200 and 2400 l/mm grating). I think the library require a selection for our purposes. Indeed, there are also stars not suitable for calibrating (I saw it includes the Be star Pleione with 29 spectra, several variable stars with many spectra each, etc.)

Paolo
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

Hi Robin and all,

I replaced the entire database of the spreadsheet tool "Reference Miles-star finder", done by me and Marco Leonardi, with stars for which there are Elodie 3.1 library profiles. I didn't apply any filter, so all the 1388 stars are considered.

The basic version (compact, only 2 Mb) do not contain spectra. It may be used to search a valid reference stars for the target. It provide, in addition to other basic star information, the Elodie profile name. If more than one profile is available in the library, the cell with the spectrum name shows a red mark. Moving the mouse cursor over the cell, it pops up a label with the other profiles names for that star. Elodie 3.1 profiles in FITS format are downloadable from the links reported above by Robin.

Image

In this example we are looking for a ref-star suitable for V694 Mon close to its transit at meridian. Excluding the first entry (HD046380) because is an emission star, we could use HD071155 (30 Mon, A0Va) with a small airmass difference.

You can download the spreadsheet here (32 and 64 bit versions from Excel 2010 onwards, let me know if it don't work properly):
http://quasar.teoth.it/html/Elodie/ELODIE31_basic.zip

The full version is similar but includes all Elodie 3.1 R=10000 profiles and you can plot them as for Miles library spreadsheet. In Excel I'm able to manage only profiles in DAT format, so I convert them from FITS. Profiles are also rescaled to 1 at 6615-6620 A.

Image

A small routine was added to "cut and interpolate" the zero values which I found in the tellurics ranges (to understand why). The operation is done on the basis of preceding and successive non zero values (simple linear interpolation). Note that all positive values are unchanged. Pressing the "Show" button you can see the resulting profile in red color. The "Export DAT" button writes both profiles on the sub-folder EXPORT. Saved profiles can be used for calculating the response curve with ISIS, VSpec and others spectro tools. It is convenient because the "telluric holes" are removed, but you can also use the original FITS profiles.

Image

Full version package weighs about 300 Mb. Before sharing it (if useful in some ways) I would like to be sure that there are not problems for credits/copyrights since I use the library spectra (also modified respect to the original FITS). Who can help me to understand the situation?

Thank you,
Paolo
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Robin Leadbeater »

Hi Paolo,

Wow! I have only just caught up with this. Fantastic work!

I found this on the ELODIE archive website concerning the use of the data

Publications making use of the ELODIE Archive should refer to : Moultaka et al. (2004) PASP, 116, 693-698
Publications making use of the ELODIE Library should refer to : Prugniel and Soubiran (2001) A&A, 369, 1048

from
http://atlas.obs-hp.fr/elodie/news.html

so it should be fine to use the data provided you reference the source and describe how it has been modified so that users can include this in any papers they may write. I see there is a list papers where the ELODIE library (and archive) have been used
http://atlas.obs-hp.fr/elodie/Archival_ELODIE_Pubs.html

I agree that the user would need to take care to avoid potentially variable stars. (The same may apply to some of the MILES stars. I tend to stick to main sequence stars where possible which are unlikely to be variable.)

We could probably weed out duplicate spectra, either by choosing a typical one or by making an average spectrum, perhaps rejecting any which show significant variability between spectra.

I see also that the ELODIE library is included in a larger list of resources in HyperLeda which looks a good place to find other sources of reference spectra.
http://leda.univ-lyon1.fr//11/spectrophotometry.html

Thanks!
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

Hi Robin, sorry for the late reply (unfortunately I had a flu during the last days...) and thanks again for all information about this interesting library. In the meantime Marco Leonardi and me have sent a mail to Prof. Philippe Prugniel. He has been very kind and provided us other advices. He told that the zero values may result from the conversion of "undefined" numbers (not all the formats/system support the IEEE 754 representation).

When I plot any library spectra with the professional tool Specview, there are no "holes" in the profile:

Image

If this will be addressed (I hope so), we'll easily create a new set of DAT spectra using the powerful ISIS batch processing commands (as we did for the present set). For the moment, if you want to use anyway the profiles, you can download the full version of the spreadsheet here (300 Mb):

http://quasar.teoth.it/html/Elodie/ELODIE31_spec.zip

As you rightly pointed out, not all spectra are valid for relative flux calibration. Prof. Prugniel told us to remain highly critical of the spectra we intend to involve in the flux calibration process. I also think that considering only main sequence stars, not Be or variable stars, helps us to prevent errors in the continuum. Anyway the spreadsheet contain now all the Elodie stars/spectra but we can confirm the use on the basis of the spectral type (listed) or other information retrieved for the star.

Please report me and Marco any errors, problems or other feedback. Above all, if you find a way to load in ISIS the library profiles without having those zeros in the flux.

Paolo
Paolo Berardi
Posts: 578
Joined: Thu Sep 29, 2011 10:51 pm

Re: ELODIE 3.1 - 1388 high res flux calibrated reference spe

Post by Paolo Berardi »

A correction. The "holes" are present on FITS profiles loaded and displayed with Specview too. Simply the tool doesn't draw the lines to zero or other limit. Zooming in they are visible:

Image

So the "missing" data are really there. An hex dump of FITS files shows several FF blocks where the zeroes should be.

Paolo
Post Reply