PNV J20233073+2046041 mag 6.8

Information about outbursts of eruptive stars, Be activity, ...
Martin Dubs
Posts: 141
Joined: Fri Jan 06, 2012 9:16 pm
Location: Maienfeld, Switzerland

Re: PNV J20233073+2046041 mag 6.8

Post by Martin Dubs »

Hello Francois,

if you are interested in the continuum and the continuum is well defined (this is not obvious in the case of Nova Del 2013) then your method of determination of the weighted flux of the full spectrum and the spectrum with emission lines removed, calculating a magnitude correction factor and applying the flux equation to the continuum spectrum should give a correct result. It gives the flux of the pseudo continuum and if e.g. the pseudo continuum at 550 nm agrees with the true spectrum at 550 nm you get the correct result by applying the flux conversion in ISIS. However, in my opinion it is unnecessary complicated. The method I propose is based on the definition of magnitude, a weighted (with the response function, in particular Bessell_V) integral of the stellar flux, which includes emission lines. It can be applied without any assumptions on what is the continuum. The method even works if no continuum is present or for stars which do not have a well defined continuum, such as M stars. A small error is introduced in my method because the measured magnitudes are additionally weighted by atmospheric extinction, which reduces the blue intensities somewhat, and the calculated flux is based on the extraterrestric spectrum. The uncertainty is increased by not knowing the used filter and CCD efficiency curve for the magnitude measurement. This difficulty is avoided by Christians direct flux measurement by comparison with a reference star at the same air mass, without using a possibly biased magnitude measurement.
I have studied the papers by Skopal. Their goal is somewhat different. They are interested in color indices U - B and B - V for the continuum, in order to classify the stars. For this it is necessary to remove the emission lines. The equation for the magnitude correction is the same as I use for the flux calculation. A similar situation would arise if you are interested in the spectrum of a star inside a gaseous nebula. Depending on the slit area on the sky you measure a smaller or larger contribution from the nebula in addition to the star, so if you are interested in the properties of the star you would subtract the contribution of the nebula first.
I am not sure if this clarifies the issue somewhat.

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

Re: PNV J20233073+2046041 mag 6.8

Post by Robin Leadbeater »

Hi Martin,

Thank you for putting this method on a firm foundation. This will be a useful when observing under variable sky conditions and it could even be used to flux calibrate past spectra.

I agree that there is no need to remove the emission or absorption lines as these are included in both the photometry and spectroscopy flux measurements. The main potential problem area is the use of filters with different passbands by the photmetrists

Nova Del has been a difficult subject for photometrists. There has been an interesting thread on the AAVSO forum.
http://www.aavso.org/nova-del-2013-photometry
The Baader V filter for example has a longer tail on the red side compared with other filters which is giving about additional 10% flux when I last checked due to the larger proportion of H alpha flux included. The usual correction factors applied to transform photometric data produced by different observers with different equipment probably breaks down when applied to complex spectra like this. In this respect at least, doing spectroscopy is more straightforward :D

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Robin Leadbeater
Posts: 1926
Joined: Mon Sep 26, 2011 4:41 pm
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Robin Leadbeater »

Hi Martin
Martin Dubs wrote: A small error is introduced in my method because the measured magnitudes are additionally weighted by atmospheric extinction, which reduces the blue intensities somewhat, and the calculated flux is based on the extraterrestric spectrum.
If measured properly, the photometric data should have already been corrected for atmospheric extinction (and transformed to approximately correct for the specific filter/camera response, though this is going to be approximate at best for a spectrum like Nova Del) Not all observers do this however

Cheers
Robin
LHIRES III #29 ATIK314 ALPY 600/200 ATIK428 Star Analyser 100/200 C11 EQ6
http://www.threehillsobservatory.co.uk
Francois Teyssier
Posts: 1520
Joined: Fri Sep 23, 2011 1:01 pm
Location: Rouen
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Francois Teyssier »

Cher Martin,
J'ai -enfin- pris le temps d'étudier ta méthode à tête reposée.
Je n'avais pas vu que tu ne passais pas par le continuum, mais que tu intègres la magnitude V sur l'ensemble de la bande.
J'admets sans réserves que non seulement ta méthode est correcte, mais surtout beaucoup plus robuste, notamment parce qu'elle évite le calcul du "continuum" qui est effectivement très incertain sur un tel objet.
Pourrais-tu faire une page web ou un pdf qui servirait de référence ?

J'ai repris ta méthode avec le spectre du 20/09
en utilisant V = 8.09 (SRIC-AAVSO)
puis comparé le spectre obtenu avec le spectre calibré que j'avais obtenu par la méthode spectrophotométrique de Christian (standart HD196544)
C'est parfait; tellement parfait que j'ai recommencé le traitement une deuxième fois, pensant avoir commis un erreur.
NovaDel2013_FluxComp_20130920_798.png
NovaDel2013_FluxComp_20130920_798.png (7.23 KiB) Viewed 8939 times
Il faut maintenant trier les "bonnes" photométrie de la base AAVSO


François
Christian Buil
Posts: 1431
Joined: Mon Sep 26, 2011 6:59 pm
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Christian Buil »

Tout le problème est en effet dans le trie des bonnes magnitudes en fait.

Le calcul est de fait réversible. Autant il est possible de trouver une magnitude en
intégrant le signal spectrophométrique dans l'intervalle spectral d'un filtre,
autant il est possible d'étalonner un spectre en flux à partir d"une magnitude V.
Le processus est parfaitement réversible, heureusement (mais attention,
en ayant en main les équations de rattachement - la constante des magnitudes
en particulier - voir à la fin - noter que Martin indique C = 13.70 et que moi
j'ai trouvé indépendement C = 13.730 à partir de mesures expérimentales - pas mal
quant même ;) ).

(Martin, c'est toi qui à évalué cette valeur de 13,70, où est-ce pris dans une publication ?).

Le passage magnitude-V vers flux en erg/cm2/s/A est par exemple la base de la transformation
d'un spectre MILES en flux relatif vers un flux absolu, lui même à la base de mon
exemple sur HD196544 par exemple (voir la formule de conversion V -> erg).
Tout ce tiens et c'est normal.

Personnellement, j'aime bien travailler directement sur des données spectrophotométriques,
plutot que transiter par la magnitude qui est une information un peu indirecte.
Mais j'admet que si on ne dispose que de la magnitude V, on peut utiliser cette information.
Le calcul est assez simple, il suffit de dérouler à l'envers le calcul de la partie 2 de
ma page, et Martin l'a bien décomposé. Je suis d'accord avec Martin, un outil spécifique
qui intégre le signal de la bande V dans le spectre observé pour extraire le fux en erg
serait commode lorsque le spectre est complexe, comme c'est le cas dans la situation de
cette nova. Je vais réfléchir à cela pour une prochaine version, c'est facile à implémenter.

Donc le problème, ce concentre sur l'exactitude des magnitude V (AAVSO, ...). Il doit
y avoir du bon et du moins bon. Noter un truc important à mes yeux. On donne souvent
la bande V des fitres Johnson, Bessel... Mais attention la bande effective est le produit
de la réponse du filtre et du détecteur (et aussi de l'atmosphère, du télescope...).

La simple forme du profil du filtre n'est pas suffisante. Et c'est là je pense une
des sources d'erreurs sur les mesures de magnitude. L'application d'un profil Bessel sur un spectre
étalonné en flux (ergs) est moins ambigue je pense (encore que, les filtres de Bessel
sont aussi associés à la courbe de réponse d'un détecteur). L'important en fin de compte,
c'est la qualité des équations de rattachement à un système photométrique. Et là, je
serait bien curieux de savoir ce que font nos amis de l'AAVSO. Pas sur que tout le monde
fasse cela rigoureusement (en tout cas pour ce qui est des magnitudes visuelles...).

Vraiment désolé pour la réponse en Français sur une question vraiment fondamentale
et cet échange très intéressent.

Christian

PS : j'ai ajouté une partie 5 dans ma page, sur l'exploitation complémentaire
des données spectrophotométriques (rendement instrument)
Martin Dubs
Posts: 141
Joined: Fri Jan 06, 2012 9:16 pm
Location: Maienfeld, Switzerland

Re: PNV J20233073+2046041 mag 6.8

Post by Martin Dubs »

Hello Christian,

the determination of the calibration constant gave me a lot of headaches. The problem is that there are more magnitude systems than astronomers, if you look up for Johnson UBVRI you find several different response curves, e.g. here:
http://ulisse.pd.astro.it/Astro/ADPS/Systems/index.html
I chose Bessell_V because it corresponds well with some Johnson V curves and to get consistent values with your calculations. No need to increase the confusion by introducing yet another system.
In addition there are different values for the flux of reference stars.
I finally adopted a value which gives a good agreement between the spectra in calspec database (from ISIS) and the V magnitudes indicated there, using Bessell-V, also from ISIS:
calibration constant.PNG
calibration constant.PNG (13.92 KiB) Viewed 8925 times
The good thing is that the equation works independent of spectral class, so I assume that for the HST calibration they use a quite similar response curve.
Using the definition of Vega system I found values closer to your value of 13.73. The difference comes maybe also from the fact that the center of gravity of the Bessell curve is at 548 nm and not at 555 nm, the absolute flux specified at different wavelengths (549 nm in Astro Virginia, 548 nm in Oke Schild 1970, 555 nm in Hayes Latham 1975). It is not surprising therefore to get slightly different values. I honestly do not know what is the current correct Vega flux value, traceable to standards of the metric system.

When I talk about the response curve, I mean the product of detector response * filter transmission, normalized to a peak value of 1, even if I do not say it every time I use it. To be precise, the detector response has to be given in A/W for a flux scale in erg/sec/cm^2/A. That some authors use photon flux instead of energy flux only increases the confusion. It introduces an additional factor of lambda into the integral.

I have gone through this experience before, when trying to specify the efficiency of solar cells for my former employer. There the specification is for a sun spectrum at a certain air mass, but the actual sun rarely has this spectral distribution and solar flux measurement is just about as difficult as in stellar astronomy. Lamps used to simulate solar radiation are even worse.

It would be really nice if the method could be implemented in ISIS, even it is not as precise as your method of flux measurement. Robin also has expressed this wish. Unfortunately I rarely have a photometric sky and quite bad seeing most of the time, being located at the bottom of the Rhine valley with day and night winds blowing up and down the valley.

Regards, Martin
Mike Potter
Posts: 32
Joined: Sat Jan 05, 2013 9:39 am
Location: Baltimore, MD
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Mike Potter »

2013_09_2425.jpg
2013_09_2425.jpg (52.68 KiB) Viewed 8914 times
Here's one of my spectra of Nova Del, this one from September 25th. I'm not 100% certain of the wavelength calibration as I only have a Ne source lamp. I was able to identify some fairly faint lines, partly with the help of Robin's article here, but the probability that I misidentified some of the lines is not zero! Still, using the lines I found I was able to get a fairly good fit (+/- 0.04 Angstroms) with a low-order Legendre polynomial. I'm currently adding the necessary keywords to the FITS headers for submission to the database. I'll also post an image of one of my calibration spectra along with a list of lines I was using to determine the wavelength calibration.

Mike Potter
Baltimore, MD
Terry Bohlsen
Posts: 191
Joined: Thu Sep 29, 2011 7:40 am

Re: PNV J20233073+2046041 mag 6.8

Post by Terry Bohlsen »

Dear All
I have been away for a week so have not taken any spectra during that time. I took another spectra with a LISA last night 29/9/13. There seems to have been quite a change in the profile of the H lines with a split in the profile now evident over the last week.
I will take more tonight.
Cheers

Terry
Image
Terry Bohlsen
Armidale NSW
Australia
Christian Buil
Posts: 1431
Joined: Mon Sep 26, 2011 6:59 pm
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Christian Buil »

Flux calibtared eShel spectrum taken the Sept. 30.7 compared with spectrum taken the Sept. 24.8:

Image

Image

The intensities of all features decrease (also Halpha) (dust absorption ?)

Same but in relative intensity for compare to Terry spectra:

Image

Image

Magnitude evaluation the 30.921 September :

B = 9.076
V = 9.006
R = 7.477

Image

Christian
Francois Teyssier
Posts: 1520
Joined: Fri Sep 23, 2011 1:01 pm
Location: Rouen
Contact:

Re: PNV J20233073+2046041 mag 6.8

Post by Francois Teyssier »

Evolution rapide,
notamment He II ,[OIII], CIII/NIII, [OI]
Tandis que les raies Fe II, CII déclinent
Ndel.png
Ndel.png (10.15 KiB) Viewed 8840 times
Post Reply