spectro-aras.com

For eShel user's, a special version :

http://www.astrosurf.com/buil/isis/eshe ... index.html

Cher collègues utilisateurs de eShel, quelques possibilités supplémentaires dans la V5.9.2 de ISIS. Je vous propose de tester et de voir si cela vous apporte un bénéfice. J'ai retravaillé en particulier l'extraction optimale des spectres en ajoutant quelques réglages supplémentaires. J'ai ajouté aussi le résultat d'expériences acquises en travaillant avec les caméras CMOS, s'il vous vient l'idée d'utiliser ce type de matériel un jour ou l'autre. Voir ici :

http://www.astrosurf.com/buil/isis/eshe ... index.html

Christian

Christian,

Thanks for the significant update!
While exploring this new version I have come across something a bit confusing. It appears that the per order response that I had calculated using version 5_9_0 does not work with the latest version. Here is a plot of the individual orders for Vega using my per order response and an AOD of 0.15 from version 5_9_1.
And here are the individual orders re-processed with version 5_9_2. Note it makes no difference if order merging is checked or unchecked.
The merged full spectra are the same for both versions (provided order merging is checked).

Is this to be expected? In other words do we need to re-calculate our per order response to use this new version.
I am confused because the old response.dat values for each order seamed to make sense i.e. close to what I would expect for my instrument response.
The new response.dat files will be completely different from my expected response.

Tim

Tim,

I tested ISIS V5.9.2 with your own data and I get a good result by selecting the "per order" option and by selecting "Order merging adjustment" (simultaneous action). See the eShel dialog box:



Here's the result :



It is important to select "Order merging adjustment" when using a response "per order", because the response files do not have a normalization relation between them.

And so, if you do not select "Order merging adjustment" you actually get this effect:



To be able to deselect "Order merging adjustment" option it is mandatory to work with a global response file (not per order). It makes sense (but I admit, it requires some explanation).

Note that today, I prefer this global method, more expeditive. The procedure is simple and standard:

- First step - I normally process the spectrum of a reference star, but without providing the name of the response file (at this point, you can select or not "Order merging adjustment", this has no very significant effect).

- Second step - I calculate the response file by dividing the spectrum obtained at the first step by the theoretical spectrum of the reference star.

- Third step - I normally process the science spectra with the global response file found. Here you can select or deselect the "Order merging adjustment" option without any problem, depending on your needs.

You arrive at a correct result now (per order + Order merging selected) ?

Christian

Tim & all,

For a full compatibility with old processing, I added (V5.9.2b) the ability to set the temperature of the tungsten lamp used. Important: the default value in ISIS is 2750 K. Adopt this value to be consistent with the previous processing (this parameter play when calculating the instrumental response):



To check that you have the correct version after downloading, refer to the Setting tab:



Christian

Christian,

It was not the merged spectra that had me concerned. I agree that the new version of ISIS gives the same result for the merged spectrum even if the old per-order response is used.
What is not the same are the INDIVIDUAL orders that are produced. With versions 5-9-1 and earlier the individual orders line up and overlay the merged spectrum quite closely (when a per-order response is used) - my first plot in previous post. With the new version they do not line up at all - my second plot.
To me this seams not ideal when submitting individual orders - say to BeSS.

I just checked with Version 5-9-2b and it is the same. In fact it does not seam to matter what value I put in for tungsten temperature the results for merged spectra and individual orders are always the same.

Tim

Merci Christian pour cette nouvelle version d'ISIS.

J'apprécie particulièrement le nouvel outil "Trace zones" qui permet de bien vérifier les zones traitées par le logiciel.


Bien pratique pour trouver les bons paramètres avec d'autres spectro que l'eShel.

Tim, I understand your problem.

(the addition of the user's tungsten lamp temperature adjustment is an answer to a different problem than yours - it was not good for this parameter to be internally hard encoded in ISIS).

An important difference between V5.9.1 and V5.9.2 is the normalization of the blaze function. Look the files blaze_32.dat, blaze_33.dat, blaze_34.dat ... produced by version 5.9.1: all the functions are normalized to the unit. Now, look the blaze functions produced by version V5.9.2: they are calculated in ADU (no normalization).

There are some reasons for this choice (it facilitates the introduction of non-optimized fusion for example, ...).

The step between individual orders are largely introduced by the Planck function (black-body curve) of the tungsten lamp (the slope increases in blue with a cold halogen temperature).

There are two important points to note:

- Regarding BeSS, each order is always considered as a spectrum independent of the others, as if they were taken with different instruments in a way. This is the philosophical principle of order cutting. When the echelle BeSS are validated, we never look at the inter-order connection. For curiosity, try to look in BeSS the professional spectra taken with the Musicos spectrograph: the blaze function is not even corrected and order merging is really impossible!

- ISIS V5.9.2 is consistent with ISIS V5.9.2 (!). If we compute the order by order responses with V5.9.2 and then process with V5.9.2 order by order, you will get a good result. It is of course essential.

The problem concerns your old data: processed with ISIS V5.9.2 while the responses files were calculated with V5.9.1. Here, I admit, there is a problem of compatibility with the old processing configuration in the peculiar situation of order per order response correction (but I repeat, even in this situation, the spectra retain their scientific value for BeSS).

There are two solutions: (1) adopt the global response scheme method, (2) re-compute the individual response files (per order) by using V5.9.2 (and very sorry for the inconvenience and the extra work to be done).

Christian

Hello Christian,

Christian Buil wrote: For a full compatibility with old processing, I added (V5.9.2b) the ability to set the temperature of the tungsten lamp used. Important: the default value in ISIS is 2750 K. Adopt this value to be consistent with the previous processing (this parameter play when calculating the instrumental response):

Does this user defined temperature apply across ISIS or just when using the Echelle processing mode ?

Thanks
Robin

Christian,

Thanks for your patience with me
I now understand the differences in version 5_9_2 (and the reasons).

It won't be a burden to re-calculate the per order response as I was doing that on a fairly routine basis anyway.
With my directly coupled spectrograph there is a reason I prefer using a per-order response. The light from the tungsten lamp enters the spectrograph in a different way than the light from a star. The result is that the blaze functions produced are not exactly correct. By using a per-order response this effect can be removed. Fiber fed instruments largely avoid this problem.

Thanks again,
Tim