LISA processing quick start

Version en Français

It is assumed in this tutorial that you're starting from scratch with ISIS. It describes the basic operation for process spectra from LISA spectrograph. The operations are conducted by steps.

In the "Configuration" tab, specify the folder containing the images to be processed.

Tip: Create a directory per night, a better way to find your data. In the example, the date of the observing night is May 11, 2011.



In the tab "General processing", give clearly information about your instrument, your observing site, and your username. These informations will be included in the FITS files spectra you will produce.

Tip: ISIS retains these informations and delivers it to the next work session. So you have to enter data only once, unless of course you change an instrument for example.

Here, t
he telescope is a Celestron 11.

The camera is a ATIK314L + model.

The observation site is Castanet, a village in the suburbs of Toulouse.




Enter the name of the object to be processed. This name reference the object in the FITS headers.

We will process in this tutorial a sequence of spectra of a bright star:  HD103287. It is the star gamma Ursae Majoris (magnitude 2.4). As name object, you might well have adopted "Gamma UMa".

This star is also known by a its name "Merak". It is strongly recommended not to adopt this non scientific designation (can cause problems during a catalog search about the star).

Finally, in this example, "HD103287" or "Beta UMa" are good choices. At any case, the ideal is to adopt a catalog number (HD, HR, SAO, ...).


We will precise the spectral parameters of images to be processed.

We are in the "General processing" tap.

We dispose of 9 spectra of HD103287, exposed each 1.5 second (a short time to avoid detector saturating - recall LISA is a high performance spectrograph!).

The individual images are named: HD103287-1, HD103287-2, ..., HD103287-9.

Enter the generic name of these images "HD103287-". Also indicate the number number of images (9).

Tip: ISIS can find itself this number if you click the little button that is located right of the text zone before.

Provide the name of the offset image. Click here for an explanation on how to obtain the offset image.

Provide the name of the dark image. Click here for an explanation on how to obtain dark image.

Provide the name of the flat-field image. Click here for an explanation on how to obtain flat-field image.

An image of the neon LISA internal lamp spectrum was taken just after the acquisition of the 9 star spectra. Note the explicit name chosen ("hd103287_neon). This calibration file is required. The neon image spectrum expsoure time is 5 seconds. Fill the corresponding text box.

Note: It is a good idea to acquire neon spectra always with the same exposure time. Because ISIS retains all parameters from one session to another, you will probably not need to give a new informations if you take your neon images with the same exposure time (here 5 seconds).



Give the pixel size of your CCD camera (microns unit).

Here the CCD camera is an ATIK314L+ model equipped with a Sony ICX285 CCD. The pixel size is 6.45 microns.




To perform calculations, the software need the position of a line in the image of the 2D star spectrum. The line in question is the red line of hydrogen at 6562.8 A. It is easy to spot on the right side spectrum.

Open the tab "Image display", then load the first spectrum of HD103287 acquisition sequence.

Adjust as needed the visualization thresholds to see the spectrum and locate Balmer absorption lines.

Measure the X and Y coordinates of Halpha line with the mouse pointer. We do not attempt here a very high accuracy. The position can be estimated at around 3 or 4 pixels precision.

Note carefully the values found. In our example X = 1163, Y= 623.



The operation can be divided into two parts:

1 - associate a wavelength at each point in the observed spectrum as accurate as possible. This is the wavelength spectral calibration.

2 - correct the intensity profile of the observed spectrum of attenuations induced by the instrument (telescope + spectrograph) and the atmosphere. It is the flux calibration.

We analyze here the the flux calibration problem. The ultimate goal is to find the actual intensity distribution into spectrum star as it would be observed from space with a spectrally neutral instruments. A neutral spectrally  instrument reduces the same way blue, green or red part of the spectrum. Somehow, the instrument (and the atmosphere) acts as a color filter.

Our goal is to isolate the instrumental spectral attenuation law. To find it, simply divide the observed spectrum of our star by the theoretical spectrum of the same star (space spectrum).

The choice of the star HD103287 is critical. The continuum is well defined, clean of spectral lines  (type A0V) and also, theoretical profile is available in the literature. Better yet, you'll find in the small spectral database associated with ISIS (download here the ISIS database - 8.4 MB and unzip a folder of your choice).

Indicate the database path in the "Setup" tab.

Now go to the "Profile display" tab. Click on the "Database" button. Select the star on the NAOA list and click "Display" button.

These reference spectra can be compared to the observed spectrum, but for a good result, remember to smooth the reference at the same spectral résolution of LISA data ("Filter" button, and adopt a coefficient between 5 to 6).

You can also use the generic profile of a A0V star, extracted from Pickles data base.

Tip: Pickles database is accessible directly, because it is integrated in the program itself.





From the ISIS database, load HD103287 reference spectrum, or a theoretical A0V spectrum (Pickles) and save it with the name "reference" - for example. ISIS then created the "reference.dat" file in your working directory.




It is now time to process spectrum.

We will proceed in two parts.

During the first pass, ISIS will calculate parameters needed for spectral calibration (evaluation of dispersion law) and the instrumental response (the computed curve describe instrumental attenuation of different parts of the spectrum).

On the second pass, ISIS uses these calibration parameters to deal finally the spectrum of HD103287.

Consider the first pass.

Go to the tab "LISA processing". We have 9 individual spectrum of the star. Rather than operate a single image (the first for example), a good idea is to calculate the average of these 9 images. We call the average image "raw" (by example). On the hard drive, in the working folder, you will find now a file image called RAW.FIT.

Tip: The mean operation is optional but still recommended because it improves the quality assessment calibration parameters.



Compute calibration parameter of your LISA spectrograph.

For the observed 2D spectral image item, give name of the average image: raw.

Enter the image coordinates of Halpha line, measured previously.

Check the box indicating that it will compute spectral response.

For the name the reference spectrum, give "reference" (spectrum of an A0V Pickles profile from ISIS database).

The smoothing factor affects the degree of noise filtering in the response pattern. Adopt the value 4 for the moment. For high coefficient value the smoothing effect is great. A value of 4 or 5 is usually a good choice.

Finally, give a name to the spectral response profile calculated. We choose here simply "response" (ISIS generate "response.dat" file).

If you like, you may also name the file "response_hd103287" for recall you that it was calculated from the star HD103287.

Press the "Go" button. The calculation takes few seconds.



Control the spectral calibration error. In our example the numerical RMS error is near 0.45 Angstroms (one standard deviation). We can consider this result as satisfactory (the error should be below 1 A with the LISA spectrograph).

ISIS also returns the coefficients of the dispersion law (polynomial of degree 3). The order 1 coefficient gives an idea of mean inverse spectral dispersion (here 2.55 A / pixel).

ISIS use stallar Balmer lines for calibrate blue part of the spectrum and neon spectrum for calibrate red part of the spectrum (mix method).

You can also see immediately spectral response curve calculated by opening the "Profile display" tab. The shape is typical of LISA spectrograph and its calibration system. (color temperature of internal tungsten lamp).





We can now calculate the final spectrum of star HD103268.

Open the "General" tab. The software has calculated for you some parameters. They are green surrounded by in theright screenshot. You normally do not touch it.

To calculate the full calibrated spectrum of HD103287, simply press the button GO.

At the end of calculation, ISIS product the final spectral 1D files (FITS and DAT format). The name of the files is constructed from association of object name and start date of the observation.

Note that in the end, ISIS gives an approximate value of spectral resolution (here R = 730 approximately).



Open the "Profile display" tab. Name spectral profile is already pre-filled. Just click on "Load" (or <Return> in the text zone file name).

You can also superpose reference spectrum with the tool "Compare". The superposition is excellent (in blue spectrum observed spectrum, in red Pickles spectrum).





Since the calibration parameters are calculated, the processing of a new spectrum (and more generally all the session of your night) is extremely simple and fast.

For example, suppose that the target is the variable star SS Cygnii.. We acquired 12 elementary spectra image exposed each 300 seconds.

The images are named sscyg-1, sscyg-2, ... sscyg-12.

Was performed at the end of image acquisition session a spectral calibration image with LISA internal neon lamp. The image is called sscyg_neon.

Download the first image of the series via the "Image" display" tab and note approximately the vertical coordinate near the middle of spectrum trace. Here we find Y = 594. Then open the tab "General" tab and report the value.. The rest is unchanged.

You click Go, and the result appears in less than one minute. You can skip to the next star as easily.


Click here : read the description of additional advanced options (hot pixels detection, cosmic rays filter, ...)