Extracting and Fitting Sources from a Data Cube

Extracting sources

Extracting sources — point or extended — will probably be necessary at some point during your analysis of your data. There are a number of ways to do this, which you use depends on your science goals.

(A)
Identify the spaxels that contain your interesting information (e.g. by looking at the data in a visualisation tool), and then manually extract those spaxels. You may be able to use the visualisation tool for this, if not then all you need to do is identify which aperture numbers you wish to extract (using your eyes, the displayed data, and the position table) and then use IRAF or IDL to pull out of the RSS fits file or cube the relevant spectra/spatial region.

(B)
For more sensitive extraction of point-sources you may wish to perform a PSF-based extraction (see below). This is something to consider if your spatial sampling is not perfect and you wish to improve the flux performance of your extraction (i.e. to mitigate the effects of bad samping, such as discussed for Integral). It is also worth considering if you are working on a faint point source and wish to improve the SNR of the extracted spectrum, if you need to separate out a point source from background emission or if you want to improve the accuracy of extracting the spectra of two distinct sources that are very close together. For any type of extraction you will have better results if you have a even and contiguous spatial sampling.

As with imaging, the instrumental spatial PSF is set by a combination of the instrument+telescope and the seeing during your observing. Spaxels that are separated from each other by less than the PSF are not independent of each other.

We hope to include more information on this technique at a later date, and scripts for doing this type of work. However, if anyone has done this before, could they please write something here?

Deblending of AGN and host galaxy emission

Two different approaches can be followed to seperate AGN and host galaxy emission in IFS data.
1. The IFS datacube is treated as a stack of monochromatic images and each of them is modeled with an appropriate 2D analytic function including a PSF component for the AGN, which has been extensively used to analyze broad band images of QSO hosts. This technique is used for example in Sanchez et al. (2004), Wisotzki et al. (2004), Lipari et al. (2009), and Husemann et al. (2010) for IFS data.
The disadvantages of this technique are that the S/N of the individual monochromatic images is quite low and is not always sufficient to get a good fit and that the complex distribution of the ionized gas usually does not follow the smooth light distribution of the stars on which the 2D model function are based on.
2. The IFS datacube is treated as a set of spectra in which the shape of the AGN spectrum is assumed to be same in all spaxels (in the absence of atmospheric dispersion) only scale in absolute flux according to the PSF. The brightest spaxel usually provide a high S/N spectrum of the AGN that can be used to create a pure AGN datacube with the PSF to be subtracted from the observed data to obtain a pure host galaxy datacube. This technique was introduced by Christensen et al. (2006) for the study of high-redshift QSOs. One problem of this technique is that it is difficult to obtain a pure AGN spectrum without any contamination by the host galaxy from the stars or the ionized gas.

In both cases the PSF needs to be determined from the AGN itself since the field of view for most of the current IFUs is still too small to capture a bright star simultaneously. For unobscured AGN, in particular QSOs, the broad emission lines allow to construct an empirical PSF directly from the observations as outlined by Jahnke et al. (2004).

We developed a software tool called QDeblend3D based on the second option outlined above. We improved the algorithm of Christensen et al. 2004 with an iterative scheme that is able to at least partially decontaminate the AGN spectrum from any underlying spatially resolved host galaxy emission based on the host galaxy residuals after each iteration. The GUI of QDeblend3D provides a simple datacube viewer written in Python and PyQt to interactively select wavelength regions of the broad lines and to set all the parameters for the iterative deblending algorithm. Since two cubes can be viewed at the same time, the deblended datacubes (AGN and host galaxy) can be compared with the original one to optimize the parameters for the deblending algorithm on the fly. More details of the adopted technique and on the program itself can be found in the QDeblend3D User Manual. Comments and suggestions for further improvements are very much appreciated. We also plan to incorporate the 2D analytic modeling technique into the program.

Two papers using QDeblend3D are currently in preparation and will be submitted soon, stay tuned.

PSF fitting techniques

This is a section we are still working on. Keep you eyes peeled!

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License