Spatial-Spectral fitting of clusters with ASCA

The large and energy-dependent wings on the ASCA telescope spatial response make analyzing spectra of extended sources a challenge. If the source spectrum varies with position then any attempt to model this must include the scattering by the telescope of photons from one part of the image to another. The ascac model is a first attempt in XSPEC to solve this problem by joint spatial-spectral fitting.

To set up the model first extract spectra from the source in annuli centered on the peak emission. There should be a maximum of 5 annuli, each 3 arcmin or more in width. In the spectrum extension of each file set the keyword XFLT0001 to an integer followed by two reals. The integer should label the annuli, starting in the center, ie if there are three annuli then the central one should have a value of 1, the next one out 2, and the outermost 3. The two reals are not used but are required for XSPEC v9 (this bug is fixed in v10 and the reals are not required).

Read the spectra into XSPEC as distinct datagroups along with background files and RMFs but do not read in any ARFs. The ascac model includes the ARF for the ASCA telescopes. Now set up the model with ascac being the first component eg ascac * phabs ( mekal ). The model normalizations for the datagroups should be linked together but any other parameters can differ. The ascac component now takes the models for each annular spectrum, calculates the energy-dependent scattering between annuli, and modifies the model spectra appropriately. The ascac model calculates and stores a matrix of annulus cross-terms for each energy so provided the ascac parameters (ie the assumed spatial distribution of the emission) are not changed the model runs reasonably fast. The recommended mode of operation is to fix the ascac parameters at some value and then fit for the spectral parameters in the annuli. Different values of the ascac parameters can be tried in order to find out how sensitive the spectral parameters are to the assumed spatial distribution.

Important note. If channels are ignored after defining the model then ascac will get confused. To fix this change a parameter in the surface brightness model. This will force a recalculation of the psf effects for the new set of channels.


Keith Arnaud

Xspec Xspec Home Page


HEASARC Home | Observatories | Archive | Calibration | Software | Tools | Students/Teachers/Public

Last modified: Monday, 11-Apr-2005 12:34:41 EDT