next up previous contents
Next: Preparing for XSPEC, XRONOS Up: ROSAT Xselect Guide Previous: Timing Analysis   Contents


Spectral Analysis

Extracting a Spectrum

When all desired filters are in place, then you are ready to extract a spectrum. This spectrum should be created from the PI column in the event file (the keyword for pha should be set to PI, if this is not the case, 'set phaname PI') the option for the data to be regrouped is offered when the spectrum is saved (grppha will allow you to reset the grouping at a later stage, so answering yes to the grouping is not going to constrain your spectral analysis).

xselect> extract spectrum

xselect> save spectrum source.sp

When using a RDF PSPC dataset you will see that the temporary spectral file made by xselect is 500 channels, and can be examined using 'plot spectrum'. As noted in section 3.1, when you save an RDF spectrum it is truncated to 256 channels which can be fit in xspec.

When extracting an HRI spectrum, you will extract the full 16 channels spectrum, and you will be offered a default rebinning down to a single channel, on saving this spectrum. It is recommended that this rebinning be used, if the user wishes to fit the HRI data using the standard single channel HRI matrix (seee section 6.1 for details on response matrices available from legacy).

The saved spectrum can be read into xspec directly. You will see that the spectral file contains a small map image in the primary array, showing where on the detector the events in the PHA or PI file originated. This weighted map is known as the WMAP, and is used by pcarf for generating an accurate ancilliary file which is used in conjunction with the photon redistribution matrix in the spectral fitting process.

Spectral filters

A variety of filtering commands are available as qualifiers for a spectral extraction. These are

xselect> filter time

For example using

xselect> filter time file good_times.xsl

allows a spectrum to be extracted using the 'good time intervals' previously defined and saved to a file. The command

xselect> filter region

is almost always required in analysis of imaging data, to define the region for which your spectrum is being extracted. Region files can be defined using an saoimage display of the field of interest, as detailed in Chapter 3. For example

xselect> filter region binary_bgd.reg

will use the region file binary_bgd.reg as a spatial filter for subsequent extracted spectra.

xselect> filter phase

allows a phase to be defined, and thus good time intervals are calculated and used as a time filter for the extracted spectrum. Similarly,

xselect> filter intensity

allows the user to extract a spectrum over a user-defined count rate


xsel:ROSAT-PSPC > filter intensity
> Enter ranges for intensity filtering (i.e. .01-.02,1-5) >[] 10-15
Making GTI to implement intensity selection with Boolean expression:
 * RATE>=  0.1000E+02&&RATE<=  0.1500E+02

These filters can be cleared in the usual way, e.g.

xselect> clear time good_times.xsl

Extracted spectra can be model-fit in xspec, or manipulated (e.g. to make ratios of high versus low state spectra ) using mathpha (Chapter 7).

next up previous contents
Next: Preparing for XSPEC, XRONOS Up: ROSAT Xselect Guide Previous: Timing Analysis   Contents
Michael Arida 2001-09-20