|
|
Spectral Fitting of Ginga Data
The process by which spectral information can be extracted from
Ginga LAC data is complicated. This is due to both the
modification of the intrinsic source spectrum by low-energy
absorption, and the complexity of the LAC response (i.e. how
the energy resolution (18% at 6 keV, scaling as E-½ throughout the full energy range), gas
and window efficiencies affect the energy and intensity of the
incident X-rays). The LAC response can be described by a matrix, each
row of which corresponds to the response of a monoenergetic X-ray
beam, as a function of pulse height channel.
The background subtracted pulse height data are then a convolution
of the modified source spectrum with the LAC response. However,
restoration of the incident spectrum from the observed data, by
multiplication of the inverse of the LAC response matrix, would yield
a noise-dominated solution. Instead, a trial source spectrum, or
model, is convolved with the LAC response, which can then be compared
to the real data. The goodness of the fit is determined using a ![[chi^2]](chi2.gif) test (i.e.
Bevington & Robinson 1992), and the spectral fitting process
iterates until a minimum in is reached. For weak sources it might be necessary to
rebin the pulse height channels at high energies. Note that the first
three pulse height channels are ignored, as the data in them is of
doubtful quality.
The rest of this section is intended to give the user enough
information to determine the Galactic hydrogen column density in a
specified direction of the sky, to correct spectral data for
fluctuations in the CXB, to convert HDS spectral files to a FITS
format suitable for the xspec spectral fitting package (which
is part of the XANADU suite of software), and to analysed HDS spectral
files using the LDAS program spf.
[Next: Galactic Column]
[Previous: HK Data]
[Back to Contents]
|
