Thermal Plasma Continuum-Simulation Example
dph Tue Oct 30 2001
In the ISIS examples, I showed how to "Fit continuum thermal emission model to `line-free' spectral regions", and promised to follow-up with more detailed simulations. Here they are.
I used a 32 ks AR Lac observation as a baseline. I used a 10-temperature component model peaking near 20 MK (logT = 7.3) to approximate the observed spectrum, then used this model to generate a model spectrum and model continuum. Then that model was used to simulate an observation (via "fakeit"; see the Tutorial Example, Generating fake grating data). The same temperature distribution was used to model the continuum.
Finally, I fit the "line-free" regions (as per Thermal Plasma Continuum-Fitting Tutorial), and compare the fit and true continuum to each other, and the simulation, folded model, and fit continuum.
The isis/slang script is Contin_sim.sl.
Some graphic examples follow (full postscript output from the script is here). (Note: the faked data have been smoothed a bit (1-2 pixel sigma) by a gaussian convolution.)
Comparisons of the continuum model and fit.
The continuum model (red) and the single-temperature fit to line-free
regions (green) (both folded through the instrument response). The
bottom panel shows residuals (fit - true). The deviations are about
10% (but less than a count/bin max).
Comparison of the faked data, predicted continuum, and model continuum (all folded through the instrument response)
MEG -1 and +1 regions, showing faked data (line spectrum), the predicted
continuum (green), and the model continuum).
The MEG -1 order O VII triplet region.
The HEG +1 short-wavelength region. The little upward bumps on the continuum are weak-line pseudo-continuum features. Little downward bumps are bad pixels, and the big dip is a chip-gap. The continuum here is very important for putting limits on Fe K features at 1.85A.
The HEG +1 order rgion near the Si XIII triplet.
The HEG +1 order region near the Ne X Ly-alpha resonance line.
Comparison of the faked data, model data, and fit continuum (also folded)
The MEG -1 order, 9A region. Green is the folded model (noiseless),
red the faked data, and white is the continuum fit.
MEG -1 order, 15A region, and detail in the Ne IX triplet reigion.
The MEG -1 region of O VIII 19A, O VII triplet 22A, to N VII 25A. Also, detail in the O VII region.
HEG -1, 12A detail near Ne X + Fe blend.
Conclusions:
Continuum modeling is useful. A 1- or 2-temperature model fit to line free regions does give a good first estimate of the local continuum throughout the spectrum. There are some regions where the apparent continuum is higher than the true continuum, or where noise makes it difficult to place a continuum accurately.
Once an approximate continuum has been used in measurement of line
fluxes, and a better plasma model is derived, that model can be used
iteratively to determine a new continuum, new line fluxes, etc.
{Please send comments to:
dph@space.mit.edu}