Noisy Spectrum

[VikashMaan]

Hii, I am working on erosita data release 1 and i came across a few interesting sources which can be new potential compact object systems. (Based on my analysis which includes the cross match between erosita dr1 and gaia dr2 ) .To confirm the nature of the compact object i need to fit the X-ray spectra with some models.I downloaded the source data products from

https://erosita.mpe.mpg.de/dr1/erodat/c ... 4_002_c010

which is one of the sources...The problem is source spectrum data is very noisy even for the energy ramge of 1.2-2.2 keV and hence its of no use to me to conclude something about the nature of this source. Data is still noisy even if I do the grouping of 50 or 200..I am attaching the source grouped spectra from the file
em01_117102_020_SourceSpec_00004_c010.fits . Can you please give me some clarity on what need to be done to get good spectra which can be fitted with some models?

[VikashMaan]

Here is the grouped spectra file (grouping of 50)

[JeremySanders]

You need to look at brighter sources to get better spectra! For very faint sources you will have to make some strong model assumptions, for example where you freeze your parameters or add strong priors. You might only get a flux depending on model.

Take a look at this advice: https://asd.gsfc.nasa.gov/XSPECwiki/low_count_spectra - you probably shouldn't group your spectrum and should use the cstat statistic. There's not a lot you can do with a spectrum with 55 counts (see ML_CTS_1 in the catalogue), however.

You may also want to look at this Bayesian spectrum fitting packagem, BXA: https://johannesbuchner.github.io/BXA/index.html

The scope of fitting faint spectra lies outside eROSITA specifically. There's also an Xspec group on facebook which might be of help, or consult someone doing similar science.

[JeremySanders]

PS In Xspec, I suggest the "setplot rebin" command to visualise spectra with low numbers of counts with low grouping. e.g. "setplot rebin 10 50", to rebin for display to S/N=10 and a max of 50 channels. This doesn't affect the fit, but allows you to visualise the fit.

[VikashMaan]

Thanks for the reply sir