---

---

Spectral Analysis

Combining Imaging or Grating Spectra

The general dilemma for spectra!

In general merged observations should not be used for spectral analysis, i.e. to extract merged spectra and responses. The merged event list does not contain soffucient information to generate correct response files. The recommended technique for the spectral analysis case is to generate separate PHA, RMF, and ARF files for each source or observation and to analyze them simultaneously.

The Caveat: Extracting Spectra from Merged Datasets explains in detail the problems related to extracting spectra from merged observations and in which cases "reasonable" results can be obtained.

---

Combining spectra

If combining spectra is useful for scientific analysis (for example to identify faint features or combine diffraction orders) or for illustrative purposes, it can be easily achieved with current CIAO scripts. This is possible both in the case of spectra of the same source extracted from different observations or spectra from different sources (i.e. when performing "stacking analysis").

The scripts also combine the corresponding responses allowing users to perform follow-up analysis using modeling and fitting packages. Scripts exist for both imaging spectra obtained with the ACIS instrument or the zeroth-order of a grating observation, or grating spectra obtained with the LETG or HETG gratings.

The following threads illustrate the use of the relevant scripts.

1. Imaging spectra

   Coadding Spectra and Responses (main script combine_spectra: how to combine multiple imaging source spectra, and the associated background spectra and response files.

   Note the following CAVEATS relative to background spectra:

   • Combining background spectra with wildly varying spectral extraction region areas may yield misleading uncertainty estimates; i.e., some extractions will be over-represented while others will be under-represented.
   • If the background rates contributing to a source are significantly different in the individual spectra to be combined, it is recommended that these spectra remain separate and be modeled simultaneously—otherwise, the modeling results of the combined source spectrum could be biased towards the observation(s) with the highest background rate(s).
   • Combining spectra is mathematically trivial and the issue really has to do with combining the responses. Since the response can change over time, with temperature, and with the location on the detector, you will need to examine how the shape of the combined response results are compared to the individual response results; significant distortion of the combined response can lead to questionable results.
2. Grating spectra

   Two main operations are generally performed on grating spectra: combining various diffraction orders from a single observation and/or combining spectra from multiple observations.

   The following threads illustrate both usage of the same script:

   • Extract Coadded and Grouped Nth-Order Source & Background Spectra and ARFs (main script combine_grating_spectra: how to create a co-added positive and negative order grating spectrum and responses for ACIS/HETG, ACIS/LETG, or HRC/LETG data.
   • Add Grating Spectra and Average ARFs (main script combine_grating_spectra: how to combine an arbitrary number of grating spectra and create averaged response files (ARF and RMF).