next up previous contents
Next: Broad band spectrum using Up: Cookbook Previous: Light-curves   Contents


Spectral extraction of variable sources

A strong limitation of the spiros imaging and spectral mode lies in the underlying assumption that the source flux is constant over the whole observation period. If this assumption is wrong and a relatively bright source does vary significantly, the results obtained in these modes will be biased.

One possible indication that a source flux change with time is when the analysis residuals remain large, even after all cares have been taken in the analysis (see next section). In these cases, we recommend to switch to the spiros TIMING mode and to let some of the suspected sources vary during the observation.

In this section an illustrative example of the data analysis methods to be used in case of variable sources is shown. A further detailed discussion with examples is provided in two separate documents: ``SPI Compact Source Analysis'' and ``SPI Complex Case Analysis'', both available from the Documentation section of the ISDC homepage [11]. We strongly recommend the user to read these documents.

The example presented here is based on observations of the black-hole V404 Cygni of Revolution 1557. The list of ScWs we are now interested is from 15570002 to 15570052.

You should follow the steps described in [*] and [*]. Then, you have to reconstruct the image in the energy range  keV following the procedure described in [*]. The background of these observations has to be modelled with a flat field with ``1'' number of pointings with constant background (due to significant solar activity, see Roques & Jourdain arXiv:1601.05289).

The ``Diff/STD'' values listed in the table ``Contributions to CHI2 parameter by pointing exposure'' at the end of the spi_sa_''UT date''.log log file obtained after running spi_science_analysis are very high ( for several pointings). These high values do not mean that all the pointings must be thrown away, but rather, they indicate that the sky model has not been defined properly. In order to define a robust sky model, the timing step has to be performed before the spectral step.

First, you have to modify the catalogue of sources obtained durimg the imaging step. We already know that V404 Cygni is variable on timescales of  ScW. Moreover, in the field of view there are two bright sources, Cyg X-1 and Cyg X-3, also variable on timescales of  ScW (if these properties are not previously known, they can be inferred from an accurate study of the ligthcurves of the sources of the field of view; see document: ``SPI Complex Case Analysis'', available from the Documentation section of the ISDC homepage [11]).

To inform spiros that the sources have to be considered as ``variable'' and to define the variability timescale, the following parameters contained in source_cat.fits must be set:

In this example, you should set SEL_FLAG=2, VAR_MODL='variable', and the fifth column of VAR_PARS=0 for all sources of the catalogue (i.e. V404 Cygni, Cyg X-1, and Cyg X-3; Fig. [*]), in which the bottom panel shows the expanded field corresponding to VAR_PARS column.

Figure: Relevant paremeters of source_cat.fits for timing analysis.
Image cat_var1


Image cat_var2

When the catalogue is ready, you can launch spi_science_analysis and give source_cat.fits as input catalogue. Then, open the spiros option GUI and set spiros mode to TIMING. Click on the timing button and set mode to WINDOW and the timing scale to 0. Click ``Ok'' and then ``Run''. The ``Diff/STD'' values listed in the table ``Contributions to CHI2 parameter by pointing exposure'' at the end of the spi_sa_''UT date''.log log file are now more reasonable. At this point, you have to remove bad pointings with high ``Diff/STD''. We suggest to remove pointings with very high ``Diff/STD'', e.g. and re-run the SPI analysis. At this first step, pointings 13 16 and 18 21 can be removed. The resulting ``Diff/STD'' values are slightly lower than before. At the second step, you can remove pointings with ``Diff/STD'' (i.e. pointings 2-7, 17, 22-23, 25, in addition to the pointings previously removed). In fact, in fields of view with very bright and variable sources (as in this case) the criterion ``Diff/STD'' can be relaxed (see ``SPI Complex Case Analysis'', available from the Documentation section of the ISDC homepage [11]). After this step you should check again the ``Diff/STD'' values. If they are reasonable (``Diff/STD'' ), you can perform the spectral analysis.

You have to launch spi_science_analysis, click on ``Energy definition button'', set ``Regions energy boundaries:'' 20,400, and ``Number of bins in each region:'' -25. The spectral analysis method presented in Sect. [*] does not take into account the variability of the sources. Therefore, you should click on ``Spiros options'' button, set spiros mode to TIMING, and then click on the timing button and set mode to WINDOW and the timing scale to 0. In addition, ``Bins for src location'' in ``Spiros options'' window (corresponding to the parameter spiros_srclocbins) has to be set to ALL (instead of FIRST), when the sources in the field of view are variable. Click ``Ok'' and then ``Run''.

spiros creates a count spectrum for each source of the catalogue. Since you are interested in V404 Cygni, you can have a look on the structure of the spectrum_Ginga_2023.fits file with fv. It contains, in addition to GROUPING, 35 extensions (spectra), i.e. one for each pointing plus the total spectrum (the last one, extension 36 in this example).

spiros in TIMING mode does not create the response file of the spectrum automatically. It has to be created with the OSA tool spirmf as follows:

spirmf rw-grp-dol="../../ic/spi/rsp/spi_rmf_grp_0004.fits[GROUPING,1,BINTABLE]" \
ebounds-dol="energy_boundaries.fits[SPI.-EBDS-SET]" \
outfile="spectral_response" \
single=yes update=yes clobber=y \
phafile="spectrum_Ginga_2023.fits"

The response file (spectra_response.rmf.fits) has been created. The values of some keywords in the header of spectrum_Ginga_2023.fits have to be modified:

fparkey "25" "spectrum_Ginga_2023.fits[36]" DETCHANS
fparkey "spectral_response.rmf.fits" "spectrum_Ginga_2023.fits[36]" RESPFILE

fchecksum spectrum_Ginga_2023.fits update+ datasum+

where DETCHANS=25 is the number of spectral bins.

Similarly to what was done in Sect. [*], you can fit the  keV spectrum
spectrum_Ginga_2023.fits\[36] with the xspec, for example with the model power*highecut. ``plot lda del'' produces the following spectral display:

Figure: V404 Cygni spectrum from cookbook example.
Image v404_spec_osa11


next up previous contents
Next: Broad band spectrum using Up: Cookbook Previous: Light-curves   Contents
inthelp@sciops.esa.int