Introduction

This guide discusses the basic differences between CIAO's srcflux tool and ACIS Extract. It is not meant to be an exhaustive comparison of all the capabilities, algorithms, and outputs of each tool, but rather is meant to be a quick guide for acis_extract users who want to use srcflux.

Both tools produce estimates of a source's count rate and estimates the flux with uncertainties. As part of this processing they produce several data products for which users may find additional uses. Under the hood, acis_extract uses the same CIAO tools as srcflux along with a few tools from HEASOFT/FTOOLs package. Therefore it is unsurprising that the results from the two tools are consistent.

Relevant threads:

• Calculate source count rates and fluxes
• Calculate source count rates and fluxes for combined datasets

Summary

Here is a quick summary of the key features of both srcflux and acis_extract.

1. srcflux does not provide this type of analysis. However, it has been upgraded to allow users to run arbitrary custom analysis scripts, "plugins", and include the results in the output *.flux file. Users cat write and run plugins to fit spectra and compute new source positions or any additional analysis they wish to perform (compute hardness ratios, radial profiles, period folding, etc.)

These are just the key features of both tools. They each produce additional outputs that are considered to be diagnostic or informative rather than essential outputs.

Basic differences

$ ds9 acis_evt2.fits -region out_broad.flux
  -or-
$ ds9 acis_evt2.fits -catalog import fits out_broad.flux

$ pset srcflux bands="0.5:2.0:1.49,2.0:8.0:3.0,0.5:8.0:2.3"

Matching data products

Below is a table showing the mapping of the important data products created by both tools, with some notes about key differences.

1. The srcflux files are prefixed with ${outroot}_${id}_${band}_, where ${outroot} is the user supplied output filename root, ${id} is the source number going from 1 to the number of sources (same as the COMPONENT value in the output .flux file), and ${band} is the band name or energy limits if not one of the CSC named bands.
2. All acis_extract files are stored in a sub directory ${coordinates}/${obsid}, where ${coordinates} is the sexagesimal representation of the input RA and Dec and ${obsid} is the OBS_ID of the observation.
3. srcflux creates a single exposure map that encloses both the source and background regions, whereas acis_extract creates individual cutouts from the input exposure map around the source and background regions separately.
4. srcflux creates the exposuremap at the characteristic energy for each energy band, whereas acis_extract uses the single exposure map provided by the user for all bands.
5. srcflux uses the CIAO definition of BACKSCAL: which is the ratio of the geometric area of the extraction region to the total field area. acis_extract uses a different definition of BACKSCAL related to the sum of the exposure map pixels in the extraction region.
6. The acis_extract extract.reg file has both the source polygon and the background circle
7. By default acis_extract does not produce background response files (ARF and RMF). srcflux does created these files by default, but they can also be skipped.
8. srcflux uses mkwarf to create weighted response file (even for on-axis sources). acis_extract uses mkarf to create point-like responses (even for off-axis sources)
9. srcflux adjusts the effective area for the fraction of the PSF in the source region using the arfcorr tool, which uses the REEF circular approximations to the PSF. acis_extract simulates the PSF at multiple energies and corrects the effective area by interpolating the PSF fraction it derives at those energies.
10. The acis_extract light curve is for the total energy band. The srcflux light curves are for each individual energy band.
11. The acis_extract light curve contains the total source region count rate. Each srcflux light curve has separate source, background, and net count rates.
12. The acis_extract lightcurve also contains grouping information (ie to produce bins with a minimum number of counts). It also contains the median energy of the events during the time bin.
13. srcflux also creates the adaptively binned lightcurve created using the Gregory-Loredo algorithm to check for variability and reports a variability probability. acis_extract computes a variability probability using a KS test.

There are also other additional diagnostic and intermediate files created by each tool.

Matching source properties

Both tools produce source properties for a each individual observation, and for the merged/combined dataset.

Comparing Results

We now show some comparisons of running srcflux and acis_extract.

For srcflux we used psfmethod=marx and set the confidence limit to 68% (conf=0.68). To save time we also skipped creating background response files, ie bkgresp=no.

For acis_extract we followed the thread shown in section 3, Getting Started, of the acis_extract users guide, with the additional step of running ae_flatten_collation to provide the simplified final output xray_properties.fits file.

The best value to compare is the photon flux values (photon/cm^2/sec) since it is has all the corrections for exposure and aperture area/PSF fraction.

$ srcflux acisf04395_repro_evt2.fits pos=csc.fits bkgresp=no psfmethod=marx conf=0.68 clob+

For a second comparison we show the results comparing srcflux and acis_extract for a merged dataset. For this example we chose the 3 observations of the Open Cluster Collinder 197

The data were reprocessed with chandra_repro, and combined with merge_obs. We then ran wavdetct on the combined image and detect 258 sources. When ran srcflux and acis_extract with the same setup and parameters as for the single observation test.

Below is a plot comparing the broad band photon flux values

As before, the X-coordinates are the srcflux photon flux estimates and the Y-coordinates are the acis_extract photon flux estimates. The srcflux values are the MERGED_NET_PHOTFLUX_APER values taken from the combined .flux file, the acis_extract values are the PhotonFlux_t values from the collated xray_properties.fits file. The diagonal line is 1:1 (two values are equal). The points are color coded by the srcflux TOTAL_COUNTS, and the number labels are the srcflux source ID number.

There are a few outliers that are worth examining, although the discrepancies are not as significant as in the single observation example.

Region 53

Region 53 is one of the brightest sources. The srcflux photon flux estimate is somewhat higher than the acis_extract estimate. The per-observation apertures are shown below

The images are for the observations in left-to-right order of observation date: OBSID 22689, 25099, 24757. The srcflux circles are in yellow and the acis_extract polygon are green. The regions are nearly identical for this nearly on-axis source. However, when we look at the fluxes for each individual observation

the source is variable. In this case a single combined flux estimate is not meaningful. The difference then may arise in how variable sources are treated. Users would need to be cautious when publishing this source's properties.

Region 215

The acis_extract photon flux estimate for region 215 is higher than the srcflux estimate (located near the center of plot). Again we start by examining the source regions

The per-observation images are in left-to-right date order. The yellow regions are the srcflux circles and the green polygons are the acis_extract apertures. The source is located at the edge of the field of view for obsid 24757 (the line segment in the right image is part of the FOV boundary). In this case, acis_extract did not produce any results for this observation (not even a region) whereas srcflux used the available one count to produce an upper limit. The upper limit in this case is slightly pulling down the overall mean value from the other two which is the reason for the difference between the acis_extract and srcflux results.

Conclusions

This guide provides information that users familiar with acis_extract may find helpful when using the srcflux tool. It is far from an exhaustive comparison of the features of each tool, but instead focuses on the results obtained with the default settings. Both tools produce a cache of useful data products for individual sources in separate observations as well as combined/merged products for sets of observations. While there are subtle differences in how the two tools work, the end results are shown to be in excellent agreement. The outliers have been shown to be a consequence of the different default aperture algorithms or due to exceptional characteristics of the data.