Estimate Source Counts in an Event File

A Simple Example

Display the file:

unix% ds9 acis_1838_evt2.fits &

and create regions by left-clicking on the image. To designated the annulus as a Background region double click on it and select Property → Background; it will then be drawn with a dashed line. Then use the the Get Info... option in the Region menu to find out the dimensions of the regions. More information on creating and modifying regions is given in the Using CIAO Region Files thread.

This example uses a single source region circle(4060.8724,4239.33,20) and a background region annulus(4060.8724,4239.33,86,114), as displayed in Figure 1.

Figure 1: Source and background regions

[Version: full-size]

Figure 1: Source and background regions

The source region is defined as a single circle: circle(4060.8724,4239.33,20); the background region is an annulus: annulus(4060.8724,4239.33,86,114).

To extract counts in the source region and compute net counts:

unix% punlearn dmextract
unix% pset dmextract infile="acis_1838_evt2.fits[bin sky=circle(4060.8724,4239.33,20)]"
unix% pset dmextract outfile=1838_simple.fits
unix% pset dmextract bkg="acis_1838_evt2.fits[bin sky=annulus(4060.8724,4239.33,86,114)]"
unix% pset dmextract opt=generic
unix% dmextract
Input event file  (acis_1838_evt2.fits[bin sky=circle(4060.8724,4239.33,20)]): 
Enter output file name (1838_simple.fits): 

The contents of the parameter file may be checked using plist dmextract.

The counts and net_counts may be found by using dmlist:

unix% dmlist "1838_simple.fits[HISTOGRAM][cols counts,net_counts]" data

--------------------------------------------------------------------------------
Data for Table Block HISTOGRAM
--------------------------------------------------------------------------------
 
ROW    COUNTS               NET_COUNTS
 
     1               7792.0      7746.2857142857 

CIAO Analysis menu in ds9

The net counts can also be obtained directly in ds9 using the CIAO analysis menu in ds9, aka dax.

From ds9 select the top-level Analysis menu and then CIAO. This is the top level menu of all the CIAO tasks that are available in ds9.

To get the net counts, simply go to Statistics and then Net Counts.

ds9 will then run dmextract with the data that are loaded and with the regions displayed. The results are displayed in a new text-window as shown in Figure 2.

Figure 2: dax Net Counts

Figure 2: dax Net Counts

Output from the Analysis → CIAO → Statistics → Net Counts task.

It is important that the background annulus be designated as background so that the tools can compute the proper net counts.

Regions vs. Stacks of Regions

Region descriptors may also be input via files, rather than typed on the command line:

unix% cat source.reg 
circle(4060.8724,4239.33,20)

unix% cat background.reg 
annulus(4060.8724,4239.33,86,114)

unix% dmextract infile="acis_1838_evt2.fits[bin sky=region(source.reg)]" outfile=1838_simple_2.fits \
      bkg="acis_1838_evt2.fits[bin sky=region(background.reg)]" opt=generic

However, if you want to extract counts from a number of source regions contained in a single file, then you must input the region file as a stack. If you have two region files:

unix% more stack.reg
circle(4060.8724,4239.33,20)
circle(4244,4094,6)

unix% more stackbgd.reg
annulus(4072.96,4248.00,40,60)
annulus(4244,4094,10,30)

then to compute net counts in each region separately:

unix% dmextract infile="acis_1838_evt2.fits[bin sky=@stack.reg]" outfile=1838_stack.fits \
      bkg="acis_1838_evt2.fits[bin sky=@stackbgd.reg]" opt=generic

Examine the output with dmlist, as before:

unix% dmlist "1838_stack.fits[cols counts,area,bg_counts,bg_area,net_counts,net_err]" data

--------------------------------------------------------------------------------
Data for Table Block HISTOGRAM
--------------------------------------------------------------------------------
 
ROW    COUNTS               AREA                 BG_COUNTS            BG_AREA              NET_COUNTS           NET_ERR
 
     1               7792.0      1256.6370614359               5057.0      6283.1853071796              6780.60        89.4107376102
     2                  3.0       113.0973355292                119.0      2513.2741228718              -2.3550         1.8002708130

If you use the sky=region(stack.reg) syntax:

unix% dmextract infile="acis_1838_evt2.fits[bin sky=region(stack.reg)]" outfile=1838_region.fits \
      bkg="acis_1838_evt2.fits[bin sky=region(stackbgd.reg)]" opt=generic

dmextract will interpret the list of regions as a single, connected region. Using this syntax with verbose > 0 will print a warning:

# dmextract (CIAO): dsDMEXTRACTREGCOMPWERR -- WARNING:Region #1 
contains more than 1 component. Only the first component will be 
described in the region columns of the output file.

This returns a single row with the sum of the counts in all the individual regions:

unix% dmlist "1838_region.fits[cols counts,area,bg_counts,bg_area,net_counts,net_err]" data

--------------------------------------------------------------------------------
Data for Table Block HISTOGRAM
--------------------------------------------------------------------------------
 
ROW    COUNTS               AREA                 BG_COUNTS            BG_AREA              NET_COUNTS           NET_ERR
 
     1               7795.0      1369.7343969651               5176.0      8796.4594300514      6989.0228571429        88.9972030746

Exposure Corrections

Exposure maps may be applied to both source and background regions. In this case, in the calculation of the net counts from the source, the background counts are normalized not only by the ratio of the areas of the source and background regions, but also by the ratio of the mean exposures in the source and background regions (be sure to read the caveat at the end of this section). The following threads give complete instructions on generating exposure maps:

• ACIS Exposure Map for a Single Chip
• ACIS Exposure Map for Multiple Chips
• HRC-I Exposure Map
• HRC-S Exposure Map

Both normalized [cm²*counts/photon] and unnormalized [cm²*sec*counts/photon] exposure maps may be used as input to dmextract. Bin the data to a FITS image and generate exposure maps that are congruent to that image. To isolate the S3 chip in this dataset:

unix% get_sky_limits flux_2.3_bin1.expmap verb=0

unix% pget get_sky_limits dmfilter
X=3638.5:4748.5:#1110,Y=3844.5:4953.5:#1109

unix% dmcopy "acis_1838_evt2.fits[bin x=3638.5:4748.5:#1110,y=3844.5:4953.5:#1109]" 1838_s3.fits

Then run dmextract and examine the results:

unix% dmextract infile="1838_s3.fits[bin sky=@stack.reg]" outfile=1838_stackexp.fits \
      bkg="1838_s3.fits[bin sky=@stackbgd.reg]" exp=flux_2.3_bin1.expmap bkgexp=flux_2.3_bin1.expmap opt=generic

unix% dmlist "1838_stackexp.fits[cols counts,area,exposure,bg_counts,bg_area,net_counts]" data
 
--------------------------------------------------------------------------------
Data for Table Block HISTOGRAM
--------------------------------------------------------------------------------
 
ROW  COUNTS   AREA    EXPOSURE        BG_COUNTS   BG_AREA   NET_COUNTS          
 
    1               7739.0               1252.0      7851.3069951925               5201.0               6272.0      6700.7901785714
    2                 86.0                113.0      7851.3069951925                 33.0               2516.0        84.5178855326

The NET_COUNTS are calculated as the counts in the source region [COUNTS] minus the counts in the background region [BG_COUNTS] (appropriately normalized by the areas [AREA/BG_AREA] and the mean exposure maps in source and background regions) and divided by the mean exposure map in the source regions.

Caveat on Exposure Corrections

Normalizing counts by mean exposure in regions may lead to errors if there are large exposure variations in the region which are not accompanied by similar variations in counts. Consider a bright point source at the center of a large region whose exposure varies strongly near the boundaries. That loss of exposure is not reflected in the counts, which are concentrated near the point source, but would strongly affect the mean exposure. In cases such as this, it is better to flat field the image by the exposure map and than extract flat-fielded counts in the region. A variance map should also be computed and used to calculate errors in the region.