Last modified: December 2022

URL: https://cxc.cfa.harvard.edu/ciao/ahelp/dmextract.html
AHELP for CIAO 4.15

dmextract

Context: Tools::Core

Synopsis

Make a histogram table file (e.g. PHA file, lightcurve file) from a table column. Generate count histogram on supplied regions for a spatial table or image file.

Syntax

dmextract  infile outfile [bkg] [error] [bkgerror] [bkgnorm] [exp]
[bkgexp] [sys_err] [opt] [defaults] [wmap] [clobber] [verbose]

Description

`dmextract' creates a histogram from a column of data in a table. Both "scalar" (PHA, TIME, etc.) and "vector" (DET, SKY, etc.) columns are supported. dmextract thus includes the capability to create PHA files, lightcurves, and spatial radial profiles.

In the case of scalar columns, a simple linear binning of the data is available, while for vector columns ('spatial extraction') the user can define the bins via a set of regions. Support for background correction is available for both scalar and vector columns. Vector column extraction also supports exposure corrections and images as input.

1. Histogram of a Scalar Column

`dmextract' takes as input a table or set of tables, often an event list. It generates an `extraction', a histogram of the data binned on any one column in the table. The resulting file is also a table, containing the histogram and associated errors and rates. For a scalar column, the extraction is defined using the CXC Data Model (DM) binning syntax (see "ahelp dm") in the input file string (see examples).

1.1 PHA File Extraction

The default mode of dmextract ("opt=pha1") is to generate a HEASARC/OGIP-compatible Type I PHA file. For X-ray event data, the PHA (pulse height) is an instrumental quantity related to the photon energy; the mapping from PHA to energy varies with detector position and sometimes time, and we also use the PI (pulse invariant) value, which is the PHA corrected to a standard energy scale. The PHA files are used in combination with response matrices (RMF files), which are made for either PHA or PI binning. See the threads on spectral fitting for advice on how to use PHA files.

1.2 TYPE 2 PHA File Extraction

Using "opt=pha2" makes a Type II PHA file, in which each line of the file contains a complete spectrum with errors. This option is used when the input is a stack of virtual files. One line of the output file is created for each input extraction, so the infile in this case will usually be a stack. For columns other than pha and pi, opt=generic2 makes a similar file without PHA specific keywords.

1.3 Lightcurve Extraction

Using "opt=ltc1" is appropriate when creating a lightcurve by binning on the TIME or EXPNO columns. Normally, COUNT_RATE is taken to be counts per bin unit per total observational good time. Choosing opt=ltc1 ensures that COUNT_RATE is counts per total time associated with the bin, rather than total time associated with the observation.

If one is binning on either TIME or EXPNO, the exposure per bin is properly calculated based upon the selection criteria. For example, if one has an observation with the nominal 3.24104 sec frame time and 3.2 sec live time, and then chooses to bin in units of 10 exposure numbers, the count rate per bin will be the total counts per bin divided by 32 seconds.

If one is binning on the TIME column, the exposure is further merged with the intersecting good time intervals, taken from the first GTI in the case of multiple GTIs (e.g., for regions that span multiple ACIS chips).

Choosing "opt=ltc2" is the lightcurve equivalent of "opt=pha2" or "opt=generic2" in that a list of input files can be extracted into multiple rows of a single Type II FITS file. The individual lightcurves, however, must be binned on the same grid.

The bkg parameter definition (below) has information on creating background-subtracted lightcurves.

1.4 Generic Histogram Extraction

Using "opt=generic" is appropriate when binning on a column other than PHA, PI or TIME. This generates a file which is similar to a Type I PHA file, but doesn't contain keywords specific to PHA data. For instance, you can bin an aspect solution file on RA to see the histogram of the RA distribution.

1.5 Binning Specification

The complete binning specification is "[bin col=min:max:step]". For instance,

% dmextract "evt1.fits[bin pha=1:2048:2]" out.pha

You can leave out any of the min,max,step, for instance

[bin pha=:2048], [bin pha=::2], [bin pha=5::]

Default values will be filled in from the maximum valid range of the column (in the case of a FITS file, the values of the TLMINn, TLMAXn columns) and the default binning (the CXC specific keyword TDBINn, if present). Using an open-ended range can lead to unexpected results at the binning endpoints, such as zero-count bins and zero exposure values. If the allowed range is very large (e.g. a range covering all possible long-integer values), it may cause the tool to fail due to lack of machine resources.

If all values are omitted, e.g. [bin pha], and the optional "defaults" parameter is supplied, dmextract will look in the defaults file for a suitable binning for that column.

All types of output files will have the columns CHANNEL, COUNTS, and COUNT_RATE. Type II output files will also have the columns SPEC_NUM and TOTCTS.

1.6 Weighted Binning Specification

Users can specify a column to use to weight the counts by using this syntax: "[bin col=min:max:step;weight]". For instance,

% eff2evt evt1.fits fluxed_evt1.fits
% dmextract "fluxed_evt1.fits[bin energy=500:7000:0.0146;flux]"
fluxed_spectrum

All of the same columns are created as before but rather than than the "COUNTS" being the number of counts, it is the sum of the weight column values for all the rows that fall in each histogram bin.

1.7 WMAP in Histogram Files

It can be useful to encode extra information about the extraction in the output file. For the ASCA mission, the idea of a WMAP (weight map) was introduced: associated with the extraction table is an image of the extraction region. Specifically, the PHA FITS file has in its primary header an image containing a low resolution map of the source in detector coordinates. This allows downstream software to determine the appropriate weighting for calibrations which depend on detector position (for instance, effective areas may depend on the off-axis angle). The dmextract wmap parameter allows the user to define such a wmap; a standard example involves adding an energy filter to the binning command:

wmap="[energy=500:2000][bin det=8]"

Using this filter, the wmap will better represent the event distribution from 500-2000 eV. The wmap option is currently only useful when making a PHA file, although in principle it could be used by software making exposure calculations for other quantities.

1.8 Systematic Errors

`dmextract' optionally writes a SYS_ERR keyword to the file header, using the value of the sys_err parameter provided by the user. The SYS_ERR keyword denotes the fractional systematic error associated with the data; thus "sys_err=0.05" indicates a 5 percent error. This error may be added in quadrature by downstream software to the statistical errors provided in the extraction.

1.9 The Defaults File

The dmextract defaults file is compatible with the XSELECT mission database (mdb) file. It is a text file containing lines of the form telescop:instrume:key value where telescop and instrume are the values of the FITS TELESCOP and INSTRUME keywords, and dmextract recognizes keys of the form colname_binning. Example:

Chandra:ACIS:pha_binning 1:4096:2

For example, to add a default for HRC time binning of 20s bins, one might add

Chandra:HRC:time_binning  ::20.0

Error conditions:

2. Extraction on a Vector Column (Spatial Extraction)

The second method of extraction for `dmextract', binning on a 2-dimensional quantity like 'sky(x,y)' or 'det(detx,dety)' can take an event or image file and a set of regions to extract and sum the counts in each of those regions. The user can supply a background file or regions to remove from the source to create a total count rate.

2.1 Spatial Extraction Syntax

There are several options for the extraction syntax. Using

filename[bin sky=shape(x0,y0,...)]

indicates a count extraction on the sky column with a single bin specified by the shape (for example circle(4096,4096,20)). The user may supply a stack of regions, with each region a separate line in an ASCII file region.lis:

filename[bin sky=@region.lis]

or, for the special case of the ANNULUS shape, can define a set of nested annuli with

filename[bin sky=annulus(x0,y0,rmin:rmax:step)]

NOTE: this syntax will only work in a command line, not in a region file.

Note that in no cases are default binning parameters allowed for vector column extraction (e.g., filename[bin sky] or filename[bin det] are not allowed and will result in a fatal error). A valid region specifier must be included (e.g. sky=circle...).

2.2 Background and Exposure Files

The source extraction will count up the number of events within each region. Errors will be calculated on the counts based on either Gehrels or gaussian option. The count rate is determined by using the exposure time for the file. If a user supplies a background file with regions, or a fixed background (counts/pixel/s), this value will be used to perform a background correction on the source counts. The information on the background counts will be included in the output as well as net counts and net rates. Background information can be normalized. The default is to normalize the counts with respect to the relative exposure times and geometric areas in the input and background files.

The user can optionally supply an exposure map that corresponds to the input file. The regions chosen in the input will be extracted and a weighted exposure correction over that region will be applied to the counts. The background file can also have an exposure correction applied to it. An EXPOSURE (and BG_EXPOSURE if relevant) column will be added to the output table.

2.3 Error Calculation

dmextract assumes symmetric errors in all cases, which bound a "1-sigma" or 68.23% confidence interval. The tool offers three methods for error calculation (see parameter "error").

If counts in each bin are sufficiently large (say, N>~20-30), Gaussian statistics are appropriate,with the 1-sigma error given by

error = sqrt(N)

For low counts per bin, the Gehrels approximation (Gehrels, 1986, ApJ, 303, 336) to confidence limits for a Poisson distribution are useful. In this case, the 1-sigma error is approximated by

error = 1 + sqrt(N + 0.75)

To be precise, the above error corresponds to the 84.13% upper confidence limit for a Poisson distribution. The corresponding lower limit (error=sqrt(N-0.25)) is somewhat smaller, and dmextract adopts the larger error to be conservative. It should also be noted that for very low counts per bin, the initial assumption of symmetric errors is inadequate, even with the Gehrels approximation. Users should consider rebinning to avoid such cases. At the minimum, they should treat errors calculated in such cases with caution.

Finally, users may provide their own error estimates in the form of a variance image computed at the same pixel scale as the input file. The region from the input file is extracted from the variance image, and the sum of the variances in the region is used for error calculations.


Examples

Example 1

% dmextract "in.fits[sky=annulus(4095.0,4096.0,60.0,100.0)][bin
pi=::100]" out.pha

Perform a scalar extraction on the single input file in.fits, binning on column PI from channels TLMIN to TLMAX (both read from input.fits) with a step size of 100, and output to a type I PHA output file out.fits. Include only counts within the specified sky-coordinate annulus.

Example 2

% dmextract "in.fits[sky=circle(4095.0,4096.0,43.0)][bin
pha=1:3000:10]" out.pha

Perform a scalar extraction on the single input file in.fits, taking photons from a circle around a particular source position specified in sky pixel coordinates, and binning on column PHA from channels 1 to 3000 with a step size of 10, and output to a type I PHA output file out.fits.

Example 3

% dmextract "in.fits[sky=annulus(4095.0,4096.0,60.0,100.0)][bin
pi=::100]" \
out.pha wmap="det=8" clobber=yes

As with example 2, but create a WMAP block in the PHA file. The WMAP is created by binning the detector coordinates by a factor of 8 of those photons within the specified sky-coordinate annulus.

As clobber is set to yes the tool will over-write any existing output file - in this case out.pha - rather than exiting with an error.

Example 4

% dmextract "in.fits[sky=annulus(4095.0,4096.0,60.0,100.0)][bin
pi=::100]" \
out.pha wmap="[energy=500:2000][bin det=8]"

As with example 3, but only use those photons with energies between 500 and 2000 eV when creating the WMAP.

Example 5

% dmextract "evt.fits[bin pi]" "pha.txt[opt kernel=text/dtf]" type=pha1

Create a Type 1 PHA file in DTF text format from a FITS file, propagating the full header; see "ahelp dmascii" for more information on working with ASCII files.

Example 6

% dmextract @inlist @outlist opt=generic

Perform extractions on the input files in1.fits and in2.fits listed within the ASCII file inlist, binning on column DETX from channels 1000 to 1500 with a step size of 10, and output to corresponding type I output files out1.fits and out2.fits listed within the file outlist. The contents of the two "stack" files are:

unix% cat inlist
in1.fits[bin detx=1000:1500:10]
in2.fits[bin detx=1000:1500:10]

and

unix% cat outlist
out1.fits
out2.fits

Example 7

% dmextract @inlist out.fits opt=generic2

Identical to the previous example except output to the single type II file out.fits.

Example 8

% punlearn dmextract
% dmextract "evt.fits[bin sky=circle(4095.0,4096.0,43.0)]" \
out.fits error=gehrels

Performs a single source count (vector) extraction in a circle centered at 4095,4096 with a radius of 43. There is no background correction specified (the default value for bkg is empty) and the errors reported are calculated with the Gehrels approximation. Note that because sky is a vector (2D) column, using the syntax "bin sky" means this is a count extraction. Since only one region is specified, the output table will have a single row, the counts within that region.

Example 9

% dmextract "evt.fits[bin sky=annulus(4095.0,4096.0,0.0:100.0:10)]" \
out.fits error=gehrels clobber=yes

This uses a set of annuli centered at 4095,4096 with radii 0, 10, 20, 30,.. 100, and so results in a radial intensity profile table.

Example 10

% punlearn dmextract
% dmextract "evt.fits[bin sky=circle(4095.0,4096.0,43.0)]" out.fits \
bkg="evt.fits[bin sky=circle(4000.0,4096.0,40)]" \
error=gaussian bkgerror=gaussian

Perform a single source count extraction on the input file evtt.fits, counting photons from a circle around a particular source position specified in sky pixel coordinates, and background subtracting the normalized counts from the background region. Gaussian errors are reported for the counts and rates.

Example 11

% dmextract "evt.fits[bin sky=@region.lis]" out.fits exp=expmap.fits

If region.lis is a file that contains 1 region per line, e.g.:

circle(4233.5,3753.5,28)
circle(4233.5,3753.5,18)
circle(4233.5,3753.5,10)
circle(4233.5,3753.5,5)

then dmextract will perform a spatial source count extraction on the input file into bins specified by this list. The exposure map is opened and a weighted exposure correction is found for each region and applied in the calculation of the net counts.

Example 12

% dmextract "evt.fits[bin sky=@region.lis]" out.fits error=variance.fits

Performs a spatial source count extraction on the input file. Errors for the counting are determined by the variance image "variance.fits". For each region, the variance in the total counts extracted is the sum of the variances in the corresponding pixels in the variance map. The number of output bins is equal to the number of regions in region.lis.

Example 13

% punlearn dmextract
% dmextract "evt.fits[sky=circle(4114,4173,4)][bin time=::100]" \
out.fits opt=ltc1

Extracts data from a circular region, and then creates a lightcurve spanning the entire observation that is binned on a 100 second time scale.

Example 14

% dmextract @input.list @output.list opt=ltc1

with input.list being an ascii file with the lines:

acis_evt2.fits[sky=region(ps.reg)][bin time=::1000.]
acis_evt2.fits[sky=region(back.reg)][bin time=::5.]

and output.list being an ascii file with the lines:

ps_output.fits
back_output.fits

Extracts lightcurves from two separate regions defined by the region files ps.reg and back.reg (for example, for a point source and background region), and places the output lightcurves into Type I FITS files called ps_output.fits and back_output.fits, respectively. The first lightcurve uses 1000 second bins, while the second uses 5 second bins.

Example 15

% dmextract @input.list output.fits opt=ltc2

with input.list being an ascii file with the lines:

acis_evt2.fits[sky=region(ps.reg)][bin time=::1000.]
acis_evt2.fits[sky=region(back.reg)][bin time=::1000.]

Extracts lightcurves from two separate regions defined by the region files ps.reg and back.reg (for example, for a point source and background region), bins the data into 1000 second time bins, and places the two lightcurves into separate rows of a Type II FITS file. The lightcurves have the same binning, but one is not subtracted from the other; they remain as individual lightcurves.

Example 16

% dmextract "evt2.fits[sky=region(ps.reg)][bin time=::200]" \
bkg="evt2.fits[sky=region(back.reg)]" \
outfile=ps_bsub.fits error=gaussian bkgerror=gehrels opt=ltc1

Extracts a lightcurve from a region defined by the region file ps.reg, bins it into 200 second time bins, and applies gaussian errors. A background lightcurve is also generated using the region file back.reg, and errors using the Gehrels approximation are calculated. Note that binning criteria are not input for the background, as dmextract uses the same criteria as for the infile.

The ps.reg lightcurve and background lightcurve use the same GTI information, as they are extracted from the same event file. If the regions are taken from different chips (for example, the ps.reg lightcurve is extracted from the backside illuminated S3 chip, which is serving as the aimpoint chip, while the background lightcurve is extracted from the other backside illuminated chip, S1), then a CCD_ID filter should first be applied to the event file being used for extraction of the background lightcurve. The background lightcurve is then subtracted from the ps.reg lightcurve, and the errors from the two lightcurves are combined. The results are placed in the Type I FITS file ps_sub.fits.

Example 17

% dmextract "evt2.fits[sky=region(ps.reg)][bin time=::0.2]" \
bad_idea.fits opt=ltc1

Extracts a lightcurve from a region defined by the region file ps.reg, and bins it into 0.2 second bins. The output is placed in the FITS file bad_idea.fits. If evt2.fits is not a data file taken in CC-mode, but rather is a Timed Event (TE) mode file, 0.2 seconds is shorter than the frame exposure time. In this case, dmextract will generate a warning that the lightcurve is being extracted with bin times shorter than the exposure time.

Example 18

% dmextract infile="hrc.fits[sky=region(regsrc.reg)][bin time=::50]" \
bkg="hrc.fits[sky=region(regbkg.reg)]" exp=dtf1.fits bkgexp=')exp' \
outfile=hrc_lc.fits opt=ltc1

Extracts an HRC lightcurve from a region defined by the region file regsrc.reg, and bins it into 50 second bins. Since the exp and bkgexp parameters are set, a weighted exposure correction will be applied to the counts. EXPOSURE and BG_EXPOSURE columns will be added to the output file.


Parameters

name type ftype def min max reqd stacks
infile file input       yes yes
outfile file output       yes yes
bkg file input       no yes
error string   gaussian     no  
bkgerror string   gaussian     no  
bkgnorm real   1.0     no  
exp file input       no yes
bkgexp file input       no yes
sys_err real   0        
opt string   pha1        
defaults file ARD {$ASCDS_CALIB}/cxo.mdb        
wmap string            
clobber boolean   no        
verbose integer   0 0 5    

Detailed Parameter Descriptions

Parameter=infile (file required filetype=input stacks=yes)

The input virtual file or stack, e.g. event list, modified by a dmextract binning command.

Any table or stack of tables is valid input, with the

table[bin scalar_col=min:max:step]

type of extraction or the

table[bin vector_col=region_list]

type of extraction.

Images can also be input with the latter type of extraction:

image[bin vector_axis=region_list]

There are three methods of stack inputs allowed.

PHA/Histogram extraction TYPE I files:

For each file in the input, there should be a file specified in the outfile stack.

PHA/Histogram extraction TYPE II files:

There can be multiple input files, but only one output file. This will create a type II file as described above.

Radial Profile/Source Extraction:

For each file in the input, there should be a file specified in the outfile stack. However, if only 1 output file is specified, dmextract will place all output information in the single file with the header information for the first input file.

Parameter=outfile (file required filetype=output stacks=yes)

The output histogram file.

The output file. This is a histogram of counts and rates for each bin (a bin is a PHA channel in the case of a pha file, histogram bin in the case of a generic scalar extraction, or a region in the case of a vector, spatial, extraction).

Please see the infile parameter for a detailed description of outfile stacks.

Parameter=bkg (file not required filetype=input default= stacks=yes)

Background file with regions or a numeric value.

This is a string that is one of:

This parameter may also be used to create a background-subtracted lightcurve. Not only is background area correctly accounted for, but also different GTIs are used for the source and background datasets (though it may be necessary to add a ccd_id filter to get the correct background GTI, if the background is taken from a different chip). The lightcurve is corrected bin-by-bin for the background rate during that time interval; the background binning is forced to be the same as the infile binning. Columns are created in the output file for NET_COUNTS, NET_RATE, and NET_ERR. The latter combines the statistical error associated with both the source and background lightcurves.

If instead of supplying a file name you supply a constant value in units of counts/sec, that number is used as the constant background rate for the lightcurve.

Parameter=error (string not required default=gaussian)

The error method for determining error on the counted items.

These are the methods for error determination. The current options are "gaussian", "gehrels", or a variance image. A variance image is not yet supported in histogram or pha binning, instead gaussian will be used.

Parameter=bkgerror (string not required default=gaussian)

The error method for the background error on the background count.

These are the methods for background error determination. The current options are "gaussian", "gehrels", or a variance image. A variance image is not yet supported in histogram or pha binning, instead gaussian will be used.

Parameter=bkgnorm (real not required default=1.0)

Background normalization for spatial count extraction.

This fudge factor allows the user to adjust the background normalization relative to the source normalization in the calculation of net counts, for a spatial extraction. When the factor is 1.0, the background counts are normalized by the ratio of exposure times and spatial areas. If bkgnorm is different from 1, the normalized background counts are further multiplied by the value of bkgnorm.

Note that the value of bkgnorm is not saved as a keyword in the output FITS file header; however, it is recorded as part of the HISTORY entry in the file header.

Parameter=exp (file not required filetype=input default= stacks=yes)

Exposure map for the input file for spatial extraction.

The exposure map that corresponds to the input file will have the information in the input regions extracted and the weighted exposure correction over this region is applied to the counts. If a stack of exposure maps is supplied, there must be the same number of files as the infile stack.

The units of the exposure map are either cm**2 sec, cm**2 (normalized by time), or dimensionless. For Chandra exposure maps generated with the mkinstmap and mkexpmap tools, the source and background exposure maps must BOTH be either unnormalized (units of cm**2-sec) or normalized (units of cm**2).

Exposure map and background exposure map units must be compatible. Otherwise, an error will be reported and the calculation will continue without the exposure correction.

In lightcurve mode, the dither_region FRACAREA correction can be input to dmextract to correct for the area off the chip. It is input in the same manner as the HRC dtf_stat file:

% pset dmextract exp="dr_out.fits[cols time,dtf=fracarea]"

Parameter=bkgexp (file not required filetype=input default= stacks=yes)

Exposure map for the background file for spatial extraction.

The exposure map that corresponds to the background file will have the information in the input regions extracted and the weighted exposure correction over this region is applied to the counts. If a stack of background exposure maps is supplied, there must be the same number of files as the bkg stack.

The units of the exposure map are either cm**2 sec, cm**2 (normalized by time), or dimensionless. For Chandra exposure maps generated with mkinstmap and mkexpmap, the source and background exposure maps must BOTH be either unnormalized (units of cm**2-sec) or normalized (units of cm**2).

Exposure map and background exposure map units must be compatible. Otherwise, an error will be reported and the calculation will continue without the exposure correction.

If the source and background regions are from the same ObsID, the background exposure map may be set as:

% pset dmextract bkgexp=")exp"

Parameter=sys_err (real default=0)

Fixed systematic error

Parameter=opt (string default=pha1)

The output file type format: one of pha1, pha2, generic, generic2, ltc1 or ltc2.

Each of the following options is discussed in detail in the DESCRIPTION section (above).

There are three main kinds of output generated by dmextract. The 'generic' output is a histogram table with each row corresponding to a different bin. The 'generic2' output is a set of histograms, one per row, with the histograms stored in array columns.

In ltc modes, a lightcurve will be generated. Binning should be on the 'TIME' column to be sure to get the correct GTIs. Only data from the first GTI will be used if multiple GTIs are present, e.g. in ACIS datasets. Other "time" quantitities, e.g., TIME_RO (for "Continuous Clocking" mode files where event times have been corrected) or PHASE (e.g., if such a column has been created with dmtcalc) will not automatically be treated like the 'TIME' column, even when using the ltc modes.

Most users will want 'opt=generic' for most applications. However, when doing a spectral extraction on PHA or PI, you need to use 'opt=pha1' or 'opt=pha2' to generate a compliant PHA type I or II file for use with Sherpa, ISIS, or XSPEC.

Parameter=defaults (file filetype=ARD default={$ASCDS_CALIB}/cxo.mdb)

The mission database file

The mission database file gives default binning instructions for certain extractions; for instance, it knows that PI extractions on ACIS files should use the binning 1:1024:1, instead of the default binning 0:1024:1 that dmextract would assume from the TLMIN/TLMAX values in the file header if no mission database file were specified.

The MDB file is just a text file, and you can generate your own MDB file to define your preferred defaults for various extractions.

Parameter=wmap (string default=)

The filtering/binning to use to make a WMAP

If you specify a wmap binning with "infile=inval wmap=wmapval", dmextract will make an image in the output file header which will be equivalent to the image generated by 'dmcopy "inval[bin wmapval]" wmapfile' (note that any binning specified in inval - such as '[bin pi]' - will be removed when creating the WMAP image). If you specify "wmap=default", a value from the mission database file specified by the "defaults" parameter will be used. For Chandra ($ASCDS_CALIB/cxo.mdb) this is equivalent to 'wmap="chip=32"'. One can also apply filtering commands in the creation of a wmap, which are appended to those applied to the 'infile'. This requires a slightly more complete syntax, which is identical to the standard datamodel filtering syntax, e.g:

wmap="[energy=500:2000][bin det=8]"

This filter will create a wmap which reflects the 500-2000 eV emission of the region used to create the PHA file.

Parameter=clobber (boolean default=no)

Specifies if an existing output file should be overwritten.

Parameter=verbose (integer default=0 min=0 max=5)

Specifies the level of verbosity (0-5) in displaying diagnostic messages.


Output Columns: radial profiles

When you calculate region counts, or radial profile, using dmextract with a "[bin sky=...]" directive and an exposure map, there are a lot of output columns, many of which are simply related to each other. We can regard the basic output values as:

Value Definition
EXPOSURE The effective exposure time, corrected for deadtime but not for spatial sensitivity variations
AREA Source region area in pixel units
RMID Average radius value in pixel units
MEAN_SRC_EXP Mean effective area in cm**2. Same as the mean value for a *normalized* exposure map in source region.
COUNTS Total source counts
ERR_COUNTS Error on COUNTS; sqrt-N or 1+sqrt(N+3/4)
BG_AREA Background region area in pixel units
MEAN_BG_EXP Mean background effective area in cm**2. Same as the mean value for a *normalized* exposure map in bg region.
BG_COUNTS Total background region counts
BG_ERR Error on BG_COUNTS; sqrt-N or 1+sqrt(N+3/4)

From these we can derive the other output values. (Here we assume that the exposure times for background and source are the same, and there is no extra background correction factor. If this is not true, simply modify BG_AREA to compensate.)

COUNT_RATE = COUNTS / EXPOSURE
BG_RATE = BG_COUNTS / EXPOSURE

NET_COUNTS = COUNTS - (AREA / BG_AREA) * BG_COUNTS
NET_ERR = sqrt(ERR_COUNTS^2 + ((AREA/BG_AREA) * BG_ERR)^2)
NET_RATE = NET_COUNTS / EXPOSURE
ERR_RATE = NET_ERR / EXPOSURE

FLUX = COUNTS / (EXPOSURE * MEAN_SRC_EXP)
NET_FLUX = FLUX - (AREA / BG_AREA) * BG_COUNTS / (EXPOSURE * MEAN_BG_EXP)
NET_FLUX_ERR = sqrt((ERR_COUNTS / MEAN_SRC_EXP)^2 +
	       ((AREA / BG_AREA) * (BG_ERR / MEAN_BG_EXP))^2) / EXPOSURE

SUR_BRI = NET_COUNTS / AREA
SUR_BRI_ERR = NET_ERR / AREA
BG_SUR_BRI = BG_COUNTS / BG_AREA
BG_SUR_BRI_ERR = BG_ERR / BG_AREA

SUR_FLUX = NET_FLUX / AREA
SUR_FLUX_ERR = NET_FLUX_ERR / AREA

RMID = (R[0]+R[1])/2.0

Additional columns

The following additional columns are provided, as transforms applied to other columns. These columns convert values in pixels to ones in arcsec (if the appropriate WCS information is provided in the input file):

Column Based on
CEL_R R
CEL_RMID RMID
CEL_AREA AREA
BG_CEL_BRI BG_SUR_BRI
BG_CEL_BRI_ERR BG_SUR_BRI_ERR
CEL_BRI SUR_BRI
CEL_BRI_ERR SUR_BRI_ERR
CEL_FLUX SUR_FLUX
CEL_FLUX_ERR SUR_FLUX_ERR

For instance, the following will display the inner and outer radii in arcsec and the surface brigntness in count/square arcsec (use the "cols" option of dmlist to find out more information on these columns):

unix% dmlist "prof.fits[cols cel_r,cel_bri]" data,clean

Output Columns: lightcurves

When you calculate a lightcurve with opt="ltc1" or "ltc2", the basic output values are listed below (the background-related quantities are only created when the bkg parameter is set):

Value Definition
TIME_BIN Time of the center of the frame
TIME_MIN Minimum time value in bin
TIME Time of the center of the frame
TIME_MAX Maximum time value in bin
COUNTS Total source counts
STAT_ERR Error on COUNTS; sqrt-N or 1+sqrt(N+3/4)
AREA Source region area in pixel units
EXPOSURE The effective exposure time in the source area
COUNT_RATE Count rate
COUNT_RATE_ERR Error on the count rate
BG_AREA Background region area in pixel units
BG_EXPOSURE The effective exposure time in the background area
BG_COUNTS Total background region counts
BG_ERR Error on BG_COUNTS; sqrt-N or 1+sqrt(N+3/4)
BG_RATE Background count rate
NORM_BG_COUNTS Background counts in the source area
NORM_BG_ERR Error on NORM_BG_COUNTS
NET_COUNTS Net Counts
NET_ERR Error on NET_COUNTS
NET_RATE Net count rate
ERR_RATE Error on NET_RATE

The net counts, rates, and associated errors are derived as:

COUNT_RATE = COUNTS / EXPOSURE
COUNT_RATE_ERR = STAT_ERR/EXPOSURE

BG_RATE = BG_COUNTS / EXPOSURE

NET_COUNTS = COUNTS - [(BG_COUNTS/BG_AREA/BG_EXPOSURE)*(SRC_AREA)*(EXPOSURE)] 
NET_ERR = sqrt(ERR_COUNTS^2 + ((AREA/BG_AREA) * BG_ERR)^2)
NET_RATE = NET_COUNTS / EXPOSURE
ERR_RATE = NET_ERR / EXPOSURE

Additional columns

The following additional columns are provided, as transforms applied to other columns. These columns provide a time column where 0 refers to the center of the first bin.

Column Based on
DT_MIN TIME_MIN
DT TIME
DT_MAX TIME_MAX

For instance, the following will print the light curve using the DT rather than TIME column:

unix% dmlist "lc.fits[cols dt,count_rate]" data,clean

Exposure Time for Multi-chip Observations

Setting the exposure time to the value of the EXPOSURE keyword in the header to calculate "RATE" and "FLUX" columns may lead to erroneous results in spatial extraction with multi-chip images and exposure maps, if the exposure times of the various chips differ significantly. If unnormalized exposure maps are used, the "FLUX" columns will be correct, but the "RATE" and "MEAN_SRC_EXP" columns may not be correct if they refer to regions on chips other than the one used to set the EXPOSURE keyword, and the exposure on that chip differs from that used to set the EXPOSURE keyword. For normalized exposure maps, the "MEAN_SRC_EXP" column will be correct, but the "RATE" and "FLUX" columns may not. A simple, albeit tedious, work-around is to generate image and exposure maps separately for each chip, so that the EXPOSURE keyword accurately reflects the exposure time for that chip.

The BACKSCAL value for Chandra ACIS data

As of CIAO 4.5, the $ASCDS_CALIB/cxo.mdb file used by dmextract to define "standard" settings in a manner analagous to the XSELECT MDB file, contains a fixed field size for ACIS data of 8192 square pixels. Other detectors - such as HRC - still use the minimum and maximum values recorded for the X and Y columns in the data subspace of the file. This change was made to allow spectra extracted from reprojected data - which may have a different range for X and Y due to the reprojection - and still have consistent area estimates.

The BACKSCAL keyword value is calculated to be the ratio of the extraction area to the field area, so for ACIS data it is

source area / (8192*8192)

whereas for other instruments it is

source area / (rx*ry)

where rx and ry are the ranges of the X and Y columns in the events file, that is the TLMAX-TLMIN values for each column.

Changes in CIAO 4.11

The RMID column

When used to create a radial profile, the output now contains the RMID column, which gives the mid-point of each annulus in pixels. If the input data also has a celestial coordinate system then a "virtual" column called CEL_RMID is also added, with the mid-point given in the appropriate units.

Changes in CIAO 4.15


Bugs

dmextract requires a "sky" vector to bin on x and y

If the x and y columns are not associated as a "sky" vector, dmextract can't bin on them:

unix% dmextract infile="img.fits[bin (X,Y)=@ann.reg]" ...

# dmextract (CIAO): dsDMEXTRACTREGWCSWERR -- WARNING: Expected 2 dimensions for the sky descriptor. Skipping translation of region to WCS. 
or
# dmextract (CIAO): dsFINDCOLUMNERR -- ERROR: Column 'sky' was not found
in file 'HDU2'.

Workaround:

Edit the file header to create a sky vector.

unix% cat dmhedit.lis
#add
MTYPE1 = POS
MFORM1 = X,Y

unix% punlearn dmhedit
unix% dmhedit img.fits dmhedit.lis
Problem with pgrid() stack syntax

There is a problem using the pgrid() stack syntax when making radial profiles.

unix% dmextract "image.fits[sky=pgrid(4096,4096,0:1000:100,30:60:30)]" sector.fits opt=generic

While the tool runs to completion, the output is not always correct.

Binning on celestial coordinates

Trying to bin on celestial coordinates can yield incorrect results

% dmextract "events.fits[bin cel=circle(150.57256,2.4995918,0.09')]" outfile=/tmp/test.fits op=generic

which may produce an incorrect radius column.

Users should only bin in physical or logical coodinates.

Caveats

ACIS: Multiple GTIs

When extracting a lightcurve using opt=ltc1|ltc2, dmextract uses the information from the input file's GTI. ACIS data contains a separate GTI for each active CCD (up to 6); however, they are typically nearly identical except for a small time offset due to the how the CCDs are read-out. dmextract uses the last GTI it finds in the file. Since the GTIs are nearly identical (especially compared to the time resolution of the lightcurve) the difference in GTIs is usually irrelevant.

However, there are some ACIS observations where the GTIs are very different for each CCD. These are mostly due to very bright sources imaged with or with out the gratings that cause telemetry to saturate and results in dropped exposures. Since exposures from individual chips are dropped separately, the GTIs can be very different for different CCDs. For example OBS_ID=3505 has very different GTIs for each CCD due to a large number of dropped exposures.

% dmlist 3505/secondary/acisf03505_001N003_flt1.fits.gz blocks
 
--------------------------------------------------------------------------------
Dataset: 3505/secondary/acisf03505_001N003_flt1.fits.gz
--------------------------------------------------------------------------------
 
     Block Name                          Type         Dimensions
--------------------------------------------------------------------------------
Block    1: PRIMARY                        Null        
Block    2: FILTER                         Table         0 cols x 0        rows
Block    3: GTI7                           Table         2 cols x 1        rows
Block    4: GTI4                           Table         2 cols x 3566     rows
Block    5: GTI5                           Table         2 cols x 3981     rows
Block    6: GTI6                           Table         2 cols x 2826     rows
Block    7: GTI8                           Table         2 cols x 2748     rows
Block    8: GTI9                           Table         2 cols x 3998     rows

This may lead to an error in the lightcurve exposure values and thus in the count_rate if the user specifies an extraction region that is on a different CCD or one that spans multiple chips.

Workaround:

Users can specify a ccd_id filter to select a single CCD when they run dmextract. This will essentially select the correct GTI. Users who need to compute a lightcurve for multiple CCDs need to run dmextract separately for each CCD and then combine them together.

See Also

calibration
ardlib
concept
subspace
dm
dm, dmascii, dmbinning, dmfiltering, dmopt
psf
psf
tools
acis_bkgrnd_lookup, acis_fef_lookup, acis_set_ardlib, addresp, apply_fov_limits, aprates, asphist, combine_grating_spectra, combine_spectra, dither_region, dmarfadd, dmcoords, dmcopy, dmextract, dmfilth, dmgroup, dmimghist, dmlist, dmregrid, dmstat, eff2evt, fullgarf, get_fov_limits, get_sky_limits, hrc_bkgrnd_lookup, make_instmap_weights, mean_energy_map, mkacisrmf, mkarf, mkexpmap, mkgarf, mkgrmf, mkinstmap, mkpsfmap, mkrmf, mkwarf, psextract, psf_project_ray, readout_bkg, rmfimg, sky2tdet, specextract, tgextract, tgextract2