XMM-Newton SAS Home Page
XMM-Newton Science Analysis System


evselect (evselect-3.58.5) [xmmsas_20041122_1834-6.1.0]

Spectrum Extraction Extraction Stage (extractor) Light Curve Extraction Home Index

Meta Index / Home Page / Description / Extraction Stage (extractor)

Image Extraction

Image extraction is enabled by setting the name of the output file using the parameter imageset. The name of the columns to use for the $x$ and $y$ axes of the image are set using the parameters xcolumn and ycolumn. The range of the columns used for image extraction is determined using the same algorithm as for the energy column in spectral extraction, and is controlled and determined independently for the $x$ and $y$ columns. The parameters used for the $x$ column are ximagemin and ximagemax, with similarly named parameters for the $y$ column.

There are two options available for binning of images. By default, the binning factor for the image is set such that the image will be of a certain size. The size of the image is controlled by the parameters ximagesize and yimagesize, and is set to $600\times600$ by default. Note that non-integral binning of integer-valued columns is not allowed. Thus when extracting an image from integer-valued columns using ximagesize and yimagesize, the image binning factor will be rounded up to the nearest integer number, so that the actual extracted image may be smaller than what was requested. It is also possible under these circumstances that the $x$ and $y$ bin sizes could be set to different values, even though a square image was requested. In this case a square image can be produced by setting the parameter squarepixels to true. Then the $x$ and $y$ bin sizes will both be set to whatever is the larger of the two values.

Alternatively, one can control the binning factor directly by setting the parameters ximagebinsize and yimagebinsize.

By default, an image is accumulated simply as counts of the number of rows (events) falling within each pixel. However, if the parameter zcolumn is set, then the image is accumulated by summing up the values of the specified column for rows that fall within each pixel.

By default, evselect tries to determine the best data type for the image created, under the assumption that the smallest data type possible should be used. If zcolumn is not set, an integer-valued image will be created. If zcolumn is set, a real-valued image will be created. The default type of image can be over-ridden by setting the parameter imagedatatype.

Any World Coordinate System (WCS) information available for the columns used in image extraction will be propagated to the output image. If no WCS information is available for a column, then the CTYPE keyword for this axis of the image is set to the name of the column, and the other WCS keywords are set to reflect the range and binning of this column, as is done for SPEC* keywords in spectral extraction. If the image extraction columns do have WCS information, then one can request evselect to shift the image so that it is centered on a particular point in the sky. This is activated by setting the coordinate center using the parameters raimagecenter and decimagecenter. Coordinates must be given in decimal degrees.

As well as the standard WCS keywords described above, evselect also writes keywords to the image describing the ``physical'' coordinate system, ie the native coordinate system of the columns used for extracting the image. In order to maintain backward compatibility, there are currently two sets of redundant keywords written for this purpose. The first set uses the LTMn_m and LTVn keywords to specify the transformation, where $n$ and $m$ can take on the values of 1 or 2. The transformation from the physical $(X_{ph}, Y_{ph})$ to the image $(X_{im}, Y_{im})$ coordinates is

\begin{displaymath}
\left( \begin{array}{ccc}
LTM_{11} & LTM_{12} & LTV_1 \\
LT...
...\begin{array}{c}
X_{im} \\
Y_{im} \\
1
\end{array} \right)
.
\end{displaymath}

As there is no rotation in the transformation, $LTM_{12}=LTM_{21}=0$.

The second method uses the multiple WCS conventions to specify the physical coordinate system for the image. In this scheme a second set of WCS keywords is written, except each of the standard keywords has a letter appended to it, which we have chosen to be ``L''. In addition we specify the following keywords to identify this coordinate system:

WCSNAMEL = 'PHYSICAL'
WCSAXESL = 2


Spectrum Extraction Extraction Stage (extractor) Light Curve Extraction Home Index

XMM-Newton SOC/SSC -- 2004-11-22