A detector map is used to provide a grid of positions over which to calculate and flux weight the ARF. The default detector map is 'flat' which has a weighting of 1.0 for each detector map pixel and is rectangular having sides which just encompass the region specified in the DSS. Only detector map pixels falling within the source region are included in the ARF calculation, e.g. for a circular source box, the pixels in the corners of the detector map will be ignored. The number of detector map pixels is set automatically unless explicitely set with the parameters detxbins and detybins. The calculated number of pixels in each dimension is constrained to be between 5 and 80, and is set proportionately to the total extent of the extraction region. The total extent is taken from the included and excluded regions and for example can be surprisingly large if detected sources have been removed by excluding regions.
Some fine regions, e.g. a narrow annulus at a large off-axis angle, may not be well represented by the default map. In these cases more map pixels can be introduced by specifying e.g. detxbins=100, detybins=100, but beware that the task execution time increases with the number of bins.
arfgen supports 3 types of detector map, selectable through the detmaptype parameter:
This is the default internal representation where all pixel values are set to 1.
Given that all the pixels are of the same value, the function of the flat detector map is essentially to describe the spatial grid points which the task should consider during the computation of the ARF.
The map bounds are by default matched to the extent of the selected regions, but can be set explicitly by setting withdetbounds true and then choosing values for the parameters detxoffset, detyoffset. The map is centred on the source position. The automatically calculated number of pixels can be overridden using the parameters detxbins and detybins.
The calculation of the flux fraction lost to chip gaps and bad pixels is purely geometrical with this map, i.e. it corrresponds to the fraction of the source extraction region which is not exposed.
This map is the default for point sources. The code uses the vignetting explicitely at the position of the point source and does not average over the detector map. The calculation of flux lost to chip gaps is made with the energy-dependent PSF for point sources regardless of which detector map is used.
For extended sources a flat map should be used unless a detector map image is supplied. The code also calculates the fraction of flux
This is an internal map based on a model of the PSF taken from the CAL.
The extent and coarseness of the detector map may be specified on the command line similarly to that of the 'flat' map.
Given that the PSF is energy dependent, the energy at which the PSF model is taken may also be specified by the user, through the parameter psfenergy.
This is a user-defined detector map. This allows the user to have control over the spatial distribution of the source which is particularly useful for extended sources.
This must be in the format of an image array of a dataset, and that image array must contain WCS information that describes the celestial coordinates of the image.
Often the best image to use will be a detector coordinate or x/y coordinate image of the same celestial region taken with another EPIC camera. e.g. If the spectrum has been extracted from EPIC-pn data then an EPIC-MOS1 image will give the spatial distribution of the extended source and will contain data where the PN has chip gaps. Alternately an EPIC-pn image may be the best option for an EPIC-MOS spectrum. Images from other missions may be used and Chandra or ROSAT pointed images covering the whole of the extraction region should give good results. ROSAT all-sky-survey images, however, have a large PSF which will usually not be adequate for representing flux present in fine structure such as chip gaps. To use the input image to correct for chip gaps and bad pixels/columns set the parameter badpixmaptype=dataset.
See the technical note, XMM-SOC-CAL-TN-0027, for an assessment of the accuracy of this method for recovering the intrinsic flux of an extended source.
XMM-Newton images can be generated in xmmselect by selecting the DETX/DETY or X/Y columns in the main GUI dialog of that task, clicking the 'Image' button and then specifying the preferred image binning in the subsequent evselect dialog. Alternatively, it could be generated directly from evselect - section 3.4 provides an example of doing this.
A detector map of this form can be passed to the task through the detmaparray parameter.
Note that in order for arfgen to generate an output consistent with the data, the detector map must enclose the regions over which the spectrum is accumulated; one or more warnings are raised if the extent of the selected regions exceeds the bounds of the detector map along either the X and/or Y axes.
XMM-Newton SOC -- 2021-11-30