This article provides a brief overview of how astrometric and photometric solutions were computed using the Terapix tool scamp. It is a modified version of the document prepared for the T0004 release. Note that CFHT have re-computing their photometric calibrations for all data, so the photometric solutions used in T0005 are not necessarily the same as T0004.
Input images for scamp were selected according to the following criteria:
Date taken: between May 26, 2003 and Feb. 16, 2008.
Terapix quality class : A or B
Exp. time: > 60 sec. No Pre-wide images were included, but short photometric exposures have been used during the astrometric calibration process.
Seeing : < 1.3" , except for u* (<1.4")
Airmass: < 1.7
Skyprobe value: <2.0 Mag. (security limit)
Note that during coaddition with swarp images a further cut by seeing and airmass is applied.
Because of the large numbers of images involved (several thousand per field), different procedures were adopted for each part of the survey field. These are explained below.
In all cases, we use the CFHT's photometric flags to indicate to SCAMP which images are photometric.
In all cases, the astrometric reference catalogue used is 2MASS. 2MASS was found to produce typical residuals of on order of 0.2'' rms absolute, considerably better than achieved with USNO in the preceding release. For the high signal-to-noise sources, the residuals are around 0.05''.
A typical scamp command can be found here.
Note that we use a global '.ahead' file which defines what the relative positioning of all CCD frames. This .ahead is supplied as part of the SCAMP distribution.
Astrometric solutions were computed simultaneously for all images in all filters for each field. I.e, all CFHTLS W1 images were used together, all CFHTLS W2 were used together and so forth.
Note that for the wide data, we do not use the data from the pre-survey fields. Much of the the pre-wide survey data is not photometric and including this dataset produces more problems than it solves.
For all deep images we add four astrometric calibration images around the central stacks (or, in the case of D1, surrounding WIDE survey field images).
For the 25% best-seeing images, we compute the astrometric and photometric solutions simultaneously for all five filters.
For the 85% best-seeing images, we compute astrometric and photometric solutions filter-by-filter (the large number of astrometric "contexts" means that it is not possible to compute a solution for all filters simultaneously due to the large memory requirements).
Comparison with SDSS survey fields
In the fields where we overlap with the SDSS we have compared our photometry to Sloan photometry. These results are presented at this page .
The following table shows an example of the output from SCAMP for the W1 survey field. The other fields are similar. The square box in the centre of the field represents the size of one MEGACAM 0.186'' pixel. The 'internal' solution represents the field-to-field solution, the 'external' solutions show the residual with respect to 2MASS.