For each month all broadband filters (Z,Y,J,H,K) have associated correction tables and overview plots presented. These are based on the most recent calibration that uses individual detector magnitude zero-points (also computed monthly).
The coordinates used in both tables and plots are standard coordinates (units are degrees) with respect to the tangent point of the telescope optical axis for that pointing (see for example Smartt "Spherical Astronomy"), rather than detector pixel coordinates. This was necessary because the science product images and catalogues are a mixture of 1x1, 2x2 or 3x3 interleaving and different dither patterns, and hence varying pixel scales and offsets.
The orientation of the plots is N to the top and E to the left with the detectors from bottom right #1,#2,#3,#4 in a clockwise direction. The ascii tables list the standard coordinates of the centre of the coarse grid used (approximately 1.2 x 1.2 arcmins), the magnitude correction to be applied at the centre of the bin, and the no. of points used. The raw magnitude corrections have been slightly smoothed to reduce the shot-noise and to ensure a smooth correction surface over each detector.
To apply the corrections we suggest using something like bilinear interpolation to compute the offset at an arbitrary detector coordinate. The correction is derived from the "average" difference between the local magnitude zero-point of 2MASS stars and the detector magnitude zero-point and therefore is applied additively.
We caution that since this is derived directly from 2MASS stars ie. relatively bright stellar objects it may not be directly applicable to faint stellar objects or non-stellar sources.
We are conducting further tests to ascertain what causes this "illumination correction" and hence whether it is directly applicable to all types of detected objects. More details on the derivation of these tables can be found here.
All the illumination correction tables are available from here.