On this page you will find an analysis of the November 2009 re-release of the MD08 data taken in June 2009. The example difference image shown is of Skycell 102 and was created by subtracting a convolved stack image from a target image (the unconvolved stack is also shown for completeness). Sources identified in the difference image by the IPP are marked with circles: GREEN CIRCLES indicate sources where no bad-flag is set, RED CIRCLES indicate sources where a bad-flag is set. These circles are also superimposed on the stacked and warp images for comparison's sake.

(Note that in this particular skycell, only one bad flag was raised - PM_SOURCE_MODE_MOMENTS_FAILURE.)

A comparison of a detail taken from the full difference image (white rectangle) with the corresponding detail taken from the stacked images reveals that bright sources have been heavily over-subtracted and are found as negative sources in the difference image. Any fresh transients located in the target images will therefore appear as negative sources in the difference images and the IPP will fail to detect them.

Full images of MD08 Skycell 102. From left to right : WARP, UNCONV STACK, CONV STACK, DIFFERENCE - click for full resolution

Detail of the MD08 Skycell 102 image (white rectangle shown in full images). From left to right : , WARP, UNCONV STACK, CONV STACK, DIFFERENCE - click for full resolution

Problems and questions

1. The subtraction isn't working well yet. Input warp is 1.4" and the stack has been convolved to 2.0". Is the problem related to the mismatch between the two ? Can a better result be obtained with ppSub with two matched warps ? We can check this with Paul this week.

2. As Eric discussed at last DRAVG, our first aim is to verify that the pixels in the image are reasonable. So top priority is solving the convolution and kernal matching. But after, we need to look at the IPP photometry - clearly too many spurious sources which do not have a "bad flag" set (the Green sources).