A set of questions that have arisen from the analysis of the MDS 07 and 08 images at Durham.

Image examples related to the questions raised here can be found at this wiki page

These questions are preliminary and will be updated as we learn more.

1) The width of the tracks removed by Magic are often extremely wide (10s of pixels). For example, in the 19 warped frames for skycell 077 in MD08 (g-band), between ~20-50% of pixel data in frames are masked out. What is the prospect of getting Magic working within the specification of only a few percent masked pixels? Note that bad read-out regions also result in large areas of individual frames being masked out.

2) We have measured the PSF of the stacked images (g-band MD08) to have a width of up-to and over 2 arcsecs whereas the data distribution table lists the image quality as 1.2 arcsecs. Why is it so bad (especially as the stack is made from only the 50% best data) and why is the quoted 1.2 arcsec wrong?

3) The stacks have very variable sky levels. This appears to come from the fact that each warp from which the stack is made has a slowly varying sky level. As masking means that different regions of the stack are partial sums of different subsets of the warped frames this leads to patchy sky in the stack with sharp discontinuties. This will prevent detection of faint extended sources. Are there plans to do a local sky subtraction in the warps or otherwise greatly improve the sky subtraction.

4) The noise in the sky background in the stacks is mottled, ie correlated over a scale comparable to the extended sources we want to detect. This is possibly an inevitable consquence of convolution to a very broad 2 arc sec PSF that is done to each warp prior to stacking, but looks far worse than those with experience of stacking expected.

5) In good seeing (~1.2 arc sec PSF) warps the images of stars are clearly not round. Some are triangular (!) others look trailed (but not coherent across whole frame and so not trailing). Is this poor focus, or an issue with read-out? What are the prospects for improvement?

6) Many reasonably bright (but unlikely to be saturated) galaxies (17 mag) have cross shaped masks applied to their centres. It looks like they've been flagged as bright stars and an erroneous attempt has been made to mask out diffraction spikes.

7) At least some of the z-band warped images appear striped, which we think is an artifact of the correlated read noise that Ken mentioned. Is that the case and if so why is it apparent in long exposure MDS frames, especially in z where the sky brightness is high?