The Variance Maps have patterns.

We investigated the variance maps for different stages/processing:

• chip (md)
• warp (md)
• stack (md and sas)

There are periodic patterns associated with warp and stack weight maps, which are not seen on the chip weight maps.

Here is a zoom of one of the unconv.wt.fits for SAS footprint: (stack_Id= 336346). You can see an odd periodic background. The yellow circles are the false detections.

Here is a zoom of one of the unconv.wt.fit stacks for a recent MD refstack, it not quite as strange, but still, there is an underlying odd, periodic background.

Chip stage weight maps do not have this background (click to zoom in, this is for chip_id 310753 and XY53)

The full chip:

The weight map of one of the cells in XY53

The detrended fits of that same cell in XY53

Warp stage (for the same chip) has a very strange background for the weight map (click to zoom in)

One of the skycells (the weight map, same exposure, same cell, but warp stage)

A zoom of the weight map for that cell - the background has a periodic pattern

And one of the warps (same cell, it looks okay)

We think something weird is happening with the variance/weight maps in the warp stage - this might cause the funny false detections in the SAS.footprints, and likely causes problems elsewhere...