The DP0.1 Data Products page quotes the DESC DC2 paper as saying that “each patch is 4100x4100 pixels”. However, after grabbing some patches with the butler, I see that this description doesn’t straightforwardly correspond to what patches actually look like. A typical patch in the middle of a tract takes up 4200x4200 pixels, a patch on an edge of a tract (and not on a corner) has 4100x4200 (or 4200x4100) pixels, and a corner patch has 4100x4100 pixels. Neighboring patches within a tract overlap by 200 pixels.
All of this is more consistent with the following scheme: Each patch has an “interior” region of 4000x4000 pixels. The interior regions of all the patches tile the tract without overlapping. Each patch is additionally surrounded by a “fringe” of 100 extra pixels on every side (except for patches on the edge or corner of a tract, which lack fringes along the tract edges). The fringes extend into the interior regions of neighboring patches, so that neighboring patches within a tract have a total of 200 pixels of overlap.
Is this correct?