skycells needs to allow for overlapping edges.

[kpat.jhu]

A buffer of 15 pixels on all edges of the sky cells from the stacks are masked. I found this first in MD08, a quick check by Mark Huber and I of MD09 and MD07 showed the same effect. The anomaly is illustrated on the PS1 wiki at

​http://ps1sc.ifa.hawaii.edu/PS1wiki/index.php/Stack_Anomalies#Edges_of_sky_cells_are_masked

The effect was noticed when examining a galaxy at the edge of a sky cell. Examination of the RA,DEC values of the pixels indicates that the sky cells meet seemlessly - little or no gap nor overlap in the vertical edges of the sky cells. However, with this 15 pixels masking of each sky cell edge we lose a strip 6 arcsec wide (2 * 15 * 0.2 arcsec/pixel) of un-imaged sky. It would be good to turn off this unnecessary masking.

In a reply to my original post, Paul-Price noted that some masking is needed for convolution purposes (presumably to remove convolution edge effects). That may be, but it would then seem that a better way to proceed is to make the sky cells bigger and then trim unnecessary masked pixels before making the final product.

Note - I am not familiar with IPP so my pick for "component" is a guess, and I have not picked keywords.

-- Gerhardt Meurer

[Paul Price]

This is due to a current limitation in the 'skycells' command within Ohana.