The size of the sampling pixel is 7.5/285 degrees on a side. It was chosen because it
The pixels were created in a reasonable manner, but are not exactly square. Please refer to rdpixels.f for the exact details, but the algorithm finds the nearest integer number of pixels of the requested size in a band of declination, and then computes the limits as constant alpha and delta. The zero point for each zone of pixels is offset so that if there were the same number of pixels in adjacent zones, then the centers of one zone would be under/above the edges of the next zone.
The USNO-A catalog is read, and the pixel for each entry is computed. Each new entry is evaluated according to the following rules, which are contained in getqual.f and each.f
The quality is defined by the following algorithm.
After defining q, a penalty is added for a weird color.
These limits are approximately defined by F0 to K5 spectral type.
The distance from the center of the pixel is not quite the geometric one because of the desire to choose a brighter star in dense regions. The algorithm proceeds as follows.
The goal of the algorithm was to find a spatially uniform sample of solar type stars, and to choose brighter ones in crowded regions. I hope that this algorithm is a reasonable approximation. Using this algorithm, 54,787,624 stars were selected, leaving approximately 10% of the pixels empty. The breakdown according to quality is as follows.
Quality Number of entries ----------------------------- 1 34,454,509 2 5,170,963 3 6,014,824 4 2,264,809 5 4,745,419 6 2,137,100The format is the same as the USNO-A catalog.
-Dave Monet (firstname.lastname@example.org)