Digitized Sky Survey - Data Organization

The DSS data files are organized as follows :

There are several ASCII Plate List files which contain limited metadata for each of the plates in a set of grouped survey plates. The primary information are the plate identifiers and the nominal Right Ascension and Declination of the plate center. The applications which retrieve DSS images use this information to narrow the plate selection down to only those which may cover the requested region. The separate Plate Lists files are used to separate the various surveys e.g. Red, Blue or Near-IR.

Each survey plate has a detailed FITS-like header which contains all the metadata for the plate and in particular the astrometric calibrations. Once the application has a small list of possible plates, it computes the exact location of the requested image on each of the possible plate and will choose the best plate to extract the image from.

Each survey plate scan was either 14000x14000 (1.7" pixels) or 23040x23040 (1.0" pixels) in size. Before compression using the Htransform technique each plate was subdived into a 30x30 grid of non-overlapping tiles which were individually compressed. Whilst this required a lot of cpu to compress the data, it makes decompression much faster - the application only needs to decompress the handful of individual tiles required to reconstruct an image at the requested location.

All of these data files are on-line at the STScI and image retrieval requests are handled by the MAST webserver which runs the GETIMAGE program to read these datafiles.

~8000 Original plate scans are ~8TB, ~4x10^12 pixels, compressed to ~1TB