An On-Line Database of APS POSS Images Astronomical Data Analysis Software and Systems V ASP Conference Series, Vol. 101, 1996 George H. Jacoby and Jeannette Barnes, eds. An On-Line Database of APS POSS Images C. S. Cornuelle, G. Aldering, A. Sourov, P. Thurmes, R. M. Humphreys University of Minnesota Abstract: We are making available our digitized images from the APS scans of the red and blue first-epoch Palomar Observatory Sky Survey (POSS I) photographic plates, forming an on-line image database. Image requests are through the APS homepage at URL http://aps.umn.edu/ and are initially limited in field width and sky region. To create this database, images from each scanned POSS I plate are processed into a set of coordinate-referenced index and pixel files. Database management software then refers to each object's pixels within a requested sky region by plate, by sub-plate, and finally by each scanned image. These images are then assembled into a mosaic FITS format file, whose header contains astrometric and photometric information. While the current APS Catalog of POSS I includes only matched images, the image database will include images above 25 microns in diameter. Thus for a given sky region, users have access to both the object (star and galaxy) catalog data and to the actual pixel data. This project is sponsored by NASA ADP contract NAS5-32670. 1. The Image Database An on-line database of images scanned from both the red (E) and blue (O) plates of the National Geographic-Palomar Observatory Sky Survey (POSS I) is accessible via a WWW browser this November. This forms a superset of the objects listed in the on-line APS Catalog of POSS I (Cornuelle et al. 1994), and thus is a valuable complement to the star and galaxy catalog. Table 1 compares these two resources.   Noteworthy are the differing goals of the two data representations. The on-line object catalog is designed to be a high-reliability resource for objects and their associated image parameters. It is for this reason that it has data only on those images appearing on both the red and blue plates. However the Image Database (IDB) is designed to be a complete record of POSS I images. Thus all discrete images with any chance of corresponding to real objects are included. Another important feature is that the two resources use the same astrometric base (Aldering et al. 1993) and are thus tied together positionally. This means that queries on both the object catalog and the IDB will yield consistent and intercomparable data sets within a given sky region, which should be an aid in source identification and other work. 2. Creating A Database Of Discrete Images The first step is of course the simultaneous scanning and digitizing of the O- and E-emulsion plates using the Automated Plate Scanner (APS). A complete description of the APS scanning procedure is given in Pennington et al. (1993). This has been completed for all POSS I fields with b above 20 degrees from the plane. All above-threshold pixels from both plates for a given field are recorded into images. This method produces definite storage and distribution benefits due to immediate compression and indexing. The threshold is 65% of the background inverse transmission, as determined from interpolation on a 750m grid. These pixel transmission values then have saturation offsets removed and are converted into more photometrically useful pseudo-densities as 0.8 arcsecond pixels. The extent of each image on the plate is stored for later use in speeding the IDB querying. The image database manager is designed to take advantage of the two-dimensional nature of the requested images, as represented by the X and Y extent of each. An above-threshold image is stored in X-order in one of 30 horizontal stripes imposed on the given plate. Its extremal plate X and Y coordinates are also stored in a key file, one per stripe. 3. Handling Image Database Queries Requests for images are managed from an HTML-formatted WWW form page. On retrieval, each database query is converted from RA, declination, and ``box'' width to X and Y ranges on one or more O- or E-emulsion plates. The appropriate key files are interrogated and a list of above-threshold images having any pixels within the request area is generated. The pixels for these images are then retrieved and assembled into a mosaic, using the astrometric plate solution to ensure an accurate final raster, of which Figure 1 is an example.    Figure 1: APS mosaic image. Individual above-threshold images are compiled into an astrometrically-correct mosaic. Pixel size is approximately 1.7 arcseconds. Figure 1: PS 577 Kb Each mosaic raster is trimmed and converted to a FITS file with a header containing useful keywords. Note that for queries including overlaps between adjacent POSS I plates an image is produced for each plate involved. This then enables the user to choose the most suitable rendering of each object near the edges of plates. A low-resolution GIF image preview is then created, and this plus a link(s) to the FITS file(s) are returned to the user's WWW browser as a form-generated page. The image preview is intended as basic confirmation of the user request. Another option is the simultaneous querying of the APS on-line object catalog. The listing of cataloged objects is then made available in the same location as the FITS file(s). 4. Future Directions The APS project will be installing a server CPU and separate storage for the IDB project in order to provide a stable platform large enough to manage the database and handle queries. More POSS I fields will be rapidly processed into the IDB. Currently, astrometry within each image is a simple FITS gnomonic projection. We will make each plate solution (Aldering 1994) available in the header in order to provide precision matching that of the APS catalog, about 0.30 arcseconds per plate. We plan to tie into the IDB the APS stellar photometry (Larsen & Humphreys 1994; Humphreys & Larsen 1995) for each POSS I field, much of which is well-calibrated, and as well each density-to-intensity relation (Odewahn 1995) for galaxy magnitudes. In the interim, users may query the APS catalog over identical sky-region ``boxes'' to determine O and E magnitudes (and classifications) for each IDB object in the returned image which also appears the object catalog. Also planned is a program of galaxy image classification in T-types, which will also be placed on-line. Scanning in pairs of the Luyten and POSS I E-emulsion plates has begun, with the goal of generating a Proper Motion Database. When completed, this will form the third leg of an internally-consistent set of star and galaxy data produced with the APS and available on-line to the community. This project is supported by NASA ADP contract NAS5-32670. References: Aldering, G., Humphreys, R. M., Odewahn, S. C., Cornuelle, C. S., & Thurmes, P. 1993, BAAS, 183, #73.02 Aldering, G. 1994, BAAS, 184, #38.04 Cornuelle, C. S., Thurmes, P., Musukula, S., Humphreys, R. M., Aldering, G., & Odewahn, S. C. 1994, BAAS, 184, #27.03 Humphreys, R. M., Landau, R., Ghigo, F. D., Zumach, W., & LaBonte, A. E. 1991, AJ, 102, 395 Humphreys, R. M., & Larsen, J. A. 1995, AJ, 110, in press Larsen, J. A., & Humphreys, R. M. 1994, ApJ, 436, L149 Odewahn, S. C. 1995, PASP, 107, 770 Pennington, R. L., Humphreys, R. M., Odewahn, S. C., Zumach, W., & Thurmes, P. M. 1993, PASP, 105, 521 Wed Jul 3 07:33:40 MST 1996