From Megan: For your commentary (sorry to miss this morning's meeting): Using SDSS, GSC2, catalogs and reprocessing Scientific rationale The use of large catalogs derived from large, robustly calibrated surveys may allow an automatic improvement in astrometry. Cross-referencing between survey data and images may allow manual improvement in astrometry. Source identification, including redshift estimates and other parameters, is more likely if one has access to multi-bandpass observations. Auxiliary data such as spectra may already be available for some of the sources. Such cross-referencing expands the utility of the HST data and the catalog/survey data. Astrometry with HST now is accomplished by centering on two guide stars in the FGS pickles, with roll-angle determined by star trackers looking out to either side of the telescope. As a result, roll-angles are usually fairly well determined, with much more uncertainty in the absolute pointing since the position of only 1 of the guide stars is used for the astrometric solution (the other is there to confirm and guide). The distortion of HST images is usually very well known, and the stability of the HST is phenomenal, so repeat observations using the same guide stars are often dead-on. However, absolute astrometry is difficult. Post-observation improvements could be achieved through comparison with the GSC2 or SDSS (and possibly the USNO A2 survey, although that one is not "in-house".) Both the GSC2 and SDSS use the Hipparcos Tycho-2 data for the locations of the brightest stars, so both catalogs are presumably tied to the same standard reference frame (ICRS?) Improved flux calibration from comparisons with Sloan or GSC2 is probably not possible. Both of these surveys are ground-based, and the HST absolute photometry is therefore far more reliable. But if the user is studying the brighter objects in the field, Sloan provides 5 bandpasses of information, GSC 2 bandpasses (with accurate proper motions). Creating an equivalent HTM Tesselation ID for every HST pointing would speed up coordinate-based queries (because the query hits one field rather than two.) The HTM coordinate system is rapidly becoming a standard coordinate for all-sky surveys. Background notes (rough): The Early Data Release (EDR) for Sloan has very small area coverage, about 460 square degrees. (54,000 spectra and almost 14 million objects). WFPC2 pointings search in approximate sky coverage of the EDR (Equatorial fields): 351-360 RA -0.5-0.5 dec 7 pointings 145 exposures 0-59 RA 61 357 145-236 RA 67 332 the other 2 Sloan pointings (much smaller) contain zero WFPC2 observations. Total: 135 pointings, 834 exposures with WFPC2 in the EDR out of a total of 7,266 unique pointings and 46,552 WFPC2 exposures (11/28/2001). The Sloan astrometry for cataloged objects is from USNO where available, Tycho 2 otherwise. Absolute astronometry claimed is <0.1-0.15" but in practice varies from field to field depending mostly on the source photometry. More objects/square arcmin than GSC2 (~8.5/arcmin^2 for the EDR, and recall this is at high Galactic latitude). Sloan images ("corrected frames") have WCS astrometry, good to ~pixel or 0.4" because of the limitations of the WCS. Typical Sloan seeing is 1.5", photometry is in the 3-5% range, magnitude limits are typically 22 mag. Scientific uses of Sloan catalog for HST observations: Source identification and associated data. Photometry of objects at wavelengths other than that observed by HST. Fiber optical spectra available for a subset of objects. Secondary source for astrometry - but has more numerous objects inside a given field of view than do the catalogs with the primary astrometric standards. Photometric calibration - not fundamental, groundbased, more systematics than HST. (Reverse calibration from HST to Sloan much more useful.) Political note: We have no funding to do Sloan development. Existing work is now at maintenance level only and the only additional effort will be to transfer a copy of the database to our local disk and machines (to minimize contact with Fermilab computers.) All of this work would have to be in the context and goal of improving HST data. Guide Star 2.2 435 million stars and non-stars (61% non-stars) 1" resolution scans of DSS at two epochs & 3 bandpasses. Full sky coverage. HTM sky tesselation standards, same as Sloan. magnitude limits ~ 18.5-19.5 bright stars from Tycho-2 catalog (Tycho-2 is a brightest 2.5million star cat.), contains proper motions astrometric errors ~ 0.3" (0.35-0.75 worst case range at the edge of the field) reference catalogs Tycho-2 and "ACT" which is an all-sky survey (maintained by USNO but independent of USNO A2?) photometry 0.2-0.25 mag, but are photographic (may require more serious color terms to compare to HST photometry). astrometry - GSC2 stars are everywhere, on average 3/sq arcmin, but much fewer at high Galactic latitude more at low Galactic latitude. It may be possible to use GSC2 stars in the FOV to check and even improve the astrometry of some HST images. Since field distortion is well-known, a few good reference stars to check the zeropoint of the astrometry may be quite useful. Comparison w/USNO shows relative uncertainties ~ 0.5", but should be on the same standard system, "ICRS". GSC2.2 has a web-based cgi-interface to query by positions. Unique STScI capability: Improvement of WCS header keywords in HST imaging data and the HST catalog may be immediately possible, particularly for WFPC2, ACS, WF3. Catalog-based data products or tools that can produce such products would be very useful to HST observers and archival researchers. I can visualize a search tool that would also enhance Sloan science by allowing "follow-up" HST observations of specific regions of SDSS sky coverage. Such observations would add morphologies, independent photometry, and an additional epoch for variability studies to the SDSS. Drawbacks: For SDSS Early Data Release, relatively few (~2%) of the WFPC2 images would be affected. However, as ACS and WFPC2 images accumulate and the Sloan releases cover much more sky, this drawback will become less of an issue. Perhaps only a few SDSS objects and even fewer GSC2 objects will lie in any given WFPC2 or ACS FOV. However, on average, an extragalactic ACS FOV will contain ~100 SDSS objects. Sloan itself is an extragalactic/Galactic halo survey (high Galactic latitude), so Galaxy observations would be unaffected. (On the other hand, access to all-sky catalogs do not have this limitation.) Implementation Roadmap 1. Figure out which products and services would be scientifically worthwhile and interesting. Prioritize and edit the list of such services by their scientific payoff and by their synergy with other SHARE recommendations. An example of a partial list: PIPELINE SERVICES - a catalog of SDSS objects in the FOV (no cross-correlation). - a resource either linking to products related to those SDSS objects or bundling those products into a single HST product or product bundle (e.g. SDSS spectra FITS files) - population of a keyword that indicates that "yes" this field is included in Sloan coverage. (SLOAN=YES or SLOAN=NO, and update with Sloan releases, or make it release dependent, like SDSS_EDR=YES) (Other partial-sky but large surveys could be included in such a table, like FIRST, 2MASS.) - populate an HTM coordinate keyword for each observation, and store that coordinate in the catalog as well (indexed) to speed up coordinate searches and position-base cross-correlation processes. - cross-correlation services with a pre-requisite of an object catalog with coordinates and fluxes or on-the-fly identification of objects & properties: - automatic selection of cross-correlations between the SDSS or GSC2 catalog and update of the WCS keywords or population of relative shifts and rotation keywords. (You may want to use flux as a rough matching criteria/filter for coordinate standards.) - population of a extended HST/SDSS/GSC2 database table linking astronomical catalog objects to objects in the HST image. SOFTWARE SERVICES AND SOFTWARE-FRIENDLY PRODUCTS/FILES - Make the catalog products produced in the pipeline image-display "friendly": use a format that will allow, for example, the VTT or ds9 or OASIS or Aladdin to display the catalog, make interactive overlays that could do further searches or display a configurable list of catalog fields with a click on the object. (This format might be an XML table or a FITS table or a comma delimited ASCII list -- not defined yet, and it should be configurable anyway, not hardwired.) - Configure/modify the VTT to read pipeline products (requested and even modified by the user) or execute object catalog queries (those object catalogs do not yet exist however) to display object catalog results. - Allow users to do a quick spectral analysis of SDSS spectra (SpecView?) such as wavelength estimates, line IDs, template cross-correlation with standard templates, flux estimations for lines and continuum, line width estimation. (Note that the tool may not need to be invented...) 2. Assess the resources required to produce those products and services, including an assessment of existing tools and resources within and outside the Institute. Existing solutions might be preferred. Also include in the assessment how much coordination between divisions and branches are required to produce these products and services. Some of the tasks are simple but because they require coordination across divisions to pull off, may be in practice hard. Some basic Sloan & GSC2 queries are already available through the web, like a coordinate-based cone search. Something somewhat more sophisticated would be needed to figure out which of the objects in a 40" arcmin radius search actually falls in the WFPC2 FOV, and anything beyond that (figuring out what, if any, fraction of the HST image lies in Sloan coverage) will require additional development resources. The resources needed to do this work are not minor. However, the technical difficulties are not huge. Querying the SDSS and the GSC2 by position, limited by the approximate FOV of the HST imager (approximated by a circle = a cone search) is a solved problem; a web interface or a software program can call the scripts. Cross-correlation of relatively few bright sources in an image is also not a hard problem, and there are WCSTools that might solve some of this very quickly: see Doug Mink's site at http://tdc-www.harvard.edu/software/wcstools/. These tools include already catalog links to specific catalogs such as HST GSC II, 2MASS, USNO-A2. See also "imWCS" for finding stars, cross correlating and updating WCS in the header. Creating an HTM keyword and backpopulating the archive catalogs is very straightforward. Conversion routines already exist, so adding this step to the existing pipeline is not hard, nor is backfilling the database by using data already there (no need to retrieve data). STScI could achieve some improvement of their data services by investigating existing tools and either using them (or refer to them in data documentation) or updating them for use with HST data, if needed. 3. Prioritize these items in the context of existing work on the archive, the pipeline, and in the instrument groups. The effort in doing these items could be stretched out over multiple years. The improvement in the WCS of headers may push some of this effort to the front of the queue, if comparison with external catalogs is deemed useful (and I can't imagine why it wouldn't be.) But tasks like creating an HTM keyword are only difficult because they require coordinating the attention of multiple groups: the pipeline developers, the instrument groups, and the archive. The biggest pay-off with Sloan comes in a year, ~Jan 2003, with the next release of Sloan data.