IPAC 2MASS Working Group Meeting #110 Minutes

IPAC 2MASS Working Group Meeting #110 Minutes 1/07/97

Attendees: R. Beck, T. Chester, R. Cutri, T. Evans, J. Fowler, L. Fullmer, T. Jarrett, G. Kopan, B. Light, H. McCallon, S. Wheelock, J. White


  1. Status Reports


  1. Status Reports

    R. Beck

    1. Kept up with software deliveries and test runs; the last deliveries were just before Christmas break.
    2. Created a preliminary version of the archive to dlt tape of the production output data.
    3. Created a preliminary version of the restore of archive data from dlt to disk.
    4. Made a new delivery of the monitor software. Improvements include a working output errors option with 'time' output for each subsystem. The space option now works for rodan.
    5. Restored the USNO catalog from cds to lugosi:/l1/USNO.cat.

    T. Evans

    Mapcor v 1.0 is now delivered to rodan, and is the version used in the present 2MAPPS version. It contains all of the planned functionality except:

    1. Seeman seeing data is not yet convolved with the confusion radius used in finding faint sources confused with brighter ones.
    2. "Table switches" used to force mapcor to use lookup table values instead of calculated values are not yet operational.
    3. Detailed stats to be output on demand, aren't.
    4. Summary file and quality output are not yet written. The newly-output aperture magnitude curve-of-growth plots (and numbers) have shown that while the differential magnitudes flatten out as the apertures get large, they stay around 0.01 mag instead of going to 0. B. Light is investigating this.

    At this point focus is shifting to the DB task. Most of the point and extended source data fields to go into the DB have been defined in the GAL0x SIS's. Lists of quantities and/or data sets that also need to go into the DB are now needed from the team, especially those that appear in summary or auxiliary files but perhaps not in the quality output files, or that are known to be very interesting for initial analysis.

    The FSC and our 2MASS simulation DB will soon (hopefully by next Tuesday) be loaded into INFORMIX. Anyone interested in playing around in the DB, or in looking at our present VERY PRELIMINARY DB access documentation, contact J. Mazzarella or T. Evans for information on how to get to them.

    J. Fowler

    DFLAT is probably ready for 2MAPPS version 1.0, although one lien remains, involving adapting to high-density regions; the hybrid stack trimming algorithm may work acceptably as is, but this will not be known until real data are processed.

    BANDMERGE is lacking statistical analysis and QUALITY output; otherwise, it appears to produce proper output for the pipeline and could serve in 2MAPPS 1.0 with corresponding liens.

    DARKS currently generates darks, masks, and responsivities, including separate evening/morning analysis and comparison. Coding is not yet finished on selection of which results to use, and testing is incomplete. Furthermore, expansion of DARKS' tasks to include soft-saturation analysis is being considered, and this may be too much to get into the 2MAPPS 1.0 delivery (it would also necessitate a RDFRAME redelivery to accomodate a new way of representing the soft saturation information).

    L. Fullmer

    The CONMAN function now reads a corners file for one night to determine pairs of overlapping scans. The overlap area of each pair is defined and source-lists from the paired scans are compared within the overlap area. For each scan a table of unmatched sources is produced as well as a table of matched sources with match information, including position and photometric differences. For the night a confirmation summary table lists match statistics for each scan. More statistics are being added to this summary table.

    A plan for presenting QUALITY nightly quick-look information in HTML pages viewable with a Web browser is being developed.

    T. Jarrett

    GALWORKS algorithm no. 1 three-channel code is 90% complete. Algorithm no. 2 is 30 - 40% complete. Analysis is continuing (with J. Rosenberg) of the SA57 protocam data; SA57 data is being converted to '3-channel' equivalent.

    G. Kopan

    FREXAS summary - FREXAS coding is complete except for improvements in the uncertainty modeling, special handeling for saturated source centroids and uncertainties, and re-generation of psfs for the production cameras.

    PIXPHOT summary - the major functions of PIXPHOT are complete. Remaining tasks include:

    1. updating the noise estimator to a histogram technique
    2. implementing an adaptive fwhm for the find step based on the seeing estimates returned by seeman
    3. possible coordinate conversion of galworks avoidance areas to band scan coordinates for prophot deblending avoidance
    4. implementation of distortion correction in picman additional diagnostics and quality checks
    B. Light

    A basic deblending algorithm has been turned on, and is currently being tested by running simulated pairs of stars through PROPHOT. The simulated data consist of grids of stars with separations from 1.1 to 2.0 pixels, the pair of sources having a ratio of brightnesses 1:1, 3:1, and 10:1. Initial guesses at the parameters used in deblending were made, and thus will result in less-than- optimal deblending. Preliminary analysis of the crude deblending shows that pairs (w/ 3:1) can be deblended 50-80% of the time down to separations of 1.6-1.7 pixels. Parameters will be refined and tested on all sets of data. As necessary, the algorithm will be enhanced.

    The non-convergence of large-aperture magnitudes (see T. Evans report) is being investigated. A very likely source of the curve-of- growth problem is in the sky estimation. A review of how sky pixels (along with bad/saturated pixels) are accumulated by the code was made, and no obvious source of the error was found. The sky level returned by the DAOPHOT-extracted code is now being investigated; specifically, will the median provide a better estimate? A subset of the pipeline (PIXPHOT+MAPCOR) is being run with an altered version of the APPHOT/PROPHOT code to check on this.

    H. McCallon

    The entire USNO-A catalog and some sample fields of the real Tycho catalog have recently arrived at IPAC; code is currently being written for POSMAN to use them. The coding for the USNO-A is complete and it appears that it may not be necessary to restructure the catalog to fit the 2MASS tile structure. Timing tests in high- density fields are needed to be sure. Work on the Tycho-specific code is just beginning. After the catalog work is complete, many tasks remain, including:

    1. Refinement of the TwoPeg Pattern Matcher code to improve efficiency and robustness
    2. Generation of QA parameters, levels of exit errors and summary file
    3. Code to model the changes in atmospheric refraction and distortion over each frame
    4. Adding code in POSPTS to do a linear fit of U-scan to J2000 using point sources from Bandmerge (up to now POSPTS has simply been applying the fit from POSFRM to the bandmerged point sources)
    5. Adding code in POSPTS to do associations of the final point sources from BandMerge with the USNO-A and to compare positions with the POSFRM point source reconstruction for QA.
    6. Design and code for the position uncertainty modeling of reconstructed frame positions in POSFRM and point source positions in POSPTS
    7. Major rewrite of the POSMAN SDS
    An additional ongoing task is modification of the routines borrowed from SLVALL (in the haste to have something pipeline-ready for the Dec. 3rd delivery) for maximum robustness and maintainability. In addition to trying to make the code as bullet-proof as possible, the plan is that comments at the beginning of each routine will define input and output variables (including units) and specify how each variable is passed. This should make it much easier to modify these routines later on. So far, approximately 1/3 of the routines have been brought up to this standard.

    J. White

    Work has continued on the DLT tape software. Since documentation for the TAR and MT utilities is minimal, considerable time has been spent experimenting with them using the DLT drive to determine details of their performance, particularly at EOT. A test using the h/w compression on the DLT 4000 tape drive after preconditioning the scan (by differential subtraction) resulted in a compression ratio of less than 1.4. Scan 083 from 95-05-03 (currently resident on rodan) was used as the test case. This ratio is disappointingly low. We had hoped to stack two compressed nights onto a single DLT without cascading. Tape cascading s/w is not currently included in either the tape load or tape write scripts.