AGENDA
Subsystems that will be highly skeletal or emulated are: DARKS, POSFRM, and POSPTS. DARKS may be emulated by use of darks and responsivity images computed offline. POSFRM will consist of a shell that allows existing algorithm demonstration code to process the three bands separately, so that the output files will be generated but will not be based on simultaneous analysis of three-band information. The PFPREP module will not have dynamic allocation of detection records and will use detection count information in the FREXAS statistics file output (SIS FRX02) to determine cutoff magnitudes to use for limiting input to the currently fixed-size arrays (the Read1 arrays will be expanded to handle galactic plane densities). A skeletal version of POSPTS has been written and tested; this version applies the transformations derived by POSFRM to the point source list generated by BANDMERGE to yield a version of the point source list whose only expected serious deficiency is that no optical catalog associations will have been made (the not-as-serious deficiencies will be the lack of position improvement and the leftover degradation from having used transformations provided by the provisional POSFRM substitute; nevertheless, the quality of the point source information is expected to be quite good).
All of the other subsystems will have some liens, some more serious than others. TAPELOAD is essentially ready for production, including its RDFRAME module. EXEC/PCP is expected to provide full support for the subsystems involved. PIXCAL/DFLAT/FREXAS will be close to final shape; the main lien will be adaptive handling of high density scans. PIXPHOT will probably still operate in band-dependent coordinates instead of U-Scan. SEEMAN will estimate seeing, but those time dependent estimates will not be used to choose time dependent PSF's within the scans. PROPHOT will be using a reduced set of PSFs (the full set is not scheduled to be ready until well into the shakedown phase of the survey). MAPCOR will be in a very preliminary stage, able to merge the Read1 detections with the Read2-Read1 extractions, but not doing aperture corrections and doing little if any "purging" of persistence, ghosts, etc. (this is consistent with the MAPCOR schedule; T. Evans has just issued version 2.2 of the MAPCOR SDS, the first version intended to be signed off). GALWORKS is expected to be able to generate its primary outputs, i.e., extended source lists and band filled point source lists.
The limitations in the processing of any given subsystem are not to be compensated by special coding in other subsystems, where "special" means not intended for the real 2MAPPS system. While some coding that will not contribute to the final 2MAPPS system is inevitable, the amount of time spent for such code is to be minimized. This exercise is not to be permitted to create a significant 2MAPPS schedule slip. The module requiring the most special purpose code for this exercise is POSFRM; the investment in creating a substitute version is a necessary evil because of the dependence of downstream subsystems on the information it will provide.
Subsystem deliveries for this version of 2MAPPS will be made via the delivery method documented in the FDD (section 2.8). This involves creating a subsystem directory under /2massc/del/src, copying all appropriate files to it from the development area (normally a similar directory under /2massc/dev/src), running the make file, and notifying G. Laughlin of the availability of the subsystem executable(s) for integration into /2massc/ops/bin.
Parameters inspected included the statistical behavior of the trimmed average standard deviations, the flattened frame noise, and the residual antipersistence artifacts in the coadded images. The pure RMDR yielded the best results in all cases except the highest density scans, where the best method had negligible superiority in frame noise (where the pure-RMDR was not the best, neither of the other methods was consistently best).
The decision for now is to use the pure-RMDR method for all flattening, with 20 samples being trimmed and 22 being averaged.
The handling of Read1 frames has been modified after discussion with G. Kopan. The DFLAT/FREXAS interface (SIS DFL02) has been modified so that FREXAS returns the median value for each flattened Read2-Read1 frame, and this is scaled by the ratio of exposure times (0.05/1.3) to obtain a value to which the corresponding flattened Read1 frame's median is forced. This solves a Read1 aperture photometry problem caused by anomalous behavior of the Read1 bias level.