Minutes for Madison Team Meeting - Mar 21 & 22, 2002 Present: Churchwell, Whitney, Watson, Meade, Dickey, Marston, Mathis, Indebetouw, Babler, Wolff, Benjamin, Jackson, Clemens, Wolfire, Lazarian, Ashman Next telecon: Tuesday April 9, 2002, 11 am CST, 608-265-1000 #4112 Thursday 21 Mar., 2002 Report on SSC activities - Tony For the May Review: - Include a demo of the post-pipeline software - Validation Observation Analysis - Details of expected final products The IRAC SSC pipeline is not yet automated. Can generate products by hand. We have some output from them. The fits headers in the output they gave us are not identical to the final product. Still culling the info in the fits header. Let Tony know if we need something in the fits header, like temperatures, volts, etc. There is no banding correction. Banding is not characterized. Some data is available on the IRAC LSWG web site (we have it). The fits keywords for distortion are better known, explained in a memo. Banding discussion: Occurs in ch 3 & 4. Hard to put point sources in ground- based testing so not much data to characterize it. Not an obvious linear effect. It's a sensitivity effect. Observing at a different angle in our survey would help. Will there be an effort to characterize banding after launch? Yes, after IOC they hope to characterize it better. Probably not in SSC pipeline at least for first year. We would need to ask for early IOC data to start working on our own banding estimate. The First Look Survey data will be available to the public. There is a galactic & extragalactic component to the FLS. 6% of pixels in field may be affected by banding. Affects completeness and reliability. We should mention at the SSC Review that we will be looking into banding; show how important it is to us. It's a function of intensity & location. Mention at SSC meeting that we will want appropriate IOC data to help us with banding. ERO: Tony wants to write an ERO proposal to observe M16. Will include our team. ERO is part of IOC. GLIMPSE Science and Science Spec. Document Star Formation, SEDs - Barb Barb talked about her & Ed's writeup which is in the Science Spec Doc made available to the team by Remy before the meeting. Her models are for low mass and high mass star formation convolved with the IRAC bands. No PAH's in the models. The early massive stars should stand out. Need distances. Want to get real data - ultra compact regions. GRS, complementary data (GBT, Arecibo) Low mass star formation regions not a primary goal. It's complementary to the Evans group. Summary: - Colors well separated from MS (0.2 mag flux errors OK) - CMDs can be used for luminosity functions - UCHII regions so bright that if IRAC is less sensitive, no problem - problem will be saturation of most luminous sources - Can detect 1 Lsun objects out to 2 kpc - Images => Stellar density Galactic Structure - Dan Molecular Ring, Spiral Arms in inner disk, & the central bar. We will get sample of SFR. Points to be addressed in the document will be areal coverage, angular resolution, #bands, sensitivity, photometric quality, SFR id ISO spectra of young stars show complex features in mid-IR due to molecules and ices. So interpretation of IRAC bands will be complicated. BU folks are building an instrument to observe L & M. AGB-carbon-OH/IR stars - Bob Check out Bob's web page: http://wisp11.physics.wisc.edu/~benjamin/glimpse/glimpseobj.html SF regions are tracers for galactic structure. There are 1100 HII regions in our survey region. Other tracers - AGBs - older objects Bob showed a color-color diagram (Fig 7) from the Egan et al 2001 paper (ApJ, 122, 1844). Carbon stars have a narrow range in lumonsity. Not clear if we will see the bar. These objects vary on time scales of a year- a reason to have 2 visits separated in time by months. 2MASS data will be very useful. ISM - John M. Continuum - gas dynamics, PDRs surrounding bright sources, 5.8 band only slightly affected 1) Unidentified IR bands 3.28, 6.2, 7.7, 8.8, 11.4 2) Scat light - not expected to be large 3) Unresolved stars Determining the extinction law is vital for dereddening all GLIMPSE data. 1) Using reddened stars of known intrinsic colors 2) Dark globules, using bkg stars seen through them Need intrinsic colors of objects with huge optical extinction Two possibilities: 1) Extinction may be constant over IRAC bands - Lutz using ISO (big footprint) 2) Extinction changes over IRAC bands - theory Mathis et al Need dereddening for Galactic Structure and far away SFRs. A really good object is the cluster Westerlund1 l=339.6, b=-0.4. (NGC3603, M17 are embargoed) There are 15 stars of known spectral types (B1V - G0Ia) It's twice as reddened as any other cluster...2 KPC distance. Fluxes are m(3.5 microns) ~8-13.5 Has a messy color-color diagram. A(V)~11 mag; A(3.5micron)~0.55 mag;A(8.5 micron)~0.2 mag But variations in the extinction law from place to place. Tony says what IRAC gets for calibration observations we can get. M16 reference: Sugitani et al 2002, ApJ, 565, L25. WR stars - Tony See the data specs document for Tony's WR writeup. Evolved stars from massive O stars. Sharra 1999 predicts 1000's of WR but haven't found many yet. They are indicators of recent SF. Colors of WR's in IRAC bands well known? Tony presented a color-color plot from his writeup showing where the WRs will likely be. Spatial information much better than MSX. There is a new WR catalog out by Karl van de Hucht with ~200 WRs. We can hopefully find WR candidates and the community can do followups. Documents (<10 pages each) - for us, SSC, public use Science & Data Requirements - John M will lead Observing Concepts Validation - Remy + Chip + Christer + Quality Assurance - Remy + Pipeline - Barb + Legacy Products - Remy + Simulations (for team use only) - Remy Ed distributed a possible outline of the Science & Data Requirements document. Would be good to have a table or matrix of breakpoints and C & R levels in executive summary. Completeness requirement, artifact requirement. Threshhold for accepting data. What if an amplifier goes flakey? How many pixels will that affect? Or a cosmic ray storm? SSC doesn't inspect every frame. Simulations - Remy completeness, reliability, photometric accuracy, background levels, saturation, instrumental artifacts, implications for observing scheme, etc. See Remy's writeup in the data spec document for more details Instr. effect Complete Reliability Phot Acc Remedy Crowding/ Confusion Undersampled PSF better dither Artifacts change obs strategy and/or increase passes High bkg S/N increase passes Expected confusion limit - Condon criteria ~28000 sources/sq deg. Can calculate source density using 2MASS & MSX. We hope to do better than the Condon criteria. 9 pixel separation between sources looks OK. ~100,000 sources/sq deg. We have been using DAOPHOT on simulated data produced by Matt Ashby's IRAC simulator. Pointing error, poisson noise, cosmic rays, minor artifacts, muxbleed are included in Ashby's ISDS. Remy has added banding for ch 3 & 4, a correction for saturated star wings, and world coordinate system info. The calibrated image uses diffuse bkg from MSX (scaled). Sources are from the 2MASS PSC with faint sources and brighter sources added. Would be good to use Martin Cohen's colors for a couple of areas. DAOPHOT doesn't pick up extended sources. Should put in galaxies and extended sources into the simulated data. The residual images show extended objects. Try to identify colors of galaxies. Completeness and reliability different for different regions in the survey. Can boost reliability but then lose objects. Flag local source density to get reliability...depends on gradient of the bkg. Tony says diffuse bkg affects us. Point out to SSC that we shouldn't be held to 2MASS standard. Bands 1 & 2 have been researched with DAOPHOT, single passes. How to deal with bkg for reliability. flag if exceeds a certain threshhold. Quantify bkg levels in terms of reliability. Slope is important. Steep slope - bkg flag. Bkg: give level and slope of rms. Mosaiced image results with DAOPHOT just starting. Trying to figure out the mosaicer. Validation Observations - Christer, Remy What will the verification observation test? - Are C & R & astrometric & photometric accuracy breakpoints reached? - Is the PSF sufficiently sampled? - Sufficient passes to have few artifacts - Is the S/N high enough to be confusion rather than noise or bkg limited? *We are doing 10 passes with range of subpixel dithers, two at as different a PA as possible.* What is capability as a function of source density? *We are using a range of source densities in the W51 field.* Does the AOR writer function properly? *AORs generated w/nearly the same code.* Can we detect asteroids? *7 hour span of observations* At the confusion limit, is there a difference between 2 & 10 passes? Is the primary instrument artifact banding? Try different banding cases in the simulations to help determine breakpoints. Observing Plans - Christer, Remy Half-step vs 2-3 visits (simulations) 2 Observing Strategies Can we solve the banding problem with 2 visits? -------------------------------------------------------------- l=10 - 25 25 - 39 39 - 51 51 - 62 7 16 25 38 change in roll angle l=-10 -25 -25 -39 -39 -51 -51 -62 15 24 36 51 change in roll angle -------------------------------------------------------------- 250 100 50 25 = pixels lost to banding 0 10 20 30 = change in roll angle assume 1.2% taken out by banding. 10 degrees helps a lot. After 10 degrees, marginal gain. --------------------------------------------------------------------- using a 12 pixel overlap 1/2 frame 2 visits 395 hours survey 10-62 10-53 using SPOT v5.0 coverage 10-66 10-57 SPOT v6.0 Can we get data with banding info from IOC data? Mention it to the SSC. We are doing different overlaps for the verification observation: 8,10,12,15 & 20 pix. Tony says there is up to 5" drift over 1/2 hour, then it's taken out. Gyro drift bias. can be out of commanded position by 5". Data is delivered by the SSC about a week after an IRAC campaign is done... The schedule gets filled in every 3 weeks. Launch +9 months is the first delivery date for us. Find known objects in the W51 verification region to do color checks, color-color plots. try to find extinction law. BU could get colors: J,H,K,L,M (Flagstaff) Survey tracking - Christer Christer has a prototype survey tracker interface at /www.astro.wisc.edu/~watson/sirtf/track.shtml Scalable - full survey down to a few frames. Has an automatic global view. Symposium or Workshop on SIRTF Science with Emphasis on Galactic Science (Stars, Galactic Structure, ISM) Tentative workshop/symposium (pending launch) in Madison, August 2003. GO proposals due 5 months after launch. ----------------------------------------------------------------------------------------- Friday 22 Mar. Overview of database design and how it will interface with the pipeline - Mike Solaris - multiprocessor- shared memory. 2-3 TB storage using IDE/RAID arrays. We will use Oracle 9i, using partitioned tables. "Spreads" the data over several logical units Think of the database as file cabinets. There are primary keys and foreign keys. Use perl scripts & SQLLDR (Oracle application) Includes CFITSIO interface SQLLDR - direct load, fast loading capability. parallel load. GLIMPSEdb - Retrieval - SQL (Structured Query Language) commands Spatial indexing - Most user searches involve celestial coordinates - Hierarchical Triangle Mesh of SDSS project Pipeline - Barb, Marilyn, Remy, Brian Barb went through the pipeline flow chart. See http://www.astro.wisc.edu/sirtf/Team/flowchart6.pdf for the latest version of the flowchart. A post-pipeline status (accomplishments, priorities for May meeting, and "to do" by Jan '03) was distributed. Data volume: 297 AORs, 75902 pointings, 303608 images Timing tests- rough estimates currently but here they are - Through Level C (automated) (assume 10 processors) - process ~12 AORs/day - Through Level C (QLVT) (assume 4 people, 2 hrs/day, 4 frames/min) - process ~2 AORs/day Near-term goals for the pipeline - Error trapping - Interaction with the database & QLVT - Bandmerging & Mosaicing - Catalog entries Flags for catalog entries - Remy Remy presented a suggested set of flags for the GBSC & GPSC. Image flags, Source flags, and Method flags. A description of these flags can be found in the data specs document. Several modifications were made to the flags after the team discussion. Pipeline demo - Mike, Marilyn, Barb, Remy An AOR of 4 rows by 4 columns was run through the pipeline up through the level C photometry stage. Both linux and solaris computers were used for the demo. It took about 10 minutes to run through the 64 images using the 8 processors on 4 linux machines (running DAOPHOT) and the solaris machine doing the other processes. See http://www.astro.wisc.edu/sirtf/Team/Processing/pipeline.mar19.doc.html for more details of the pipeline demo. Quick Look Validation Tool (QLVT) demo - Mark Mark gave a demo of the current version of the QLVT. See http://www.astro.wisc.edu/sirtf/Team/Processing/qlvt/index.html for a look at a somewhat older version of the program. The QLVT displays all 4 IRAC bands and corresponding MSX & 2MASS fields, can overplot point source catalog info, overplot mask information such as bad pixels, cosmic rays, saturated pixels, pixels affected by banding, latent pixels, etc. It can blink two images, and has lots of other neat features. Mark has delivered the QLVT to the SSC.