1. SINGS Team Meeting — Baltimore, October 16, 2015JDS 1 CUBISM or How I Learned to Stop Worrying and Love the Cube J.D. Smith

2. SINGS Team Meeting — Baltimore, October 16, 2015JDS 2 Today’s Menu  SINGS IRS Program Refresher  CUBISM — your friendly neighborhood cube-builder.  From here to there  Assembling cubes from BCD inputs. ...and back again  Spectral maps and line extraction.  Leftovers — casserole anyone?

3. SINGS Team Meeting — Baltimore, October 16, 2015JDS 3 SINGS IRS Program  LL radial strips: 4–16x11 steps, both slits at once:  76.2 hrs; 936.7 MB.  SL: 1x9, subslits SL1 & SL2 pointed separately:  Nuclear: 16.1 hrs; 224.6 MB.  Extra-Nuclear: 36.6 hrs; 224.6 MB.  SH & LH : 3x5 (SH:  2 cycles):  Nuclear: 61.3 hours; 561.6 MB.  Extra-Nuclear: 62.1 hours; 549.1 MB.

4. SINGS Team Meeting — Baltimore, October 16, 2015JDS 4 Mapping Data  The IRS spectral maps range in size from 15 to 176 DCE’s per slit.  Each galaxy’s nucleus is covered by ~85 individual slit pointings.  SL maps will offer “satellite” wings, since each sub-slit is mapped sequentially over the same area.  LL maps will have non-uniform sub-slit coverage.

5. SINGS Team Meeting — Baltimore, October 16, 2015JDS 5 Background  CUBISM is a tool for constructing spectral cubes ( , , — one pixel per location/wavelength) from both low and hi- res IRS spectral mapping data sets, the release-form of which will consist of:  The Cube-Builder Back End.  CubeView and CubeProject Interface Components.  Written entirely in IDL †, drawing on existing elements of SCOREX. † Except for one auto-compiled, fault-tolerant component to be written in C for speed. Currently targets: IDLv5.5. Will target: IDLv5.6.

6. SINGS Team Meeting — Baltimore, October 16, 2015JDS 6 Development Synopsis  Low-Res  High-Res, progressing through three stages: 1. Rough cube, no de-fringing, error weighting, adjustable grid spacing, or refined image combination. 2. Proper weighting for combination from pixels at different pointings and spectral orders. 3. Possible additions: tunable grid spacing, optimized slit-image combination.  Simulated data was/is critical.  SIM data progressed alongside the cube builder to provide increasingly robust checks.

7. SINGS Team Meeting — Baltimore, October 16, 2015JDS 7 Cube Builder  Assemble cubes directly from BCD data set, minimizing interpolation — pixels take shortest route possible from BCD to cube.  Single cube pixel:  Data from different pointings (e.g. half-slit offsets)  Data from different orders on the array (bonus order/high resolution modules)  Provides “pixel accountability.”  Controls error “blooming” caused by unnecessary correlating operations (e.g. fractional pixel shifts or rotation).

8. SINGS Team Meeting — Baltimore, October 16, 2015JDS 8 Cube Builder  Permits flexible grid spacing.  Will accommodate pointing uncertainties and geometric distortions in the slit image (if measured).  Allows Resolution vs. S/N tradeoffs — exploring alternative cube “sky” grid spacing.  Flux–conserving, drizzle-like (with many more clip operations, but unit drop size).  Separate cube for each sub-slit. Drizzle

9. SINGS Team Meeting — Baltimore, October 16, 2015JDS 9 Cube Building — Details 1. A few to dozens of BCD’s are read into a cube “project”. Headers are fully parsed. 2. Logical row/column numbers, plate scale, slit and slit dimensions stored for locating samples on the mapping grid. 3. Calibration set object loaded from disk  Cached order position, wavelength solution, PR clipping areas and polygons, etc.  User-selectable. New aperture clips automatically stored.

10. SINGS Team Meeting — Baltimore, October 16, 2015JDS 10 Cube Building — Details 4. Build order and aperture selected (normalized “slit-length”, linearly varying across order). 5. Spatial dimensions of the cube calculated from map step coordinates & slit size. 6. Pixel-based WAVSAMP samples obtained from calibration object for chosen aperture. 7. Merge data for combining wavelength- overlapping orders pre-computed from order(s) selected and calibration inputs.

11. SINGS Team Meeting — Baltimore, October 16, 2015JDS 11 Cube Building — Details 8. Wavelength dimension of the cube computed from merge data. 9. Accounting list: BCD pixel  Cube pixels, with overlap areas. Slow to compute: cached (big). Associated with BCD record (compute individually). 10. Each order and each PR considered in turn. Partial pixels de-rotated (slit- rotation, position-angle deviations within map, etc.), offset and clipped to sky grid (can’t trust pointing reconstruction).

12. SINGS Team Meeting — Baltimore, October 16, 2015JDS 12 Cube Building — Details 11. Accounting cube filled and (possibly) merged with existing accounts. 12. Reverse accounting histogram constructed from merged accounts: Cube Pixel  BCD Pixels. 13. There’s no step 13. 14. Reverse accounts used to combine appropriate fractions of contributing BCD pixels in cube. Total area also accumulated. 15. Area-weighted cube complete.

13. SINGS Team Meeting — Baltimore, October 16, 2015JDS 13 Cube Builder

14. SINGS Team Meeting — Baltimore, October 16, 2015JDS 14 Maps and 1D Spectra  Both spectral maps (e.g. [Ne II], PAH 11.3µm) and (non-BQD) 1D spectral extractions (e.g. resolved nucleus) can be made directly from the cube.  Flexible map creation: arbitrary band-pass.  Simple square aperture, expanding aperture, or full parallel aperture photometry 1D extractions:  Can be tailored for changing extraction environment.  Potential for automatic aperture selection (brightest knot).

15. SINGS Team Meeting — Baltimore, October 16, 2015JDS 15 Archival Format  Discussions with FITS spec-writers led to:  FITS Cube/Error Cube/Flag Cube in subsequent FITS Extensions (  -headers — STScI convention).  Wavelength map for each cube plane in a single column of a binary table in a separate extension, pointed to by PS3_{0,1} with CTYPE3: ‘WAVE-TAB'.  Wavelength sampling still unresolved:  Linearize?  Small or vanishing overlap…  Other archive product formats TBD.

16. SINGS Team Meeting — Baltimore, October 16, 2015JDS 16 Archival Format

17. SINGS Team Meeting — Baltimore, October 16, 2015JDS 17 From SIM to Cube

18. SINGS Team Meeting — Baltimore, October 16, 2015JDS 18 From SIM to Cube

19. SINGS Team Meeting — Baltimore, October 16, 2015JDS 19 From SIM to Cube

20. SINGS Team Meeting — Baltimore, October 16, 2015JDS 20 From SIM to Cube

21. SINGS Team Meeting — Baltimore, October 16, 2015JDS 21 Interface  CubeView & CubeProject — defer to demo.  Fundamental Interface Doctrine: All cube operations can be performed without the interface (scripting):  E.g. after 100 cubes have been delivered, the calibration data gets a major update: re-do them all by hand? No, simply script it.

22. SINGS Team Meeting — Baltimore, October 16, 2015JDS 22 CUBISM Schedule  How we’re doing:  Overall, quite well.  Cube improvements more involved, unexpected “deeper” issues discovered  interface lagged.  February, 2002: First internal alpha team release.  December, 2004: First full release to the SSC (to coincide with first data cubes available), with SSC-scientist level documentation.

23. SINGS Team Meeting — Baltimore, October 16, 2015JDS 23 The Leftovers  MIPS SED.  Assorted cube improvements planned.  Interface gravy.  Unknowns:  Fringing.  Deep cube building issues.  Questions and uncertainties for the SSC.

24. SINGS Team Meeting — Baltimore, October 16, 2015JDS 24 MIPS Coordination  MIPS SED mode can produce spectral maps too!  In coordination with the MIPS team, MIPS SED data will also be processed and displayed in CUBISM.  Analogous to a single Low-Res IRS order (e.g. SL1), but with more involved detector issues (offsets, varying gains, etc.).  MIPS-specific issues will be factored out (G. Bendo).

25. SINGS Team Meeting — Baltimore, October 16, 2015JDS 25 Cube Improvements  Position-based offsets: bizarre combinations, like SL1_cen, followed by SL_cen at another epoch (and PA). Exact telescope/calibration agreement on slit center required.  Position optimization — treating reconstructed positions in aggregate.  Full error treatment, and alternative BCD pixels  Cube pixel combination choices (area weighted: A, A  2, A  2 t, clipped, median, etc.)

26. SINGS Team Meeting — Baltimore, October 16, 2015JDS 26 Cube Improvements  Auto-aperture selection (guided by SIMIII).  Noise floor for data rejection.  User definable bad pixels.  Updated and optimal extractions:  Expanding aperture.  In-place sky removal ala parallel aperture photometry.  Images stacked according to line fit (e.g. de-blending line pair).

27. SINGS Team Meeting — Baltimore, October 16, 2015JDS 27 Interface Additions  Pixel-account querying and feedback.  User-definable bad pixels.  Saved extraction band-pass sets.  Graphical manual/auto spectral aperture selector.  Circular aperture, expanding cube aperture extraction selector.

28. SINGS Team Meeting — Baltimore, October 16, 2015JDS 28 Spectral vs. Spatial  Pixel-level ambiguity between spatial & spectral resolution (slicers undersample).  Plan: carry 1  n & 2  n for comparison.

29. SINGS Team Meeting — Baltimore, October 16, 2015JDS 29 Fringing  IRS has fringes (<15% peak-peak) in both high-resolution modules at R~50, originating in the detector substrate.  Photometric accuracy of line maps impacted.  Multiple de-fringing efforts underway.  1D vs. 2D de-fringing, in or after the IRS pipeline?  CUBISM favors 2D de-fringers, but will evaluate all options when the smoke clears after launch.

30. SINGS Team Meeting — Baltimore, October 16, 2015JDS 30 Questions & Uncertainties  For the SSC:  Flux calibration: different for extended objects (sensitive to slit throughput function)?  Reformulated WAVSAMP: pixel-based?  Error plane reliability and simulations.  Wait and see:  Pointing inaccuracies: Drift, Reconstruction errors.  Background subtraction strategy (SV).

31. Fin

33. SINGS Team Meeting — Baltimore, October 16, 2015JDS 33 Raw SCORE (SH Prototype) Atmospheric Emission Spectrum