---------------------------------------------- IRS/SINGS Validation on NGC 7331 ---------------------------------------------- ======================================================================== Summary - Critical Tests: The IRS data for NGC 7331 meet all of the critical IRS tests for the SINGS project: data accountability and integrity; pointing; sensitivity; and LL background. SH and LH cubes have not yet been created, largely as a result of the different high-res plate scales expected vs. achieved, and remaining pipeline issues regarding the treatment of "rogue pixels". As a result, the centering of the high res cubes cannot yet be determined. However, the SH map appears to be centered based on the relative emission strength of the emission in the various high res pointings; Our expectation from low-res maps and other evidence is that the pointing will be on target, and as a result we no longer need to peak-up on the source or nearby 2MASS star: Spitzer's excellent blind pointing is more than sufficient for our map alignment. - Non-critical Tests: CUBISM is progressing nicely. We can make excellent SL and LL cubes. SH and LH cubes are next on the docket. There are some minor issues that need to be resolved; they are listed in Appendix 1. The preliminary wavelength calibration looks promising. Some work is pending improved- and readily-available calibration files from the SSC. Preliminary cross-instrument and cross-observatory flux calibration work has been carried out. SL and ISOCAM data agree within a factor of two. There are no major fringing or stray light issues which prevent SINGS from proceeding. - Program Adjustments: Several adjustments to the SINGS IRS observing strategy were motivated by better than expected on-orbit performance of the IRS. The validation data emphasize the power of SL; the SINGS project would benefit from larger (2x) SL maps. The extra time required to perform this expansion can be obtained entirely from savings realized as a result of the fact that LL is wider and longer than reported in the original SOM -- our LL maps now require fewer slit-perpendicular and slit-parallel steps to cover the same area. Even when expanded, our SL maps do not go far out enough to provide reliable measurements of the zodiacal background, in contrast to the long LL radial strips. Though zodiacal light will not impact the spectrum in the bright central regions of most SINGS galaxies, extended low-level disk emission will be affected. A plan to use the time saved by omitting map peak-up for an internal calibration of the zodiacal background is outlined in Appendix 2. Other minor adjustments included expanding SH nuclear maps for a small number of nuclear ring targets, introducing a limited set of off-source SL zodiacal background observations, and re-allocating a number of pre-selected extra-nuclear targets to reserved status. All adjustments are detailed in Appendix 2. ======================================================================== Appendix 1 -- IRS/SINGS validation on NGC 7331: Critical and Non-Critical Test Results ============== Critical Tests ============== The following assessments are critical for further acquisition of IRS/SINGS data to continue as currently specified by the AORs. (1) Data Accountability and Integrity - Compare requested and expected number of raw DCEs and processed BCDs, mask files, extracted spectra, and ancillary data files. (did we receive all the expected files?) RESPONSE: We received all of the expected and necessary data. - Were there any corrupted data that will prevent our completion of any of the critical analysis tasks listed below (#2-4)? If so, do we need a re-transmission of data from JPL or the spacecraft? If data were corrupted on board (due to CR's, instrument anomalies, etc.) do we need to re-observe all or part of NGC 7331? RESPONSE: The data were not corrupted so as to prevent validation. - Were there significant saturated spectra that make validation of the data or data-taking strategy impossible? Has stray light from the peak-up arrays severely contaminated the SL2 spectra (see also #8, below)? RESPONSE: The data were not saturated so as to prevent validation, and peak-up stray light was not severe enough to prevent validation. (2) Pointing - Compare requested and reconstructed pointing for all IRS DCEs. * Is the center of the map consistent with the nucleus or HII region? Are there global offsets beyond the reported pointing accuracies of SIRTF and the IRS? These could indicate a problem with the pointing for NGC 7331 or the pointing reconstruction at the SSC. RESPONSE: The SL and LL maps created with CUBISM nicely demonstrate centering on the nucleus. SH and LH cubes have not yet been constructed, but are expected to show similarly good alignment. There is no global offset larger than the reported pointing precision. * Are the positional offsets consistent with those requested over the entire map? This should be evaluated between slit positions and also between map legs. Requested positions are indicated by header keywords (RA,DEC,PA)_RQST in the BCDs. Reconstructed positions, based upon the current Observer, are indicated by header keywords (RA,DEC,PA)_FOV and (RA,DEC,PA)_SLT in the BCD (bcd_fp.fits files). RESPONSE: Spitzer's good pointing means that cubes can be built reliably even using the commanded positions, as demonstrated by the constructed maps of the ring. The errors in the reconstructed position keywords have been addressed in the Jan 04 pipeline build, and so it should now be possible for either commanded or reconstructed positions to be used to build cubes, with similar results expected. (3) Sensitivity (and mapping redundancy) - Do the reported sensitivities of all four modules fall within the tolerances of the existing program? This can be answered without reference to NGC 7331 data. RESPONSE: The sensitivities of our data, and the signal-to-noise achieved for each IRS module, are generally in good agreement with our expectations, and with the sensitivities reported in the IRS SOM: SH: 3sigma: 2e-18 W/m^2 Goal: 5e-18 W/m^2 SOM: 2e-18 W/m^2 0.42e-17 W/m^2 for 30x2=60s + 4x redundancy S/N: 6 and 7 for [Ne III] and [Ne II] LH: 3sigma: 5e-18 W/m^2 Goal: 5e-18 W/m^2 SOM: 5e-18 W/m^2 1e-17 W/m^2 for 60s @ 34um + 4x redundancy S/N: 10 and 25 for [SIII] and [SiII] SL: 3sigma: 0.4-4 mJy for off-source ("background") Goal: 3 mJy SOM: 0.4/1.3 mJy 0.6/1.8 for 60/14s extranuc/nuc + 2x redund S/N: 5 to 20 ranging from nuc to interarm continuum LL: 3sigma: 1-4 MJy/sr Goal: 1-2 MJy/sr SOM: 1-8 MJy/sr 2-16 MJy/sr for 30s + 4x redundancy S/N: - Do our validation data suggest a deviation from the reported values which would compromise the program and/or force a revision in data-taking strategy (see #4 and #7, below)? RESPONSE: No, all sensitivities meet our goals. - If pointing and offsets were as commanded (see above), and the instrument response is as expected, is there adequate redundancy and slit overlap in our nominal data-taking strategy to ensure high S/N cubes and maps? This is particularly important for the SL spectral maps. RESPONSE: Our 1/2 slit step strategy is validated by the integrity of the assembled low-resolution cubes. Our LH and SH validation maps were based on pre-launch data regarding slit size, and as such were composed of steps which were larger than the actual 1/2 slit size. This has been corrected in the AORs to reflect true 1/2 slit steps. The larger on-orbit LL slits have also resulted in a significant time savings, since only 10 slit-perpendicular LL steps are required to match the usable length of the SL slit. (4) Removal of LL Slit Backgrounds [less critical than 1-3, above] - To what extent can data in the "wings" of the LL spectral maps be used to assess and remove the local backgrounds? Are the LL flanking fields spatially flat? Do we need to revise our data- taking plan to include real off-galaxy spectra? RESPONSE: The LL emission drops off fast enough radially and the additional coverage in the LL satellite wings is extends far enough out that a very good estimate of local backgrounds can be obtained with the present LL mapping strategy. ================== Non-Critical Tests ================== IRS/SINGS data can continue being taken, and non-critical issues will be addressed concurrently. (5) CUBISM Functionality - Verify that CUBISM can read all the BCDs, extract the necessary header information and assemble the cubes. * Are there any artifacts in the cubes or extracted maps which cannot be explained via data-taking (e.g. stepping) anomalies addressed above? RESPONSE: There are unmasked and/or uncorrected hot pixels in all modules' data. This will not warrant a change of observing strategy, and is currently being addressed by the SSC. * Can anomalies in the cubes or maps be removed by modifying CUBISM in a reasonable amount of time -- without affecting our delivery schedule? RESPONSE: Yes, all data anomalies are being treated either in the pipeline, or by modification to the cube-building algorithms. * Is the pointing information for each BCD accounted for correctly, such that multiple epochs of data will not pose problems? RESPONSE: Incorrect reconstructed positions are a minor (and hopefully temporary) inconvenience, preventing us from testing map astrometry refinement using the reconstructions of all steps in aggregate. * Are there any irregularities in the regions of order overlap? If so, are these indicative of a problem in CUBISM, or the IRS pipeline products -- e.g. flat fielding or stray light removal, (see #8, below). Can the overlapping scan legs be "seamlessly" combined to produce a map? RESPONSE: SH shows continuum oscillations and strong order mis-match in the multi-order spectra. There are flat-field residuals near the order extremities in the high-res BCDs that may be related, and the SSC has been notified. * Are the delivered uncertainty images useful for weighting and rejecting data from the extremities of the order/slit? RESPONSE: Valid uncertainty images are not yet produced by the IRS pipeline for testing, but will be very important for optimal cube building. * Is there a significant difference in the maps when using the reconstructed vs. the requested offsets? RESPONSE: We could not perform reconstructed map builds because of the reconstructed positional keyword errors. (6) Calibration Products - Are all requisite calibration products available in the correct formats? These include wavelength and order position solutions, permanent mask files, and flux calibration files/keywords. RESPONSE: There a few outstanding issues with calibration files. In particular, we have a crucial need for the ORDFIND and WAVSAMP calibration files which contain the polynomial fits to the order positions and wavelength solutions (the "a", "b", and "c" coefficients). These basic calibration files are used as inputs to the WAVSAMP, and until 10/03 were listed among the supported calibration products. To date, we have received some updated version of these files from P. Morris by private communication, but would urge returning these important calibration products to supported status. - Does the calibration match pre-launch expectations? If not, will significant deviations require changes to the established cube-building algorithms? RESPONSE: Wavelength calibration has been cursorily checked. We have examined the fitted central wavelengths to several lines in SH and LH maps, and the results are encouraging. Accounting for a redshift of 816 km/s, we find: ------------------------------------------------- line rest expected observed diff obs-exp lam/R um um um % um um ------------------------------------------------- SIII 33.482 33.573 33.535 0.10 -0.038 0.056 SiII 34.815 34.910 34.906 0.01 -0.004 0.058 NeII 12.814 12.849 12.853 0.03 0.004 0.021 NeIII 15.555 15.597 15.605 0.05 0.008 0.026 ------------------------------------------------- Note the single rather large wavelength offset (~1.5 pixel) in order 12. Flat-fielding problems may exist in SH as discussed above. LH and LL exhibit problems resulting from flat field issues at the very ends of the slit; in the case of LH, data at the ends has been set to NaN, which adversely affects the cube creation. The SSC has been notified and has committed to resolving this issue. (7) IRS - IRAC - MIPS cross-calibration - Verify calibration consistency through direct comparison to ISO maps. RESPONSE: Flux calibration accuracy was validated to first order using NGC 7331 ISOCAM maps from the ISO archives in filters LW2 (5-8.5 um, corresponding to SL2), LW3 (12-18 um, corresponding to SH), LW6 (7.0 - 8.5 um, corresponds to SL2), and LW7 (8.5 - 10.7 um, corresponding to SL1). LW8 (10.7-12.0um) and LW9 (14.0-16.0um) data suffer from a flat-field artifact superposed on the nucleus, so were disregarded. Zodiacal emission was subtracted using the estimates from Spot, and only rough alignment of the ISOCAM maps with the spectral cubes was obtainable. The measured fluxes measured within matching apertures agree within a factor of ~2 with the ISO data in all filters, with residual errors dominated by uncertainty in the application of the point-source flux calibrations. A primary remaining concern is the conversion of these point-source calibrations embodied in FLXCON header keywords to a useful extended-source calibration. This conversion depends on the details used in the construction of the FLXCON values which were not available to us. - Do the IRS flanking peak-up fields sample backgrounds match what we derive from the slit perimeter fields (see #4, above)? Use SPOT visualization to verify peak-up and slit perimeter overlap. RESPONSE: The LL maps provide significant background at the expected level. The unexpected sensitivity and power of SL means that even the satellite fields in each of the sub-slit centered SL1/SL2 pointings contain emission from the galaxy. This has motivated the inclusion of selected background pointings (see Appendix 2). In the LL maps, the expected background flux is obtained in the extremities of the slit, with uncertainties contributed by the same problems outlined in the last response. - Generate maps over relevant IRAC and MIPS filter bandpasses and compare to IRAC and MIPS data directly after extracting to the same projected aperture. This will require co-ordination with IRAC/MIPS groups. Is there a systematic offset in one of the derived products that is indicative of a change in sensitivity or an effect not documented by the IRAC, MIPS or IRS teams during IOC or SV? RESPONSE: This has not been done, but good independent flux measurement exist for each instrument's data. (8) IRS Pipeline - Is fringing adequately removed by the pipeline flat-fielding? RESPONSE: Without a rigorous analysis, it appears that the pipeline processing will be sufficient. However, we need to build high-res cubes before this issue can be properly addressed. Fortunately, we can already tell that this will not be a show-stopper, as first order flat-fielding removes the bulk of the fringe amplitude. We are very interested in 2-D fringe-removal techniques, and encourage their development. - Are Peak-Up stray light and high-res crosstalk removal modules functioning well enough to use corrected BCD products as inputs to CUBISM? RESPONSE: Yes. The BCD products are acceptable for cube-building. We are seeing some scattered light in the peak-up arrays, but this is not an issue for validation and does not constitute a problem for SINGS. At the 5% level it would prevent the use of the PU images obtained during spectral mapping as a "parallel" imaging mosaic data set. ======================================================================== Appendix 2 -- IRS/SINGS validation on NGC 7331: Summary and Rationale for Observing Adjustments Several adjustments were made to the SINGS IRS observing strategy to capitalize on improved on-orbit performance and instrument parameters. - SL Nuclear Expansion, and LL Step Adjustment The validation data emphasize the power of SL; the SINGS project would benefit from larger (2x; i.e. 1x18 steps) SL maps. This expansion will be applied to all SINGS targets. The extra time required to perform these observations can be obtained almost entirely from the savings realized as a result of the larger LL slits -- our LL maps now require fewer slit-perpendicular and slit-parallel steps to cover the same area and match the SL coverage. The sense of the SL expansion is along the radial strip direction, such that the area of SL and LL overlap is being doubled. No expansion is performed for extra-nuclear SL maps. Total SL Expansion Cost: 14.0 hours. Total Saved by optimized LL Maps: 14.0 hours. - SH Nuclear Expansion on Selected Galaxies with Nuclear Rings Several SINGS galaxies show bright, extended emission in a ring around the nucleus, which would not be covered by the default 3x5 SH map. By using ISOCAM and H-alpha images to investigate such nuclear extensions, 6 galaxies were identified which offer real expected benefit from a SH expansion to 6x10 steps (4x the area). The 6 galaxies so identified are: N1097, N1512, N1705, N4321, N6946, and N7552. Total Cost: 7.0 hours - Peak-Up dropped for all Spectral Maps We will remove all peak-ups for all spectral maps. Spitzer's blind pointing is good to 1" (1-sigma) and potentially a bit better than that. Nuclear and extra-nuclear peak-up also present the real possibility of failure if another bright source is present in the ~1 arcmin peak-up field. Offset 2MASS star peak-ups could be fine, but blind pointing is sufficient for our alignment needs. Total Saved: 5.8 hrs - Off-source SL Zodiacal Measurements in Selected Galaxies The SL maps, despite being expanded on the nucleus, will contain little usable background. In high-flux regions at the centers of the maps, the zodiacal background will affect the spectra very little. In the extended disk and in fainter objects, however, fluxes can be comparable to the zodiacal level, which varies in our sample from f(12um)=13 -- 60 MJy/sr, depending on ecliptic latitude and time of observation (the range includes the min and max over all possible observation dates; N7331 had f(12um)~20MJy/sr). Such faint disk spectra will suffer in two ways from the unsubtracted background: 1. The continuum slope will be affected, as it is superposed on a T=~250-280K blackbody. 2. Residual 10um silicate emission in the Zodiacal light (feature strength 1-10%) will be inappropriately attributed to galactic emission. This variable excess emission will fill in intrinsic silicate absorption, or, in the absence of absorption, alter the continuum level between the 7.7um and 11.3um PAH features. Reach, Morris et al. (2003) find significant variation in both the effective blackbody temperature and silicate emission strength with ecliptic latitude and solar elongation. The silicate feature is not presently accounted for in any zodiacal light model, including the one utilized by Spot. While we have no control over the solar elongation for our observations, we can internally calibrate the variation with ecliptic latitude by attaching off-source SL pointings to a small subset of the sample, chosen to span a range of latitudes. A small SL map composed of 3 steps with 2 60s cycles at each step will cost 11.3min per source, and will provide a high S/N background spectrum, as verified by the 2x60s + 2x14s off-source background pointings obtained along with the NGC7331 validation data. Although a very detailed measurement of the zodiacal background will be obtained over the course of the mission, this ensures timely availability of background spectra for use with SINGS data, and can be folded into a larger effort to quantify spectral variations in the background. Ten sources uniformly distributed in ecliptic latitude were chosen for SL background observations. Total Cost: 1.9 hours - Dropped All Nuclear Maps for Galaxies without Nuclei Two galaxies which absolutely showed no ISOCAM nuclear emission, HoII and NGC6822, had all nuclear maps dropped. Total Savings: 2.6 - Drop Pre-Selected Extra-Nuclear Targets if Necessary Savings is 1.3 hours per target. GRAND TOTAL ----------- 263.5 hours, which includes -expanded SL maps for galaxy nuclei -expanded SH maps for 6 galaxy nuclei -10 background measurements for SL -no extranuclear targets dropped