Some Thoughts on Cross Calibration of Space Based Infrared Telescopes

1. Some Thoughts on Cross Calibration of Space Based Infrared Telescopes Sean Carey Spitzer Science Center

2. Outline • Introduction • Calibrating the speaker • State of infrared calibration • A Holy Grail – An infrared zero point that can be agreed upon • That longer wavelength Great Observatory • Some cross calibration examples drawn from personal experience • Spitzer instrument cross calibration • Cross calibration to help with instrument features • What can we do next

3. Relative versus Absolute Calibration and Cross Calibration • Relative calibration provides correct flux ratios (colors) • Absolute calibration places a good relative calibration to physical units, relative calibration referenced to a good photometric standard • Hard to do! • Calibrate to NIST standards • Cross calibration is comparison between instruments • If instruments share a common absolute cal methodology then cross calibration does not probe the absolute calibration • Have to understand photometric reference of each instrument • Literature can be incomplete / conflicting • Bottom line is that current state of the art absolute calibration is good to no better than 2% in infrared (> 2 mm) and worse at longer wavelengths (> 30 mm)

4. Does Science Depend on Absolute Calibration • Most science can get away with an incomplete absolute calibration • Except for dark energy experiments • but Spitzer observations are routinely limited by abs cal same for JWST • ~1% error in flux compared to model will have little effect in derived source temperatures, distances/luminosities • But biases between instruments need to be understood when measuring colors, fitting SEDs, looking for excesses • Most transiting exoplanet science is immune as long as the differential signal in a band is robust • Knowledge of host star limits estimation of planet temperatures, radii and atmospheric composition • Spectral typing / radius measurement uncertainties should be considered

5. Personal Background and Potential Biases about Calibration • Worked on MSX data processing • Indoctrinated in the Price et al. (2004) methodology • MSX: 35 cm, 5 bands (4-21 mm), • Responsible for current state of IRAC calibration as Instrument Team lead • IRAC makes use of the Cohen spectral templates • PI of Inner Galactic plane mapping with MIPS (MIPSGAL) • Worked closely with MIPS team at SSC and Rieke et al. (2008) methodology • Career at l > 2 mm • Mostly and blissfully ignorant of calibration issues in optical • If calibration is a religion, agnostic when it comes to infrared calibration • No preferred calibration standards

6. Issues with Infrared Absolute Calibration • Vega is a bad choice of primary flux standard • Infrared excess due to circumstellar disk at > 5 mm • Pole on rapid rotator so also a bad fit to A0V in V-band • Normalized “Vega” A0V Kurucz model used instead • Much confusion in the literature on how to extrapolate from V band to infrared • Authors use conflicting “Vega” fluxes in infrared • Rieke et al. (2008) find offsets between instrumental calibration relative to MIPS 24 mm • Used A stars and solar analogs • Most of each offset can be attributed to different zero points • 2MASS 2% low, IRAC 1.5% low, IRAS 25 mm 2% high • Conflicts with Price et al. (2004) result which included emissive spheres and same model Vega

7. More Issues with Absolute Infrared Calibration • Uncertainty in truth of standards used to transfer photometric zero point • A0V standards • Can be bright (good for previous telescopes/spectroscopy) • Use some form of Kurucz model • Can have circumstellar disks in IR (10-15% of field stars) • Assume most are better behaved than Vega • KIII standards • Molecular absorption features in mid-IR (CO, SiO) complicate analysis • Solar analogs • ~2% variation between models • White Dwarfs • Haven’t been used in IR and are faint • Best current strategy is to use ensemble of calibrator types

8. Hope for the Future • Use of better models particularly replacing Engelke functions for l > 20 mm • Castelli and Kurucz (2004) • MARCS (Decin & Eriksson 2007) • Better templates through improved IR spectra (Engelke et al. 2006) • Reconciles 3.6 mm KIII derived flux conversion with AV derived flux conversion for IRAC • 7% discrepancy reduced to 1.5% • More observations of more standards • Joint HST/Spitzer calibration campaign (see later talks)

9. Status of Spitzer • First cycle of warm observations is almost complete • IRAC has been on for 345 days without incident • Current warm calibration accuracy is 3% • Flux conversions and various other calibrations varied slightly and as mostly as expected • Intra-pixel responsivity variations more significant • Have identified function form of responsivity variation in warm data and are now applying it to cryogenic data • Arrays are more non-linear • MIPS final reprocessing has finished • Final calibration for cryogenic IRAC is winding up • IRS closeout is progressing

10. Warm / Cryogenic IRAC Cross Calibration Warm 3.6 mm (top) Cryo 3.6 mm (bottom) Same functional form Different amplitude

11. Engelke / Cohen Comparison • IRTF SpecX data of NPM1p68.0422 (K2III) calibrator for IRAC • Red is ratio of spectra / Cohen template • Blue is ratio of spectra / Engelke template mm

12. IRAC AV / KIII Calibration Offset • In Reach et al. (2005) difference between Predicted/Observed between AV and KIII calibrators was 7.3%, 6.5%, 3.6% and 2.1% for 3.6, 4.5, 5.8 and 8.0 mm • Revised templates improve discrepancy • 4.5, 5.8 and 8.0 mm analysis in works

13. Spitzer Instrument Cross Calibration • Each instrument used different calibration methodology • No cross calibration requirement • IRAC used 4 AV stars as primary calibrators (Reach et al. 2005) and Cohen et al. templates and zero points using the “Vega” template verified by Price et al. (2004) • Zero points of 280.9, 179.7, 115.0 and 64.13 Jy. • A 3% absolute accuracy is quoted. • MIPS used 22 A stars, Ks – [24] = 0 and a “Vega” zero point (Engelbracht et al. 2007) • Zero point of 7.17 Jy • 4% absolute accuracy is quoted • IRS calibration is based on MARCS models of Decin et al. (2004). HR 7341 (K1III) is used as the primary standard. • 5% absolute accuracy is quoted in Instrument Handbook

14. Spitzer Cross Instrument Calibration • Gizis et al. (in prep) compared IRAC to IRS and IRS to MIPS • 8 mm photometry of the IRS calibrators, HR 7341, HR 2194 (A0V) and HR 6606 (G9III), to IRAC magnitudes inferred from IRS spectra and IRAC response function • Combination of SL1 and SL2 orders for IRS • The photometry for all three sources agrees to better than 1% • 24 mm photometry was compared to IRS synthesized photometry from the LL module for HR 2194 and HR 6348 (K1III) • IRS 2.2%±1.0% fainter than MIPS 24 mm • In agreement with Rieke et al. (2008) comparison of IRAC and MIPS

15. MIPSGAL Point Source Check • Color excess between predicted and observed for 20 AV and 7 KIII • 3 A stars have 24 mm excess • 24 mm 3% brighter • Similar result noted by SAGE legacy team

16. Archival Search for Non-linearities • To support the observations of JWST calibrators it is important to verify if possible that there are Spitzer observations are linear at low well depth • Difficult measurement to make • No indication that Spitzer arrays (InSb, Si:As, Si:Sb) exhibit count-rate non-linearity • Some indication from FEPS team (Carpenter et al. 2008) that IRAC and MIPS photometry is a function of frametime • Not monotonic in IRAC frametimes • Not repeated in IST tests of IRAC relative calibration with frametime

17. IRAC Si:As Extended Source Calibration • The 5.8 and 8.0 mm arrays exhibit significant internal scattering (and droop) resulting in extended sources having larger measured fluxes than they should • Cohen et al. (2007) and the IRAC IST independent measured this effect • Using HII regions and comparing 8.0 mm to MSX 8.3 mm • Extrapolating MSX 8.3 mm to IRAC 5.8 mm assuming a PDR like SED • Using elliptical galaxies and extrapolating 2MASS Ks to IRAC wavelengths • Measured effect is significant ~30% at 5.8 and 8.0 mm

18. IRAC Internal Scattering

19. MIPSGAL to MSX Extended Source • Compared MSX 21 mm to MIPSGAL 24 mm surface brightness for M16 • Data smoothed to 20 arcsec MSX resolution • Color corrected using model spectra for star forming ISM (Flagey 2007) • Correlation goes as the expected l1.5 for dust emission due to SFR

20. Calibration at Longer Wavelengths • Harder to do as stars are faint at l > 30 mm • Either need very sensitive telescope (JWST) • Or use diffuse emission (much larger uncertainty) • For MIPSGAL 70 mm needed to correct for significant non-linearity and gain offset • Used transformation from IRIS 60 mm data to MIPSGAL 70 mm • IRIS version of IRAS data tied to DIRBE (Miville-Deschênes & Lagache 2005) • Applied color correction using model of dust emission and 60/100 color • MIPSGAL 70 mm used as sanity check on PACS 70 mm maps from HiGal

21. MIPS 70 mm Galactic plane No gain correction Per stim-flash correction Global / IRIS correction

22. MIPSGAL 70 mm / HiGal 70 mm

23. What’s Next • Take more cross calibration data • Some IRAC observations of HST calibrators finished • Cryogenic IRAC for AKARI standards need analysis • AKARI spectra for IRAC calibrators need analysis • Warm IRAC observations planned of subset of DIRBE calibrators • HST observations of infrared KIII calibrators • Observations of standards tied directly to NIST standards • ACCESS and SNDICE for example • Need the calibration community to come up with a unified zero point concept • Possibly a calibration summit to resolve current differences