1. Jet Propulsion Laboratory California Institute of Technology 1 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Near Infrared CO 2 Spectral Database Charles E. Miller, Linda R. Brown, and Robert A. Toth Jet Propulsion Laboratory, California Institute of Technology, 4800 Oak Grove Dr., Pasadena, California 91109 D. Chris Benner, V. Malathy Devi The College of William and Mary, Box 8795, Williamsburg, Virginia 23187-8795, U.S.A Acknowledgments The research at the Jet Propulsion Laboratory (JPL), California Institute of Technology, was performed under contract with National Aeronautics and Space Administration. We thank NASA’s Upper Atmosphere Research Program for support of the McMath-Pierce laboratory facility. CEM thanks NASA’s Tropospheric Chemistry and Atmospheric Composition programs for support. The material presented in this investigation is based upon work supported by the National Science Foundation under Grant No. ATM-0338475 to the College of William and Mary. The authors express sincere appreciation to M. Dulick of NOAO (National Optical Astronomy Observatory) for the assistance in obtaining the data. We also thank Gregory DiComo for assistance in setting up the multispectrum solution.

2. Jet Propulsion Laboratory California Institute of Technology 2 10 th HITRAN Database Conference Cambridge MA June 22, 2008 According to Herzberg… “The spectrum of carbon dioxide has been studied exhaustively by a large number of investigators.” The Spectrum of CO 2 Below 1.25  J. Opt. Soc. Am. 43, 1037 (1953)

3. Jet Propulsion Laboratory California Institute of Technology 3 10 th HITRAN Database Conference Cambridge MA June 22, 2008 According to Herzberg… “The spectrum of carbon dioxide has been studied exhaustively by a large number of investigators.” The Spectrum of CO 2 Below 1.25  J. Opt. Soc. Am. 43, 1037 (1953) Toth et al., JQSRT 109, 906 (2008) Toth et al., J. Mol. Spectrosc. 246, 133 (2007) Malathy Devi et al., J. Mol. Spectrosc. 245, 52 (2007) Toth et al., J. Mol. Spectrosc. 243, 43 (2007) Malathy Devi et al., J. Mol. Spectrosc. 242, 90 (2007) Toth et al., J. Mol. Spectrosc. 239, 243 (2006) Toth et al., J. Mol. Spectrosc. 239, 221 (2006) Miller et al., CR Physique 6, 876 (2005) Miller et al., J. Mol. Spectrosc. 228, 329 (2004) Miller et al., J. Mol. Spectrosc. 228, 355 (2004)

4. Jet Propulsion Laboratory California Institute of Technology 4 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Return global X CO2 data with 0.3% precision Miller et al., JGR 112, D10314 (2007)

5. Jet Propulsion Laboratory California Institute of Technology 5 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Remote Sensing of GHGs at the Sub-1% Level Challenges Spectroscopic Databases CO CO 2 O O2O2 Column Abundance Path Dependent X CO2 Path Independent Mixing Ratio Measured Spectra Ratio

6. Jet Propulsion Laboratory California Institute of Technology 6 10 th HITRAN Database Conference Cambridge MA June 22, 2008 How well can we retrieve CO 2 ? - Circa 1990 Wallace & Livingston, J. Geophys. Res. D 95, 9823 (1990) Wallace and Livingston’s seminal work on CO 2 remote sensing [1990] with the Kitt Peak FTS revealed deficiencies in the CO 2 spectral database ( HITRAN 1986). Insufficient NIR Spectroscopic Reference Standard Accuracy 1. Incomplete knowledge of spectrum 2. Inadequate position knowledge 3. Intensities known to 5 – 20% unc. 4. Unvalidated air-widths 5. No pressure shifts

7. Jet Propulsion Laboratory California Institute of Technology 7 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Improved Solar Spectra Retrievals circa 2002 Kitt Peak solar data reanalyzed Improved retrieval algorithm Improved HITRAN 2000 database ( HITRAN 1992 + CO 2 DND list ) Results Systematic residuals in spectra +5.8% bias between observed and in situ column amounts 0.5% precision in column CO 2 "Remaining errors are dominated by deficiencies in the spectroscopic line lists" Yang et al., Geophys. Res. Lett. 29(10) GL014537 (2002)

8. Jet Propulsion Laboratory California Institute of Technology 8 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Washenfelder et al. (2006) Park Falls WI The TCCON Prototype Washenfelder et al., J Geophys. Res. 111 D22305 (2006) Ideal 1:1 Line Uncorrected New data acquisition hardware and methodology (based around Bruker 125 HR) Results 0.1% XCO 2 precision Systematic residuals persist +2.12% bias for 30013 +2.40% bias for 30012 “Systematic differences attributed to known uncertainties in the CO 2 line strengths and pressure broadened widths”

9. Jet Propulsion Laboratory California Institute of Technology 9 10 th HITRAN Database Conference Cambridge MA June 22, 2008 CO 2 Nomenclature Vib. Band Notation follows the HITRAN convention ABCDE where A = No. v 1 quanta B = No. v 2 quanta C = v 2 vib ang mom D = No. v 3 quanta E = 1 : normal E  1: Fermi res. Isotopomer Nomenclature: 16 O 12 C 16 O  626 16 O 13 C 16 O  636 16 O 12 C 18 O  628 16 O 12 C 17 O  627 16 O 13 C 18 O  638 16 O 13 C 17 O  637 18 O 12 C 18 O  828 18 O 12 C 17 O  827

10. Jet Propulsion Laboratory California Institute of Technology 10 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Kitt Peak FTS used for lab studies

11. Jet Propulsion Laboratory California Institute of Technology 11 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Improving Laboratory Accuracies Requires Precise Knowledge/Control of the Experimental State Pristine new cells – no contamination Temperature monitoring inside the cell Isotopic enriched samples Mass spectrometric standard samples Stable spectrometer performance Goal for Experimental Uncertainties: Pressure:  0.01 Torr (if P > 10 Torr) Temperature:  0.1 K Path:  2 mm (0.1%) Composition :  0.05% SNR:>1000 Resolution:0.011 cm -1 100% Trans:  0.1% 0% Trans:  0.1% Positions:  0.0001 cm -1 Intensity :  0.1% (Relative) Four Temp Probes (PRT) going Inside the Cell

12. Jet Propulsion Laboratory California Institute of Technology 12 10 th HITRAN Database Conference Cambridge MA June 22, 2008 1. Determining the Complete Spectrum Linear Log Accurate CO 2 remote sensing to 0.3% requires knowledge of all absorption features that contribute to the CO 2 absorption spectrum at the level of approximately 0.1% of I max Examination of the known NIR CO 2 features on a LOG scale shows that transitions from many weaker bands contribute detectable absorption to the spectrum Completeness will be a critical requirement for the spectral database Simulations from HITRAN04

13. Jet Propulsion Laboratory California Institute of Technology 13 10 th HITRAN Database Conference Cambridge MA June 22, 2008 1. Determining the Complete Spectrum Path = 97 m Pres = 2.06 Torr Temp= 294 K C 2 H 2 in 2nd cell to calibrate line positions 30011 30012 30013 30014 Miller & Brown, J. Mol. Spectrosc. 228, 329 (2004) 16 O 12 C 16 O = 626

14. Jet Propulsion Laboratory California Institute of Technology 14 10 th HITRAN Database Conference Cambridge MA June 22, 2008 626 626: 15% 16 O 12 C 16 O 628: 48% 16 O 12 C 18 O 828: 33% 18 O 12 C 18 O 2ν32ν3 1.Determining the Complete Spectrum: Characterize Isotopologue Transitions In natural CO 2 16 O 12 C 18 O < 0.4 % 18 O 12 C 18 O < 0.0004 % Note: 628 has 2 x more lines than symmetric isotopologues (626, 828) due to different spin statistical weights. The 2 ν 3 band of 628 is allowed, but not for 626, 828. Note: These 828 bands are not in HITRAN 2004 Toth et al., J. Mol. Spectrosc. 243, 43 (2007)

15. Jet Propulsion Laboratory California Institute of Technology 15 10 th HITRAN Database Conference Cambridge MA June 22, 2008 1.Determining the Complete Spectrum: Characterize Isotopologue Transitions Toth et al., J. Mol. Spectrosc. 243, 43 (2007) 828628 626 626: 15% 628: 48% 828: 37% The region below 6920 cm -1 would be transparent in models neglecting 18 O species Note: These 828 lines are not in HITRAN 2004

16. Jet Propulsion Laboratory California Institute of Technology 16 10 th HITRAN Database Conference Cambridge MA June 22, 2008 2. Improved Line Positions Absolute Uncertainties < 0.0001 cm -1 Line position differences of the experimentally measured line positions of Miller & Brown and Vander Auwera et al. Miller & Brown, J. Mol. Spectrosc. 228, 329 (2004)  = 5x10 -5 cm -1

17. Jet Propulsion Laboratory California Institute of Technology 17 10 th HITRAN Database Conference Cambridge MA June 22, 2008 3. Measured line intensities of 125 Bands Retrievals : Voigt line shape & line-by-line fitting of individual spectra % Differences between HITRAN 2004 and new band strengths Toth et al., J. Mol. Spectrosc. 243, 43 (2007) 21 bands of 628 8 bands of 627 25 bands of 828 626 628 Toth et al. J. Mol. Spectrosc. 239, 221 (2006) Reported 58 band strengths of 626

18. Jet Propulsion Laboratory California Institute of Technology 18 10 th HITRAN Database Conference Cambridge MA June 22, 2008 3. Measured line intensities of 125 Bands Toth et al. J. Mol. Spectrosc. 239, 221 (2006) 626 Intensities for NIR CO 2 bands from multiple laboratories agree at the sub-1% value A more accurate intercomparison requires specific line shape specification –Speed dependence –Line mixing

19. Jet Propulsion Laboratory California Institute of Technology 19 10 th HITRAN Database Conference Cambridge MA June 22, 2008 4. & 5. Self-broadened widths and pressure-shifts 15 bands of 626 Toth et al., J. Mol. Spectrosc. 239, 243 (2006) Self- Widths Note vibrational dependence Self- Shifts Fermi Triad and ν 2 +2 ν 3 4700 – 5400 cm -1 Fermi Tetrad and 3 ν 3 6000 – 7000 cm -1 (in cm -1 /atm) m = J" for P branch, J"+1 for R branch

20. Jet Propulsion Laboratory California Institute of Technology 20 10 th HITRAN Database Conference Cambridge MA June 22, 2008 4. & 5. Air-broadened widths and pressure-shifts 626 Toth et al., J. Mol. Spectrosc. 246, 133 (2007) Air Widths Note vibrational dependence Air- Shifts Fermi Triad and ν 2 +2 ν 3 4700 – 5400 cm -1 Fermi Tetrad and 3 ν 3 6000 – 7000 cm -1 (in cm -1 /atm) m = J" for P branch, J"+1 for R branch

21. Jet Propulsion Laboratory California Institute of Technology 21 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Validate lab results with atmospheric data Top trace: HITRAN 2004 Right trace: Current Best line list JPL MkIV (G. Toon) 29 km Tangent Height Observed and calculated balloon-based FTS spectra

22. Jet Propulsion Laboratory California Institute of Technology 22 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Small Changes in Widths Affect Retrievals at High Airmass Test Line List A Test Line List B

23. Jet Propulsion Laboratory California Institute of Technology 23 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Accuracy of ± 0.3% using new Voigt line list Differences Between In Situ and FTS Column CO 2 (Ground- and Balloon-based) [After Sen et al., 2006 HITRAN Conference] Bruker 125 HR:Park Falls Z min : 0.47 km; SZA: 38.7  MkIV (JPL) Z min : 31.65 km; SZA: 91.7  Region Used HITRAN 2004 JPL 2008 HITRAN 2004 JPL 2008 2.1  m 9%+0.3%7%+0.2% 1.6  m 4%-0.1% 1.58  m -1%-0.3% Precision ~0.1% [Washenfelder et al. 2006]

24. Jet Propulsion Laboratory California Institute of Technology 24 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Active Remote Sensing of CO 2 Requires Even Greater Line Shape Accuracy Candidate transition: R(30) of 20013  00001 @ 2050.967 nm (4875.748 cm -1 ) ASCENDS ASCOPE GOSAT-II P = 269.03 Torr L = 0.347 m T = 297.04K.

25. Jet Propulsion Laboratory California Institute of Technology 25 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Improved Multispectrum Fitting [Benner et al., JQSRT 53, 705 (1995)] Fit all lines and spectra simultaneously Use quantum mechanical constraints for positions and intensities Increases sensitivity to subtle effects in line shapes Updated capabilities include non-Voigt line shapes, line mixing, speed dependence (Benner et al., in preparation) Line Positions: n i = n 0 + B(J(J+1)) + D(J(J+1)) 2 + H(J(J+1)) 3 + … n i resonant frequency n 0 band origin B, D, H rotational constants Jrotational quantum number Line Shape Parameters:  i = a 1 + a 2 m + a 3 m 2 +a 4 m 3 + ….. Measured half-width at half-max at each line position Line Intensities: S i = ( i / 0 )(S v /L i ) exp(-hcE i ″/kT)[1-exp(hcv i /kT)].F S i,observed individual line intensity S v vibrational band intensity, L i Hönl-London factor, where l i = (m 2  l″ 2 )/|m| for CO 2 m = J″+1 for the R branch, m =  J″ for the P branch J″lower-state rotational quantum number. langular momentum quantum number. Q r lower state rotational partition function at T0=296 K E i ″ lower state rotational energy F Herman-Wallis factor = [1+A 1 m+A 2 m 2 +A 3 m 3 ]

26. Jet Propulsion Laboratory California Institute of Technology 26 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Line Shape Problems! Line Mixing Occurs in CO2 P and R Branches Miller et al. Comptes Rendus Physique 6 (2005) 876-887.

27. Jet Propulsion Laboratory California Institute of Technology 27 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Multispectral Fitting of the 30012 Spectrum Malathy Devi et al. J. Mol. Spectrosc. 242, 90 (2007).

28. Jet Propulsion Laboratory California Institute of Technology 28 10 th HITRAN Database Conference Cambridge MA June 22, 2008 CO 2 Line Mixing Coefficients Line mixing observed at 6220 cm  1 even though this band has no Q-branch, no perturbations and adjacent lines are spaced by ~ 1 cm  1 Rosenkranz Off diagonal relaxation matrix

29. Jet Propulsion Laboratory California Institute of Technology 29 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Line Mixing & Speed Dependence Observed for Self- and Air-broadened Spectra

30. Jet Propulsion Laboratory California Institute of Technology 30 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Conclusions Accurate remote sensing of CO 2 is critical for climate change science CO 2 remote sensing poses a significant spectroscopic and algorithm challenge –This is NOT YET a solved problem Consideration of strong 16 O 12 C 16 O (626) transitions alone is insufficient –Must include hot bands –Must include 16 O 13 C 16 O (636), 16 O 12 C 18 O (628), etc Line shape choice is crucial to simulate high quality spectra within their experimental uncertainty –Non-Voigt line shapes improve fits 30% - 50% vs Voigt fits –Line Mixing is needed to remove systematic residuals

31. Jet Propulsion Laboratory California Institute of Technology 31 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Kitt Peak Co-Conspirators Mike Dulick (KPNO) $$$ NASA, NSF Chris Benner (W&M) Malathy Devi (LaRC) Not shown Linda Brown Bob Toth (JPL)

32. Jet Propulsion Laboratory California Institute of Technology 32 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Backup

33. Jet Propulsion Laboratory California Institute of Technology 33 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Isotopic Fractionation in Martian CO 2 0.2% precision desired PFS/Mars Express (2004) Grassi et al., Planet. Space Sci. 53, 1017 (2005) Measured & modeled PFS/Mars spectra ********

34. Jet Propulsion Laboratory California Institute of Technology 34 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Unanticipated Behavior for High-J Transitions High-J transitions may show large (>10 -4 cm -1 ), unexpected deviations from their predicted positions due to –Poor spectroscopic parameter extrapolations –Perturbations not observed at low-J Rare isotopologues and hot bands are especially susceptible to these problems since they are much more difficult to characterize accurately 636 This work – R92 638 626 Miller & Brown, J. Mol. Spectrosc. 228, 329 (2004) Miller et al., J. Mol. Spectrosc. 228, 355 (2004)

35. Jet Propulsion Laboratory California Institute of Technology 35 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Uncharacterized High-J Perturbations May Lead to Gross Retrieval Errors Short scans of CO 2 covering the perturbed R74, R76, R78 and R80 lines in the 20012-00001 band of 626. The calculated positions refer to unperturbed locations calculated from parameters derived from lower J transitions. Toth et al., J. Mol. Spectrosc. 239, 221 (2006)

36. Jet Propulsion Laboratory California Institute of Technology 36 10 th HITRAN Database Conference Cambridge MA June 22, 2008 Filling the 2 um Atmospheric Window (1/2) 636 626 NEW CO C2H2C2H2 C2H2C2H2 13 CO 2 constitutes only ~1% of the natural CO 2 Isotopic substitution shifts the band centers in the Fermi triad region such that the 13 CO 2 bands effectively fill the 2 um (5000 cm -1 ) atmospheric windows –Significant radiative impact under saturated absorption conditions The allowed 2v 3 band of 638 (NEW) is seen in the 4300 – 4700 cm -1 window