1. A database for water transitions from experiment and theory Jonathan Tennyson HITRAN meeting Department of Physics and Astronomy Harvard University College London June 2006 The Earth seen in water vapour by NASA’s GOES satellite

2. Why water vapour? Molecule number 1 in HITRAN Major (70%) atmospheric absorber of incoming sunlight Even H 2 18 O is fifth biggest absorber Largest (60%) greenhouse gas Atmospheres of cool stars Combustion Life !?

3. UCL strategy for a reliable, complete (300K) linelist Strong lines: water-air spectra, variable path-length Weak lines: water vapour spectra, longest path-length & integration times possible Isotopologues: Isotopically enhanced samples (Kitt Peak, CRDS) Completeness/assignments: High quality variational calculations

4. IUPAC Task group A database of water transitions from experiment and theory Water lines at room temperature (HITRAN) Hot water Isotopologues Line profiles Theory Validation Database Meet room P226 “Tea Room” Weds from 2.30 pm Thurs until lunch

5. transitions 0  30,000 cm . linelist for room temperature (C, 296 K) & hot (H) water. C complete for intensities > 10 –29 cm molecule –1 in natural abundance. Singly & doubly substituted isotopologues: HD 16 O, H 2 18 O, H 2 17 O, D 2 16 O, HD 17 O, and HD 18 O. No triply substituted isotopologues, no tritium. Line profiles: function form? Broadening parameters γ and δ. Dependence: pressure (0 – 3 atm), temperature (200 – 300 K) Experimental & computational data. Parameters for self-, N 2, O 2, air, and H 2 broadening. Scope

6. Database Master database to be prepared for each isotopologue. Should capture origin & time-dependence of measured and computed values. Both ‘old’ and ‘new’ data archived and accessible. Flexible in terms of data structures HITRAN “button”

7. Master file strategy Use most complete (not necessarily best) as Master file eg BT2 Augment with data from other sources: expt, other theory Store all known data: use error analysis to combine Clear data history Files structured by function: levels, transitions (+ mixings?) Distributed data? Some functionality in-built eg HITRAN button

8. 50,000 processor hours. Wavefunctions > 0.8 terabites 221,100 energy levels (all to J=50, E = 30,000 cm  ) 14,889 experimentally known 506 million transitions (PS list has 308m) >100,000 experimentally known with intensities  Partition function 99.9915% of Vidler & Tennyson’s value at 3,000K New BT2 linelist Barber et al, Mon. Not. R. astr. Soc. 368, 1087 (2006). http://www.tampa.phys.ucl.ac.uk/ftp/astrodata/water/BT2/

9. Comparison with Experimental Levels BTPS Agreement:% Within 0.10 cm -1 48.759.2 Within 0.33 cm -1 91.485.6 Within 1 cm -1 99.292.6 Within 3 cm -1 99.996.5 Within 5 cm -1 100.097.0 Within 10 cm -1 100.098.1 Number of Experimental Levels: 14,889

10. 1715470339003.8927003.799 2000.092 4.01E-032.78E-223.89E-04 6.71E-01 1303831179098.5307098.116 2000.415 1.56E-031.01E-221.41E-04 5.59E-01 17084604710486.1388485.481 2000.657 4.69E-021.12E-211.56E-03 7.84E+00 16077614510939.5328938.685 2000.848 4.83E-038.33E-231.16E-04 9.34E-01 161115154407.2212406.299 2000.922 2.77E-025.25E-207.34E-02 5.35E+00 060165054407.3552406.297 2001.058 3.26E-022.06E-202.88E-02 6.30E+00 14160404611384.2459383.183 2001.062 6.66E-038.35E-231.17E-04 1.86E+00 16078706010955.9148954.726 2001.188 1.69E-022.88E-224.03E-04 3.27E+00 071197096034.9924033.695 2001.297 7.29E-041.43E-222.00E-04 1.22E-01 151104507512912.87110911.526 2001.344 3.36E-021.40E-221.96E-04 7.68E+00 Raw spectra from DVR3D program suite

11. ABC D EFGH IJK 43432111508730.1369980211138 43433111518819.7739620401166 43434111528918.53621500211210 43435111538965.4961300211156 43436111548975.1451752001148 43437111559007.8688941011138 43438111569082.4138911201166 43439111579170.3438711011156 43440111589223.4441580021148 43441111599264.4898152001166 43442111609267.08831605011210 43443111619369.8877220211174 43444111629434.0025470401184 43445111639457.2726551011174 43446111649498.0127280021166 43447111659565.8900231201184 Energy file : N J sym n E/cm -1 v 1 v 2 v 3 J K a K c

12. 1448481461833.46E-04 1153091085207.42E-04 1960181984131.95E-04 703177031.13E-02 1491761501231.69E-04 81528787342.30E-01 80829782378.83E-04 2096722108762.51E-01 2070262032412.72E-04 1889721849711.25E-01 1524711533991.12E-02 39749374791.46E-07 10579158826.90E-05 34458356171.15E-03 Transitions file: N f N i A if 12.8 Gb Divided into 16 files by frequency For downloading

13. Master file strategy: Inclusion of Experimental (+ other theoretical) data Added to record. Data classified: Property of level  Energy File Experimental levels (already included) Alternative quantum numbers (local modes) Property of transition  Transition File Measured intensities or A coefficients Line profile parameters Line mixing as a third file? Location of partition sums?

14. AuthorPotenialE max /cm -1 J max transitionsavaiable H 2 16 O BT2Shirin (2003)3000050   H 2 17 O ShirinFIS3 (2006)26000?10   H 2 18 O ShirinFIS3 (2006)26000?10   HDO TashkunPS (1997)26000?  ? D 2 16 O ZobovShirin (2004)1400030   HD 18 O   HD 17 O   Linelists available for Master databases

15. Main characteristics (poster by Attila Csaszar) Dual database of rovibrational energy levels and rovibrational transition with well-defined uncertainties Complete collection and storage of all relevant spectroscopic data for all major isotopologues of water Critical evaluation of data which will always carry their own pedigree (e.g., bibliographical references, important measurement conditions, metadata) Inclusion of intensities, line widths, and line broadenings in the database, possibly including refinement of relevant parameters Global multi-dataset optimization Curation, organizational, data-mining and displaying tools Allow immediate (and automatic) consistency analysis of newly reported data before data deposition Allow „experiments” with what-if scenarios (important in order to predict what extra information new experiments might provide All supporting programs written in C++ and Java Sensitivity analysis of uncertainties Reproduce all known and well-defined experimental data (time-dependence) Predictions are rigorously quantified by their respective uncertainty bounds Minimal chance of leaving feasible regions of parameters HITRAN „button” to produce the best available data in HITRAN form for modeling studies

16. IUPAC Task group A database of water transitions from experiment and theory MEMBERS: Peter Bernath (Waterloo, Canada); Alain Campargue (Grenoble, France); Michel Carleer (Brussels, Belgium); Attila Császár (Budapest, Hungary); Robert Gamache (Lowell, U.S.A.); Joseph Hodges (NIST, U.S.A.); Alain Jenouvrier (Reims, France); Olga Naumenko (Tomsk, Russia); Oleg Polyansky (Ulm, Germany); Laurence Rothman (Harvard, U.S.A.); Jonathan Tennyson (London, U.K.); Robert Toth (JPL, U.S.A.); Ann Vandaele (Brussels, Belgium); Nikolai Zobov (Nizhny Novgorod, Russia)

17. Spectroscopy of water Nikolai ZobovBob Barber Boris Voronin Roman Tolchenov Lorenzo Lodi Paolo Barletta

18. www.worldscibooks.com/physics/p371.html

19. Labelling BT2 energy levels J=25J=26J=27J=28J=29J=30J=31 eeeeeee 8888888 6,171.5956,647.0597,139.127,647.6508,172.4878,713.4839,270.484 7,026.7167,533.3698,055.928,594.1689,147.8729,716.78710,300.639 7,715.4498,187.4398,675.839,180.8339,702.38610,240.43510,794.934 7,729.1468,262.5548,811.359,375.0529,953.43110,546.25511,153.286 8,297.7718,860.8009,437.2810,027.24310,630.63011,247.23111,876.321 8,668.2329,174.1059,695.1510,231.12810,781.64511,346.33011,925.255 8,679.0399,278.6219,892.8510,519.90711,158.55811,773.13712,333.738 9,041.3869,628.65410,200.5910,706.82111,230.92011,808.10512,468.284 9,240.9609,712.02410,236.4010,864.54811,512.63812,179.57712,748.026 9,417.4299,951.83010,500.8011,064.10111,641.40312,201.87112,836.462 9,560.23210,139.88010,658.5011,157.11111,671.77712,233.01312,865.287 9,709.82610,176.03110,737.0811,283.24511,799.69512,332.02312,880.080 9,830.64710,298.57510,782.8211,351.59811,983.27712,632.01413,297.793 10,003.35810,570.61311,150.0211,741.33912,341.64812,919.54613,487.796 10,147.81610,728.50311,301.8811,831.22312,376.62712,969.28213,596.429 10,283.87210,786.03011,325.9111,939.68712,569.51213,186.45513,758.743 10,380.25510,978.41411,558.7412,086.27412,628.90013,215.06013,876.022 10,549.89111,046.59411,593.3812,206.87512,751.21013,310.61913,884.951

20. J=31o8J=32o8 9270.484000311 9843.328000321 10300.6390003132910899.13200032330 10794.935010311 11365.856010321 11153.2930003152711774.27700032528 11876.8390003172512507.25801032330 11925.2770103132912528.73400032726 12333.738020311 12912.970020321 12487.1330003192313151.95200032924 12748.027100311 13309.624100321 12837.1080103152713443.559001320 12880.079001310 13454.79501032528 13019.63900031112113697.050000321122 13487.8230203132914067.28502032330 13574.23900031131914236.72801032726 13598.8670103172514253.538000321320 13758.7461003132914345.56110032330 13884.9510013122914473.88700132230

21. Room temperature H 2 16 O lines Strong line data about 9000 cm -1 Compatability between mid and near infrared intensities Weak lines throughout whole spectrum Far infrared? Solution strategy largely experimental plus careful analysis?

22. Hot water (up to T=3000+ K) New complete linelist available from UCL Accuracy? Experimental assignments New experiments? H 2 16 O only? (Some experiment for HDO and D 2 O) Line profiles? Solution strategy: largely theoretical with validation by experiment

23. Isotopologues H 2 18 O, H 2 17 O, HDO lines patchy in visible D 2 16 O not well known above 10000 cm -1 Any interest in other isotopologues? Room T only? Line profiles? Solution strategy Isotopically enhanced experiments

24. Line profiles Broadening by which species? water, O 2, N 2, air, H 2,…..? T dependence? P dependence? (up to 10 atm?) Solution strategy Theory validated by high quality experiment?

25. Validation between experiments atmospheric spectra Theory vs experiment other

26. Distribution and storage HITRAN Web database eg Spectroscopic databank at Tomsk Publication or other means of distribution?

27. So what is the problem? Water is well studied (30,000+ lines in HITRAN) But Water spectra have huge dynamic range Difficult to work with experimentally Spectra very dense: baseline hard to characterise Strong lines usually saturated (water-air spectra) Line profiles important (water-air & water-water) Weak lines can be significant (pure water spectra) Line assignment difficult (Variational Methods)

28. P. Macko, D. Romanini, S. N. Mikhailenko, O. V. Naumenko, S. Kassi, A. Jenouvrier, Vl. G. Tyuterev and A. Campargue, J. Molec. Spectrosc. (in press).

29. P. Macko, D. Romanini, S. N. Mikhailenko, O. V. Naumenko, S. Kassi, A. Jenouvrier, Vl. G. Tyuterev and A. Campargue, J. Molec. Spectrosc. (in press).

30. P. Dupre, T. Germain, A. Campargue, N.F. Zobov, O.L. Polyansky, S.V. Shirin, R.N. Tolchenov and J. Tennyson, J. Molec. Spectrosc. (to be submitted).

31. Polyad structure in water absorption spectrum Long pathlength Fourier Transform spectrum recorded by R Schmeraul

32. R. Schermaul, R.C.M. Learner, J.W. Brault, A.A.D. Canas, O.L. Polyansky, D. Belmiloud, N.F. Zobov and J. Tennyson J. Molec. Spectrosc., 211, 169 (2002).

33. Weak lines: new experimental measurements MSF data (NERC) : 8m cell, pure water vapour Schermaul, Learner et al. Bruker F.T.S. Range : 9000-12 700 cm -1 T : 295.7 K p(H 2 O) : 22.93 hPa pathlength ~ 800.8 m Number of lines : 7923 Number of new lines : 1082 Schermaul, Learner et al. Bruker F.T.S. Range :11 700-14 750 cm -1 T : 294.4 K p(H 2 O) : 23.02 hPa pathlength ~ 800.8 m Number of lines : 5316 Number of new lines : 1534 Weak water linesVery difficult to record Only a few weak lines in HITRAN Also data in 6000 - 9000 cm -1 region

34. Weak lines: new experimental measurements REIMS data, 50 m cell, pure water vapor (also water-air) Coheur et al., Fally et al. Bruker F.T.S Range :13 000 - 25 020 cm -1 T : 291.3 K p(H 2 O) : 18.32 hPa pathlength ~ 602.32 m Number of lines: 9353 Number of new lines : 2286 Merienne et al. Bruker F.T.S Range : 9 250 - 13 000 cm -1 T : 292 K p(H 2 O) : 23.02 hPa pathlength ~ 602.32 m Number of lines : 7061 Number of new lines : small HDO ! Full assignment nearly complete

35. Water vapour spectrum: new assignments in the blue Long pathlength FTS M. Carleer, A. Jenouvrier, A.-C. Vandaele, P.F. Bernath, M.F. Marienne, R. Colin, N.F. Zobov, O.L. Polyansky, J. Tennyson & V.A. Savin J. Chem. Phys., 111, 2444 (1999)

36. MSF spectra : line parameter retrieval using GOBLIN Residue of fit residue from fit

37. Intensity comparison for weak lines: MSF vs Rheims

38. Reliable intensities required for satellite retrievals MSF data (ESA) : 8m cell, water-air spectra Schermaul, Learner, Brault, Newnham et al. Bruker F.T.S. Range : 9000  12 700 cm -1 T : 295.7 K (also 253 K) p(H 2 O) : 10.03 hPa Pathlength: SPAC 4.938 m LPAC 32.75m, 128.75m, 512.75m Number of lines : 7923 Number of new lines : 1082 See poster by Tolchenov

39. Intensity data compared to HITRAN-96 by polyad for spectral region 8500 – 15800 cm -1 HITRAN underestimates intensity of strong lines! D Belmiloud et al, Geophys. Res. Lett., 27, 3703 (2000). Numbers are ratio of total intensity to Hitran96 PolyadIntegrated absorbance Spectral linefits Ab Initio calculation Correction Giver et al.  1.261.310.92  1.191.211.041.14  1.261.25 1.09  1.061.040.96

40. Intensity comparison for strong lines: ESA vs Hitran 2000 Comparison with data of Brown et al (2002) Comparison with data from Hitran 96

41. ESA spectra: line parameter retrieval 129m water-air spectrum residue of fit Still problems with fit See poster...........

42. Validation using atmospheric spectra Atmospheric spectra due to Newnham & Smith (RAL)

43. Water isotopmers in the visible Fourier transform spectra in Kitt Peak archive up to 15 000 cm -1 H 2 18 O: M. Tanaka, J.W. Brault and J. Tennyson, J. Molec. Spectrosc., 216, 77 (2002). H 2 17 O: M. Tanaka, O. Naumenko, J.W. Brault and J. Tennyson to be published Cavity ringdown spectra from Amsterdam about 17 000 cm -1 H 2 18 O: M. Tanaka, M. Sneep, W. Ubachs & J. Tennyson, J. Molec. Spectrosc., 226, 1 (2004). H 2 17 O: being analysed at UCL HDO: Brussels/Rheims spectra of Coheur et al being analysed in Tomsk

44. Missing absorption due to water: First estimates Theory Experiment Radiative Transfer Model Atmospheric absorption  In the red and visible :  Unobserved weak lines have a significant effect : ~ 3 Wm -2  Estimated additional 2.5-3 % absorption in the near I.R/Red.  Estimated additional 8-11 % absorption in the ‘Blue’ ?  Underestimate of strong lines even more important : ~ 8 Wm -2  Estimated additional 8 % absorption in the near I.R/Red.

45. Missing absorption due to water: Outstanding issues  In the near infrared and red:  Contributions due to H 2 18 O, H 2 17 O and HDO.  Possible role of water dimer (H 2 O) 2.  In the blue and ultraviolet:  Are H 2 16 O line intensities also underestimated?  Contribution due to weak lines