1. Evaluation of the Experimental and Theoretical Intensities of Water- Vapor Lines in the 2 µm Region Using Spectra from the Solar- Pointing FTS Iouli Gordon, Laurence Rothman, Lorenzo Lodi, Jonathan Tennyson, Geoffrey Toon, Linda Brown Columbus, OH, 21 June 2012

2. The 4500-5000 cm -1 region This spectral region is used for several purposes: Water vapor retrieval by ground-based FTIR spectrometers (NDACC, TCCON). OCO and GOSAT CO 2 nadir measurements using the 4800-4900 cm -1 region. One has to account for interfering water as well determine concentrations of water to estimate H 2 O broadening of CO 2 lines. MkIV balloon occultation retrievals of H 2 O, CO 2, CH 4 and N 2 O in the troposphere

3. 447-m WLEF tower FTS near Tall Tower in Park Falls WI Total Column Carbon Observing Network (TCCON)

4. Park Falls, WI Solar-pointing FTS Assessment of water vapor spectral parameters

5. Data in HITRAN and other sources The lines are from the 3ν 2, ν 1 +ν 2, ν 2 +ν 3 bands and they originate from the SISAM database: R.A. Toth, “Linelist of water vapor parameters from 500 to 8000 cm -1,” http://mark4sun.jpl.nasa.gov/h2o.html Contrary to a common believe HITRAN data from SISAM is largely a semi-empirical calculation, supplemented with experimental value whenever calculation was not available. Other experiments claim differences 3-10 % for stronger lines. Jenouvrier et al, “Fourier transform measurements of water vapor line parameters in the 4200-6600 cm -1 region,” JQSRT 105, 326- 355 (2007), Mikhailenko et al, “Water vapor absorption line intensities in the 1900–6600 cm -1 region,” JQSRT 109, 2687, (2008)

6. New ab initio calculations of water intensities (claimed unprecedented uncertainty for 90% of the lines) Lodi L., Tennyson J., Polyansky O. L., J Chem Phys, 135, 034113-10 (2011).

7. New ab initio calculation versus BT2 3ν23ν2 4700-5000 cm -1 region

8. Toth calc&exp vs ab initio

9. New line list Took HITEMP database with 10 -29 cm/molecule cutoff. HITEMP is a BT2 dataset with HITRAN data substituted in when possible as well as line positions generated from empirical energy levels. Substituted Lodi et al intensities into HITEMP Introduced line shifts for HDO from Jenovrier et al. Four different water vapor linelists were used in evaluation: ATM - kludged linelist, based on Toth 2003, used by MkIV & TCCON HIT - HITRAN 2008 NEW - a new linelist FUD - new linelist with some of the widths and shifts fudged

10. The 4500-4550 cm -1 region

11. The 4600-4650 cm -1 region

12. The 4850-4900 cm -1 region

13. RMS is the Root Mean Square variation of the residuals. VSF is the VMR Scaling Factor Summary of Spectral Fits

14. H 2 O VMR Scale factors The factor by which the H 2 O amount retrieved using a particular window & analyzed using a particular linelist compares with the overall mean. Absolute values are arbitrary. The new linelists (NEW, FUD) give larger H 2 O amounts than earlier linelists (ATM, HIT)

15. Conclusions The new linelist produces H 2 O amounts that are 0-5% larger than HITRAN, the exact amount depending on the interval. Individual lines will have larger excursions. In total, 95/2497 lines were adjusted based on poor residuals to atmospheric spectra to generate the fudged linelist.. The new intensities are better than HITRAN08 -- more consistent from line to line -- so that fitting windows with multiple lines you get smaller RMS residuals. In 10/10 spectral intervals, the new linelist gives better fits than HITRAN08

16. In 6/10 spectral intervals, the new linelist gives better fits than ATM. In 9/10 intervals the fudged new linelist (FUD) gives better fits than ATM. This analysis says nothing about the absolute accuracy of the intensities. The main problem with the new linelist is now the widths and shifts. Conclusions (continued)

17. Acknowledgements  OCO-2 ABSCO team  NASA