Water Pressure Broadening: A Never-ending Story

Abstract

Numerous papers on pressure broadening of water have been published, however, with differences exceeding stated uncertainties. The present work attempts to initiate a process towards better defined uncertainties. Self- and airbroadened water vapour spectra were recorded with a Bruker IFS120HR in the range from 1250 to 1750 cm^-1. 45 measurements covered the temperature range 208 – 316 K, air pressures from 0 – 1000 mbar, sample pressures 0.0004 – 5 mbar, and absorption paths 0.16 – 84 m. To allow for quality assurance, care was taken to cover most lines by several measurements with different opacity and linewidth. Furthermore, the data cover five orders of magnitude in intensity. Transmittance spectra were corrected for detector non-linearity and channeling. Line-by-line analysis was done on a microwindow basis resulting in positions, linestrengths and linewidths using the Voigt function as monochromatic lineshape. In the line fitting, especially of the 50 mbar measurements, the residual spectra show systematic deviations due to Dicke narrowing which are mostly smaller than 1%. The further data reduction was done with a new software tool which reads the broadening data and fits air and self pressure broadening parameter and temperature exponent. The software compiles the experimental data for each transition and selects the parameters to be fitted. The software also allows for various data quality assessments which are presented in this talk. This quality assessment revealed that 50 mbar measurements had to be excluded from the data analysis due to the Voigt profile being not capable of yielding pressure- and opacity-independent broadening parameters for these measurements. An important quality assessment is the intercomparison with hot cell measurements which were obtained a few years ago using a different cell, beamsplitter, detector, higher air pressure (500, 1000 mbar) and a different flow setup. The line broadening agreed within 1% on average. Finally, the differences of measured and calculated spectra were investigated showing almost pure instrumental noise and thus indicating that the data reduction is not introducing systematic errors. In total the number of air broadening parameters is 985 (ν₂: 44% H₂O, 17% H₂¹⁸O, 10% H₂¹⁷O, 12% HDO, 2ν₂-ν₂ H₂O: 17%) and of temperature exponents is 672. The resulting data show that in contrast to ozone the pressure broadening cannot be expressed in polynomials in the quantum numbers. Intercomparison with HITRAN 2004 and other authors, especially Toth, will be discussed.