Measurement of H2O Broadening of O2 A-Band Transitions and

Mar 27, 2012 - Department of Chemistry and Biochemistry, James Madison University, MSC 4501, Harrisonburg, Virginia 22807, United States. ‡. Materia...
9 downloads 0 Views 863KB Size
Download PDF
Article pubs.acs.org/JPCA

Measurement of H2O Broadening of O2 A-Band Transitions and Implications for Atmospheric Remote Sensing E. M. Vess,† C. J. Wallace,† H. M. Campbell,† V. E. Awadalla,† J. T. Hodges,‡ D. A. Long,‡ and D. K. Havey*,† †

Department of Chemistry and Biochemistry, James Madison University, MSC 4501, Harrisonburg, Virginia 22807, United States Material Measurement Laboratory, National Institute of Standards and Technology, 100 Bureau Drive, MS 8320, Gaithersburg, Maryland 20899, United States

‡

ABSTRACT: We present laboratory measurements of H2O-broadened 16O2 A-band (b1Σ+g ← X3Σ−g (0,0)) absorption spectra acquired with a laser-based photoacoustic spectroscopy method. This absorption band is widely used in a variety of high-precision atmospheric remote sensing applications. We report H2O broadening parameters for six of the strongest transitions in this band, and we show that these measured values are nominally 1.5−2 times greater than the corresponding air-broadening parameters. Simulations of atmospheric transmission spectra in the O2 A-band that incorporate our measured H2O broadening parameters indicate that H2O present at concentrations typically found in the Earth’s atmosphere can influence the columnintegrated transmission relative to the dry air case. Further, because of spatial and seasonal variations in humidity, failure to account for the enhanced H2O pressure broadening effects can lead to concomitant biases in atmospheric O2 A-band retrievals of quantities such as surface pressure and path length in greenhouse gas retrievals. Robichaud et al.8,9 used the frequency-stabilized cavity ringdown spectroscopy (FS-CRDS) technique to measure O2 Aband P-branch lines for self- and air-broadened cases over the range of angular momentum quantum number J ≤ 32. They determined line intensities as well as broadening, collisional narrowing, and pressure shifting parameters based on fits of Galatry profiles to the measured line shapes. Consistent with earlier measurements by Ritter and Wilkerson,10 these and additional FS-CRDS measurements summarized by Long et al.11 firmly establish that the failure to account for collisional narrowing (typically through use of the Voigt profile in spectrum analysis) tends to underestimate the Lorentzian width by several percent for pressures less than ∼50 kPa. Tran and Hartmann12 developed an O2 A-band absorption model, which considers Voigt profiles for isolated transitions in addition to line mixing and collisional-induced absorption effects. They showed that line mixing effects are important near and above atmospheric pressure conditions. This line mixing effect tends to reduce absorption in the line cores and increase absorption in the wings. In many atmospheric O2 A-band measurements, the line cores are highly saturated because of the long path lengths involved, and consequently, the line wings are the most useful spectral regions to be analyzed. Thus, in this context, it is important to quantify pressure broadening in this regime, with

1. INTRODUCTION Transmission spectra of the O2 A-band (b1Σ+g ← X3Σ−g (0,0) located near 13 100 cm−1) play an important role in atmospheric remote sensing. Because the mole fraction of O2 in dry air is uniform and well-mixed in the Earth’s atmosphere, measured O2 A-band spectra can be used to retrieve total surface pressure1,2 and path length, for which the latter quantity can be used to determine cloud height and aerosol optical depth.3,4 In many cases, measurements of atmospheric O2 Aband spectra demand low-uncertainty spectroscopic line parameter data such as line intensities, pressure broadening parameters, and pressure-shifting coefficients (e.g., 0.25% target precision for the NASA atmospheric-CO2-monitoring satellite, OCO-2).5 In 2005, Yang et al.6 presented an analysis of ground-based Fourier-transform spectroscopy measurements of O2 A-band solar absorption spectra. Utilizing available line parameter data and the Voigt line profile for calculating the shapes of individual transitions, their atmospheric transmission model overestimated and underestimated absorption in the near wings and far wings, respectively. They concluded that limitations in understanding of the O2 A-band spectrum, which includes line shape effects associated with isolated lines as well as line mixing effects, limited the precision and accuracy of their retrievals. In recent years, many laboratory studies have worked to improve spectroscopic reference data for the O2 A-band to build and expand on what is contained in atmospheric spectroscopy databases such as HITRAN.7 To this end, © 2012 American Chemical Society

Received: February 6, 2012 Revised: March 26, 2012 Published: March 27, 2012 4069

dx.doi.org/10.1021/jp301194j | J. Phys. Chem. A 2012, 116, 4069−4073

The Journal of Physical Chemistry A

Article

having a precision ±0.001 cm−1 and a standard uncertainty of 0.002 cm−1. In this fashion, the observed spectrum was determined by the measured quantities ν, WPP, and RPAS, and these data were fit to theoretical line profiles as discussed below. Multiple O2 spectra were acquired over a range of H2O-in-air sample concentrations. A commercial dew point generator containing a water-based saturator was used to regulate the humidity in a flowing stream of ambient laboratory air. The hygrometer was located downstream of the PAS cell, and the total volumetric flow rate was ∼500 cm3/min. The flowing sample was drawn through the system causing the pressure in the PAS cell to be ∼5 kPa below atmospheric pressure. The mole fraction of H2O, XH2O, produced by the dew point generator ranged from approximately 6.00(8) mmol mol−1 to 24.0(2) mmol mol−1 (Tdp = 0.5 °C ± 0.2 °C to 19.5 °C ± 0.2 °C). Measured dew point temperatures were converted to the water vapor mole fraction using a temperature-dependent vapor pressure correlation for H2O and the dew point hygrometer pressure.21 As a cross-check, the partial pressure of H2O was determined by optically measuring the peak absorption coefficient of the (101) ← (000) 20,0 ← 30,3 transition at 7181.156 cm−1 as a function of dew point temperature. We found that, at the dew point temperatures used in this study, the instrumental dew point temperature corresponds to the true partial pressure of H2O within the stated uncertainty of the manufacturer (±0.2 °C). The PAS cell pressure P, was measured with a capacitance diaphragm gauge having a standard uncertainty of 160 Pa (∼0.16%) and was assumed to be equal to the dew point hygrometer pressure. The PAS cell temperature, T, was measured with a custom NIST-calibrated thermistor, which has a combined standard uncertainty of ∼20 mK. This temperature uncertainty accounts for the estimated difference between the sample gas and cell temperatures. All measurements were made between T = 292−294 K and P = 95−98 kPa. 2.2. Spectral Fitting Methods. As was previously mentioned, the choice of spectral line profile affects the line parameters that can be extracted from experimental data. This point is especially relevant in the case of the O2 A-band at subatmospheric pressures, where collisional narrowing is known to be important. The Galatry profile incorporates Dopplerbroadening, collisional broadening, and collisional narrowing and gives a more accurate representation of O2 A-band line shapes than does the Voigt profile (which does not include collisional narrowing) in the low-pressure domain (P < 50 kPa). We note, however, that, at elevated pressures such that the Lorentzian width (which captures collisional broadening) is significantly greater than the Doppler width, the Galatry profile essentially reduces to the Voigt profile. In the present work, although the measurements were made near atmospheric pressure conditions, we chose to fit Galatry profiles to our experimental O2 A-band spectral line shapes,22 consistent with previous FS-CRDS studies. We modeled each measured spectrum as an isolated line because there was no evidence in the fit residuals indicating the need to account for interfering lines or the effects of line mixing. In this case, α(ν) is given by the product of the O2 number density, the temperature-dependent line intensity, and the normalized Galatry line profile g(ν;ν0,δνD,Γ,νnar), where ν0 is the line center, δνD is the Doppler full width at half-maximum (fwhm), Γ is the Lorentzian fwhm, and νnar is the narrowing frequency. With the inclusion of a two-parameter linear

the most important collisional partners being those that are most abundant in air: N2, O2, Ar, and H2O. Collisional broadening by these species is typically accounted for by transition-dependent air broadening parameters, which, when combined with total pressure, can be used to model the Lorentzian width. Over the years, numerous groups8,10,11,13−17 have quantified J-dependent air-broadening parameters in detail for the O2 A-band for dry gas samples. Nevertheless, despite the ubiquity of H2O in the Earth’s atmosphere, we are unaware of any reported measurements or theoretical calculations for foreign broadening of O2 A-band transitions by H2O. In this study, we present measurements of O2 A-band transitions broadened by H2O. Our goal was to accurately measure the H2O broadening parameters for several transitions and to quantify the extent to which broadening by this species affects atmospheric transmission spectra in the O2 A-band. To this end, we used photoacoustic absorption spectroscopy (PAS) to measure the line shapes of six 16O2 A-band transitions in the wavenumber range 13 090 cm−1 to 13 110 cm−1. We find that foreign broadening of these lines by H2O is approximately twice as large as that of air. Finally, we quantify the role of H2O broadening with regard to column-integrated transmission spectra for the O2 A-band, and these results are discussed in terms of atmospheric applications and remote sensing measurement requirements.

2. METHODS 2.1. Experimental Details. Experiments were carried out in the James Madison University Regional Undergraduate Laser Laboratory. The PAS measurements were made using a custom acoustic resonator having a Q-factor of about thirty and an internal path length of 20 cm, which accommodated flowing samples of air that was conditioned over a range of humidity levels. The PAS system has been both described18 and characterized18,19 previously in detail. Recently, we have utilized the same setup to measure broadening of near-infrared CO2 transitions by H2O.20 The principal difference between the present experiment and our previous CO2 line shape study20 is the choice of a probe laser wavelength and the beam path within the PAS cell. In the present study, we used a single mode, continuous wave, external cavity diode laser (ECDL) as a tunable light source, nominally centered at wavenumber, ν = 12 900 cm−1. The laser had an output power of ∼2 mW, and its intensity was modulated at frequency f mod using a mechanical chopper. To maximize signal strength for the microphone of the PAS cell, f mod was tuned to the acoustic resonance frequency (∼1616 Hz). The component of the PAS microphone signal amplitude at f mod was measured using a lock-in amplifier referenced to this frequency. With this setup, the PAS output signal, RPAS, from the lock-in amplifier was directly proportional to the sample absorption coefficient, α(ν), given by RPAS = α(ν)WPPCsys, where WPP is the peak-to-peak modulated laser power and Csys is the photoacoustic system constant [28.4 V/(W cm−1)]. The low output power of the ECDL limited the signal-to-noise ratio (SNR) of the measured absorption spectra. To boost spectrum SNR, we used a doublepass alignment of the probe laser through the PAS cell, which increased the effective WPP by approximately a factor of 2. For each line shape measurement, the ECDL wavenumber was step-scanned across an individual O2 A-band transition by fine-tuning the laser resonator via an externally driven piezoelectric transducer. The absolute wavenumber of the laser was measured using a commercial wavelength meter 4070

dx.doi.org/10.1021/jp301194j | J. Phys. Chem. A 2012, 116, 4069−4073

The Journal of Physical Chemistry A

Article

and the H2O-broadened width Γw. Assuming an ideal gas mixture with a partial pressure of water vapor given by Pw, then Xw = Pw/P and

background term (ao,bo) to account for a nonzero spectrum offset and residual absorption from neighboring lines, the measured spectra were fit according to

Γ = 2[γairP + (γw − γair)Pw ]

RPAS(ν)/WPP = Ag (ν; ν0, δνD, Γ, νnar) + ao + bo(ν − ν0) (1)

(2)

where γair and γw are the pressure broadening coefficients (onehalf fwhm per unit pressure). By fitting eq 2 to the measured Γ(Pw) values, we assigned the slope and intercept to be 2(γw −γair) and 2γairP, respectively. See Figure 2 for this method of

where the quantity A equals the product of line intensity, O2 number density, and PAS system constant. Equation 1 was fitted to the measured spectra with the following constraints on δνD and νnar: δνD = (ν0/c)(8kBT ln(2)/m)1/2 and νnar = ηnarrP. Here, kB is the Boltzmann constant, c is the speed of light, and m is the molecular mass of 16 O2. The transition-dependent collisional narrowing parameters, denoted by ηnarr, were taken directly from Long et al.11 The other parameters in eq 1 were floated to yield a fitted value of Γ for each spectrum. An example measured O2 A-band absorption spectrum with fitted Galatry line profile and fit residuals is shown in Figure 1. The SNR for this measurement, defined as the peak signal divided by the standard deviation in the baseline residuals, was approximately 350:1.

4. RESULTS We modeled the fitted Lorentzian width, Γ, of each spectrum as the mole-fraction-weighted sum of the air-broadened width Γair, Figure 2. Measured Γ vs Pw for the P9Q8 16O2 transition. Error bars on the plot represent Type-A standard uncertainties. In this study, all Γ values were corrected to the reference temperature Tref = 296 K using the expression Γ(Tref) = Γ(T)(T/Tref)n with the corresponding temperature coefficients, n, measured by Brown and Plymate36 and archived in HITRAN 2008. Total pressure, P (see eq 2), was roughly 96.6 kPa for all measurements.

analysis for the P9Q8 transition. The resulting pressure broadening coefficients determined in this manner (referenced to a temperature of 296 K) are shown in Table 1. The combined uncertainties of γH2O and γair were obtained by a quadrature sum of Type A (statistical) and Type B (systematic) uncertainties. The Type A uncertainties in γw were estimated from the measurement scatter and the linear regression of the Γ vs P data using known practices of uncertainty analysis23,24 and were ultimately driven by the SNR of the fitted spectra. In this experiment, we estimate that the largest Type-B uncertainties in γw came from the measurements of Pw and ν. The combined uncertainties for γw were primarily Type A and ranged from 5% for P9Q8 to 45% for P5Q4. The relatively large Type-A uncertainties, which could be dramatically reduced by using a higher power laser, precluded the accurate measurement of any transitions weaker than P5Q4. Interestingly, we observe that γw is systematically and substantially larger than γair. We find the ratio γw/γair to be between 1.4 and 2.0. The P7P7 and P7Q6 transitions have the largest ratio of 1.9 and 2.0, respectively, whereas the P5P5 and P5Q4 have the smallest ratios near 1.5. Our current measurement fidelity is not high enough to resolve subtle changes in γw as a function of J for the O2 A-band. To our knowledge, there has been only one literature study, by Fanjoux et al.,25 which explored the effect of collisional broadening by H2O on the spectrum of O2 in detail. However, the work of Fanjoux et al.25 was done using coherent anti-Stokes Raman spectroscopy at temperatures between 446 and 990 K. A direct comparison to the results in the present work requires a

Figure 1. Top panel: example experimental spectrum (symbols) of the P9Q8 16O2 A-band transition taken near 13 093.656 cm−1 and fitted Galatry profile (line). This spectrum, which was obtained by taking six passes across the transition with an acquisition time of 20 min for each pass, has a SNR of ∼350:1. Bottom panel: fit residuals given by open symbols (measurements − fit). The solid curve represents the smoothed residuals data after applying an 8-point adjacent averaging filter. This curve qualitatively agrees with the absolute value of the derivative of the line profile; this suggests that the scatter in the residuals near line center may be driven by imprecision in the measured laser frequency. Experimental conditions were P = 96.9 kPa, T = 293.0 K, and XH2O = 9.69 mmol mol−1 (∼1% by volume). Constrained parameters included the Doppler width (fwhm) = 0.0284 cm−1 (calculated using the measured temperature) and the narrowing frequency = 0.0010 cm−1 (calculated using the reference data of Long et al.11). The Lorentzian width, peak position, and transition area were allowed to float in the fit. 4071

dx.doi.org/10.1021/jp301194j | J. Phys. Chem. A 2012, 116, 4069−4073

The Journal of Physical Chemistry A

Article

Table 1. Summary of 16O2 A-Band Transitions and Measured Air- and H2O-Broadening Coefficients for This Studya transition

ν011 (cm−1)

exptl uptime (hr)

γw/γair

γw (cm−1/atm)

γair (cm−1/atm)

γair‑Long11 (cm−1/atm)

P9P9 P9Q8 P7P7 P7Q6 P5P5 P5Q4

13 091.710358 13 093.655833 13 098.848243 13 100.821748 13 105.616870 13 107.628463

60 35 40 20 30 20

1.6(2) 1.5(1) 1.9(1) 2.0(4) 1.4(3) 1.5(7)

0.079(10) 0.076(4) 0.095(7) 0.105(21) 0.074(14) 0.079(36)

0.0495(5) 0.0497(5) 0.0507(5) 0.0510(6) 0.0532(5) 0.0539(6)

0.0490 0.0490 0.0507 0.0507 0.0531 0.0531

a The line positions and air-broadening coefficients are those reported by Long et al.11 The experimental uptime is the total time spent collecting experimental spectral line shapes for each transition over the range of H2O concentrations considered. Values in parentheses represent 1σ uncertainties in the last digit (e.g. 0.079(10) cm−1/atm is equivalent to 0.079 cm−1/atm ± 0.010 cm−1/atm).

significant temperature extrapolation and yields a value for the H2O broadening parameter that is approximately 1.08 times larger than the corresponding air-broadening parameter. A quantitative assessment of these two data sets is difficult, but results in Table 1 do not appear to agree with the work of Fanjoux et al.25 However, for comparison, we note that previous work on foreign broadening of CO220,26−30 by H2O has shown that, similar to our findings, γw is between 1.5 and 2.2 times larger than the corresponding γair for transitions near 4.30 and 1.57 μm. We also did a series of cross-checks in order to validate our measured γw parameters. First, our measured air-broadening parameters were compared against those recently reported by Long et al.11 As mentioned above, this study also was the source for the Galatry profile narrowing parameter that we used in our spectrum fits. We find that differences between our γair values and those of Long et al.11 are nearly within the standard uncertainty of the present measurements. Absolute values for γair from both studies are presented in Table 1. The mean relative difference between γair values is −0.8%. As a further check on the fidelity of our wavenumber measurements, the measured P9P9 transition line position was subtracted from the corresponding vacuum wavenumber reported by Robichaud et al.31 to give the pressure shift, and this result was then divided by P to yield the pressure-shifting coefficient. The resulting value was −0.00208(41) MHz/Pa, and it is in agreement with the low-uncertainty value of −0.00207 MHz/Pa based on measurements by Robichaud et al.9 and archived by Long et al.11 A series of atmospheric transmission spectra were computed at zenith angle of 51.5° in order to assess the importance of the observed H2O broadening of the O2 A-band on future remote sensing measurements. These calculations divided the atmosphere into 70 1 km thick slices and employed the Total Carbon Column Observing Network (TCCON) a priori profiles for Lamont, OK, which include pressure, temperature, and H2O concentration.32,33 The O2 dry air mole fraction was taken to be constant at 0.2094. Galatry line profiles were utilized for all O2 A-band transitions using the line shape parameters of Long et al.11,34 Water-broadening was set to γw = 1.8γair for each transition; this is a nominal value based on the experimental results in Table 1 of the present study. The A-band spectrum at each altitude was calculated for a 12 975−13 170 cm−1 wavenumber interval at the resolution of the JPL Table Mountain spectrometer (i.e., 0.054 cm−1). See Long and Hodges35 for further details on this A-band atmospheric model and the impact of various line shape effects. Figure 3 shows the calculated effect of H2O broadening of the O2 A-band atmospheric transmission spectrum. On August 1st, 2008, when the surface humidity exceeded a mole ratio of

Figure 3. Top panel: simulated, column-integrated atmospheric transmission spectrum in the O2 A-band region for August 1st, 2008, near the TCCON site in Lamont, OK, at a solar zenith angle of 51.5°. Middle panel: difference in transmission, in the simulated spectrum if H2O broadening is included versus neglected, for a humid summer day. Bottom panel: difference in transmission for a simulated spectrum occurring approximately five months later in Lamont, OK, on a dry winter day. To account for H2O broadening, a common factor of γw = 1.8γair was used for all transitions.

2.7%, the observed H2O broadening is expected to lead to a maximum difference in the observed transmission of ∼0.4%. The root-mean-square (rms) deviation across the entire simulation range is 0.16%. In contrast, on December 31st, 2008, for which the surface humidity was 0.8%, the effect of water-broadening is much smaller with a maximum difference 4072

dx.doi.org/10.1021/jp301194j | J. Phys. Chem. A 2012, 116, 4069−4073

The Journal of Physical Chemistry A

Article

in the observed transmission of just ∼0.1% and an rms deviation of 0.04%. Note that similar calculations performed at a solar zenith angle of 80° produced nearly identical results, indicating that the results discussed herein have minimal airmass dependence. In order to assess the impact of H2O broadening of the A-band at higher humidity levels, a series of transmission spectra were calculated using TCCON a priori profiles for Darwin, Australia, from December 30th, 2008. Because of the high water mixing ratio at the surface (3.3%), the maximum difference in transmission (i.e., including H2O broadening vs neglecting it) was ∼0.5% with an rms deviation of 0.22%.

(8) Robichaud, D. J.; Hodges, J. T.; Brown, L. R.; Lisak, D.; Masłowski, P.; Yeung, L. Y.; Okumura, M.; Miller, C. E. J. Mol. Spectrosc. 2008, 248, 1−13. (9) Robichaud, D. J.; Hodges, J. T.; Lisak, D.; Miller, C. E.; Okumura, M. J. Quant. Spectrosc. Radiat. Transfer 2008, 109, 435−444. (10) Ritter, K. J.; Wilkerson, T. D. J. Mol. Spectrosc. 1987, 121, 1−19. (11) Long, D. A.; Havey, D. K.; Okumura, M.; Miller, C. E.; Hodges, J. T. J. Quant. Spectrosc. Radiat. Transfer 2010, 111, 2021−2036. (12) Tran, H.; Hartmann, J. M. J. Geophys. Res., Atmos. 2008, 113, D18104. (13) Gordon, I. E.; Kassi, S.; Campargue, A.; Toon, G. C. J. Quant. Spectrosc. Radiat. Transfer 2010, 111, 1174−1183. (14) Predoi-Cross, A.; Hambrook, K.; Keller, R.; Povey, C.; Schofield, I.; Hurtmans, D.; Over, H.; Mellau, G. C. J. Mol. Spectrosc. 2008, 248, 85−110. (15) Predoi-Cross, A.; Holladay, C.; Heung, H.; Bouanich, J.-P.; Mellau, G. C.; Keller, R.; Hurtmans, D. R. J. Mol. Spectrosc. 2008, 251, 159−175. (16) Yang, S.; Canagaratna, M. R.; Witonsky, S. K.; Coy, S. L.; Steinfeld, J. I.; Field, R. W.; Kachanov, A. A. J. Mol. Spectrosc. 2000, 201, 188−197. (17) Havey, D. K.; Long, D. A.; Okumura, M.; Miller, C. E.; Hodges, J. T. Chem. Phys. Lett. 2009, 483, 49−54. (18) Gillis, K. A.; Havey, D. K.; Hodges, J. T. Rev. Sci. Instrum. 2010, 81, 064902. (19) Havey, D. K.; Bueno, P. A.; Gillis, K. A.; Hodges, J. T.; Mulholland, G. W.; Van Zee, R. D.; Zachariah, M. R. Anal. Chem. 2010, 82, 7935−7942. (20) Wallace, C. J.; Jeon, C.; Anderson, C. N.; Havey, D. K. J. Phys. Chem. A 2011, 115, 13804−13810. (21) IAPWS Revised Release on the Pressure Along the Melting and Sublimation Curves of Ordinary Water Substance, 2008. http://www. iapws.org. (22) Galatry, L. Phys. Rev. 1961, 122, 1218−1223. (23) Taylor, B. N.; Kuyatt, C. E. Technical Note 1297; NIST: Gaithersburg, MD, 1994. (24) Bevington, P.; Robinson, D. K. Data Reduction and Error Analysis for the Physical Sciences; McGraw-Hill: Boston, MA, 2002. (25) Fanjoux, G.; Millot, G.; Saint-Loup, R.; Chaux, R.; Rosenmann, L. J. Chem. Phys. 1994, 101, 1061−1071. (26) Rosenmann, L.; Hartmann, J. M.; Perrin, M. Y.; Taine, J. Appl. Opt. 1988, 27, 3902−3906. (27) Rosenmann, L.; Perrin, M. Y.; Hartmann, J. M.; Taine, J. J. Quant. Spectrosc. Radiat. Transfer 1988, 40, 569−576. (28) Sung, K.; Brown, L. R.; Toth, R. A.; Crawford, T. J. EOS Trans. AGU 2008, 89, A41D-0136. (29) Sung, K.; Brown, L. R.; Toth, R. A.; Crawford, T. J. Can. J. Phys. 2009, 87, 469−484. (30) Wang, H.; Cao, Z.; Wang, L.; Gao, W.; Zhang, W.; Gong, Z.; Gao, X. High Power Laser Part. Beams 2010, 22, 1982−1986. (31) Robichaud, D. J.; Hodges, J. T.; Masłowski, P.; Yeung, L. Y.; Okumura, M.; Miller, C. E.; Brown, L. R. J. Mol. Spectrosc. 2008, 251, 27−37. (32) Kalnay, E.; Kanamitsu, M.; Kistler, R.; Collins, W.; Deaven, D.; Gandin, L.; Iredell, M.; Saha, S.; White, G.; Woollen, J. Bull. Am. Meteorol. Soc. 1996, 77, 437−471. (33) Wunch, D.; Toon, G. C.; Blavier, J. F. L.; Washenfelder, R. A.; Notholt, J.; Connor, B. J.; Griffith, D. W. T.; Sherlock, V.; Wennberg, P. O. Philos. Trans. R. Soc., A 2011, 369, 2087−2112. (34) Long, D. A.; Havey, D. K.; Yu, S. S.; Okumura, M.; Miller, C. E.; Hodges, J. T. J. Quant. Spectrosc. Radiat. Transfer 2011, 112, 2527− 2541. (35) Long, D. A.; Hodges, J. T.; 2012, manuscript in preparation. (36) Brown, L. R.; Plymate, C. J. Mol. Spectrosc. 2000, 199, 166−179.

5. CONCLUSIONS In this study, the effect of H2O broadening on O2 A-band spectra was measured using a laser-based PAS technique. H2Obroadening parameters with average relative combined standard uncertainties of ∼20% were obtained for six transitions of 16O2. Our measurements indicate that H2O is an efficient broadening species with γw values between 1.4 and 2.0 times as large as γair. Using a line-by-line model, atmospheric O2 A-band transmission spectra were simulated to quantify the impact of broadening by H2O. In general, increasing H2O number density results in reduced absorption near transition line centers and enhanced absorption in the far wings. The calculated peak fractional differences in the simulated spectra for Lamont, OK, are approximately 0.4% during summer and 0.1% during winter, and are as high as 0.5% for summer in Darwin, Australia (a representative tropical region). This effect appears to exhibit minimal air mass dependence. In order to avoid seasonally and spatially dependent biases, high-precision remote sensing missions that probe the O2 A-band need to consider the effects of water broadening.

■ ■

AUTHOR INFORMATION

Notes

The authors declare no competing financial interest.

ACKNOWLEDGMENTS This work was supported primarily by the National Institute of Standards and Technology (NIST) through a measurement science grant (70NANB10H250) and the NIST Greenhouse Gas and Climate Science Measurements Program. Measurements were carried out at James Madison University. Debra Wunch (California Institute of Technology) is thanked for providing the TCCON profiles, which were accessed at https://tccon-wiki.caltech.edu/ on 13 Dec 2011.

■

REFERENCES

(1) O’Brien, D. M.; English, S. A.; Da Costa, G. J. Atmos. Oceanic Technol. 1997, 14, 105−119. (2) O’Brien, D. M.; Mitchell, R. M.; English, S. A.; Da Costa, G. A. J. Atmos. Oceanic Technol. 1998, 15, 1272−1286. (3) Heidinger, A. K.; Stephens, G. L. J. Atmos. Sci. 2000, 57, 1615− 1634. (4) Stephens, G. L.; Heidinger, A. J. Atmos. Sci. 2000, 57, 1599−1614. (5) Miller, C. E.; Brown, L. R.; Toth, R. A.; Benner, D. C.; Devi, V. M. C. R. Phys. 2005, 6, 876−887. (6) Yang, Z.; Wennberg, P. O.; Cageao, R. P.; Pongetti, T. J.; Toon, G. C.; Sander, S. P. J. Quant. Spectrosc. Radiat. Transfer 2005, 90, 309− 321. (7) Rothman, L. S.; Gordon, I. E.; Barbe, A.; Benner, D. C.; Bernath, P. F.; Birk, M.; Boudon, V.; Brown, L. R.; Campargue, A.; Champion, J. P. J. Quant. Spectrosc. Radiat. Transfer 2009, 110, 533−572. 4073

dx.doi.org/10.1021/jp301194j | J. Phys. Chem. A 2012, 116, 4069−4073