Absolute concentrations (cm–3) of highly vibrationally excited hydroxyl (OH) are derived from measurements of the volume emission rate of the υ = 9 + 8 states of the OH radical made by the SABER instrument on the TIMED satellite. SABER has exceptionally sensitive measurement precision that corresponds to an ability to detect changes in volume emission rate on the order of ~5 excited OH molecules per cm3. Peak zonal annual mean concentrations observed by SABER exceed 1000 cm–3 at night and 225 cm–3 during the day. Measurements since 2002 show an apparent altitude-dependent variation of the night OH(υ = 9 + 8) concentrations with the 11 year solar cycle, with concentrations decreasing below ~ 95 km from 2002 to 2008. These observations provide a global database for evaluating photochemical model computations of OH abundance, reaction kinetics, and rates and mechanisms responsible for maintaining vibrationally excited OH in the mesopause region.
 The hydroxyl radical (OH) is formed in Earth's upper mesosphere following the reaction of atomic hydrogen (H) and ozone (O3). This reaction preferentially populates high-lying vibrational states of OH just below the dissociation limit, starting at υ = 9 and extending down to υ = 6. The remaining vibrational states are populated by collisional and radiative cascade from these high-lying states. The SABER instrument [Russell et al., 1999] on the NASA TIMED satellite observes OH emission near 2 µm from the strong Δυ = 2 sequence of radiative transitions of OH. The 2 µm channel of SABER is sensitive to emission from the υ = 9 and υ = 8 states of OH. This channel was chosen specifically to observe the nascent OH vibrational concentrations immediately after the reaction of H and O3 to facilitate the derivation of atomic oxygen concentrations and also the heating rate due to this reaction [Smith et al., 2010; Mlynczak, 1999].
 The SABER instrument scans the Earth's limb and records vertical profiles of infrared radiance (W m–2 sr–1) as a function of tangent altitude from approximately 400 km to just below the hard Earth surface. The measured OH limb radiance is inverted to yield volume emission rates of energy (erg cm–3 s–1) assuming the emission is in the weak-line limit of radiative transfer. The standard SABER data product is the volume emission rate of OH (υ = 9 + 8). The volume emission rates are used here to derive the absolute concentrations of the highly vibrationally excited OH molecule. The extensive SABER data set beginning in 2002 is used to analyze the concentration of highly vibrationally excited OH in the mesopause region both day and night. The emission rates are observed to decrease between 2002 and 2008 and increase again thereafter, implying solar cycle dependence in the OH volume emission rates and the concentration of excited OH molecules.
 The methodology for deriving absolute concentrations of highly vibrationally excited OH molecules is given in the next section. Results are given in section 'Results', followed by a summary in section 'Discussion and Summary'.
 The SABER OH channel at 2 µm has a bandpass (5% relative transmission points) between 1.939 and 2.217 µm (5157 to 4509 wave numbers). This bandpass is sensitive to emission almost entirely arising from the 9➔7 and 8➔6 transitions. Only a few percent of the 7➔5 transition of OH is captured within the SABER 2 µm channel filter. The shape of the SABER spectral response function for this channel initially weights the volume emission rate. A small correction factor is applied to the derived emission rate to account for the shape of the SABER spectral filter, as described by Mlynczak et al. . It is emphasized that SABER simultaneously measures emission from the υ = 8 and υ = 9 states in one channel. The derived volume emission rate profile represents the sum of the emission from these two states. Any differences in the altitude distributions of the individual υ = 8 and υ = 9 state populations cannot be determined from these measurements.
 The volume emission rate measured by SABER is the emission rate of energy and is given by the product of the concentration of excited molecules and the Einstein coefficient for spontaneous emission of radiation and the photon energy hcω. Here h is Planck's constant, c the speed of light, and ω the photon energy in cm–1. For the SABER channel near 2.0 µm, the observed emission rate V (photons cm–3) is given by
where A9-7 and A8-6 are the Einstein coefficients for the Δυ = 2 transitions from υ = 9 to υ = 7 and from υ = 8 to υ = 6, respectively, and hcω is the mean photon energy in the band.
 The absolute concentration of the sum of the OH υ = 9 and υ = 8 states is obtained by dividing the volume emission rate by the Einstein coefficients for the two transitions. A9-7 and A8-6 are essentially identical, allowing the right-hand side of (1) to be divided by a single A-value, leading to an expression for the sum of the concentration (cm–3) of the OH(9) and OH(8) vibrational states. This approach results in the SABER data set of OH volume emission rates to be used to assess the global distribution of highly vibrationally excited OH molecules in the mesopause region.
 In the SABER analyses [e.g., Smith et al., 2010] values of 118.3 and 117.2 s–1 are used for the 9➔7 and 8➔6 transitions, respectively. These are taken from a set of A values provided by D. Nesbitt (JILA, Boulder, CO, private communication, 1995) early on in the development of the SABER project, which are based on the dipole moment measurements of Nelson et al. . These are also consistent with values of 111 and 115 s–1, respectively, as computed by the SABER science team using OH line and band strengths reported on the HITRAN database. They are also consistent with the theoretical results of van der Loo and Groenenboom  who computed values of 119.93 and 120.10 s–1 (G. Groenenboom, Radboud University, Nijmegen, NL, private communication, 2007). In this analysis we adopt a value of 120 s–1 for A97 and A86. These Einstein A values are for a rotational temperature of 200 K. The resulting data are the concentrations (cm–3) of the sum of the two highest-lying states of the OH molecule, which are 3.25 eV (υ=9) and 2.97 eV (υ=8) above ground. These energies correspond to values that are 200 and 187 times the nominal value of kT (k is Boltzmann's constant, T is temperature) in the mesopause region.
 The SABER instrument is exceptionally sensitive in this channel at 2.0 µm. The measured noise equivalent radiance (NER) of this channel is 1.3 × 10–6 W m–2 sr–1. Individual measured radiance profiles have peak signal-to-noise levels from 100:1 to 1000:1. The NER corresponds to a “noise equivalent volume emission rate” of about 500–600 photons s–1 cm–3, or an equivalent “noise equivalent concentration” of ~ 5 molecules/cm3. This extremely high measurement precision enables SABER to detect small changes in concentration of highly vibrationally excited OH in the mesopause region.
 SABER began routine science measurements in January 2002 after a successful launch in December 2001. There are now nearly 11 years of extant SABER data and the instrument continues to operate nominally, having collected nearly 98% of possible data to date. Shown in Figures 1 and 2 are the annual zonal mean concentrations of highly vibrationally excited OH(υ = 9 + 8) observed in 2002 and 2008, respectively. Concentrations for night and day are shown in separate figures. Day is defined as solar zenith angles (SZA) less than 85° and night for SZA greater than 95°. SABER observes over all 24 h of local time every 60 days. The data shown here are averages over the respective local times for day and night. The data are plotted on pressure surfaces, and an approximate altitude scale is indicated on the right axis of each figure. The latitude range is +/– 55° because the SABER instrument continuously observes this region. Peak concentrations are observed to exceed 1000 cm–3 at night in both years although the peak concentrations in 2002 are clearly larger. During the day the concentrations are approximately five times smaller at the peak and the 2002 concentrations are visibly larger than during 2008.
 Figure 3 shows the difference in annual zonal mean concentrations shown in Figures 1 and 2. The maximum difference is approximately 160 cm–3 near the equator, and at high latitude in the Southern Hemisphere at night, and about 25 cm–3 near the equator during the day. There is some indication of a small decrease in population near 95 km at +/– 25° latitude during the daytime. However the derived decrease in concentration of 5 cm–3 is near the limit of SABER sensitivity discussed above and may not be real. Figure 4 shows the area-weighted average (cosine latitude weighted) means of the concentrations from 2002 and 2008 shown in Figures 1 and 2. The latitude range of +/– 55° observed by SABER corresponds to ~ 82% of total atmospheric area. The means and variation shown here are likely representative of the global means. The smaller values in 2008 suggest a solar cycle dependence of the highly excited OH concentrations both night and day as the minimum of solar cycle 23 occurred in 2008.
 Figure 5 shows the absolute difference in the global mean concentrations from 2002 to 2008, both night and day. As in prior figures, positive values in Figure 5 indicate larger values in 2002. The largest absolute decreases at night occur near 87 km, which is the classic altitude for the peak of the OH emission layer. During the day the largest decrease in concentration is observed near 83 km. These figures also indicate small positive increases above ~ 92 km in the daytime and above ~ 95 km at night. However these increases are less than 5 cm–3 and are therefore within the sensitivity limit of the SABER instrument to detect. The concentrations of OH(υ = 9 + 8) at these altitudes are simply too small for SABER to detect changes possibly associated with the solar cycle.
 Figure 6 shows the time series of the difference in the area-weighted average concentrations by year from that observed in 2002, for years 2003 through 2011. Positive values in this figure indicate the OH concentrations are smaller in years subsequent to 2002. At night the largest difference of 160 cm–3 is observed to occur in late 2008 and early 2009, coincident with the occurrence of the minimum of solar cycle 23. Subsequent to 2009 and through 2011 the differences at night from 2002 become less as the OH concentrations increase again toward the values observed in 2002. The differences during the day are also observed to decrease from 2002 although the minimum is not as clearly defined as the differences are only a few times the level of the sensitivity of the SABER instrument. The results in Figure 6 again suggest a very strong influence of the solar cycle on the concentration of vibrationally excited OH in the mesopause region at night.
4 Discussion and Summary
 We have presented the first long-term, global observations of the concentration of highly vibrationally excited OH near the dissociation limit of the molecule. The SABER OH(υ = 9 + 8) concentrations are derived independently of any assumed mechanism for the production and loss of vibrationally excited OH, other than the provision of the Einstein coefficient for spontaneous emission for the 9➔7 and 8➔6 transitions. Peak concentrations for the sum of the υ = 9 and 8 states exceeding 1000 cm–3 and 225 cm–3 are observed night and day, respectively. From 2002 to 2008 the concentrations decrease and then increase from 2008 to 2011. The observed behavior strongly suggests an altitude-dependent solar cycle influence on the OH (υ = 9 + 8) concentrations (and on the corresponding OH volume emission rates). Although not shown in this paper, SABER temperatures and ozone concentrations in the mesopause region also show similar temporal variations and coincidence with the solar cycle from 2002 to 2011, thus strengthening the assertion that the observed OH(υ) variations are indeed due to the solar cycle. The SABER ozone and temperature results will be the topic of a future paper.
 A primary application of these data is in the modeling of the concentrations of vibrationally excited OH. The radiometric quality of the SABER observations, and the independence of the observed concentrations on any presumed mechanism of OH(υ) formation, ultimately places a very strong constraint on the rates of production and loss of OH(9) and OH(8). The absolute concentrations are determined by the rate of production of OH(9) and OH(8) by the reaction of atomic hydrogen and ozone and of the rate of loss of these states by collisions, reactions, and radiation. Any statistical equilibrium model of the OH vibrational state concentrations must be able to reproduce the observed SABER OH(υ = 9 + 8) concentrations shown here. One specific issue to be addressed is the rate of removal of OH(9) and OH(8) in collisions with atomic oxygen [e.g., Kalogerakis et al., 2011]. Absolute OH vibrational state concentrations are also needed to assess the overall loss of OH by reactions because it has been suggested [Mlynczak and Solomon, 1993] that above 85 km most of the OH molecules in the mesopause region are vibrationally excited. The expected larger reaction rate coefficients for vibrationally excited OH relative to that for the ground state may lead to different modeled odd-oxygen or odd-hydrogen concentrations in the mesopause region. The apparent solar cycle variability of the observed concentrations also places a strong constraint on multidimensional photochemical models and their coupled odd-oxygen and odd-hydrogen chemistry in the mesopause region.
 Lastly, the SABER OH concentrations presented here, in conjunction with those from SABER's second OH channel near 1.6 µm that observes the 5➔3, 4➔2 and 3➔1 bands, can be used with a model of OH vibrational states to assess the overall heating efficiency for the reaction of atomic hydrogen and ozone in the mesopause region. This reaction is the single largest source of heat in the mesopause region [Mlynczak and Solomon, 1993]. However, derivation of the sum of the OH(υ = 5 + 4 + 3) concentrations is not as straightforward as for the υ = 9 + 8 states presented here because the Einstein coefficients for these lower-lying states are different by nearly a factor of 3. Results related to this topic will be presented in a future paper.
 The authors wish to acknowledge continued support from the NASA Heliophysics Division through the TIMED Project.