Journal of Geophysical Research: Atmospheres

Atomic oxygen profiles (80 to 115 km) derived from Wind Imaging Interferometer/Upper Atmospheric Research Satellite measurements of the hydroxyl and greenline airglow: Local time–latitude dependence

Authors


Abstract

[1] Hydroxyl and oxygen greenline nightglow observations from the Wind Imaging Interferometer (WINDII) are used to examine the local time–latitude variation of atomic oxygen in the mesopause region. Individual hydroxyl and greenline emission profiles from over 5 years of data are converted to oxygen mixing ratio (or concentration) profiles and then binned into local times, latitudes, and seasons. The two derived oxygen profiles from each emission are then combined into a single profile that spans a significant portion of the mesopause region (80 to 115 km). A technique developed earlier that addresses the altitude variability of the emission profiles is used. This level of agreement indicates a high degree of consistency in the radiance observations and in the photochemistry used to convert the emission rates to oxygen profiles. We demonstrate that the atomic oxygen concentration or mixing ratio profiles are very sensitive to local nighttime, and we display the manner in which they vary. The local time variation is primarily due to the tidal dynamics in the atmosphere. Comparisons between our atomic oxygen data set, a simple tidal model, and the TIME-GCM show good agreement; however, the local time tidal structure of atomic oxygen from MSISE-90 shows a 180° phase inconsistency. The measured local time oxygen variations vary with season and latitude, and we show that these oscillations are stronger under equinox conditions.

1. Introduction

[2] In the mesopause region, atomic oxygen is the most important minor species and peaks in concentration at an altitude in a 10-km range about 95 km. It is created by photodissociation of molecular oxygen during the day in the lower thermosphere and transported downward to a denser region where three-body recombination with background gases converts it back to molecular oxygen [Chamberlain and Hunten, 1987]. Since this process can store solar energy, atomic oxygen is important in the determination of the atmospheric energy budget in this region [Mlynczak and Solomon, 1993; Ward and Fomichev, 1993]. In addition, the recombination process provides the energy for many of the night airglow emissions that are observed.

[3] Continuous direct measurement of atomic oxygen in the mesopause region is difficult, costly, and physically challenging. The region is too high for aeroplanes or balloons and too low for in situ satellites to make direct observations. Direct measurements have only been made from a sprinkling of rocket measurements. A summary of rocket experiments that have measured the atomic oxygen abundance in situ over the past few decades is found in a thesis by Gumbel [1997]. It was found that above the peak, the amount falls off slowly and is controlled by molecular diffusion. Since atomic oxygen is a lighter species than the background molecules, its mixing ratio increases with height. Below the peak, there is a rapid decline referred to as an “oxygen ledge” by Gumbel where chemical lifetimes are shorter and eddy diffusion is greater. The summary of atomic oxygen profiles from rocket measurements shows a strong variability; however, there is not a sufficient number of rocket flights to determine the source of this variability nor the general local time and latitudinal tendencies.

[4] In order to determine the local time and latitudinal variability of atomic oxygen, the concentrations need to be measured indirectly from measurements over a wide geographic and temporal range. Since the mesopause region is the same region as the airglow, where excited minor atmospheric constituents that are known to be involved in oxygen recombination chemistry spontaneously emit a spectrum of measurable wavelengths, it is possible to use these measurements as a probe to deduce atomic oxygen profiles [e.g., Murtagh et al., 1990; McDade et al., 1986]. Therefore, continuous satellite observations of the airglow are an ideal source of measurements for determining temporal and latitudinal variations in atomic oxygen profiles.

[5] Launched in 1991, the Upper Atmospheric Research Satellite (UARS) has provided the community with a wealth of knowledge of the airglow. Two of its instruments, the High-Resolution Doppler Imager (HRDI) [Hays et al., 1993] and the WIND Imaging Interferometer (WINDII) [Shepherd et al., 1993], have provided continuous airglow measurements for a number of years during the mission. The tuned mechanistic tidal model (TMTM) demonstrated consistency between tidal signatures in HRDI/WINDII airglow observations and HRDI winds [Yudin et al., 1998]. Work published by members of the HRDI team [Yee et al., 1997; Hays et al., 2003] has shown that there is a correlation between prominent variations in nightglow emissions and variations in atomic oxygen by comparing their airglow data with the thermosphere-ionosphere-mesosphere- electrodynamic general circulation model (TIME-GCM).

[6] In this paper, we report on local time and latitude variations in atomic oxygen derived from WINDII hydroxyl and oxygen greenline observations. A global set of oxygen profiles in the altitude range from 80 to 115 km is determined using the approach described by Russell and Lowe [2003]. From this data set, the latitudinal and local time behavior of atomic oxygen is determined by appropriately binning and then combining the profiles of the two emissions. The two nightglow emissions used in this analysis are green line emission (λ = 557.7 nm) of atomic oxygen and the P1(3) (λ = 734 nm) emission of the hydroxyl (8–3) vibration-rotation band.

[7] The chemistry of the greenline emission has been summarized by McDade et al. [1986] and a means of deriving atomic oxygen from it described. The chemistry of the hydroxyl emission was outlined by Russell and Lowe [2003], and a means of deriving atomic oxygen from this emission was also described. The relationship between the hydroxyl emission and atomic oxygen is the more complex of the two emissions, and uncertainties exist in the transition and quenching rates of the excited hydroxyl radicals. Russell and Lowe [2003] have used WINDII data to show that Goldman [1998] hydroxyl transition rates combined with sudden death method [McDade et al., 1986] of vibrationally excited hydroxyl quenching provided the best agreement with oxygen derived from the greenline. This paper confirms this earlier work, and this parameter set is used for deriving oxygen from all the hydroxyl nightglow emission data.

[8] By using two nightglow sources to derive atomic oxygen profiles, a more complete atomic oxygen profile extending from the base of the hydroxyl layer at around 80 km to approximately 20 km above the peak in the greenline layer (∼115 km) is produced for this paper. We construct a large data set of atomic oxygen profiles that cover mid to equatorial latitudes throughout the night. This enables us to determine latitudinal and local time variability in this altitude range over the span of many years.

[9] In section 2, the data analysis technique is reviewed in detail and characterized using the full WINDII data set. In sections 3 and 4, the local time/latitude dependence of atomic oxygen is presented. In section 5, a discussion of the causes of the observed variations and the rationale for attributing these variations to tidal dynamics. In this paper, modeled variations in atomic oxygen are demonstrated to be present in observed atomic oxygen.

2. Data Analysis Procedure

[10] The process to calculate atomic oxygen concentrations from the WINDII emission data has been explained and demonstrated in the companion paper [Russell and Lowe, 2003]. In it, the method for deriving atomic oxygen concentrations from hydroxyl emission rates is presented and the assumptions and rationale in the development of this method are explained. The largest uncertainty in the method is associated with the transition and quenching rates used in the analysis. By comparing the atomic oxygen derived from greenline data with that of the hydroxyl data at similar altitudes, latitudes, and local times, an optimal hydroxyl parameter set is chosen.

[11] For this study, the above-mentioned parameter set which involved Goldman transition probabilities and sudden death quenching is used to derive all the hydroxyl data from WINDII into atomic oxygen. In the absence of information about the background atmosphere at the moment the measurements were taken, it was necessary to use standard background atmospheres and temperatures from the MSISE-90 atmospheric model in the derivations of both greenline and hydroxyl emissions. Local time, longitude, and latitude as well as the F10.7 flux for every measurement point were used as inputs into this model (KP indices had no impact at these altitudes). The impact of using MSIS is discussed in the next section.

[12] The hydroxyl emission rate for the (8–3) band peaks on average around 87 km. Atomic oxygen can be derived from hydroxyl data within 6 km of this peak. Below this range, the signal-to-noise in the data becomes too low to make accurate derivations. Above this range, a contaminant in the background measurement of the NO2 continuum which was used in the hydroxyl derivation becomes stronger and is thought to affect the results [Russell and Lowe, 2003].

[13] The greenline emission rate peaks in a 10-km range about 95 km. Atomic oxygen data are derived from 5 km below this peak to 20 km above it. A combination of these two ranges produces an altitude range from 80 km to 115 km, with a 4-km overlap typically centered between 90 and 92 km. It is with this overlap that the hydroxyl data are validated against the greenline data in the first paper.

[14] One feature that is accommodated in the first paper in regard to combining or comparing the two atomic oxygen profiles from the two emission sources is the variability of the altitude of the hydroxyl and greenline peaks. Depending on local time and latitude, zonally averaged emission peaks varied in altitude and intensity. As a result, comparing the profiles at a single altitude for all space and time is not appropriate because, as the profile peak moved in altitude, the chosen altitude for comparison could sit at an area where the derived atomic oxygen data from one or the other emission is poor. To overcome this difficulty, a nominal altitude is chosen for every latitude and local nighttime bin that contained the strongest emission data from both sources. This nominal height is chosen using the technique developed by Russell and Lowe [2003]. Their technique calculated the altitude of the primary peaks for each emission and then, after binning and zonally averaging, the midpoint between the two peaks is chosen as the nominal altitude for that particular bin during that night.

[15] The companion paper [Russell and Lowe, 2003], in which the optimal hydroxyl parameter set was determined, looked at only two nights of data. These two nights were special in that the hydroxyl and greenline emissions were measured on alternating orbits allowing for a comparison between the two at similar local times and latitudes. To substantiate this choice of hydroxyl chemistry, we examine all 6 years of data at the nominal altitude for every bin using the chosen hydroxyl set as well as the two sets that were found to produce atomic oxygen concentrations closest to the chosen set: i.e., single quantum quenching with Mies [Mies, 1974] transition rates and multiquantum quenching [Adler-Golden, 1997] with Goldman transition rates.

[16] Figure 1 shows three scatterplots at the nominal height of all the oxygen mixing ratio data used for each set. Figure 1a shows the percent difference between the oxygen mixing ratio derived from the greenline emission and the atomic oxygen mixing ratio derived from the hydroxyl emission that used Goldman transition probabilities and a multiquantum quenching technique. Figure 1b shows the same thing, but the hydroxyl chemistry used Goldman transition rates and sudden death quenching. Similarly, Figure 1c involves hydroxyl chemistry that used Mies transition rates and single quantum quenching. It is clear that the average percent difference between the profiles at the nominal height is the least in our chosen atomic oxygen set (Figure 1b) and that the standard deviation about this mean is larger for the data than when we used Mies transition rates. We point out that in this study, we only examine the 9–8 vibrational quenching and can say nothing about further quenching beyond this. This means that, based on the comparison, the sudden death transition is best when examining the transition from the ninth vibrational level.

Figure 1.

Scatterplots of difference between the derived atomic oxygen mixing ratios at the nominal altitude from greenline and OH data using three different OH chemical processes. Each point in the scatter represents one latitude-longitude bin over the 6 years of data. (a) Goldman-MultiQuantum Quenching scheme. (b) Goldman-Sudden Death scheme. (c) Mies-Single Quantum Quenching scheme. Averages (thin lines) and standard deviations (dashed lines) are also shown.

[17] To illustrate the latitudinal independence of our combination technique, Figure 2 shows scatterplots of the percent difference in the bin averaged mixing ratio values and the corresponding slope of the mixing ratio profile at the nominal height. In each bin, the data are zonally averaged and only bins with at least five data points are used. Figures 2a and 2b contain bins that cover all local times and latitudes south of 20°S. Figures 2c and 2d show bin averaged differences from 20°S to 20°N, and Figures 2e and 2f contain differences north of 20°N. The plots show that there is no latitudinal bias to the scattered data. The scatterplots in the figure show how the two derived atomic oxygen mixing ratios compare at the nominal height where they are combined. In order to combine the profile in the best possible manner, atomic mixing ratios are used because those profiles are dynamically less perturbed and the slopes do not change as significantly about the nominal altitude.

Figure 2.

Scatterplots representing all the data used in this study separated into three latitude bands. Each point represents a particular latitude/local time bin as explained in the paper at the nominal height chosen for that bin. (left) Percent difference between the two derived mixing ratio profiles. (right) Percent difference between their slopes at that point. Averages (thin lines) and standard deviations (dashed lines) are also shown.

[18] The percent differences shown in Figure 2 are reasonable and fall within the ranges of the variability associated with the two sets of profiles. Figure 3 shows the standard deviation of each binned point for each derived set of atomic oxygen mixing ratios. This figure also shows that no bias with latitude exists. The hydroxyl derived mixing ratio variability is greater than the greenline derived variability. The standard deviations are for the most part less than 50%, with an average standard deviation of about 30% that is roughly constant in altitude. This is of the same order as the standard deviation in the derived difference in atomic oxygen mixing ratios. Therefore we conclude that in the absence of additional knowledge about the background atmosphere, the technique for combining the two atomic oxygen mixing ratio profiles from the two different sources at the nominal altitude for each bin is the best technique and derived values lie within the range of their standard deviations. The residual dependence on height may be associated with differences between the “real” atmosphere and the selected background atmosphere from MSISE-90. Issues with the use of temperature and density from this model represent a higher order correction and are not addressed at this time.

Figure 3.

Scatterplots representing all the data used in this study separated into three latitude bands. Each point represents a particular latitude/local time bin as explained in the paper at the nominal height chosen for that bin. (left) Standard deviation of each point for the hydroxyl atomic oxygen mixing ratio. (right) Standard deviations of the greenline atomic oxygen mixing ratio. Averages (thin lines) and standard deviations (dashed lines) are also shown.

[19] The WINDII emission data set is organized into latitude and local time bins based on the satellite's natural binning because the orbit is such that each ascending or descending latitudinal crossing on a given day occurred at approximately the same local time. The latitudinal bins covered in this study are from 52.5°S to 52.5°N with a width of 5°, centered at the equator. The local time binning covered from 17:45 hours (5:45 p.m.) local time to 30:15 hours (6:15 a.m.) local time throughout the night with a width of half an hour, centered at local midnight. As an example, the midnight equatorial bin included points in the range from −2.5° south latitude to +2.5° north and from 23:45 hours to 00:15 hours local time, and so on.

[20] The profiles from each derivation are then combined for each bin. At the nominal altitude chosen for each bin, where both emissions produced good results, the mixing ratios are averaged. In the bin that is 2 km higher, the profiles are combined using a 75/25% weighted average favoring the greenline profile since this altitude is closer to the greenline peak. In the bin that is 2 km lower, the profiles are combined using a 25/75% weighted average favoring the hydroxyl profile since this altitude is closer to the hydroxyl peak. In the few regions where there is no overlap, the two mixing ratio profiles are linearly interpolated to the nominal altitude. This approach is used throughout the entire data set.

3. WINDII: Atomic Oxygen Climatology

[21] The WINDII emission data set used in this climatology extends from late 1991 through to 1997. Owing to typical operational procedures, the collected data are not evenly distributed over those years and there are many times where there are large gaps in the data. Also, there are times when greenline emission is measured continuously, but not the hydroxyl emission, and visa versa. The UARS spacecraft was thermally designed for a launch into a planned orbit such that every 36 days, it would rotate its yaw angle and point either northward or southward. This period is known as a yaw period. During a yaw period, the local time of each ascending or descending latitudinal crossing completed a 12-hour cycle. During northward looking yaw periods, more data were collected at night at 70°N and 40°S than anywhere else and visa versa for southward looking yaw periods. This is due to the fact that the satellite was transiting from an upleg direction to a downleg direction. To deal with the many gaps and irregularities, the data are divided into these yaw periods and then collected into four annual seasons, centered on either a solstice or an equinox. Each binned season contained two or three complete yaw periods.

[22] Typically, during a yaw period, the satellite would take measurements in all latitude and local time bins. Since each season contained two or three yaw periods, there are usually enough data in each bin to construct an atomic oxygen climatology. However, there are some cases where there are little to no data. For example, since only nighttime data are used, the latitudinal range above 52.5° latitude is rarely measured. In the summer hemispheres, the local nighttime range is short, and for some seasons, little or no data are collected. For a bin to be deemed adequate, there had to be at least five data points in it.

[23] The greenline data and the hydroxyl data are binned in the fashion described above. In those seasonal, latitudinal, and local time bins where an atomic oxygen data set is available from both sources, a complete atomic oxygen profile from 80 to 115 km is created using the technique mentioned in section 2.

4. Results

[24] Atomic oxygen shows significant variation as a function of altitude, latitude, local time, and season. With the current data set dependent on four different parameters, a variety of two-dimensional views are needed to illustrate this variability. In order to bring out certain features, the actual concentration, the mixing ratio, or a percent difference from the mean may be used as the quantities of interest in the figures. The contour scaling may also be either linear or logarithmic, depending on which parameters are chosen and what is actually plotted.

[25] Where possible, the atomic oxygen data set is compared to the atomic oxygen concentration from the output of the TIME-GCM [Roble and Ridley, 1994]. The modeled data represented zonally averaged data binned on the day of either a solstice or an equinox, rather than the binning of an entire season of data as was done with the WINDII data. When examining the concentration of atomic oxygen on the logarithmic scales used, each contour is 1.25 times greater or smaller than the preceding one. When examining the mixing ratio using the logarithmic scales, each contour is 1.5 times greater or smaller than the preceding one.

4.1. Altitude–Local Time Slices of the Atomic Oxygen Data

[26] Atomic oxygen concentration and mixing ratios are plotted to show altitudinal and local time variations by keeping the season and latitude fixed. Figure 4 is a an example of one such set of contour plots in which the season is centered on the September 1993 equinox and the chosen latitude is the equator. Most bins between 80 and 115 km have good data coverage. There are six color contour plots in Figure 4. In the left-hand column, the plots show derived atomic oxygen concentration, derived atomic mixing ratio, and an example from the atomic oxygen output of the TIME-GCM. In the right-hand column, an average is determined for every altitude and the percentage differences from this mean are plotted. The plots in the right-hand column are undertaken to bring out the variability more clearly.

Figure 4.

Altitude/local time slices of atomic oxygen for the equatorial bin under equinox conditions. (a) A log plot of derived atomic oxygen concentration and (b) a plot of the deviation from the mean at each altitude. (c, d) Atomic oxygen mixing ratio in the same manner. (e, f) Atomic oxygen TIME-GCM results.

[27] Many features are evident in these contour plots. First, the derived data set closely matches the TIME-GCM, keeping in mind the differences in the binning between the data and the model. It can be seen that the peak in the concentration slowly descends until midnight, and when it passes below the lower boundary of the data set, a new phase front appears at a higher altitude and then begins to slowly descend again during the rest of the night. Below the peak altitude at around 86 km, there is a maximum at around 22:00 local time in the mixing ratio percent difference (Figure 4d) and a minimum occurs later. Above the peak altitude, near 98 km, the opposite is true and a minimum occurs at around 22:00 local time. The systematic variations are consistent with diurnal tidal variations with a vertical wavelength of ∼24 km. When compared with the TIME-GCM model, there appears to be a slight phase shift of a few hours between them.

[28] Figure 5 is similar to Figure 4 except now the season is the 1993/1994 solstice period. During solstices, the amplitudes of the local time variations are smaller and are harder to observe, especially at higher altitudes. Below 95 km, there are some similarities with equinox conditions in that the phase variation with local time can be seen. However, there is more variability. Above 96 km the phase structure is difficult to discern. This behavior is in contrast to the TIME-GCM results where this structure is evident throughout the whole altitude range. This structure is consistent with a weaker and less upward penetrating migrating diurnal tide and is consistent with the analysis of Yudin et al. [1998].

Figure 5.

As in Figure 4 but for equatorial solstice conditions.

4.2. Altitude-Latitude Slices of the Atomic Oxygen Data

[29] Atomic oxygen concentration and mixing ratios are plotted to show altitudinal and latitudinal variations by keeping the season and local time fixed. Figures 6 and 7are examples of two sets of contour plots in which the season is centered on the September 1993 equinox and the December 1993 solstice, respectively. The local time is set to local midnight. Most bins between 80 and 115 km have good data coverage. There are six color contour plots in each of these figures. In the left-hand column, the plots show derived atomic oxygen concentration, derived atomic mixing ratio, and an example from the atomic oxygen output of the TIME-GCM. For clarity, in the right-hand column, an average is determined for every altitude and the percentages from this mean are plotted.

Figure 6.

Altitude/latitude slices of atomic oxygen for local midnight under equinox conditions. (a) A log plot of derived atomic oxygen concentration and (b) a plot of the deviation from the mean at each altitude. (c, d) Atomic oxygen mixing ratio in the same manner. (e, f) Atomic oxygen TIME-GCM results.

Figure 7.

As in Figure 6 but for local midnight under solstice conditions.

[30] Again, keeping in mind the differences in the binning, the derived data set matches the TIME-GCM in form but there are differences in detail. For example, atomic oxygen decreases with height more quickly than the observations; the equatorial minimum is higher in the TIME-GCM, suggesting a longer wavelength for the model; the model variability is larger than the observations; and the observations show more structure at high latitudes than the model.

[31] It can be seen that the height of the concentration peak varies with latitude. There is a clear relative minimum in concentration at the equator at 95 km near the peak in the oxygen profile. However, below the peak, at around 85 km, there is clearly a relative equatorial maximum. The percent variation of the latitudinal variation below the peak is quite large. There is very little atomic oxygen at 85 km at mid latitudes near ±40°. From the plots of mixing ratio and concentration (Figures 6b and 7b and Figures 6d and 7d) at 85 km, there is a maximum at the equator and a minimum at midlatitudes. However, at 98 km, the opposite is true and a minimum occurs at the equator. Again, the amplitude is greater below the peak. This behavior is consistent with the migrating diurnal tide.

[32] For solstice conditions (Figure 7), the basic form of the oxygen variations is similar to the equinox below ∼95 km, but there are differences of detail. The local minimum at ∼95 km over the equator is not as localized as the one during equinox and extends upward above 100 km (Figure 7d) in a manner somewhat reminiscent of evanescent behavior. There is a marked asymmetry in the height of the local minima and maxima between hemispheres. This variation has been noted earlier in winds [McLandress, 1997].

[33] Compared to TIME-GCM results, the observations show more variability. In addition, as with equinox, oxygen decreases more rapidly at lower heights than the observations do. The TIME-GCM does model the hemispheric asymmetry appropriately.

4.3. Latitude–Local Time Slices of the Atomic Oxygen Data

[34] The local time and latitudinal amplitudes and phases of the features observed in the preceding sections are more clearly illustrated when examining the atomic oxygen data as a function of latitude and local time when the season and the altitude are held fixed. Figure 8 shows six different plots of the September 1994 equinox atomic oxygen at different altitudes. In these plots, the contouring is linear and represents the percent deviation from the mean at that altitude.

Figure 8.

Latitude/local time slices of derived atomic oxygen for many altitudes under equinox conditions. The contours represent percent deviations from the mean at each altitude.

[35] The features that are observed in the previous figures can be seen more clearly in this figure. Since it is at equinox, the oxygen is basically symmetric about the equator and the local time features observed in the earlier figures are seen in this cut. For example, by examining the equatorial or midnight cuts in these plots, the same information is found as in the contour plots mentioned in the previous two sections. The phase of these variations migrate as a function of altitude. For example, the equatorial maximum and minimum is seen migrating to earlier hours as we go up in altitude from 86 to 102 km.

[36] Not shown are the results from the solstices. At these times, there is a clear anti-symmetry between the June and December solstices. The winter hemisphere always has greater amounts of atomic oxygen at midlatitudes near 40° when compared with the summer hemisphere at 40°, and the rate at which atomic oxygen decreases from 40° toward the tropical latitudes is also much higher. This latitudinal variability is also noted in a paper by Russell et al. [2004]. In fact, the atomic oxygen concentration can drop quite drastically in only a few degrees of latitude.

[37] The form of these features and their phases remain similar from year to year. The only difference is at higher altitudes above the peak altitude of 95 km. In 1992, there is more atomic oxygen at these altitudes than in the following years. This interannual difference is likely caused by the slowly decreasing flux of radiation from the sun due to the solar cycle which peaked in 1991. A discussion of this will be given in a future paper.

5. Discussion

[38] Our data set of atomic oxygen clearly shows that this minor chemical species is extremely variable in time and space and exhibits systematic variations in amplitudes and phases as functions of altitude, latitude, local time, and season. These features are the results of our dynamical atmosphere. Our data set improves upon existing atomic oxygen reference models such as that published by Llewellyn and McDade [1996]. Their model is derived from monthly ozone profiles taken at fixed local times in the mesopause region, whereas our results include the local time behavior during the night.

[39] It is well documented that measured airglow features are associated with the diurnal 24-hour tide [Shepherd et al., 1995]. That these variations are caused by tides is supported through comparisons with atmospheric models [Yee et al., 1997; Zhang et al., 2001]. As expected, very similar results are found between the local time variation of atomic oxygen concentration at the equator seen in Figure 3 and the greenline emission rate observed by Shepherd et al. [1995]. Many papers [Ward, 1998; Angelats i Coll and Forbes, 1998] have proposed that vertical advection is the primary cause of these features. The 25-km vertical wavelength of the local time variations corresponds to the vertical tidal wavelength observed in tidal wind analyses [McLandress et al., 1996].

[40] A simple model for simulating the tides was developed by Ward [1999] using latitudinal variations with (1,1) Hough modes describing variations in vertical transport. A comparison is made between the features found in Figure 8 and those from his simple model. Figure 9 contains an example output of the percentage difference from the mean of atomic oxygen using the same contouring scheme as in Figure 8. As can been seen in the simple model, the tidal patterns are very similar in phase to that in the derived atomic oxygen from WINDII at 86 km and above. At 82 km the phases do not match. This is because the model assumes that atomic oxygen is a dynamical tracer of the motion. This assumption breaks down below roughly 85 km where the recombination rate becomes fast enough that the chemical timescale is shorter than the dynamical timescale.

Figure 9.

Latitude/local time slices of modeled atomic oxygen from a simple tidal model for many altitudes under equinox conditions. This contour is under the same scheme as that for Figure 8. The contours represent percent deviations from the mean at each altitude.

[41] These atomic oxygen variations are also compared to the MSISE-90 output for atomic oxygen under the same conditions as the collected data. Figure 10 is a set of three contour plots that show the difference between the MSISE-90 output and that of either the derived atomic oxygen or the TIME-GCM atomic oxygen that is shown in Figure 4. It is clear that the tidal structure of the MSISE-90 model is incorrect and is in fact 180° out of phase with the data, the TIME-GCM, and the simple model. As noted by Ward [1998], vertical motions dominate the tidal behavior of the oxygen and airglow and produce variations out of phase to the effects of temperature and density variations. MSIS only includes the density effect and hence produces concentration variations 180° out of phase relative to the observed variations. This will have a large impact on any model that uses diurnal variations of atomic oxygen from MSIS.

Figure 10.

Altitude/local time slices of atomic oxygen for the equatorial bin under equinox conditions. (a) Atomic oxygen mixing ratio as a percent deviation from the mean at each altitude as seen in Figure 3. (b) Atomic oxygen from the TIME-GCM in the same manner. (c) Atomic oxygen MSISE-90 results, clearly showing the temporal 180° phase difference.

6. Conclusions

[42] A latitude/local time data set which extends from the equator to midlatitudes has been constructed from atomic oxygen [80 to 115 km] profiles. Two different airglow emissions, namely the hydroxyl P1(3) line of the Meinel (8–3) band (λ = 734 nm) and the atomic oxygen greenline (λ = 557.7 nm), were used to determine atomic oxygen using the approaches described by Russell and Lowe [2003] and McDade et al. [1986], respectively. The technique of Russell and Lowe [2003] was characterized, and the differences between the oxygen profile derived from the two emissions was shown to be consistent with the observed variability in atomic oxygen. That these profiles, measured independently, should be in such good agreement also validates the instrument calibration and associated photochemistry.

[43] The two sources combined give an atomic oxygen profile that covers well above and below its peak altitude of around 95 km. It is shown that the atomic oxygen concentration is extremely variable as a function of local time and latitude. In this paper, atomic oxygen climatology is studied as a function of altitude, latitude, local time, and season. It is found that the atomic oxygen concentrations agreed well with existing models such as the TIME-GCM, although small differences in the phase of the local time variations are noticed and the background oxygen decreased more rapidly with decreasing height than the observations. The primary features in the latitude and local time variations also agreed well with a simple tidal model that used the (1,1) Hough modes for latitudinal variations in vertical transport.

[44] This data set shows that the atomic oxygen varies strongly with local time. The variability depends on latitude, local time, and season. From these data it is clear that local time variations must be accounted for in order to understand background variations in the oxygen mixing ratio. Although definitive identification of the local time variability with tidal advection is not possible due to limited local time coverage, the comparisons with models suggest that this is the case.

[45] This atomic oxygen data set is available for all seasons and over the multiyear period when the WINDII instrument was working (1991–1997). These data can be used to help modelers in validating their models, can be compared to future atomic oxygen rocket measurements, and can be compared to atomic oxygen derived from future radiance measurements. Since atomic oxygen in the mesopause region is involved in the energy budget and is a quasi-conserved tracer for transport, knowledge of its temporal and geographical behavior is essential for understanding this remote region of the atmosphere.

Acknowledgments

[46] This work was supported through an NSERC Discovery Grant and through an NSERC Strategy Grant: Global Chemistry and Climate. The WINDII project is supported by the Canadian Space Agency and the Centre National d'Etudes Spatiales of France.

Ancillary