Remote sensing of atomic oxygen in auroral rocket experiments using topside zenith viewing O/N2 brightness ratios

JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 102, NO. A2, PAGES 2475-2482, FEBRUARY 1, 1997 Remote sensing of atomic oxygen in auroral rocket experiments using topside zenith viewing O/N2 brightness ratios D. J. Strickland and T. Majeed Computational Physics, Inc., Fairfax, Virginia A. B. Christensen and J. H. Hecht The Aerospace Corporation, Los Angeles, California Abstract. A study is presented that discusses the information content of a zenith viewing brightness ratio for optically thin O and N: emissions arising from auroral electron impact excitation. The discussion focuses on the altitude dependence of the ratio and distinguishes between a topside portion and the rest which lies below. The topside is defined as the region in which the auroral electron flux is essentially constant with altitude. The lower boundary to the topside is a function of the averag energy <E> of the precipitating electrons. The profile of the brightness ratio is a function of <E> and the ratio of O to N: column densities (designated as O/N:). On the topside, the <E> dependence disappears leaving a unique relationship between the brightness ratio and O/N:. Thus profiles of both quantities have identical shapes on the topside. The shape depends on the exospheric temperature T o. Thus a model-based brightness ratio being used to compare with measurements on the topside mustake this into account. The temperature can be quantified from measurements having good counting statistics or in their absence from an atmospheric model such as the mass spectrometer/incoherent scatter (MSIS) model. Without knowledge of <E>, data/model comparisons must be restricted to the topside in order to uniquely specify O/N:. With such knowledge, however, data/model comparisons may be extended into the region below the topside. The chief advantages are better counting statistics and the decreasing dependence of the brightness ratio on T o with decreasing altitude. Derived values of O/N: may be related to O density profiles throughout the lower thermosphere with the use of a model atmosphere. MSIS-83 is used in this work following the preference of J. H. Hecht and colleagues in analyzinground-bas, ed optical data. The findings discussed above are used to analyze zenith viewing OI 844.6/N2391.4 ratio data from the Atmospheric Response in Aurora (ARIA) flights I and IV. The ARIA I data support an O concentration that is a factor of-1.1 above MSIS-83 (similar to the findings ofhecht et al. [1995]), while the ARIA IV data lead to a factor of-0.75 referenced to the ARIA IV MSIS atmosphere. Both atmospheres possess similar O/N: values below 120 kin. Within relative calibration errors between the experiments, one may conclude that the ARIA IV O concentrations throughout the lower thermosphere (120 km- 200 kin) were less than ARIA I by a factor of-0.7. 1. Introduction the abundance of atomic oxygen O and molecular nitrogen N2 from observations of oxygen and nitrogen emissions excited The work being reported is part of an ongoing auroral by electron impact excitation. We will address filter investigation based on rocket and ground-based data from the photometer data from ARIA I and IV campaigns with specific Atmospheric Response in Aurora (ARIA) program. There attention to the observed altitude profiles of the zenith viewing have been four campaigns with launches out of Poker Flat, brightness ratio of OI 844.6-nm and N2* 1NG (farst negative Alaska: ARIA I (March 3, 1992), ARIA II (February 12, group) 391.4-nm emission. Remote sensing of thermospheric 1994), ARIA III (February 2, 1995), and ARIA IV O is an active area of investigation because of its predicted (November 27, 1995). Results from the farst campaign have and observed spatial variability on regional and global scales. been published by Anderson et al. [1995] and colleagues in While solar ultraviolet radiation creates O at thermospheric companion papers. In this paper, we describe an analysis heights via photodissociation of 02, it becomes redistributed technique of general applicability and apply it to specifi cases by diffusion and mass motion in the thermosphere and of ARIA photometer observations of the diffuse aurora. Our mesosphere. Important forcing terms include upward objective is to infer properties of the atmosphere, in particular propagating tides and high latitude Joule and particle heating. Clear examples of O redistribution by high latitude forcing Copyright 1997 by the American Geophysical Union can be seen in DE- 1 UV images [Craven and Frank, 1984; Craven et al., 1994]. Heating at high latitudes introduces upwelling and downwelling motions that both deplete and enhance the relative concentration of O. Examples of theory Paper number 96JA03565. 0148-0227/97/96JA-03565509.00 2475

2476 STRICKLAND ET AL.' REMOTE SENSING OF ATOMIC OXYGEN IN AURORA papers that have addressed the process are those of Robie et The lower boundary decreases in altitude with increasing a/. [ 1987], Rees and Fuller-Rowell [ 1988], Prolss [ 1987], <E>. Figure 1 illustrates the problem by showing altitude Bums eta/. [ 1989], and Bums and Killeen [ 1991]. Remote profiles of the spherically integrated electron flux 4)(z,E) in sensing of the variability of relative O concentrations during units of electrons cm'2s4ev 4 where z and E are altitude and aurora has focused on the interpretation of ground-based energy, respectively. The upper panel shows fluxes for an brightness measurements of OI 630.0 nm, OI 777.4 nm, OI isotropic incident electron flux characterized by a pure 844.6 nm, and N 2 * 1NG 427.8 nm [Strickland et al., 1989; Maxwellian distribution having an averag energy <E> of 3 Hecht eta/., 1989, 1991, 1995]. The overall findings indicate kev and an energy flux Q of 1 erg cm'2s 4. Fluxes are shown reductions in the abundance of O relative to N2 in the lower for selected energies from 20 to 4000 ev. The lower panel thermosphere by more than a factor of two during aurora with show similar fluxes for a more energetic Maxwellian having a temporal scale comparable to the time scale for auroral an <E> value of 6 kev and a Q value also of 1 erg cm'2s 4. substorms. The fluxes were obtained by the electron transport model of In this paper we examine the brightness ratio of emissions Strickland et al. [1976] with updates most recently discussed arising from O and N with emphasis on the altitude region by Basu et al. [1993] and Strickland et al. [1993]. Our choice well above the height of the peak in the volume emission rate. of Maxwellians is based on the electron flux measurements of In this region, which we refer to as the topside (generally above 130-150 km), the ratio of emissions depends only on the ratio of the column densities of O to N (designated by O/N ) referenced to the observing altitude. This is a consequence of the near constancy of the electron energy flux 20eV ' i I with altitude on the topside. Hence a unique and proportional... 40eV i i I relationship exists between the brightness ratio and O/N 100 ev i i I,' above the observing altitude. This will be true regardless of 1000 ev I ' the distribution of the two species (whether there are... 4000 ev! departures from diffusive equilibrium or not). While emphasis is on the topside, nevertheless, we will extend our I I,: analysis below this region where the brightness ratio also becomes dependent by spectral characteristics of the 150 ß / I'... I,:,:... precipitating electrons (parameterized by us in the form of the,' / / average energy <E> in kiloelectronvolts). This is made / /'/ / possible by knowledge of <E>, which for the ARIA,,' /' /,, campaigns comes from particle measurements to be discussed later in the paper. While the immediate product from a topside O/N2 100...,... i i i i i ii brightness ratio profile is the corresponding O/N2 abundance 103 104 105 106 107 108 ratio profile, knowledge of the exospheric temperature Texo is Spherically Integrated Electron Flux (cm'2 s '1 ev '1) needed to convert this information to an overall concentration of O in and below the topside. As will be shown, the 200 brightness ratio and corresponding column density ratio are 20 ev sensitive to Texo. If the counting statistics are sufficient, a... 40 ev measured slope of the altitude profile of the O/N brightness 100 ev ratio up to at least an N scale height (>30 km) into the 1000 ev topside region can be used to infer Texo. The ratio, rather than... 10,000 ev the brightness of the features, themselves, should be used since any changes in Q (energy flux of precipitating electrons) <E> = 6 kev and mean energy <E>, while reflected in the brightness, will not affect the ratio. If the exosphefic temperature cannot be satisfactorily determined from the measurements, a model i i t : atmosphere can be used to estimate its value. The mass spectrometer/incoherent scatter model of Hedin [1983] (MSIS-83) was used in this analysis following the example of ] /! Hecht et al. [1991]. Details follow on our analysis of the,/,/'//.."/ observed altitude profile of the brightness ratio 844.6/391.4 as well as on model results that illustrate the behavior of this ratio as a function of <E>, O/N and Texo....,...,... lo 3 104 105 106 107 108 Spherically Integrated Electron Flux (cm'2 s 4 ev 4) 2. Topside Behavior of Auroral Electron Fluxes Figure 1. Spherically integrated electron fluxes versus and Optical Emissions altitude. Upper and lower panel show fluxes for incident spectra given by Maxwellians with averagenergies <E> of 3 The topside, as introduced the previous section, has an and 6 kev, respectively. The horizontal dashed lines show inexact lower boundary depending on <E> and on how much our choices for the lower boundary to the topside region. The decrease in the electron flux is acceptable at this boundary figure illustrates the slow variation of the fluxes with altitude compared to its near constant behavior at higher altitudes. in the topside region.

STRICKLAND ET AL.' REMOTE SENSING OF ATOMIC OXYGEN IN AURORA 2477 J. Sharber (personal communication, 1996) on board the ARIA I and IV rockets. Diffuse aurora was observed on these flights for which the measured electron fluxes exhibited a Maxwellian behavior with typical <E> values of 6 and 3 kev for ARIA I and IV, respectively. The horizontal dashed lines in Figure 1 illustrate our choices of topside lower boundaries for the two <E> values. An MSIS-83 atmosphere [Hedin, 1983] was used to specify the density distributions of N,., O,.,. and O with the following input parameter values: F 0.7-- 130 <F10.7 > =, Ap = 11, year equal to 1995, day equal to 331 = UT equal to 29,434 s, latitude equal to 65øN, longitude equal to 160øW. Our choices of O and N,. emission features for illustrating the behavior of their brightness ratio as a function of <E> and O/N,. are OI 844.6 nm and N,. 391.4 nm. The excitation mechanisms are electron impact on O and N,., respectively. A second source of 844.6 nm is impact on O,.. Thisource insignificant for the range of <E> values discussed here (see Strickland et al. [1989] for further information on the importance of the O,. source relative to impact on O). The OI 844.6-nm and N + 391.4-nm emissions are easily measured in aurora, experience no absorption at rocket altitudes, and are easy to model following the transport calculations that provide the electron flux (z,e). Furthermore, zenith brightness of these features was recorded on board the rockets from all of the ARIA campaigns. Figure 2 shows altitude profiles of the v zenith viewing brightness of these features for selected incident electron fluxes given by Maxwellians. The chosen 130 <E> values are 2, 4, 6, and 8 kev and Q = 1 erg cm% ' for 0 all cases. The above discussed MSIS model atmosphere was 'r, \. X ', used to derive the results in Figure 2. The selected altitude X,N', ' range is the same for which data will be presented in section o 5. The electron impact cross sections for 844.6 nm and 391.4 nm used to calculate the brightness of these features will be discussed shortly. The upper panel contains 844.6-nm...... <E> = 2 kev <E> = 4 kev <E> = 6 kev <E> = 8 kev brightness profiles and are seen to decrease with increasing,, i i i i iii i i i i i <E>. The decrease on the topside is due to lower energy 1 10 1 O0 1000 deposition in this region which in turn is due to the shifting of Zenith Viewing Brightness (Rayleighs) the energy deposition profile to lower altitudes. The decrease Figure 2. Zenith viewing brightness of OI 844.6 nm and N + at the lower altitudes with increasing <E> is due to less ING 391.4 nm for incident electron spectra given by o 1 01844.6 nm <E> = 2 kev... <E> = 4 kev <E> = 6 kev... <E> = 8 kev N2 + 391.4 nm '\,\ \\\",,, '\ \ ',, '\ I I I I I IJ I i i i i i i i J I * I I I I I I I I 10 1 O0 1000 Zenith Viewing Brightness (Rayleighs) \,\ \\",,, '\,\\ overall energy being deposited in O given the facthathe Maxwellians with average energies (<E>) of 2, 4, 6, and 8 peak region of energy deposition is moving to lower altitudes kev, energy fluxes (Q) of 1 erg cm% ' and isotropy over the where the density rati of O to N,. is decreasing. For the same downward hemisphere. reason, the 391.4-nm brightness is increasing with <E> at low observing altitudes as can be seen in the lower panel. Figure 3 shows altitude profiles of the brightness ratio of 844.6 nm to 391.4 nm. The upper panel shows four given altitude above -150 km. The region above 150 km is 844.6/391.4 curves for the four <E> values used to produce the common topside region for the selected energies and is the results in Figure 2. A fifth curve is O/N,. normalized to the where the brightness ratio has an identical shape to that of other curves at km. The unnormalized O/N,_ comes from O/N,.. The normalized O/N2 profile has been included to show the MSIS atmosphere used to calculate the other four curves. the points of departure of the 844.6/391.4 profiles from the The purpose of including the O/N,. profile will be given O/N,. shape as one moves down from the topside. For <E> = shortly. The lower panel shows four curves for a single value 2 kev, the deviation becomes noticeable near 160 km. For of <E> (4 kev) and four model atmospheres. One is the <E> = 6 kev, the deviation does not begin until reaching an MSIS atmosphere used to produce the results in the previous altitude of-130 km which is consistent with the chosen figures. The other three are variations on this atmosphere topside lower boundary in Figure 1 for this energy (130 km). produced by scaling the O density profile using scaling factors In all cases, once the deviation begins, knowledge of <E> is of 1.5, 0.5, and 0.25. The brightness ratio is directly required in order to relate the brightness ratio to O/N,.. The proportional to the O scalings on the topside (beginning at low-altitude behavior is well known and has been utilized in -130 km for <E> = 4 kev) given the near constant behavior ground-based remote sensing studies by Strickland et al. of the electron flux with altitude (in other words, the [1989] and Hecht et al. [1989, 1991, 1995]. The decrease in distributions of the densities are not important). In the upper the ratio with increasing <E> is the result of a shift in the panel, the ratio is the same over the selected <E> range at any energy deposition profile to lower altitudes which favors II

2478 STRICKLAND ET AL.' REMOTE SENSING OF ATOlVlIC OXYGEN IN AURORA Model O/N 2 (Scaled) <E> = 2 kev <E> = 4 kev <E> = 6 kev <E> = 8 kev effectiv energy range for producing 844.6-nm and 391.4-nm emission. To illustrate the point, examples of fluxes and brightness intelgrands for 844.6 nm and 391.4 nm are shown in Figure 4 (The form of the integrand is n(z) (z,e)o0 ) where n and o are density and cross section, respectively). The upper panel shows spherically integrated electron fluxes at 200 km for <E> = 3 and 6 kev along with the emission cross sections for 844.6 nm and 391.4 nm. The 844.6-nm cross section was provided by R. Meier (personal communication, 1991) who revised the magnitude of the Julienne and Davis [1976] 844.6-nm cross section with his 1.0 10.0 Zenith Viewing Brightness Ratio (OI 844.6 nm/n2 + 391.4 nm) fo=0.25 /,,,/ fo=0.50 /," I i '1 fo = 1.0 /," /, fo= 1.5 /," /,/ <E> = 4 kev / / /!!ø ;;! 00ø!,," /,/,' /,/'/ 10' I...,,,,, 1040 107 06 1 10 '." x, 0.01 0 1.00 10.00, Zenith Viewing Brightness Ratio (OI 844.6 nm/n2 + 391.4 nm) 4-' Figure 3. 844.6/391.4 zenith viewing brightness ratios. The ratios m the upper panel use the results from Figure 2. This 104 panel also contains an O/N2 profile normalized to the other profiles at km to showhere the brightness ratios depart in shape from O/N2. The ratios in the lower panel refer to the single <E> value of 4 kev are obtained for four model 1042 atmospheres differing from one another by scaling the O density profile with the factors shown in the panel. The results in this figure illustrate two important topside characteristics: insensitivity of 844.6/391.4 to the incident 1043 electron flux and direct proportionality to O/N2.... <E> = 3 kev... <E> = 6 kev / /'\ "-.. I \ "':x-,... - - 104.1_" NNOi844.6n,N, '",. [I \, '\, ", ø '1 t x, > x,_> \,,,,,1,,, i,,,i... ',,ll,'",,, 10:1,i,,, 10 10: 10 10" Energy (ev) _... <E> = 3 kev... <E> = 6 kev 10! 102 103 Energy (ev) Figure 4. Spherically integrated electron fluxes, cross energy absorption by N2 relative to O. The combined results sections, and their products. The upper panel shows fluxes at in the two panels clearly illustrate the greater difficulty of 200 km for <E> values of 3 and 6 kev (scale to the left) and determining O/N2 from observations of 844.6/391.4 below the the electron impact cross sections for 844.6 and 391.4 nm topside. The above ground-based remote sensing studies (scale to the right). The lower panel shows the product of flux addressed this problem by supplementing the above features and cross section (brightness integrand less the density) for both features and fluxes. The products for <E> = 6 kev have (with 427.8 nm in place of 391.4 nm) with OI 777.4 nm and been scaled (same factor for 844.6 nm and 391.4 nm) for OI 630.0 nm to determine both <E> and O/N2 (actually an O improved visual comparison between the two <E> cases. The scaling factor for an MSIS atmosphere) results in the lower panel illustrate the effectivenergy range In order to obtain an 844.6/391.4 ratio that is independent for excitation and insensitivity of the ratio of integrands (i.e., of <E> on the topside, the energy dependence of the electron insensitivity of brightness ratio) to changes in the incident flux must be similar from one <E> to another over the electron flux. ' 1046 105 10 47

STRICKLAND ET AL.' REMOTE SENSING OF ATOMIC OXYGEN IN AURORA 2479 own estimates of cascade into the upper state of the transition illustrates the effect on O/N2 and the corresponding (3p3P) under optically thick conditions. Further details are 844.6/391.4 brightness ratio profile by changing T,xo. The given by Hecht et al. [1991, Appendix 1]. The 391.4-nm upper panel shows the brightness ratio for MSIS-83 cros section at 100 ev is from Doering and Yang [1996] and atmospheres using input parameters appropriate to the two uses the shape of the Borst and Zipf[1970] 391.4-nm cross experiments (ARIA I' F 0.7 = 178, <F 0.7> = 210, Ap = 11, year section. The lower panel of Figure 4 shows intelgrands for equal to 1992, day equal to 63, UT equal to 50,904 s, latitude both <E> values where the intelgrands for <E> = 6 kev have equal to 65.1øN, longitude equal to 148øW; ARIA IV: F 0.7 = been sealed to better compare their shapes with those for <E>, <F 0.7> = 73, Ap = 11, year equal to 1995, day equal to = 3 kev. It is important to note that the same scaling factor 331, UT equal to 29,434 s, latitude equal to 65. has been used for both features. The effectiv energy range for equal to 148øW). The exospheric temperatures are 1025 K producing the emissions from about 20 ev to 1000 ev. and 732 K, respectively. The selected incident electron flux The relative changes between the 844.6-nm and 391.4-nm for deriving the brightness ratios is a Maxwellian distribution intelgrands from <E> = 3 to 6 kev are insignificant, thus with <E> = 4 kev. The O/N2 profile is seen to rise more illustrating the shapes of the two fluxes over the effective rapidly with altitude on the topside for the colder model. The integration range are essentially the same. cause of this effect is the faster decrease with altitude of the N2 As noted in section 1, exospheric temperature must be density relative to O for a cold atmosphere compared to a taken into account in order to interpret derived topside O/N2 warmer one. The change in the brightness ratio on the topside values in terms of overall O concentrations. Figure 5 due to changing T,xo is seen to be mirror the change in O/N2 which again illustrates the proportional relationship between these quantities. l q, longitude 3. Selecting a Lower Boundary for Data Analysis,, i,, Analysis scenarios may be divided into two categories specific to the problem of relating O/N2 brightness ratios to O/N2 values. In the first category are those situations where useful information is not available on <E> along the rocket trajectory. This arises where particle measurements have not <E> = 4 kev been made and Q and <E> are changing. Complicated structure can occur in the individual brightness profiles, especially during active aurora. Such structure may also exist ARIA I MSIS (Tex o = 1025 K) in a given O/N2 brightness ratio at low observing altitudes but ARIA IV MSIS (Tex o = 732 K) will then become smooth on the topside where neither changes in Q nor <E> manifest themselves in the ratio. For lack of knowledge about <E>, the analysis must be restricted to the topside in order to derive unique values of O/N2. One!! 8O can only guess the altitude of the topside lower boundary. In 1.0 10.0 this case, statistical arguments can be used about the Zenith Viewing Brightness Ratio (OI 844.6 nm/n2 + 391.4 nm) likelihood that <E> is large enough to place the desired starting altitude for the analysis in the topside region. For example, we may select 150 km as the "safe" starting point. This requires that <E> be above a few kiloelectronvolts as demonstrated in Figure 3. Statistically, the probability is favorable for this in auroras with Q values above a few ergs per square centimeter per second. Such auroras are common and are usually targeted in rocket experiments. The probability, of course, worsens that the observed aurora has 130 the minimum required <E> value as one lowers the starting altitude. To begin the analysis at, for example, 130 km requires that <E> be above approximately 5 kev. In the second scenario, reliable information is available on ARIA I MSIS (T o = 1025 K) <E>. This will come from either particle measurements or ARIA IV MSIS (T = 732 K) from a purely optical experiment for which <E> is essentially constant along the trajectory. In the latter situation, the brightness profiles themselves will be smooth everywhere, 8O I i i i i i i I [ i I i i i i i i and <E> may then be determined from the shapes of the 1.0 10.0 profiles as well as from profiles of their ratios (assumed to be O/N measured from the bottom of the emitting layer up to altitudes Figure 5. Calculated 844.6/391.4 brightness ratios (upper within the topside region). In either case, the topside lower panel) and corresponding O/N2 profiles (lower panel) for boundary can effectively be specified with knowledge of <E>. ARIA I and IV MSIS model atmospheres. The exospheric Unlike scenario 1, there is no need to restrict analysis of the temperatures are 1025 K and 732 K, respectively. observations to the topside since the effect of <E> on the ratio

24 STRICKLAND ET AL.' REMOTE SENSING OF ATOMIC OXYGEN IN AURORA below the topside can be incorporated into calculated ratios profile point (effectively at the lower boundary of the emitting used to interpret the data. The second scenario applies here layer). The approach taken is to assume the shapes of the O where particle data give us knowledge of <E>. Model/data and N2 density profiles and derive a scaling factor that gives comparisons will be considered down to 120 km. The only the magnitude of the O profile relative to that of N2 caveato extending the analysis below the topside is the [Strickland et al., 1989; Hecht et al., 1989, 1991, 1995]. possibility that errors will be introduced in derived O/N2 values in this region due to any deviations in the shapes of the 4. Relating O/N 2 to O Density Profiles model O and N2 density profiles from the true shapes of the profiles in the region being observed. While deviations have A model atmosphere must be introduced to map from O/N2 no impact on derived O/N2 values on the topside, errors can be to volume densities. Furthermore, model atmosphere must introduced at lower altitudes where the electron flux is no have the correct exospheric temperature. By meeting this longer constant with altitude. This issue becomes most condition, a profile the model O/N2 brightness ratio may be important for remote sensing of both <E> and O/N2 from the compared with the data since both will have the same shape. ground where brightness ratios are measured at only a single The rati of profiles yields a scaling factor that we apply to the model O density profile. The scaled profile gives a meaningful concentration of O relative to the model N2 density ' profile throughout the lower thermosphere provided the model itself gives physically realistic shapes for the two density ARIA I profiles. The derivation of an O scaling factor is the best that can be done since these shapes must be assumed. We know, 160 140 120 100 160 140 120 100 Data Model i i i i i i [ i i i i i i i i 1.0 10.0 Zenith Viewing Brightness Ratio (OI 844.6 nm/n2* 391.4 nm) / // /// / / ARIA IV Data. Model however, that the shapes are correct, at least on the topside as long as the model O/N2 brightness ratio shape agrees with that of the topside data. In order to further relate the O concentrations from one experiment to the other, it is necessary to farst compare the two model N2 density profiles since the derived O concentrations are relative to the model N2 density profiles. Near the bottom of the present region of interest (120 km) where exospheric temperatureffects are small, the MSIS atmospheres used to analyze the ARIA I and IV data give essentially the same N2 densities. Therefore, the ratio of O scaling factors provides a measure of overall change in O composition between experiments, again, assuming realistic model density profile shapes in and below the topside. Differences between model shapes of density profiles in the lower thermosphere and true shapes can arise from two sources: an incorrect model temperature profile in the lower thennosphere and departures in this region from diffusive equilibrium. The latter topic has been addressed by various modelers concerned with thermospheric response to highlatitude forcing (see the references in section 1 addressing thermospheric dynamics). An assessment of possible departures in MSIS-83 density profile shapes from the true shapes associated with the ARIA I and IV experiments is beyond the scope of this paper. For the moderate levels of disturbance that occurre during and before the experiments, it is unlikely that any significant errors will be introduced into our determinations of O concentrations based on the assumed shapes of the model density profiles. 5. Observations, Discussion, and Conclusions 1.0 10.0 Figure 6 shows measured and calculated 844.6/391.4 profiles from the upleg portions of the ARIA I and IV flights. Zenith Viewing Brightness Ratio (OI 844.6 nm/n2* 391.4 nm) A solar resonant fluorescence scattering component of 90 Rayleighs has been subtracted from the ARIA I 391.4-nm Figure 6. Measured and calculated 844.6/391.4 brightness data consistent with the treatment by Hecht et al. [ 1995]. No ratios for ARIA I (upper panel) and ARIA IV (lower panel). The calculated topside profiles are based on MSIS-83 subtraction was necessary for ARIA IV since the solar zenith atmospheres appropriate to the times and locations of the angles along the rocket trajectory exceeded values for which experiments. The calculated profiles require scalings of-1.1 solar photons could penetrate into the F- region being viewed. and -0.75 to agree with the ARiA I and IV data, respectively. Hecht et al. [1995] show a profile in Figure 13 of their paper The fits are achieved by applying these factors to the O density similar to the upper panel in Figure 6. There are differences, profiles of the two MSIS model atmospheres. however, which can be explained by the fact that Hecht et al.

_. _. STRICKLAND ET AL.: REMOTE SENSING OF ATOMIC OXYGEN IN AURORA 2481 160 loo 160 140- '~ 120- O ARIA... ARIA IV i! i i i i i [ I i i i,, i i 1.0 10.0 Zenith Viewing Brightness Ratio (OI 844.6 nm/n2 + 391.4 nm),/ I /i /I // I /' I I ARIAI(T... =1025K) experiments. Introducing differences in T, o (assuming MSIS is providing reasonable values), however, leads to a 100 t' [ ARIA IV(T... =732K) significantly different conclusion as can be seen with the aid I,! I I I I I I _._ ARI, A!V itex o 1025 I I! of the lower panel. The solid curve is an extension of the I I calculated curve in the upper panel of Figure 6 down to 1.0 10.0 Zenith Viewing Brightness Ratio (OI 844.6 nm/n2 + 391.4 nm) km. Similarly, the dashed curve is an extension of the calculated curve in the lower panel of Figure 6. The third Figure 7. Redisplay of the data for easier comparison (upper curve (dash-dotted) is common with the dashed curve in panel) and three calculated brightness profiles (lower panel) to incident electron flux (<E> = 3 kev) while then common with help interpret the data over the range displayed. The solid and the solid curve in the choice of MSIS atmosphere. Using this dashed curves are extended versions of the calculated profiles curve to interpret topside ARIA IV O/N2 in terms of the in Figure 6. The dash-dotted curve is based on the same <E> overall O concentration would lead to a serious error. It is value as the dashed curve but was calculated for the same clear, even without the aid of detail model results like those in MSIS atmosphere used to generate the solid curve. The three profiles illustrate <E> and Texo effects that must be taken into the lower panel, that there was significantly less O (by a factor account specific to our analysis that has been extended below of-.7 between experiments) during the ARIA IV experiment the topside. given the knowledge that softer precipitation was occumng and that the atmosphere was colder. In this situation, the ARIA IV data profile cannot cross the ARIA I profile for subtracted both an estimated solar resonant fluorescence calculated profile is shown down to 120 km since we know from the particle data that precipitation was similar down to this altitude as was occumng at higher altitudes. The calculated profile needs to be scaled by -1.1 in order to agree with the data. On the basis of the discussion of Figure 3 in section 2, the model O/N2 (or equivalently, the model O density profile) needs to be scaled by the same factor. A similar finding was reported by Hecht et al. [1995]. For the comparison in the lower panel of Figure 6, the ARIA IV MSIS atmosphere introduced in section 2 was used along with an incident electron flux given by a Maxwellian distribution with <E> = 3 kev which is similar to the incident fluxes derived from the particle data above 120 km. In contrast to ARIA I, the data here suggest that less O is present relative to MSIS. The needed reduction in the O profile to achieve agreement between the two curves is -0.75. Within the uncertainties associated with the data, the shape of the measured ratio on the topside suggests a larger T,xo than predicted by MSIS. One can see that without the correct value, a single factor cannot be derived from the model/data comparison. The above value of 0.75 was taken from the vicinity of 140 km. Given that the MSIS temperature is lower than suggested by the data, a larger factor is obtained on the topside, that is, above 150 km. The region near 130 to 140 km is deskable for specifying the factor in this analysis given that <E> is known for this region coupled to the fact that T,xo effects are weak here. A better comparison between the data profiles may be seen in the upper panel of Figure 7. If attention is limited to just the topside (between, for example, 160 and km) and differences in T, o between the experiments are ignored, one will conclude that the same O concentration existed for both similar O concentrations. component from the 391.4-nm profile as well as a component While the discussion in this paper has been specific to from the 844.6-nm profile (an estimate of the brightness at 844.6 nm and 391.4 nm, the application, as noted earlier, is to 200 km). For more details about the use of their particular O and N2 emissions, in general. Many other combinations of form of the profile, the reader is referred to Hecht qt al [1995]. features may be considered among commonly observed Here the profile of interest is as presented, namely, the features such as OI 557.7 nm, OI 630.0 nm, OI 135.6 nm, brightness ratio for all emission arising from auroral electron other N 2 + 1NG bands, and bands of the N2 Second Positive impact excitation. Group (2PG), N2 Vegard-Kaplan (VK), the N2 First Positive The calculated profile in the upper panel was obtained for Group (1PG), and N2 Lyman-Birge-Hopfield (LBH). In many the ARIA I MSIS-83 atmosphere introduced section 2. The cases, complications exist which increase the uncertainty chosen incident electron flux was a Maxwellian with <E> = 6 the inferred values of O/N2. For example, chemistry modeling kev, similar to the ARIA I particle measurements. The must be invoked when considering 557.7, 630.0, and VK.

2482 STRICKLAND ET AL.: REMOTE SENSING OF ATOMIC OXYGEN IN AURORA Some features are weaker than others that increase the Borst, W., and E. C. Zipf; Cross section for electron-impact excitation statistical uncertainty the measurement. Confidence in of the (0,0) first negative band of N 2 + from threshold to 3 kev, calibration factors and in cross-section measurements also Phys. Rev. A, 1 834, 1970. Bums, A. G., T. L. Killeen, G. CrowIcy, B. A. Emery, and R. G. varies from feature to feature. Specific to 391.4 nm, the zenith Robie, On the mechanisms responsible for high-latitude viewing brightness will be contaminated at high altitudes if thermosphericomposition variations during the recovery phase of the solar zenith angle permits resonant fluorescence scattering a geomagnetic storm, J. Geophys. Res., 94, 16,961, 1989. of solar photons off N 2 * in the F-region. If this presents a Bums, A. G., and T. L. Killeen, A theoretical study of thermospheric composition perturbations during an impulsive geomagnetic storm, problem, a good replacement of 391.4 nm is the 337.1-nm J. Geophys. Res., 96, 14,153, 1991. band of the 2PG system. For features that yield good counting Craven, J. D., and L. A. Frank, An apparent signature of statistics and are simple to model (i.e., for which there is a modifications to the upper atmosphere during auroral substorms, one-to-one correspondence between emission and excitation Eos, 65, No. 45, 1021, 1984. Craven, J. D., A. C. Nicholas, L. A. Frank, and D. J. Strickland, by electron impact), the only significant sources of systematic Variations in FUV dayglow brightness following intense auroral error relevant to inferring O/N2 from topside data are activity, Geophys. Res. Let& 21, 2793, 1994. calibration and relative errors between the O and N2 emission Doering, J.P., and Y. Yang, Comparison of the electron impact cross cros sections. For the case of the ARIA program, the relative section for the N 2 * first negative (0,0) band (,3914A) measured by errors in O/N2 between campaigns due to systematic errors optical fluorescence, coincident electron impact, and photoionization experiments, J. Geophys. Res., 101, 19,723, 1996. should be insignificant since the same instruments were used Hecht, J. H., A. B. Christensen, D. J. Strickland, and R. R. Meier, as well as the same cross sections. In this case, the dominant Deducing composition and incident electron spectra from grounderror can be expected to come from statistical fluctuations in based auroral optical measurements: Variations in oxygen density, the data. J. Geophys. Res., 94, 13,553, 1989. Results have been presented that provide new insights into Hecht, J. H., D. J. Strickland, A. B. Christensen, D.C. Kayser, and R. L. Walterscheid, Lower thermospheric composition changes the behavior of zenith viewing O/N2 auroral brightness ratios derived from optical and radar data taken at Sondre Stromfjord for optically thin emissions arising from electron impact during the great magnetic storm of February 1986, J. Geophys. excitation. Unambiguous values of O/N2 may be obtained Res., 96, 5757, 1991. from observations of O/N2 brightness ratios such as Hecht, J. H., A. B. Christensen, D. J. Gutierrez, W. E. Sharp, J. R. 844.6/391.4 on the topside which we have defined as the Sharber, J. D. Winningham, R. A. Frahm, D. J. Strickland, and D. J. McEwen, Observations of the neutral atmosphere between 100 region where the electron flux is essentially constant with to 200 km using ARIA rocketborne and ground-based instruments, altitude. This region begins somewhere around 130 to 140 J. Geophys. Res., 100, 17,285, 1995. km, depending on <E>, and is the transition point from joint Hedin, A.E., A revised thermospheric model based on mass dependence of the brightness ratio on <E> and O/N2 to spectrometer and incoherent scatter data: MSIS-83, J. Geophys. Res., 88, 10,170, 1983. dependence on just the latter. While determination of topside Julienne, P.S., and J. Davis, Cascade and radiation trapping effects on O/N2 from brightness ratio measurements is straightforward, atmospheric atomic oxygen emission excited by electron impact, J. interpretation in terms of O concentrations is more Geophys. Res., 81, 1397, 1976. complicated and requires knowledge of Texo. Fortunately, a Pr61ss, G. W., Storm-induced changes in the thermospheric good measurement of the topside brightness ratio allows one composition at middle latitudes, Planet. Space Sci., 35, 7, 1987. Rees, D., and T. J. Fuller-Rowell, Understanding the transport of to determine this parameter. Otherwise, one must resorto the atomic oxygen within the thermosphere using a numerical global use of a model such as MSIS. The analysis was expanded to thermospheric model, Planet. Space ScJ., 36, 935, 1988. altitudes below the topside with the aid of particle data. These Robie, R. G., J. M. Forbes, and F. A. Marcos, Thermospheric data indicated that similar precipitation occurred on the dynamics during the March 22, 1979, magnetic storm, 1, Model topside and below, at least down to -120 km for either simulations, J. Geophys. Res., 92, 6045, 1987. experiment. A more accurate specification of the overall O Strickland, D. J., D. L. Book, T. P. Coffey, and J. A. Fedder, concentration can be made when the analysis can include data Transport equation techniques for the deposition of auroral electrons, J. Geophys. Res., 81, 2755, 1976. below the topside. This requires knowledge of <E> in the Stfickland, D. J., R. R. Meier, J. H. Hecht, and A. B. Christensen, extended region which, when available, leads to less Deducing composition and incident electron spectra from grounddependence on an accurately specified value of Texo. based auroral optical measurements: Theory and model results, J. Geophys. Res., 94, 13,527, 1989. Strickland, D. J., R. E. Daniell Jr., B. Basu, and J. R. Jasperse, Acknowledgements. Supporto A.B.C. and J.H.H. was provided Transport-theoretic model for the electron-proton-hydrogen atom by NASA grant NAG5-5001 and the Aerospace Sponsored Research aurora, 2, Model results, J. Geophys. Res., 98, 21,533, 1993. Program. The editor thanks two referees for their assistance in evaluating A.B. Christensen and J. H. Hecht, The Aerospace Corporation, Los this paper. Angeles, CA 90009 (e-mail andy_chfistensen@qmai12.aero.org, jim_hecht@qmai12.aero.org References T. Majeed and D. J. Stfickland, Computational Physics, Inc., 2750 Prosperity Avenue, Suite 600, Fairfax, VA 22031. (e-mail Anderson, P. C., et al., The ARIA I rocket campaign, J. Geophys. tariq@euler.cpi.com, and dstrick@euler.cpi.com) Res., 100, 17,265, 1995. Basu, B., J. R. Jasperse, D. J. Strickland, and R. E. Daniell Jr., Transport-theoretic model for the electron-proton-hydrogen atom (Received July 1, 1996; revised November 14, 1996; aurora, 1, Theory, J. Geophys. Res., 98, 21,517, 1993. accepted November 14, 1996)