Effects of temperature dependence of reaction N 2 ( A 3 ∑ + u ) + O on greenline dayglow emission

Abstract. The greenline dayglow emission profiles measured by the Wind Imaging Interferometer (WINDII) on the Upper Atmosphere Research Satellite (UARS) are modelled using recently proposed revisions to the temperature dependent rate coefficient of the reaction N2 (A3 S +u ) + O in the glow model. The volume emission rates of greenline dayglow emissions are calculated using the Hinteregger et al. (1981) and Tobiska (1991) solar flux models. It is found that the average modelled profiles obtained using the Hinteregger et al. (1981) solar flux model with the temperature dependent rate coefficient and a quantum yield of 0.36 for the reaction N2 (A3 S +u ) + O agree to within 8% of the observed profiles in the thermospheric peak region, which shows significant improvement over the earlier results (20% smaller than WINDII results) obtained using the temperature independent reaction rate coefficient. On the other hand, the average modelled profiles obtained with a temperature dependent rate coefficient in the Tobiska (1991) solar flux model are about 12% higher than the WINDII results, whereas with the temperature independent rate coefficient the results are about 10% smaller than the WINDII results in the thermospheric peak region. The present study reveals that the emission profiles obtained using the Hinteregger et al. (1981) solar flux model, along with the temperature dependent rate coefficient and a quantum yield of 0.36 for the reaction N2 (A3 S +u ) + O in glow model, reproduce the thermospheric emission peak as observed by WINDII, a capability which eluded earlier models. These findings support the newly discovered temperature dependence of the rate coefficient of re-action N2 (A3 S +u ) with O. Key words. Ionosphere (ionization mechanisms; modeling and forecasting; general or miscellaneous)


Introduction
In recent years, a number of studies have been done to examine the greenline dayglow emission (Wittase et al., 1999; Correspondence to: V. Singh (virphfph@iitr.ernet.in)Singh and Tyagi, 1997;Singh et al., 1996;Torr et al., 1993;Bates 1990;Akmaev and Shved, 1980;Frederick et al., 1976;Feldman et al., 1971;Wallace and Mc Elory, 1966).These studies have given the theoretical models, as well as the experimental data for this important atomic oxygen emission at 5577 Å.Though the theoretical models have taken into account all the well-known sources of greenline dayglow emission, along with the latest atmospheric parameters such as collision cross sections, reaction rate coefficients and quantum yields, these models are still unable to explain the measurements.These models need further refinement so that a better agreement may be achieved between the model results and the experimental observations.Further, the global measurements of dayglow greenline are needed for comparison purposes.The available database on the greenline dayglow emission has recently been expanded as a result of observations made with the Wind Imaging Interferometer (WINDII) on board the Upper Atmosphere Research Satellite (Shepherd et al., 1993).In the thermospheric emission peak region, photoelectron impact excitation of O, dissociative recombination of O + 2 and the reaction of N 2 (A 3 + u ) with O are the major sources of O( 1 S).The cross sections, quantum yields and reaction rate coefficients for the first two processes have been found consistent with the dayglow emissions.On the other hand, the quantum yield of O( 1 S) production due to the reaction N 2 (A 3 + u ) with O has been the subject of considerable discussion in the literature (Frederick et al., 1976;Torr et al., 1993;Bucsela et al., 1998).In their analysis, Singh et al. (1996) and Tyagi and Singh (1998) used the value of 0.36 for the quantum yield, since it gave the best global fit to all of the measured profiles.However, with this value the model was found to underestimate the measured profiles in the majority of the cases in the thermospheric emission peak region.
In the model calculations of Singh et al. (1996) and Tyagi and Singh (1998), the reaction rate coefficient for the reaction N 2 (A 3 + u ) + O was used from Piper et al. (1981).Recently, Hill et al. (2000) have proposed revisions to this rate coefficient based on their analysis of middle-ultraviolet (MUV) measurements from the rocket-borne MUSTANG spectrometer and the shuttle-borne GLO spectrometer.The major finding of Hill et al. (2000) is that the rate coefficient for the reaction N 2 (A 3 + u ) with O is temperature dependent.This new finding is very important as it has a major impact on earlier dayglow model results, in particular on the greenline dayglow emission.
Recently, Witasse et al. (1999) have modelled O( 1 S) thermospheric dayglow at higher latitudes and higher solar zenith angles (76 • −84 • ) using the TRANSCAR model.This model also underestimated the emission rate in the vicinity of the thermospheric peak in two out of the three cases they considered.For solar zenith angles greater than 70 • , the thermospheric peak is found above 180 km, where the impact of the N 2 (A 3 + u ) reaction is minimal.The authors suggest that the cause of the discrepancy above 180 km might be due to the underestimation of superthermal electron impact (photoelectron flux).
A key problem in the modelling of the dayglow is the appropriate use of solar EUV flux.These solar EUV fluxes are involved in direct and indirect processes of excitation of atmospheric species which result in the production of airglow emissions.Hinteregger et al. (1981) and Tobiska (1991) are the two respective models which provide the solar fluxes under different solar activity conditions.Though these models take into account the variation of solar activity, they have different scaling techniques.Consequently, both of these models provide quite different solar EUV fluxes.One should exercise more care in using the solar EUV flux models.
In this paper, we present a re-analysis of the greenline dayglow emission profiles between 120-250 km measured by WINDII on board the Upper Atmosphere Research Satellite (UARS) in light of the proposed temperature dependence of the rate coefficient for the reaction N 2 (A 3 + u ) with O.The glow model is updated using the temperature dependent rate coefficient of the reaction N 2 (A 3 + u )+ O, and the volume emission rates (VER) of the greenline dayglow emission are calculated using the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models.

Model
Mechanisms for the production of the O( 1 S) dayglow emissions have been discussed by Tyagi and Singh (1998).In the lower thermosphere the four major sources for these emissions are the photoelectron impact excitation of atomic oxygen, dissociative recombination of O + 2 , energy transfer to atomic oxygen from N 2 (A 3 + u ) and photodissociation of molecular oxygen.In the model calculations presented below, the same parameters as used by Tyagi and Singh (1998) are being used, with the exception that the temperature dependence of the rate coefficient of the reaction N 2 (A 3 + u )+ O, as proposed by Hill et al. (2000), is now included.The new rate coefficients for the v = 0, 1 and 2 vibrational levels are (3.4±0.8)× 10 −11 (T/298) 0.5 , (5.6 ±1.3) × 10 − 11 (T/298) 0.5 and (4.8 ± 1.2) −11 (T/298) 0.5 , respectively.In the present calculations we have only used the rate coefficient for v = 0, because for higher vibrational levels the branching ratio is not known.The branching ratio can only be known if the photoelectron impact excitation cross sections for the N 2 (A 3 + u ) state are known separately for vibrational levels v = 0, 1 and 2. Unfortunately, these individual cross sections are not available in the literature.However, the total photoelectron impact excitation cross section for the N 2 (A 3 + u ) state is known, which includes all the possible vibrational levels.Consequently, we have assumed that the N 2 (A 3 + u ) state is produced in the v = 0 vibrational state.The density of N 2 (A 3 + u ) for v = 0 level is obtained under a photochemical equilibrium condition.This assumption may be of some error.However, it was the conclusion of Hill et al. (2000) that the higher vibrational levels contribute less to the production of O( 1 S), and this finding has been supported in the theoretical work of Kirillov (1997).It has been pointed out in the Introduction that the quantum yield of O( 1 S) production due to the reaction of N 2 (A 3 + u ) with O has been the subject of considerable discussion in recent years.Hill et al. (2000) have given a very detailed history of the quantum yield.The various values of the quantum yield, which range from 0.36 to 0.75, have been obtained using airglow and auroral studies by various workers (Frederick et al., 1976;DeSouza et al., 1985;Singh et al., 1996;Bucsela et al., 1998;Strickland et al., 2000).Hill et al. (2000) proposed a value of 0.47 ± 0.17 of the quantum yield for the reaction N 2 (A 3 + u )+ O. Consequently, we have used the values of 0.36, 0.47 and 0.75 for the quantum yield of O( 1 S) production due to the reaction N 2 (A 3 + u )+ O, to obtain the O( 1 S) emission profiles.These three values are used in the model to compare the O( 1 S) emission rates with the WINDII observations and to find out the suitable value which would give better agreement with the WINDII measurements.The O( 1 S) emission profiles are obtained using the production rates from the various sources discussed above and by taking into account the appropriate loss processes (Singh et al., 1996) under photochemical equilibrium conditions.The solar EUV flux plays a very important role in the production of dayglow emissions.The results are quite sensitive to the values of solar irradiance.Consequently, one should exercise care in using the solar EUV flux model.Hinteregger et al. (1981) and Tobiska (1991) are the two respective models which provide the solar EUV fluxes under different solar activity conditions.Though these models take into account the variation of solar activity, they have different scaling techniques.Consequently, both of these models provide quite different solar EUV fluxes.Therefore, it would be more appropriate to make use of both of these models in the calculations and to find out which model gives a better agreement with the measurements.
In the Hinteregger et al. (1981) model the solar EUV flux is scaled using a parameterization method based on daily 10.7 cm solar radio flux and its 81-day average flux.The reference solar spectrum is chosen from SC# 21 REF.For ionizing EUV, Hinteregger et al. (1981) have used a contrast ratio method which is based on the bin structure very similar to Torr and Torr (1985).The Lyman α and Fe XVI (335 Å) enhancement ratios are used to account for solar activity.The following equations are used for the scaling of solar flux: Where SFLUX is scaled solar Flux, and RFLUX is reference solar Flux.S 1 are scaling factosr for Fe (XVI) coronal emission (335 Å), R 1 = 1+ 0.0138 × (F10.7A-71.5)+ 0.005 × (F10.7-F10.7A+ 3.9), and R 2 = 1+ 0.59425 × (F10.7A-71.5)+ 0.3811 × (F10.7-F10.7A+ 3.9).There are 59 values for each S 1 and S 2 , which corresponds to 59 bins of solar spectrum (the values of S 1 and S 2 are given in the subroutine SSFLUX of the glow model).
where HEIEW = He I 10830 Å equivalent width HLYA = H Lyman ∝ flux.The values of the above parameters are listed in the subroutine SSFLUX of glow model.

Results and discussion
The 5577 Å dayglow volume emission rates (VER) are calculated at several latitudes and local times for which WINDII data is available in both hemispheres.For illustration purposes, we have chosen eight cases at various latitudes and local times in both hemispheres.The measured and modelled VER for these cases are shown in Figs.1-5.The VER are calculated using Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models.In Fig. 1 we have shown the comparison between the emission profiles obtained using the quantum yield of 0.36 and 0.75, along with WINDII  tained using the temperature dependent rate coefficient (Hill et al., 2000) and the Hinteregger et al. (1981) solar EUV flux model (with quantum yield 0.36) are in very good agreement with the WINDII observations, and they reproduce the ther- mospheric peak as observed from WINDII within an error of 8%.It is also quite clear from the Fig. 1 that the VER labelled as Hint.(temp.independent, qyield 0.36), obtained using Hinteregger et al. (1981) solar EUV flux model with temperature independent rate coefficient (Piper et al., 1981), are significantly smaller than the measured emission rates in the thermospheric peak region.Further, it is noticeable from Fig. 1 that the emission profiles obtained using the quantum yield of 0.75, along with the temperature independent rate coefficient taken from the Piper et al. (1981), are significantly higher than the WINDII observations for both the solar EUV flux models.These results indicate that the quantum yield of 0.75, which was proposed by Piper et al. (1981), is not suitable to explain the WINDII observations.
In Figs. 2 and 3 we have shown the comparison between the emission profiles obtained using the quantum yield of 0.36 and 0.47 (Hill et al., 2000).It is quite evident from Figs. 2 and 3 that the results obtained using the temperature dependent rate coefficient (Hill et al., 2000) and the Hinteregger et al. (1981) solar EUV flux model (with quantum yield 0.36) are in very good agreement with the WINDII observations, and they reproduce the thermospheric peak as observed from WINDII within an error of 8%.It is also quite evident from Figs. 2 and 3 that the emission profiles obtained using the quantum yield of 0.47, as proposed by Hill et al. (2000), are about 20% and 40% higher than the WINDII observations for Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models, respectively, in the thermospheric peak region.The individual components of O( 1 S) production due to various sources are shown in Figs. 4 and 5 for both solar EUV flux models.Two representative cases, one at local noontime and other in the afternoon, have been chosen.A value of 0.47 is used for the quantum yield of reaction N 2 (A 3 + u ) with O.It is noticeable from these figures that the reaction of N 2 (A 3 + u ) with O is the dominating source of O( 1 S) between 130 and 170 km.In the earlier studies of Singh et al. (1996) and Tyagi and Singh (1998), which were based on the temperature independent rate coefficient of the reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36, the contribution of this reaction was comparable with that of the dissociative recombination reaction O + 2 + e between 130 and 170 km.Further, it is quite evident from Figs. 4 and 5 that the photodissociation of O 2 becomes the major source of O( 1 S) Further, it is quite evident from Figs. 1-3 that the emission rates obtained using the Tobiska (1991) solar EUV flux model (with quantum yield 0.36) and the temperature independent reaction rate coefficient of N 2 (A 3 + u )+ O are also in good agreement (with in 10%) with the measured emission rates, whereas the emission rates obtained with the temperature dependent rate coefficient are marginally higher (about 12%) than the measured emission rates.One may also notice that the emission rates obtained from the Tobiska (1991) solar EUV flux model are about 20-30% higher than those emission rates obtained from the Hinteregger et al. (1981) solar EUV flux model in all the corresponding cases.The main reason of this difference is due to the fact that the Tobiska flux model provides about 20-30% higher values of EUV flux than that of the Hinteregger et al. ( 1981) solar EUV flux model below the 300 Å wavelength.The energy associated with solar radiation below 300 Å is responsible mainly for the production of greenline dayglow emission in the thermosphere (above 120 km).Consequently, the Tobiska (1991) solar EUV flux model would provide higher values of emission rates than those obtained from the Hinteregger et al. (1981) solar EUV flux model.
The mean percentage difference of the unrevised and the revised results with the WINDII results are shown in Fig. 6, for both the solar EUV flux models.The mean VER is obtained from the analysis of several greenline dayglow profiles (twenty in numbers) at various latitudes and local times in both hemispheres.The percentage difference between the modelled and WINDII results is calculated at a particular altitude for each profile.After obtaining the percentage difference at a particular altitude, the simple average is obtained for all the emission profiles.One can notice from Fig. 6 that the revised results obtained from the Hinteregger et al. ( 1981) solar EUV flux model are within 8% agreement with the WINDII results between 120 and 200 km, which shows significant improvement over the corresponding unrevised results.In case of the Tobiska (1991) solar EUV flux model, the VER with the temperature independent rate coefficient are 10% smaller than the WINDII results between 120 to 200 km.It is also evident from Fig. 6 that the VER obtained with the temperature dependent rate coefficient in the Tobiska (1991) solar EUV flux model are about 12% higher than the WINDII results between 120 and 200 km.One can also notice in Fig. 6 that the VER obtained using a quantum yield of 0.47 with the temperature dependent rate coefficient are about 20% and 40% higher than the WINDII results for the Hintergger et al. (1981) and Tobsika (1991) solar EUV flux models, respectively, in the thermospheric peak region (120-200 km).
The following points have emerged from the above discussion: 1.The VER obtained using the Hinteregger et al. (1981) solar EUV flux model, along with the temperature dependent rate coefficient and a quantum yield of 0.36 for the reaction N 2 (A 3 + u )+ O, are in very good agreement with the WINDII results in the thermospheric peak region.
2. The VER obtained using the Tobiska (1991) solar EUV flux model, along with the temperature dependent rate coefficient and a quantum yield of 0.36 for the reaction N 2 (A 3 + u )+ O, are about 12% higher than the WINDII results in the thermospheric peak region.
3. The VER obtained using the temperature dependent rate coefficient and a quantum yield of 0.47 are about 20% slope for coronal fluxes.HLYMOD = HEIEW ×3.77847 × 10 9

Fig. 1 .
Fig. 1.Comparison between the observed and modelled O( 1 S) VER profiles.The darkened triangles and darkened asterisks are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature dependent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36.The plus and the multiplication signs are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature independent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36.The dashed line and the darkened squares are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature independent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.75.The darkened circles are the observed VER from WINDII.

Fig. 2 .
Fig. 2. Comparison between the observed and modelled O( 1 S) VER profiles.The darkened triangles and darkened asterisks are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature dependent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36.The plus and the multiplication signs are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature independent rate coefficient of N 2 (A 3 + u )+ O and a quantum yield of 0.36.The dashed line and the darkened squares are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature dependent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.47.The darkened circles are the observed VER from WINDII.

Fig. 3 .
Fig. 3. Same as in Fig. 2 but at different latitudes and local times.

Fig. 4 .
Fig. 4. Modelled and measured greenline dayglow emission profiles for selected observing (WINDII) conditions along with the various contributions of the O( 1 S) production processes.

Fig. 5 .
Fig. 5. Same as in Fig. 4. but at different latitudes and local times.

Fig. 6 .
Fig. 6.The mean % difference in the VER between the observations (WINDII) and modelled results.The darkened triangles and darkened asterisks are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature dependent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36.The plus and the multiplication signs are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature independent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.36.The dashed line and the darkened squares are for the Hinteregger et al. (1981) and Tobiska (1991) solar EUV flux models with the temperature dependent rate coefficient of reaction N 2 (A 3 + u )+ O and a quantum yield of 0.47.