Volume 115, Issue A9
Ionosphere and Upper Atmosphere
Free Access

H Balmer lines in terrestrial aurora: Historical record and new observations by OSIRIS on Odin

R. L. Gattinger

R. L. Gattinger

ISAS, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
A. Egeland

A. Egeland

Department of Physics, University of Oslo, Oslo, Norway

Search for more papers by this author
A. E. Bourassa

A. E. Bourassa

ISAS, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
N. D. Lloyd

N. D. Lloyd

ISAS, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
D. A. Degenstein

D. A. Degenstein

ISAS, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
J. Stegman

J. Stegman

Department of Meteorology, Stockholm University, Stockholm, Sweden

Search for more papers by this author
E. J. Llewellyn

E. J. Llewellyn

ISAS, Department of Physics and Engineering Physics, University of Saskatchewan, Saskatoon, Saskatchewan, Canada

Search for more papers by this author
First published: 09 September 2010
Citations: 3

Abstract

[1] The H Balmer emissions were first identified in terrestrial aurora by Vegard (1939). The earliest photographic spectral observations are reviewed. In the subsequent decade, the intensity ratios for Hα, Hβ, and Hγ were measured, and the well-known line broadening and blue shift were established. Recently, the Hα, Hγ, Hδ, and Hɛ features have been measured by OSIRIS on Odin. The Balmer components are resolved from other auroral features using sets of synthetic spectra. The measured intensity ratios are in good agreement with an extensive set of published model calculations. The presented observations are in the polar region averaged over limb tangent altitudes from 100 to 105 km, approximately perpendicular to the terrestrial magnetic field lines, for this geometry showing Doppler broadening without obvious Doppler shifts. The OSIRIS-measured full-width at half-height of the Hα feature is 2.2 nm corresponding to an H atom velocity of 500 km s−1 and energy approximately 1.3 keV.

1. Introduction

[2] The first reported detection of H Balmer emissions in terrestrial aurora, specifically the Hα and Hβ features, was by Vegard [1939]. Vegard noted that previous observations had failed to detect the Balmer features; this conclusion is substantiated in the work of Currie and Edwards [1936], Vegard and Tonsberg [1937], and Hewson [1937]. Later, Vegard and Tonsberg [1941] added the detection of the Hγ line. Bernard [1948] tentatively identified the Hδ line at 410 nm. In a summary report, Vegard [1948] discussed the Doppler-widened nature of the H Balmer lines. Gartlein [1950] also investigated the relative intensities of the Hα, Hβ, and Hγ features and estimated the Doppler width of the Hβ and Hγ lines. Galperin [1959, 1963] reported additional observations of the H Balmer features and summarized information on the observed intensity ratios. Of historical interest, numerous measurements of auroral heights were obtained [see Harang, 1951; Stormer, 1955], but the current study is limited to auroral spectral observations.

[3] The first clear observation of the Doppler shifting of an H Balmer line in terrestrial aurora was reported by Vegard [1950] who detected a shift to shorter wavelength in the Hβ line in a spectrogram obtained at Oslo on 23 February 1950. On 18–19 August 1950 at the Yerkes Observatory, Meinel [1950a] observed a pronounced blue shift in Hα zenith observations relative to the Hα horizon observations. This was followed with observations by Gartlein [1951] who obtained similar pairs of spectra on 30 September 1950 that also showed the blue-shifted zenith Hα feature compared with the merely broadened Hα horizon feature. Omholt et al. [1962] recorded a significant temporal variation of the relative intensities of Hα and the nearby N2 First Positive Group (1PG) bands during a 30 min period during an auroral storm. These observations were near the magnetic zenith and showed both the broadened and blue-shifted Hα feature. Soraas et al. [1974] extended the observations by obtaining rocket measurements of proton precipitation energies that enabled model comparisons of H Balmer line shapes. Many observations of the relative intensities of the Hα to Hβ lines have been reported with values ranging from 1.65 to 7 [Omholt, 1971]. Recently, Galand et al. [2004] measured an Hα to Hβ ratio of 2.4 with an estimated RMS uncertainty of approximately 50%. Related measurements of the spectrum of proton aurora have been described, for example, the anomalous vibrational development of the N2+ First Negative Group (1NG) bands in proton aurora observed by Degen et al. [1972] and Sivjee [1980] as well as the apparent weakness of the neutral molecular nitrogen features [Lummerzheim et al., 2003].

[4] Eather [1967] presented an extensive review of the terrestrial H Balmer observations including a discussion of the excitation mechanisms arising from incoming energetic protons at the top of the atmosphere interacting with terrestrial molecules and atoms leading to the emission of the observed H Balmer lines. The incoming proton energy is degraded by sequentially ionizing atmospheric species as the proton penetrates deeper into the atmosphere. Through the process of charge exchange with atmospheric species the proton converts to an excited H atom giving rise to H Lyman and Balmer radiation. The energetic H atom can be reionized in further collisions with atmospheric species and the process repeats. A number of models that simulate the interaction between incoming energetic electrons and protons and the background terrestrial atmosphere have been developed [e.g., Strickland et al., 1993; Galand et al., 1997, 1998; Lummerzheim et al., 2003; Gérard et al., 2005; Simon et al., 2007]. These models have also been used successfully to simulate observed H Balmer line shapes. For example, Robertson et al. [2006] compared the measured and simulated spectral shape of Hβ in the dayside cusp.

[5] The OSIRIS spectrographic instrument [Llewellyn et al., 2004] on Odin [Murtagh et al., 2002] in one of its operating modes views the mesosphere and lower thermosphere and is thus suited for the detection of aurora. A number of the auroral spectra obtained by OSIRIS show clearly detectable H Balmer emission lines. An accurate calibration of the relative spectral response of the OSIRIS instrument [Gattinger et al., 2009] allows precise intensity ratio measurements for the H Balmer lines. Selected spectra have been averaged to improve the signal-to-noise ratio and compared with model calculations of the H Balmer line intensity ratios.

[6] Theoretical estimates of the H Balmer line ratios aimed specifically at terrestrial auroral conditions have been reviewed by Sigernes et al. [1995]. These model calculations are generally limited to the Hα-to-Hβ intensity ratio [e.g., Rees, 1982], and all predict the ratio to be greater than 5. Because the OSIRIS auroral observations include other members of the H Balmer series and do not include Hβ, the usual model calculations are not suitable for comparison purposes. Brocklehurst [1971] and Hummer and Storey [1987] developed a comprehensive model of H emission line ratios, including many lines in the Balmer series, specifically for studies of stellar atmospheres. This set of theoretical H emission line ratios is commonly used in astronomical studies to measure the spectral dependence of dust extinction [e.g., Petersen and Gammelgard, 1996]. Although not specifically designed for the simulation of energetic protons in the terrestrial atmosphere, the possibility that these theoretical ratios developed by Hummer and Storey [1987] are applicable to OSIRIS H Balmer proton auroral observations is investigated in the following sections.

2. Synthetic Spectra for the Analysis of H Balmer Observations

[7] To isolate the H Balmer lines from other auroral features, sets of synthetic spectra were prepared and scaled to match the observed spectral features. The algorithms for generating the auroral synthetic spectra, including the N2 1PG, the N2+ 1NG, the N2 Second Positive Group (2PG), the N2+ Meinel (M), the N2 Vegard Kaplan (VK), and the O2+ First Negative Group (1NG) band systems have been described previously [Vallance Jones and Gattinger, 1972, 1975; Gattinger and Vallance Jones, 1974]. The relative band intensities within the different band systems are based on the tabulations by Vallance Jones [1974], with system intensity normalizations calculated using the brighter bands observed in each system. Relative intensities of the simulated N2 VK bands have been updated with the more recent results of Piper [1993]. Vibrational populations within the system simulations have been slightly adjusted as required to achieve the best match between the spectral model and the observations.

[8] For the H Balmer series, Hummer and Storey [1987] have provided comprehensive model calculations of emission line ratios for stellar atmospheres. As mentioned above, these line ratios are assumed here in the absence of other model calculations. Their tabulations cover a broad range of electron densities beginning at 1 × 102 cm−3 and electron temperatures beginning at 1000 K. For low- to medium-intensity terrestrial aurora in the 105 km altitude region, the electron density is approximately 1 × 106 cm−3 [Vallance Jones, 1974]. Electron temperatures at 105 km vary from approximately 300 to possibly 1000 K [Vallance Jones, 1974; Abe et al., 2006]. For the comparisons presented here, the tabulated set by Hummer and Storey [1987] with a density of 1 × 106 cm−3 and temperature of 1000 K has been chosen. From this selected set of theoretical ratios, the first nine lines in the H Balmer series are listed in Table 1. Because observed Balmer ratios are often expressed with respect to Hβ, the same convention has been followed here; the ratios calculated by Hummer and Storey [1987] are used to normalize Hα to 3.19 relative to Hβ. This is necessary because Hβ is in a spectral region that is blocked by the order sorter included in the OSIRIS spectrograph.

Table 1. H Balmer Series Relative Emission Intensitiesa
Balmer Line Wavelength (nm) Theory 1000 K Theory 3000 K Galperin [1963] OSIRIS Average
Hα 656.21 3.19 3.01 3.0 ± 0.2 3.19 ± 0.15
Hβ 486.08 1.00 1.00 1.00 (1.00)
Hγ 434.00 0.45 0.46 0.8 ± 0.1 0.43 ± 0.12
Hδ 410.13 0.25 0.25 - 0.22 ± 0.06
396.97 0.16 0.16 - 0.14 ± 0.06
388.86 0.11 0.11 ?
383.50 0.086 0.081 Tentative
379.75 0.071 0.064 ?
377.02 0.062 0.053 Tentative
  • a Wavelengths are from the Balmer formula. Theory columns are from Hummer and Storey [1987]. OSIRIS error estimates include both random and systematic errors.

[9] Also included in Table 1 are the Hummer and Storey [1987] results for 3000 K; this possibly gives an indication of the temperature dependence of the calculated intensity ratios. The differences are less than 10% for the H Balmer lines analyzed in detail in the OSIRIS auroral observations. Likewise, for a 10-fold reduction in electron density, the calculated ratios change by less than 10% for Hα to Hɛ inclusive.

[10] The H Balmer line shapes are assumed to be Gaussian with the width of the Hα line determined experimentally from the OSIRIS observations (see below). The widths of the Balmer features at shorter wavelengths are scaled by the wavelength ratio relative to Hα. The area under each of the simulated H Balmer lines is normalized with the values calculated by Hummer and Storey [1987]. Combining the above considerations, an H Balmer synthetic spectrum was generated (Figure 1) for use in the following analyses.

Details are in the caption following the image
(a) The weighted average of 12 calibrated OSIRIS spectra from 7 to 8 January 2005 and 8 May 2005 clearly showing the Hα (656.2 nm) and Hγ (434.0 nm) features. The three brightest auroral features are off scale: O(1S−1D) (557.7 nm), N2+ 1NG 0-0 (391.4 nm), and 0-1 (427.8 nm) The weaker features are identified in the figures to follow. (b) The model H Balmer features from the calculations of Hummer and Storey [1987]. Simulated widths are referred to the measured Hα width in Figure 1a.

3. OSIRIS-Observed Spectra

[11] The Odin spacecraft was launched on 20 February 2001 into a circular Sun-synchronous orbit, an ascending node of 1800 Local Time, and an orbit inclination of 97.8°, at an altitude of 620 km. Operating modes include limb scans through the mesosphere and lower thermosphere, up to 115 km, and thus, auroral spectra are occasionally obtained when the limb tangent point is located within the region of the auroral oval. The OSIRIS instrument on Odin has a vertical field of view of 1 km at the tangent point, and the spectral range is from 275 to 815 nm with a spectral resolution of approximately 0.90 nm. The limb scan rate at mesospheric altitudes is nominally 0.75 km s−1 with exposure times typically 2 s. Absolute limb pointing is known to better than 0.5 km at the tangent point.

[12] Accurate relative spectral calibration of the OSIRIS instrument is essential to obtaining reliable emission ratios that are separated widely in wavelength. Standard lamps were used in the prelaunch calibrations, followed by on-orbit astronomical observations to refine the calibration that results in an estimated absolute calibration error of ±10%. Response changes during the mission are tracked by a comparison between the observed on-orbit limb scatter spectra and simulations using the SASKTRAN three-dimensional radiative transfer model described by Bourassa et al. [2008]. The model incorporates multiple Rayleigh molecular scatter and Mie aerosol scatter, Lambertian ground albedo, and molecular atmospheric extinction that includes temperature dependence. Ozone altitude profiles, required for a satisfactory simulation, are determined as part of the solution [Roth et al., 2007; McLinden et al., 2007]. In addition, the sulfate aerosol vertical profile [Bourassa et al., 2007] and the surface albedo are determined with the SASKTRAN model. The relative spectral response from the model is estimated to be better than 5% over the full spectral range from 275 to 815 nm. Using auroral emissions observed by OSIRIS at multiple wavelengths and with common upper states, Gattinger et al. [2009] confirmed the accurate relative calibration.

[13] An averaged OSIRIS auroral spectrum that exhibits the H Balmer emissions is shown in Figure 1a; the plot is limited to the portion of the spectrum that contains the Balmer lines. Twelve individual spectra from 7 to 8 January 2005 and 8 May 2005, with a solar zenith angle greater than 101°, averaged over the 100 to 105 km tangent limb viewing altitudes, are weighted by averaging the measured brightness of the Hα feature at 656 nm. The slowly changing CCD dark pattern is determined for each limb scan in the 50 km tangent altitude range, well below the approximately 105 km tangent altitude where the auroral signature typically maximizes, and subtracted from the observed tangent limb spectra. The spectra are calibrated over the full OSIRIS wavelength range. In addition to the observed strong Hα (656.2 nm) feature and the weaker Hγ (434.0 nm) feature, other bright emissions include O(1S–1D) (557.7 nm; off scale), the N2+ 1NG 0-0 (391.4 nm; off scale), 0-1 (427.8 nm), and 0-2 (470.9 nm) bands and N2 2PG bands in the 370 to 400 nm range. A number of these brighter features are used to derive the relative band system intensities for the synthetic spectral components as described below. The data gap over the 480 to 530 nm region is due to the order sorter included in the OSIRIS spectrograph design. Because Hβ occurs in the gap region, it cannot be included in the current analysis.

[14] From the averaged spectrum in Figure 1a, the observed Hα to N2+ 1NG 0-2 (470.9 nm) ratio is 4.4. Using the Hα-to-Hβ ratio of 3.19 (Table 1), this translates to an Hβ-to-N2+ 1NG 0-2 ratio of 1.4. Omholt [1971] states that when this ratio approaches unity, it is likely that the aurora is produced mostly by protons.

[15] By comparison with the synthetic spectrum of the Balmer lines (Figure 1b), only the Hα (656.2 nm) and Hγ (434.0 nm) lines are obviously present in the OSIRIS-observed spectrum. However, detailed analyses that include synthetic spectra of atmospheric species demonstrate the presence of additional Balmer lines in the observed spectrum.

4. Comparison between OSIRIS-Observed and Synthetic Spectra

[16] A detailed comparison between the averaged observed spectra in the Hα 656 nm region with the simulated spectral components is shown in Figure 2. Synthetic spectra are generated at a resolution of 0.1 nm and convolved with the OSIRIS slit function, approximately 0.9 nm. The dominant underlying band system is N2 1PG with small contributions from N2+ M bands. Scaling of the N2 1PG synthetic band system is from the average of the 7-4 and 6-3 bands on either side of the Hα emission, whereas the uncontaminated portion of the N2+ M 3-0 band at 687 nm is used for that system scaling. Removing the scaled N2 1PG and N2+ M components from the observed total leaves the resolved Hα component (Figure 2, short-dashed line). This observed Hα component is used to compare with the model Hα line width and brightness (Figure 2, dotted line). The uncertainty in the Hα brightness measurement, including systematic error, is estimated to be approximately 5% (Table 1).

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 656 nm region, expanded from Figure 1. Simulated spectra include the (dotted line) Hα feature and the (long-dashed line) N2 1PG and (dot-dot-dashed line) N2+ M band systems along with the (dot-dashed line) simulated total. (short-dashed line) The measured Hα component is obtained by subtracting the simulated N2 1PG and N2+ M bands from the observed total. The RMS error is indicated.

[17] Because the OSIRIS limb line of sight is approximately normal to the magnetic field lines, the Balmer features are not expected to be shifted in wavelength. The measured Hα half-width at half height from Figure 2 is approximately 1.1 nm, considerably broader than the early horizon measurement of 0.6 nm by Meinel [1950b]. The corresponding average H atom speed is approximately 500 km s−1 equating to energy near 1.3 keV. To attain an altitude of 105 km, the incoming proton precipitation energy at the top of the atmosphere must be approximately 100 keV [Eather, 1967]. This initial energy is degraded to near zero by successive ionizations of atmospheric atoms and molecules. The spectra presented here are for protons approaching the end of this energy loss process as the particle travels deeper into the atmosphere.

[18] A second detailed comparison, this one between the OSIRIS averaged observed spectrum in the Hγ 434 nm region and the simulated spectral components, is shown in Figure 3. The dominant underlying band system is N2 2PG (estimated error of 5%), with a small contribution from N2 VK (estimated error of 20%). Band system scaling is from the N2 2PG 0-0 band at 337 nm and the N2 VK 1-10 band at 343 nm. A small contribution from the OI 5 line at 436.8 nm [Moore, 1945; Vegard, 1956; Ralchenko et al., 2008] to the observed total is also taken into account. The simulated Hγ brightness is from the measured Hα brightness in Figure 2 (dotted line) scaled with the line ratio calculated by Hummer and Storey [1987]. The measured Hγ component, also shown in Figure 3 (short-dashed line), is obtained by subtracting the N2 2PG and N2 VK band components and atomic lines from the observed total. The difference between the model Hγ component and the measured Hγ component is an indication of the measurement accuracy; this assumes that the model values of Hummer and Storey [1987] are applicable. The measured Hγ relative brightness is 0.43 compared to the predicted value of 0.45 (Table 1). The uncertainty in the Hγ brightness measurement, including systematic error, is estimated to be approximately 30% (Table 1). Again, the Hγ feature is not obviously shifted in wavelength but merely broadened, as for Hα in Figure 2.

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 434 nm region, expanded from Figure 1. Simulated spectra include the (dotted line) Hγ feature, the (long-dashed line) N2 2PG and (dot-dot-dashed line) N2 VK band systems, (also dot-dot-dashed line) atomic lines, along with the (dot-dashed line) simulated total. (short-dashed line) The measured Hγ component is obtained by subtracting the simulated the N2 2PG and N2 VK bands and the atomic lines from the observed total. The RMS error is indicated.

[19] Further to the discussion above, the uncertainty in the measured Hγ-to-Hα ratio is investigated in Figure 4. Ratios determined from the individual spectra that are used to form the weighted average observed spectrum in Figure 1 are plotted as a function of relative brightness of the measured Hα emission in each spectrum. When the Hα brightness is small, the variability in the ratios is large because of the decreased signal-to-noise ratio in the spectra, but with increasing Hα brightness, the measured ratios approach 0.14, which happens to correspond with the calculated value of Hummer and Storey [1987].

Details are in the caption following the image
The Hγ/Hα ratios determined from the individual OSIRIS spectra used to form the average in Figure 1. Ratios are plotted in order of increasing measured Hα brightness for each spectrum. The theoretical ratio from Hummer and Storey [1987] is shown by the dashed line.

[20] A third detailed comparison, this time between the averaged observed spectrum in the Hδ 410 nm region and the simulated spectral components, is shown in Figure 5. It is important to note that the measured Hδ brightness is less than 1% of the measured OI 558 nm line, shown as off scale in Figure 1, and is approaching the noise limit of the individual measurements. Although additional uncertainty arises from the subtraction of the background continuum, having detailed synthetic spectra to quantify the contributions from the band systems improves the accuracy of baseline determination. The effects of baseline uncertainty are included in the error estimates in Table 1. The dominant underlying band system in Figure 5 is N2 VK, with a small contribution from N2 2PG and from atomic lines. As above for Hγ, the simulated Hδ emission (dotted line) is scaled from the experimentally determined Hα emission. The measured Hδ (short-dashed line), resolved from the total spectra by subtracting the model N2 VK and N2 2PG bands and atomic lines, is very similar to the simulated Hδ feature. The measured Hδ relative brightness is 0.22 compared with the predicted value of 0.25 (Table 1). The uncertainty in the Hδ brightness measurement, including systematic error, is estimated to be approximately 25% (Table 1). Although the noise component adds considerable uncertainty, it appears that the observed Hδ emission line is not significantly shifted in wavelength.

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 410 nm region, expanded from Figure 1. Simulated spectra include the (dotted line) Hδ feature, the (long-dashed line) N2 2PG and (dot-dot-dashed line) N2 VK band systems, (also dot-dot-dashed line) atomic lines, along with the (dot-dashed line) simulated total. (short-dashed line) The measured Hδ component is obtained by subtracting the simulated N2 2PG, N2 VK and atomic lines from the observed total. The RMS error is indicated.

[21] A fourth detailed comparison, this time between the averaged observed spectrum in the Hɛ 397 nm region and the simulated spectral components, is shown in Figure 6. The dominant underlying band systems are N2 VK (estimated error 20%) and N2 2PG (estimated error 5%) with a small contribution from atomic lines. Here again, the simulated Hɛ 397 nm emission (dotted line) is scaled from the experimentally determined Hα feature assuming the ratios calculated by Hummer and Storey [1987]. The measured Hɛ (short-dashed line) is resolved from the total observed spectra by subtracting the model N2 VK and N2 2PG contributions and the atomic lines. The measured Hɛ relative brightness is 0.14 compared with the predicted value of 0.16 (Table 1). The uncertainty in the Hɛ brightness measurement is estimated to be approximately 40% of the model value (Table 1). Although the noise component leads to considerable uncertainty, this is considered to be a positive identification of the Hɛ 397 nm emission.

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 397 nm region, expanded from Figure 1. Simulated spectra include the (dotted line) Hɛ feature, (long-dashed line) N2 2PG, (dot-dot-dashed line) N2 VK, (also dot-dot-dashed line) atomic lines, along with the (dot-dashed line) simulated total. (short-dashed line) The measured Hɛ component is obtained by subtracting the simulated N2 2PG and N2 VK bands and atomic lines from the observed total. The RMS error is indicated.

[22] We have also extended the search for the possible identification of weaker Balmer lines' observed shortwave of Hɛ; a further detailed comparison is shown in Figure 7 for the OSIRIS spectrum from 374 to 391 nm. This spectral region includes Hζ (388.86 nm), Hη (383.50 nm), Hθ (379.75 nm), and Hι (377.02 nm) (Figure 7, dotted line). The Hζ feature is dominated by the N2+ 1NG 1-1 (388.4 nm) band and cannot be resolved from the observed auroral spectrum. Similarly, the Hθ feature is dominated by the N2 2PG 0-2 (380.4 nm) band and cannot be positively identified. However, the Hη feature is located in the minimum between the N2+ 1NG 1-1 and N2 2PG 0-2 bands, on the shortwave shoulder of the weak N2 VK 3-13 (385.6 nm) band. Including the model Hη feature in the synthetic spectrum total improves the match with the observed spectrum and so suggests a tentative identification of Hη by OSIRIS (Table 1). Similarly, adding the Hι model component, approximately spectrally collocated with the N2 VK 2-12 (376.8 nm) band, to the total synthetic spectrum improves the fit to the observed total, likewise suggesting a tentative identification of Hι by OSIRIS (Table 1).

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 384 nm region, expanded from Figure 1, showing the (dotted line) simulated Hζ (388.86 nm), Hη (383.50 nm), Hθ (379.75 nm), and Hι (377.02 nm) features, (short-dashed line) N2+ 1NG, (long-dashed line) N2 2PG, (dot-dot-dashed line) N2 VK, (also dot-dot-dashed line) atomic lines, along with the (dot-dashed line) simulated total. The RMS error is indicated.

[23] A final detailed fit in the region of the Balmer series limit (364.56 nm) is shown in Figure 8, primarily to document the fit between the OSIRIS-observed spectrum and the simulated spectrum. The observed features in the 366 to 372 nm region are the N2 2PG 2-4 (370.9 nm) and 3-5 (367.1 nm) bands and the N2 VK 1-11 (368.4 nm) band. The match between the observed spectrum and the total of the synthetic N2 2PG and N2 VK spectral components is satisfactory. Also shown is the synthetic spectrum for the Balmer series (dotted line), with an electron density of 1 × 106 cm−3 and electron temperature of 1000 K (as above) up to n = 50 in the tabulations of Hummer and Storey [1987]. As the Balmer lines converge nearer the series limit, the overlapping of lines lead to a pseudocontinuum. Although the comparison is limited by detector noise, a matching pseudocontinuum does not appear to be present in the OSIRIS-observed spectrum. However, for an assumed model temperature of 3000 K, the predicted relative intensities of the H Balmer lines in the range of n = 50 from the tables by Hummer and Storey [1987] are nearly 3 times smaller than for the 1000 K case. A reduced model pseudocontinuum of a factor of 3 would place it near the noise level of the observed spectrum.

Details are in the caption following the image
(solid line) The OSIRIS-observed spectrum in the 363 to 379 nm region, expanded from Figure 1, showing the (dotted line) simulated Balmer lines to n = 50, the (long-dashed line) N2 2PG bands and (dot-dot-dashed line) N2 VK bands, along with the (dot-dashed line) simulated total excluding the Balmer series. The H Balmer series limit (364.56 nm) is indicated by H∞. The RMS error is also indicated.

5. Conclusions

[24] The OSIRIS imaging spectrograph on the Odin spacecraft has clearly detected the Hα (656 nm) Balmer line in numerous terrestrial auroral spectra in the 105 km region of the tangent limb. A subset of 12 of these individual spectra with the brightest Hα signal was averaged to improve the signal-to-noise ratio. This averaged observed spectrum was compared with synthetic spectral simulations to identify individual features. Simulations include the N2 1PG, N2+ M, O2+ 1NG, N2+ 1NG, N2 2PG, and N2 VK band systems; the Doppler broadened lines in the H Balmer series; and the numerous atomic lines and multiplets. A comparison between the observed averaged spectrum and the scaled and summed synthetic spectral components has led to the positive identification of H Balmer lines in terrestrial aurora not previously verified in the literature, specifically the Hδ (410 nm) and the Hɛ (397 nm) features. The measured intensities of Hγ (434 nm), Hδ, and Hɛ, all relative Hα, are in good agreement with the theoretical ratios predicted by Hummer and Storey [1987]; this suggests that the hydrogenic ion recombination model for stellar atmospheres might also be applicable to terrestrial energetic precipitating protons. In addition, the agreement between the observed and the synthetic spectra in the 380 nm region is improved by including the Hη and Hι Balmer features.

[25] The OSIRIS viewing direction is approximately perpendicular to the magnetic field lines at the limb tangent point, and no Doppler shift is observed. The measured Hα half-width at half-height is approximately 1.1 nm corresponding to an average H atom speed of 500 km s−1 and an energy of 1.3 keV.

Acknowledgments

[26] This work was supported by the Canadian Space Agency and the Natural Sciences and Engineering Research Council (Canada). Odin is a Swedish-led satellite project funded jointly by Sweden (Swedish National Space Board), Canada (Canadian Space Agency), France (Centre National d'Etudes Spatiales, Toulouse, France), and Finland (Tekes). The authors thank the reviewers for their constructive comments that resulted in an improved paper.

[27] Robert Lysak thanks the reviewers for their assistance in evaluating this paper.