Measurements of the neutral temperature in the lower thermosphere, using high resolution (0.13 nm) N2 Lyman-Birge-Hopfield (LBH) band emissions are presented. The data were taken by the High-resolution Ionospheric and Thermospheric Spectrograph (HITS) instrument aboard the Advanced Research and Global Observation Satellite (ARGOS). By fitting the LBH band profiles, temperatures in the thermosphere between ∼150 and 270 km have been retrieved. Temperatures retrieved for three geomagnetically quiet days, 28–30 July 2001 are consistent with temperatures from the Mass Spectrometer and Incoherent Scatter (MSIS) model.
 The Earth's lower thermosphere responds to changes in solar and geomagnetic activity, meaning the atmospheric conditions may vary strongly in time and space. These changing conditions combined with the coupling between neutral and ionized species make it difficult to predict the atmospheric response to space weather. Neutral temperatures (Tn) in the lower thermosphere are a key to understanding changes in the space weather. Tn changes in the lower thermosphere are related to structure and composition at higher altitudes, and knowledge of Tn provides insight into neutral and ion density variations. Heating arises from a variety of sources, including ion and neutral chemical heating, compressional heating and ionization. A temperature increase is also one of the main factors leading to enhancements in the nitric oxide (NO) emission, which is important for radiative energy loss rates below 210 km, especially during geomagnetic storms [e.g., Mlynczak et al., 2005]. Other mechanisms responsible for temperature variations include downward heat conduction [e.g., Banks and Kockarts, 1973] and adiabatic expansion [e.g., Eastes et al., 1992]. Consequently, understanding the response of the lower thermosphere to solar and geomagnetic forcing bears directly upon our ability to understand the Earth's thermosphere-ionosphere response to space weather and to assess the impact upon space-dependent systems and technologies.
 Despite its importance, our knowledge about the thermal structure in the lower thermosphere is limited by a lack of reliable temperatures. Numerous studies have used the Mass Spectrometer and Incoherent Scatter (MSIS) model [Hedin, 1987] to determine the state of the neutral atmosphere. Most of the observational data included in the MSIS model come from either below 120 km or above 250 km, and there are few direct measurements of the temperatures in the lower thermosphere. To calculate the temperature variation in the thermosphere, MSIS uses a cold boundary at 120 km and an exponential temperature increase to an exospheric temperature at high altitudes. There have been efforts to obtain temperatures from ultraviolet, visible and infrared remote sensing techniques, though the techniques that have been discussed in publications either require a better understanding of the excitation processes [e.g., Mlynczak and Marshall, 1996; Sharma et al., 1996, 1998; Sharma and Duff, 1997; Sharma and Roble, 2001] or are difficult to accomplish [e.g., Barth and Eparvier, 1993; Meier and Picone, 1994]. For example the rotational temperature of NO studied by Sharma and Duff  includes contributions from excitation processes that cause the rotational temperature to deviate from the neutral temperature.
 In this paper the neutral temperature in the lower thermosphere between ∼150 and 270 km is determined, using the rotational structure of the Lyman Birge-Hopfield (LBH) bands of N2. The LBH data were taken by the High-resolution Ionospheric and Thermospheric Spectrograph (HITS), one of the instruments aboard the Advanced Research and Global Observation Satellite (ARGOS). To our knowledge, this is the first analysis performed using satellite-based observations of N2 limb profiles at sufficient spectral resolution to determine temperatures in the lower thermosphere. In Section 2, the data utilized are presented and a description of the technique is provided. Results and discussion are given in Section 3, followed by the conclusions (Section 4).
2. Data and Technique
 The ARGOS satellite was launched 23 February 1999 into a sun-synchronous orbit (840 km) which crossed the dayside equator at 14:30 hours solar local time. HITS observed 11.0 nm sections of the 50–180 nm emissions from the Earth's atmosphere. The N2 LBH bands are observed at wavelengths greater than 127.3 nm. The high resolution (0.13 nm Full Width Half Maximum [FWHM]) HITS observations examined in this study are from the 144.0–155.0 nm passband and were obtained while the instrument scanned the limb. This passband includes LBH emissions from all vibrational levels, with the brightest band being v = 1 at 146.4 nm. The observations were made on three geomagnetically quiet days (Table 1: Ap between 3 and 7) on 28–30 July 2001.
Table 1. Geophysical Parameters Used in the Mass Spectrometer and Incoherent Scatter (MSIS) Calculations for the Three Events of 28–30 July 2001
28 July 2001
29 July 2001
30 July 2001
Note that the F10.7 values given in Table 1 are for the previous day, as this is input to the MSIS model.
 HITS measurements from the daytime airglow observed at tangent altitudes near 177 km on 28 July 2001 are shown (as an example) in Figure 1a (dashed lines). The pairs of numbers above the peaks give the vibrational levels (initial, final) in the N2 molecule. For these observations, a best fit model (see details in next paragraph), plotted as a solid line in Figure 1b, is calculated. Note that the wavelength region 150.6–151.4 nm, for which a modeled peak is absent, is not fit due to uncertainties in the detector response. The horizontal solid line gives the background level. Next, the fit for Tr and the contributions from v = 1 and 5 are recalculated using only the 145.8–147.7 nm observations (Figure 1c). This recalculation is done to decrease the statistical uncertainty in the temperature calculations, as v = 1 provides the largest signal relative to the background noise. The (1,1) band at 146.4 nm and the (5,4) band at 147.4 nm are well separated from other LBH band emissions. The next nearest LBH emission is the (4,3) band at 145.9 nm, but this band is approximately 400 times weaker than the (1,1) band. The rotational temperature Tr which gives the best fit between the modeled band shape profile and the observations is then determined. For the band profile in Figure 1c, a rotational temperature of 907 K (solid line) best fits the data. Also plotted for comparison (long dashes) is the band profile for an assumed rotational temperature of 600 K, revealing a much sharper decrease at longer wavelengths.
 The modeled LBH spectra use available molecular data [e.g., Huber and Herzberg, 1979], and magnetic dipole and electric quadrupole contributions of approximately 95 and 5 percent respectively for the (1,1) band. Eastes and Dentamaro  and Eastes  have suggested that a mechanism called collision-induced electronic transitions (CIET) may significantly change the amount of excitation to (and emission from) the LBH band system. However, CIET has no affect on the temperatures and is therefore not included when fitting the observations. The fitting routine varies Tr, populations of each vibrational band, constant background, amount of 149.3 emissions (from atomic nitrogen) and O2 photoabsorption (using a temperature-dependent O2 Schumann-Runge absorption cross section based on Ogawa and Ogawa ). To provide the most reliable fit between the calculated LBH spectra and observed HITS spectra, a C-statistic minimization [Cash, 1979] implemented in a modified Levenberg-Marquardt scheme to perform Discrete Inverse Theory retrievals has been used. According to Budzien et al. , such an approach is more appropriate than the more traditional χ2 minimization during cases with low count rates. However, in large signal-to-noise cases, we note that the C-statistic and χ2 fits are almost identical.
 In addition to an appropriate statistical fitting algorithm, the quality of the fitting approach depends on the instrumental line shape. Due to optical aberrations within the instrument, a measured instrument line profile is used. The HITS data (0.13 nm FWHM) allows a good determination of the instrumental line shape, by summing a large number of HI Lyman-α (121.6 nm) observations [Budzien et al., 2001]. A validation has been performed for the 144.0–155.0 nm passband using fits to the NI 149.3 nm multiplet.
 While the rotational temperature is retrieved from the LBH bands the atmospheric neutral temperature Tn is the desired quantity. Based upon accepted principles of molecular spectroscopy and experimental laboratory results the two should be equal. The LBH bands are electric dipole-forbidden transitions that are excited purely by photoelectron impact on N2. As excitation is more likely between states with similar inter-nuclear spacing (Franck-Condon principle), the vibrational energy of the molecule is not expected to be significantly altered. Likewise, the rotational energy changes during electron impact with molecules should be limited. The interaction times are short, relative to the rotational period, and electrons cannot efficiently transfer kinetic energy to the much more massive molecules and therefore are not expected to change the rotational temperature.
 Similarly, laboratory measurements of collisions between excited and unexcited molecules at thermal velocities [e.g., Katayama et al., 1994] find that even during collisions at thermal velocities, where one might expect significant changes due to the longer interaction time and the masses involved, vibrational energies tend to be preserved, with transitions approximately following a dipole like selection rule. Even in the collisional excitation of NO (for which the electric dipole moment is orders of magnitude larger than for N2) by oxygen atoms, the rotational temperature is relatively close to the neutral temperature [Sharma and Duff, 1997]. Finally, since N2 is a homonuclear molecule, pure rotational emission is dipole forbidden. Once collisionally thermalized, the rotational populations cannot readily relax by radiation. Consequently, a one-to-one relationship between Tr and Tn should be expected, but nothing is as convincing as verification from observations.
 Observations of a different N2 band system, the first negative bands of N2+, by Kurihara et al.  have shown agreement between Tr and Tn. The emissions they used were produced by atmospheric N2 being ionized and excited using an electron gun (their rocket flew at night). Kurihara et al.  determined profiles of the neutral temperature between 100 and 150 km from measurements of the rotational temperature of N2. A comparison with temperatures from MSIS yielded reasonable agreement. For altitudes below 135 km, the temperatures measured by Kurihara et al.  were larger than MSIS, with differences exceeding 100 K around 115 km. Near 150 km, MSIS provided slightly larger values. The LBH bands are more often used than the first negative bands for dayside observations and should provide similar temperature information.
3. Results and Discussion
 In Figure 2 the neutral temperatures from daily averages of HITS data (diamonds) 28–30 July 2001 are shown. These data are representative of the HITS observations. For each day, temperatures are derived for tangent altitudes centered on 161, 177, 194, 214, 234 and 256 km. At the lower altitudes the temperatures have been corrected for the higher temperature contributions from the higher altitudes in the line-of-sight. To perform this correction the following approach has been used: first the volume excitation rates as a function of altitude have been calculated using the Continuous Slowing Down (CSD) model [Jasperse, 1976]. Next, the total emissions expected at different tangent altitudes when limb scanning have been estimated, yielding distribution functions to correct the initial HITS temperatures. In Table 2, both the initial temperatures and the corrected ones are presented, revealing minor differences above ∼200 km (<20 K). However, at 161 km, the corrected temperatures are approximately 100 K lower.
Table 2. Temperatures (Initial and Corrected) Retrieved From HITS Observations on 28–30 July 2001
Center of the tangent altitude bin used. The bin extends to half the distance to the adjacent altitude bin.
Initial temperatures retrieved from HITS.
Corrected temperatures (for the higher altitude temperature contributions), as presented in Figure 2.
Statistical uncertainty (±1σ).
28 July 2001
29 July 2001
30 July 2001
 As indicated by the ±1σ uncertainties in Figure 2 and Table 2, the statistical uncertainties of the HITS measurements are typically between 5 and 10%. Also plotted are the temperatures from the MSIS model [Hedin, 1987]. The dotted line represents MSIS-E-90 [Hedin, 1991], while the dashed line comes from the recent NRLMSISE-00 version [Picone et al., 2002]. Both MSIS calculations are averages over latitudes between −20 and +60 degrees, approximately the latitude region from which the HITS data are taken. A reasonable match between temperatures from HITS and MSIS is seen for all three events, as 15 (16) of the 18 temperatures derived using MSIS-E-90 (NRLMSISE-00) are within the statistical uncertainty of the HITS calculations. For altitudes below 220 km, the average difference between the measured (HITS) and calculated (MSIS) temperatures is 30 K. If the higher altitudes are included, the average difference increases slightly, to 39 K. The agreement between the observed temperatures and MSIS calculations during the geomagnetically quiet period presented is similar to that obtained by Richards , in a comparison of neutral density data from the Atmosphere Explorer (AE)-C satellite with MSIS calculations. However, Richards  shows that on disturbed days, MSIS may underestimate Tn by several hundred degrees. Also, the HITS values in Figure 2 seem to show some structure beyond a simple MSIS exponential temperature profile. This may be explained by the different mechanisms responsible for temperature variations as listed in Section 1, though a more thorough examination regarding these issues is beyond the scope of this paper.
 In addition to the random statistical uncertainties in the temperature calculations of 5–10% shown in Figure 2, other systematic errors may also be present. By varying the amount of O2 absorption used in the fitting, a maximum temperature difference of ∼0.5% is calculated. Similar differences are expected for errors in sensitivity calibration and the background determined during the fitting. The instrumental line profile is another possible source of systematic error. However, possible systematic errors from the instrumental line profile are expected to be small, due to the resolution of the HITS data (see discussion in Section 2).
 In this study the neutral temperature Tn in the lower thermosphere (150–270 km) has been measured, using the rotational structure of the LBH bands of N2 obtained by the HITS instrument onboard the ARGOS satellite. By fitting model spectra to the high resolution (0.13 nm FWHM) HITS observations, a rotational temperature Tr (=Tn) is determined from the band shape and used to study the thermal structure of the lower thermosphere for three geomagnetically quiet days, 28–30 July 2001.
 This work was supported by NASA grant NAG5-12786 to the University of Central Florida.