Narrow-band filter, high-spectral-resolution (0.2 cm−1) spectro-imaging infrared observations of Jupiter's auroral zones, acquired in October 1999 and October 2000 with the FTS/BEAR instrument at the Canada–France–Hawaii Telescope, have provided maps of the emission from the H2 S1(1) quadrupole line and several lines. H2 and emissions appear to be morphologically different, especially in the north, where the latter notably exhibits a ‘hot spot’ near λIII=150–170° System III longitude. The spectra include a total of 14 lines, including two hot lines from the 3ν2–ν2 band, detected on Jupiter for the first time. They can be used to determine column densities, rotational (Trot) and vibrational (Tvib) temperatures. We find the mean Tvib of the ν2=3 state to be lower (960±50 K) than the mean Trot in v2=2 (1170±75 K), indicating an underpopulation of the v2=3 level with respect to local thermodynamical equilibrium. Rotational temperatures and associated column densities are generally higher and lower, respectively, than inferred previously from ν2 observations. These features can be explained by the combination of both a large positive temperature gradient in the sub-microbar auroral atmosphere and non-local thermal equilibrium effects affecting preferentially hot and combination bands. Spatial variations in line intensities are mostly owing to correlated variations in the column densities. The thermostatic role played by at ionospheric levels may provide an explanation. The exception is the northern ‘hot spot’, which exhibits a Tvib about 250 K higher than other regions.
Since its discovery in Jupiter's polar regions at 2 μm in its 2ν2 band (Drossart et al. 1989), the ion has been the subject of considerable interest and recognized to be an important tracer of the Jovian upper atmosphere and its coupling with Jupiter's magnetosphere, and more generally a tracer of Giant Planet upper atmosphere energetics and dynamics (see review by S. Miller et al. 2006). At Jupiter, a whole suite of observational studies were conducted in three general directions: (i) determining the mean line emission intensities, column density and temperatures (kinetic, rotational and vibrational) in the northern and southern auroral regions, so as to infer its role in the auroral atmosphere heat budget, (ii) characterizing the morphology and variability of the auroral emissions and measuring the non-auroral (mid-to-low latitude) emission, and (iii) using the emission as a wind tracer, from direct measurement of line Doppler shifts, to evaluate the large-scale circulation at sub-microbar levels. While the early observations were either single-aperture spectroscopy (with no spatial information) or pure two-dimensional imaging (with no insight in the physical parameters, except for the total emission), most of the subsequent studies made use of one-dimensional long slit spectroscopy. Still, in many cases, separating the temperature and column density parameters turned out to be difficult.
We describe here high-spectral-resolution (approx. 25 000) two-dimensional spectro-imaging observations of H2 and in the 2 μm region in both hemispheres, bearing information on the composition, temperature and potentially wind speeds, in the Jovian auroral upper atmosphere. Results on the H2 and emission spatial distribution, and on the temperatures and column abundances have been reported in detail by Raynaud et al. (2004). A summary is presented here, focusing on the physical interpretations of the main results.
2. Observations and overview of spectra
Observations were performed on 26–28 October 1999 and 10–12 October 2000 using the BEAR Imaging FTS at the 3.60 m Canada–France–Hawaii Telescope. We covered a portion of the 2ν2 band of by means of two narrow-band filters, referred hereafter as ‘F2.12’ and ‘F2.09’, giving access to the 4698–4752 and 4760–4805 cm−1 spectral ranges with a spectral resolution of 0.20 cm−1. The data consist of 14 data cubes, sampling either the northern or the southern polar region at a variety of central meridian longitudes (CML). We, hereafter, emphasize the northern cubes, which are characterized by stronger signal and higher signal-to-noise ratio. The (seeing-limited) spatial resolution is approximately 0.5″. A special reconstructing procedure was developed to avoid smearing owing to Jupiter's rotation during approximately 1 h required to record the interferograms (see Raynaud et al. 2004).
In addition to the H2 S1(1) line at 4713 cm−1, the spectra (figure 1) show the detection of a total of 14 lines (table 1). Spectral identification was based on the help of Kao et al. (1991), Neale et al. (1996), Lindsay & McCall (2001) and T. Oka. Twelve of the lines belong to the 2ν2 rovibrational band. The remaining two, at 4721.8 and 4749.7 cm−1, were identified as belonging to the 3ν2–ν2 band, for which we made the first detection on Jupiter. The presence of the hot band lines makes a separate determination of the rotational (Trot) and vibrational (Tvib) temperatures possible.
3. Spatial distribution of and H2 line emissions
Emission features in the H2 S1(1) quadrupole line at 4712.9 cm−1 and in the strong 4732 cm−1 line, observed simultaneously in the northern polar region, are clearly morphologically different (figure 2). The emission shows large spatial contrasts, which exhibits, in particular, a ‘hot spot’ near λIII=150–170°. The H2 emission is more uniform and does not show the hot spot. Instead, it exhibits a broad maximum over longitudes of 180–250°. The structure of the 4712.9 cm−1 emission is best viewed in polar projection (figure 3). The hot spot near 70°, λIII=155° is the only region where the weak 4721.9 cm−1 hot line stands out of the noise. In the southern hemisphere, where our data (not shown here, see Raynaud et al. 2004) provide the first imaging observations of H2, the H2 and distributions are more similar, with a maximum emission near λIII=30–50°. In general, the line intensities and spatial distribution that we observe are consistent with earlier investigations (Drossart et al. 1989; Satoh & Connerney 1999).
4. Temperature and column density determination, and search for correlations
The detection of several lines with different energy levels allows us to determine simultaneously the temperature and column density. Assuming local thermal equilibrium (LTE), an ortho–para ratio of 1, and that the emission is optically thin, the intensity of an (if) transition may be calculated from(4.1)where is the total column density along the line-of-sight; νif is the wavenumber; and the other line parameters are given in table 1. Equation (4.1) assumes that a single temperature can be derived from the data. Because our spectra include rotational lines from two different vibrational bands, this is equivalent to assuming that complete LTE is effectively realized (i.e. Trot=Tvib=Tkin). In fact, spectra in the ‘F2.09’ filter include only lines of the 2ν2 band; therefore, they provide Trot, the rotational temperature within the v2=2 level. While the ‘F2.12’ filter includes lines from both the 2ν2 and 3ν2–ν2 bands, the two 3ν2–ν2 lines provide the prime temperature constraint, given their high energies; these spectra therefore mostly provide Tvib, i.e. the vibrational temperature defining the relative population of the v2=2 and 3 levels. At the low pressure levels probed by , radiative de-excitation of the vibrational levels occurs on time-scales comparable to collisional de-excitation; therefore, their populations relative to the ground state are lower than in LTE. Nonetheless, Kim et al. (1992) found that the ratios between the relative populations of the various excited levels containing ν2 are similar to a LTE distribution, a situation subsequently termed ‘quasi-local thermal equilibrium (QLTE)’. We will discuss later the validity of the QLTE assumption.
Fitting the mean spectrum for each data cube indicates no gross variation of the temperature with CML (figure 4). In contrast, the mean (i.e. averaged over CML) vibrational temperature of the v2=3 state appears to be lower (960±50 K) than the mean Trot in the v2=2 level (1170±75 K), indicating an underpopulation of the v2=3 level (relative to v2=2) with respect to LTE.
For the two ‘F2.12’ cubes with the highest signal-to-noise ratio, the spectra were averaged in five intensity bins according to the radiance in the strongest 4732 cm−1 line, and the column density and temperature were determined separately for each bin. Results, shown in figure 5, indicate (i) a strong anticorrelation between the retrieved temperature and columns, noted previously (Lam et al. 1997), and that (ii) in general, the intensity variations are best interpreted as owing to variations in the column density, with a fairly constant 900–1050 K temperature. This conclusion tends to be confirmed by a search for correlations between line intensity, column density and (vibrational) temperature, which indicate a positive correlation between the first two parameters (see Raynaud et al. 2004). An exception is the region of strongest intensities (bin 5), in which the temperature is about 250 K higher than in the other zones.
Given the short time constants for rotational relaxation, the rotational temperature must represent the actual kinetic temperature. In average, we get , and an total column density of (4–8)×1010 cm−2. Interestingly, these values are significantly higher and lower, respectively, than inferred from the ν2 band by Lam et al. (1997), namely 700–1000 K and typically 1×1012 cm−2. This can be explained in the context of recent models. According to Grodent et al. (2001), the pressure range sounded by the IR emission (10−6–10−10 bar) exhibits a strong positive temperature gradient, typically 200 K per decade in pressure. The recent Melin et al. (2005) model, which makes use of that thermal model and a new method of detailed balance calculations for evaluating non-LTE effects (Oka & Epp 2004), indicates that the overtone 2ν2 and hot (2ν2–ν2 and 3ν2–ν2) bands are formed at higher levels (typically, 1000–1600 km above cloud level) than the fundamental ν2 lines, for which maximum contribution occurs near 500 km. In the Grodent et al. (2001) model, the temperatures at 500 and 1400 km are 750 and 1200 K, similar to the rotational temperatures quoted above for the ν2 and 2ν2 bands, respectively.
For the northern region, we find that the mean value of Trot(2ν2) is higher than that of Tvib (=960±50 K), as determined from the 3ν2−ν2/2ν2 line comparison. This indicates a departure from LTE, with an underpopulation of the v2=3 level with respect to v2=2, compared to the thermal equilibrium situation. A possible explanation is that the v2=2 level is anomalously populated. This is the scenario advocated by Kim et al. (1992), who invoked resonant transfer with H2. However, this mechanism was rejected by Stallard et al. (2002), who found no evidence for an overpopulation of the v2=2 level. Instead, Stallard et al. (2002) proposed that upper vibrational levels are actually populated by a ‘proton-hopping’ collision mechanism (, where an asterisk denotes the colliding molecule). In these conditions, Stallard et al. (2002) showed that the intensity of a hot band relative to a fundamental band is lower than in LTE. This is confirmed by the more detailed calculations of Melin et al. (2005), who demonstrate that (i) all nv2 levels start to depart from thermal population above approximately 700 km, and (ii) the departure from LTE is all the more severe for combination and hot bands (figure 6a). This proves that the QLTE approach is not valid and implies that (i) line intensities calculated in the LTE hypothesis can be strongly overestimated (figure 6b), (ii) vibrational temperatures determined from equation (4.1) underestimate the true kinetic temperature, and (iii) number densities may often be strongly underestimated. Melin et al. (2005) find possible errors by factors of 6–10 for the densities inferred from the 2ν2–ν2 band, and as much as factors of 20–200 for measurements based on the 2ν2 and 3ν2–ν2 bands such as ours. Their study thus provides an explanation for the low columns and the multiple temperatures reported in our study. The complexity of the problem is such that in the future forward modelling using non-LTE and a priori temperature profiles appears better suited to data analysis than ‘temperature/column retrievals’.
With the exception of the hot spot near 70°, λIII=155°, we find that the column density is the main parameter driving line emission variations. The general constancy of temperatures over the northern auroral region is probably related to the fact that is the main cooling agent at sub-microbar levels, owing to its excellent radiative properties and despite its small abundance (Miller et al. 2000). The rapid variation of the cooling rate with temperature ensures that large variations of input energy will result in only modest temperature variations. This ‘thermostatic’ effect is demonstrated in more detailed calculations by Grodent et al., who find similar ‘mean’ (column-weighted) temperatures for the diffuse and discrete auroral cases. In contrast, the northern hot spot, which is prominent at a variety of other wavelengths, exhibits a Tvib about 250 K higher than other regions. A partial explanation might invoke a homopause elevation in this region, leading to a local increase in the methane abundance and an associated decrease in the contribution of the deeper and colder component. However, this scenario raises a number of unsolved questions; more generally, the different H2 and distributions remain difficult to explain (Raynaud et al. 2004).
B. Joseph (University of Hawaii, USA). Is the ‘hot spot’ seen in H3+ stable with time?
E. Lellouch. The geometry of our observations allowed us to see it only twice (10 and 12 October 2000). However, the region in question has appeared to be peculiar at a number of wavelengths and on many occasions, showing increased thermal IR hydrocarbon emission, increased H2 Lyα brightnesses and increased X-ray emission. Therefore, it is most probably stable. The fact that it does not show up in the H2 IR emission is thus particularly puzzling.
One contribution of 26 to a Discussion Meeting Issue ‘Physics, chemistry and astronomy of H3+’.
- © 2006 The Royal Society