Volume 45, Issue 21 p. 11,547-11,554
Research Letter
Free Access

Pulsation Characteristics of Jovian Infrared Northern Aurora Observed by the Subaru IRCS with Adaptive Optics

H. Watanabe

Corresponding Author

H. Watanabe

Planetary Plasma and Atmospheric Research Center, Tohoku University, Sendai, Japan

Correspondence to: H. Watanabe,

[email protected]

Search for more papers by this author
H. Kita

H. Kita

Institute of Space and Astronautical Science, Japan Aerospace Exploration Agency, Sagamihara, Japan

Search for more papers by this author
C. Tao

C. Tao

National Institute of Information and Communications Technology, Koganei, Japan

Search for more papers by this author
M. Kagitani

M. Kagitani

Planetary Plasma and Atmospheric Research Center, Tohoku University, Sendai, Japan

Search for more papers by this author
T. Sakanoi

T. Sakanoi

Planetary Plasma and Atmospheric Research Center, Tohoku University, Sendai, Japan

Search for more papers by this author
Y. Kasaba

Y. Kasaba

Planetary Plasma and Atmospheric Research Center, Tohoku University, Sendai, Japan

Department of Geophysics, Tohoku University, Sendai, Japan

Search for more papers by this author
First published: 22 October 2018
Citations: 12

Abstract

We report narrow band-filtered imaging observations of the Jovian H3+ 3.4-μm emission using the IRCS (infrared camera and spectrograph) on the Subaru telescope taken on 25 May 2016. Approximately 1 hr of data was taken at intervals of 45–110 s, with high spatial resolution (~0.2 arcsec) using adaptive optics. In the northern polar region, we found bright patch-like emissions on the poleward side of the main oval. One of them had a pulsation period of ~10 min. We utilized an H3+ emission model to investigate the response time of the H3+ emission to abrupt and periodic variations of the precipitating electron flux. The model showed that the H3+ emission could pulsate with this timescale due to a modulated flux of the precipitating electrons in the kilo-electron-volt to tens of kilo-electron-volt energy range.

Key Points

  • We took infrared images of the Jovian H3+ aurora for ~1 hr at time intervals of 45–110 s with the assistance of adaptive optics
  • The aurora images showed a patchy structure on the polar side of the northern main oval with a pulsating interval of ~10 min
  • A model analysis showed that such a fast variation can be driven by the modulated electron flux with energy in the kilo-electron-volt to tens of kilo-electron-volt range

Plain Language Summary

We made a movie of the Jovian infrared aurora for the first time. The high spatial resolution images were observed for ~1 hr with the time interval of 45–110 s using the infrared camera of the Subaru-8-m telescope. This movie showed that the infrared aurora from the hydrogen ion molecule H3+ had patchy structures on the northern auroral region and a pulsation period of ~10 min. Our model analysis proved that such a fast variation could be driven by the modulation of the kilo-electron-volt to tens of kilo-electron-volt electrons coming into Jupiter.

1 Introduction

Jupiter has bright auroral emissions in the ultraviolet (UV) and infrared (IR) wavelength ranges. The former is emitted from H2 and H that are directly excited by precipitating auroral electrons, while the latter is due to thermally excited rotational and vibrational emissions of H3+ and H2 molecules. Both types of auroral emissions occur in three different regions: the main oval emission, polar emission, and outer emission (e.g., Grodent, 2015, and the references therein). The main oval emission is generated by the magnetosphere-ionosphere-thermosphere coupling current system that is connected to the middle magnetosphere, where deviation from the corotation of plasma that is supplied from the Io torus becomes significant. This emission is excited by electrons that precipitate into the upper atmosphere in the upward current region. In general, the dawnside main emission has a narrow-arc structure, while the duskside emission consists of patchy and/or multiple-arc structures distributed over broad areas. The outer emission has a diffuse structure and suddenly brightens within one Jovian rotation (Kimura, Badman, et al., 2015, Kimura et al., 2017). The emissions at the satellite footprints are located outside of and within the main emission region. Comparing the main emission and satellite footprint mapping to the subcorotation and corotation regions in Jupiter's magnetosphere, it is seen that the polar emission inside of the main oval is more variable in time and morphology and has a strong local time dependence (Grodent et al., 2003).

UV observations by the Hubble Space Telescope have revealed the spatial distribution and timescales of the variations of the polar emissions. Three regions are identified in the northern UV polar emissions that are fixed at a specific latitude and local time (Grodent et al., 2003). The swirl region is on the poleward-most side and includes some faint patchy features. This region has fainter emissions with a higher color ratio (i.e., higher precipitating energy) than the main oval, as shown by observations from the Hubble Space Telescope (e.g., Gérard et al., 2014) and Juno observations (Bonfond et al., 2017). Gérard et al. (2003) identified rapid brightening events from the UV spectral observations in the swirl region with timescales of 50–100 s. They estimated that the energy of the precipitating electrons ranged from 40 to 120 keV. The dawnside is referred to as the dark region and has almost no emissions, while the active region consists of the noon and afternoon sectors, which have bright arc-like structures and flares. In the active region, quasiperiodic emissions are seen on a timescale of 2–3 min with either broad or patchy morphologies (Bonfond et al., 2016, 2011). Nichols, Badman, et al. (2017) highlighted the periodic brightness variations at the duskside of the active region. They called this region the dusk active region, where the periods varied daily between ~3 and ~11 min. They also inferred from Juno's solar wind observation that dusk active region becomes bright during the solar wind compression.

It has been suggested that the morphology of IR aurora is similar to UV aurora (Clarke et al., 2004; Radioti et al., 2013). The IR polar aurora also can be distinguished into three areas. The IR bright polar region is fixed at the duskside and overlaps with the UV's active region. The IR dark region is divided into two regions according to the velocity of the H3+ auroral electrojet. The fixed-dark polar region is stagnant at a high latitude around the north pole, and the velocity of the rotating-dark polar region increases from zero as the colatitude increases (e.g., Stallard et al., 2003). The locations and morphologies of the fixed-dark polar region and rotating-dark polar region match the swirl region and the dark UV region. Multiarc structures and bright/dark spots are observed in the polar region by the Juno/Jovian Infrared Auroral Mapper (Mura et al., 2017). In particular, a small, bright, and transient spot was found on the northern polar region, and the spot was fixed to the System III (SIII) longitude. However, simultaneous observations in the UV and IR showed that the brightness of the dark polar region in the UV is close to zero, while the brightness is quite substantial in the IR (Radioti et al., 2013).

The similarities and differences in the averaged morphologies of the UV and IR aurora could represent the energy input from the magnetosphere and the heating processes in the thermosphere. The Jovian UV aurora is the result of electron excitation of atomic and molecular hydrogen on a timescale of 10−2 s. In contrast, the IR aurora is a rotational-vibrational emission of H3+ that is produced through the ionization of H2 and ion chemistry. Thus, the timescale of the IR auroral variability could be longer than that of the UV aurora (e.g., Clarke et al., 2004). Stallard et al. (2016) reported from ~16-min step images that the shortest timescale of the variation observed in the polar auroral region was ~30 min. They also reported that the UV images that were averaged over 12 min, the approximate lifetime of the H3+, showed similar structures to those observed in the H3+ images. Conversely, high temporal resolution UV observations revealed the existence of fine structures and short-time variations.

In this study, we attempted to detect the short-term variations of the Jovian IR aurora based on the narrow-line H3+ filter imaging campaign that was executed by the Subaru-8-m telescope with an adaptive optics system. The high spatial (~0.2 arcsec) and temporal (~1–2 min) resolution of our observations acquired on May 2016 enabled the detection of small features over the range of a few arcsecs and the detection of fast variations that occur over several minutes. To validate the possibility of such fast and periodic variations in the H3+ emission lines, we also utilized an auroral emission model of Tao et al. (2011) to evaluate the timescale of variation of the IR H3+ emission controlled by the modulation of precipitating electron flux.

2 Observation and Analysis

We performed the Jovian IR auroral observations on 25 May 2016 with the infrared camera and spectrograph that is a component of the Subaru telescope (Kobayashi et al., 2000; Tokunaga et al., 1998; (Proposal ID: S16A-007). The imaging sequence is similar to that utilized in our previous observations (Kita et al., 2018; Uno et al., 2014), with the adaptive optics instrument (AO188; Minowa et al., 2010) locked on the Galilean satellites. This configuration enabled us to obtain high spatial resolution images of the Jovian IR aurora as shown in Figure 1. In this paper, we analyzed the images using an H3+ narrow-band filter (central wavelength: 3.413 μm; full width at half maximum: 0.022 μm), which covers four strong H3+ emission lines at 3.42700, 3.42067, 3.41483, and 3.41274 μm (Neale et al., 1996). Each image frame consisted of ten 2-s integrations on Jupiter. Total time intervals per frame were 45–110 s, which included exposures on sky. The images were obtained from 6:53–7:46 UT with the assistance of the AO188 activation locked on Europa (diameter: ~1 arcsec). This provided the Jovian IR H3+ aurora images with high spatial (~0.2 arcsec) and temporal (~1–2 min) resolutions. The flux was calibrated using a standard star (ksi Vir, K = 4.4 mag) before and after the run.

Details are in the caption following the image
(a–d) Line-of-sight-corrected images and (e–h) their polar-projected images obtained by the Subaru infrared camera and spectrograph on 25 May 2016. The pairs of vertically aligned images are from the same data (e.g., a and e). The observation start time in UT and the central meridian longitude (CML) of each frame are written above a–d. The color bar indicates the observed intensity of the H3+ emission, and the bottom side of each panel is set to System III longitude = 180°. The dotted lines are latitude and longitude lines at 10° increments. The solid blue line is the statistical position of the main oval (Nichols et al., 2009). Io's footprint is shown in e. In f, the green line represents the main oval-1, the yellow line represents the main oval-2, and the red line represents the polar patch. The definitions of each region are explained below.

3 Results

During this 1-hr run, we observed the patch-like pulsating features on the polar side of the main oval. Figure 1 shows examples of the data set. Figures 1a–1d show the observed images of the northern polar region. Their polar projection maps are shown in Figures 1e–1h, which are adjusted by the line-of-sight correction (Stallard, 2001). As shown in Figure 1, the main oval emission, Io's footprint (latitude ~54° and SIII longitude ~140–148°), and some patch-like polar emissions (latitude ~60–80° and SIII longitude ~170–210°) are observed. The patches moved toward the dawnside along the main oval, and their intensities pulsated (see also the movie that contains all the images presented in the supporting information S1). During this observation on 25 May (DOY145) 2017, the total flux of the UV aurora observed by Hisaki was ~1 TW, and the overall auroral activity was at a low level (see Figure 1 in Nichols, Badman, et al., 2017). We note that Nichols, Yeoman, et al. (2017) reported pulsating patches with ~11-min periodicity in the UV images within the main oval region, while similar features could not be identified in the IR H3+ aurora during this observation period. We could not perform any analysis in the active UV region, which is known to be variable and have periodic pulsations on timescales of 2–11 min (Bonfond et al., 2016, 2011; Nichols, Badman, et al., 2017), because it was not possible to identify it due to the geometry of this observation.

The time series of the polar patches and the main oval emissions found in Figure 1 are summarized in Figure 2. First, we focused on the two regions of the main oval emissions: the dawnside (outlined by the green line) and the duskside (outlined by the yellow line) areas as shown in Figure 1f. These areas are separated by the SIII longitude ~180° and are hereafter called the main oval-1 and the main oval-2. Figure 2a represents the variations of the averaged intensity in these regions. The main oval-1 (green) was brighter than the main oval-2 (yellow), and the intensities in both regions varied slowly. It should be noted that the fluctuating features of less than 3 min that are seen in the intensity data of the main oval are artifacts that are caused by the variation of the bias level in each frame.

Details are in the caption following the image
The time variation of the H3+ aurora features. (a) The variation of the averaged intensity from each region defined in Figure 1. (b) The residuals of the variations of each region from the linear trends. The times at which the images shown in Figure 1 were observed are marked with arrows. (c) The Lomb-Scargle powers derived from b. The horizontal dashed and dash-dotted lines are the 95% and 99% confidence levels versus the null hypothesis of Gaussian noise, respectively. The error bars in a and b are the root mean square of the background sky flux. The green, yellow, and red lines are the flux of main oval-1, main oval-2, and polar patch, respectively.

Next, we focused on a pulsating auroral feature observed at SIII longitude ~180° in the polar region. In order to specify the spread of this structure, we performed the following analysis. We applied a 11 × 11 binning to the polar projection images that have block square grid. The image after binning has a grid scale larger than the spatial resolution under AO (~0.2 arcsec) at the position of the polar emission around ~50–75° latitude. The increasing or decreasing temporal trend was subtracted after the binning. Then we calculated the averages of each correlation coefficient between a pixel and four surrounding pixels. We defined the grid group with high intensity, high amplitude, and high correlation coefficients as polar patch. This region well covers the pulsating area. Figure 2a shows the variations of the averaged intensity of the polar patch (red line). The polar patch intensity (red) shows pulsating features with approximately 10-min intervals. The dotted lines in Figure 2a show the linear trends of the time profile fitted using the least squares method. The polar patch exhibited a gradually decreasing trend in addition to pulsating features, and the main oval-2 showed a decreasing trend too. Since the polar patch and the main oval-2 are close to the terminator, this trend might be due to the local time dependence of the solar extreme UV irradiation intensity.

In Figure 2b, the residuals of the observed time profiles of the linear trends are shown for each of the three regions, in order to examine the pulsation characteristics. The polar patch shows short-term variability with the timescales of several to several tens of minutes, while the dawnside and duskside main ovals are stable. Figure 2c represents the power spectral density of the residuals derived using the Lomb-Scargle method (Horne & Baliunas, 1986; Woodward et al., 1994). Since the time resolution was ~2–3 min in the observation duration of ~50 min, we determined that periods that are longer than 5 min or shorter than one third of the observation term are significant. The horizontal dashed and dash-dotted lines represent the 95% and 99% confidence levels versus the null hypothesis of Gaussian noise, respectively. These levels are properly applied to the tallest peak. We identified that the polar patch has significant periodicity of ~10 min.

4 Discussion and Summary

The variation of the IR auroral intensity with ~10-min cycle that is observed in this study was not observed in previous IR observations. Moreover, the position of the polar patch defined in this study appeared to be bright with patch-like morphologies in some of the UV images. However, this is not always the case (Bonfond et al., 2017, 2016; Gérard et al., 2003; Kimura, Badman, et al., 2015; Nichols et al., 2009, Nichols, Badman, et al., 2017, Nichols, Yeoman, et al., 2017). In the color ratio map by Bonfond et al. (2017), the duskside patch region exhibited a high color ratio. The intensity-time variation of these patchy regions had not previously been analyzed, and the variation with a timescale of ~10 min had not been observed in any polar region in the IR. Here we consider possible mechanisms for the blinking of H3+ lines that have a periodicity of ~10 min. To meet this objective, we evaluated the timescale of the H3+ emission using the emission model of Tao et al. (2011). This model can be used to estimate the H3+ IR emission intensity as a response to auroral electron precipitation and solar extreme UV radiation by making assumptions about the atmospheric neutral density, the temperature profiles, and the main ion chemistry reactions. Temperature change occurs over the timescale of ~103–104 s/K, transport and diffusion occur over the timescales of 104–105 s, ion chemistry resulting in sources producing IR transitions has timescale of 102–104 s, and IR transitions occur within 10−2 s (Badman et al., 2014). Therefore, ion chemistry is the candidate that can produce the ~10-min variations.

In the analysis in this report, we set the Jovian atmosphere profiles with an exospheric temperature of 1,200 K and the neutral density profiles of H, He, H2, CH4, C2H2, and C2H4 as shown in Figure 2 in Tao et al. (2011). The inflow density of the monotonic electrons with specific energies of 0.1–300 keV was initially set to ~6 × 1011/m2, and then it was switched off at time 0. We traced the variation of the H3+ emission line intensity after switching it off. The intensity variation of four H3+ lines within the 3.4-μm band filter is shown in Figure 3. In Figure 3a, the time variation of H3+ intensity decreased immediately after the electron inflow flux disappeared at 0 min. The IR intensity I was normalized by the initial value I0 and the final value I1 for each energy case as (I-I1)/(I0-I1), where I0 = 3.8 × 10−6, 4.1 × 10−6, 6.6 × 10−6, 1.6 × 10−5, 2.9 × 10−5, 2.5 × 10−5, 1.3 × 10−5, and 7.7 × 10−6 W/m2/str for the 0.1, 0.3, 1, 3, 10, 30, 100, and 300 keV cases, respectively and I1~3.8 × 10−6. This IR intensity is roughly proportional to the H3+ density at the IR emission peak altitude. Figure 3b shows the timescale of the intensity decrease, which is defined as the time period required for the intensity to fall to 1/e of each initial value. The obtained timescale varies over 6–24 min depending on the energy of the precipitating electrons; the shortest value in the precipitating electrons was 10–30 keV, which also produced the brightest IR emissions.

Details are in the caption following the image
Emission model evaluation for the variation of H3+ 3.4-μm emission lines controlled by the precipitating electrons. (a) Time variations of the H3+ emission normalized intensity after the switch-off (time = 0 min) of the precipitating electrons with the monotonic energies of 0.1–300 keV. (b) Timescale for the decrease of the H3+ emission. (c) The periodic electron flux input with various timescales τ. (d) H3+ infrared (IR) emission response to the inflow electrons with the variations shown in c. The electron energy input in d was set to 3 keV. The colors in a and b represent the electron energies of 0.1, 0.3, 1, 3, 10, 30, 100, and 300 keV. The correspondence between the color and energy is found in b. The colors in c and d represent the timescales of 1, 3, 10, and 30 min. Four lines in c are identical.

This effect can be explained as follows. In the model, the loss rate of H3+ is the sum of the dissociative recombination with hydrocarbons in proportion to [H3+][hydrocarbons] and the recombination with electrons in proportion to [H3+][e] (Tao et al., 2011). Because the electron density around the altitudes of the H3+ density peak is nearly equal to the H3+ density, the H3+ loss rate by electron recombination is proportional to [H3+]2. In this model, the altitude of H3+ emission peak under the precipitated electrons of 1, 10, 30, 100, and 300 keV are 1,400, 770, 600, 560, and 560 km above the 1-bar level, respectively. At those altitudes hydrocarbon is rare, and so the H3+ loss rate is mainly due to the electron recombination. In the range of <30 keV, higher energy electrons create a higher H3+ density peak at lower altitudes, and a higher H3+ density causes a faster H3+ emission loss. However, for energies greater than 30 keV, electrons reach the lower thermosphere but cannot create a larger number of H3+ because of the much higher loss rate by more hydrocarbons. In addition, lower temperature at lower altitudes decreases the IR emission. Therefore, the H3+ emission peak altitude remains ~560 km. In this case, higher energy electrons with energies greater than 30 keV create a lower number of H3+ ions at the emission peak altitude, and the loss rate of H3+ emission strength decreases because there is less electron recombination.

These processes contribute to the timescale shown in Figure 3b, with the minimum value of approximately 10–30 keV. Since the specific energy spectra responsible for the pulsating patches are not detected so far, we also executed the simulation using a Maxwellian energy distribution, E/Ei × exp(−E/Ei), where Ei is the characteristic energy, with an average energy 2Ei for inflow precipitating electrons. Even under these conditions, the absolute values of the retrieved intensities and the characteristic timescales were similar to those shown in Figures 3a and 3b with differences less than 26%. Using this estimation, the electrons with energies of a few keV to up to ~100 keV can be considered as the cause of the intensity variation over the timescale of 10 min because of their small decreasing timescale.

We also evaluated the response of the H3+ emission intensity to the periodic electron flux variations for the periodic timescale τ = 1, 3, 10, and 30 min. We set the electron energy to 3 keV as shown by the thick diamond in Figure 3b, which has a longer characteristic timescale than that of electrons with energies of several tens of keV. A few keV of downward electrons were observed in the Jovian polar auroral region by the JADE-E (Ebert et al., 2017). Flux variation and its apparent variation are based on temporal variation seen in UV emission (Bonfond et al., 2016, 2011). For rapid changes in input, that is, τ = 1 and 3 min, the variations of the auroral intensity are small (less than ~10% of the constant input case), which is due to the longer response time compared to the input variation. However, with the longer timescale of the inflow electron variations, that is, τ = 10 and 30 min, larger amplitude variations of more than 10% can be observed in the H3+ emission flux. Through these model estimations, we propose that the variations of the H3+ emission intensity with the timescale of 10 to a few tens of minutes could be caused by the modulation of the auroral electron flux variation with similar timescales.

The periodic flux variations of the charged particle have been observed as the periodic auroral emissions in multiwavelength. The polar UV aurora excited by precipitating electrons has periods of 2–11 min (Bonfond et al., 2011, 2016; Nichols, Badman, et al., 2017). The 40-min quasiperiodic radio bursts from southern polar region were observed by Ulysses (MacDowall et al., 1993; McKibben et al., 1993). The X-ray aurora has a hot spot in the northern polar region, which is produced by heavy ions with energies of tens of MeV. This hot spot has the periods of 12, 26, and 40–45 min, and its longitudinal and latitudinal position is similar to the IR polar patch in this study. From magnetospheric mapping, magnetopause reconnection or ultra low frequency waves driven by Kelvin Helmholts instabilities have been proposed as the mechanisms for precipitating charged particles to the hot spot (Dunn et al., 2016, 2017; Gladstone et al., 2002; Kimura, Kraft, et al., 2015). These processes could be candidates that drive the pulsations observed in this study.

The precipitating electron energy in the polar auroral region has been estimated to be from several tens to several hundreds of keV by the UV color ratio method (Gérard et al., 2016, 2014; Gustin et al., 2016). While we chose 3-keV electron for flux variation test in the discussion based on Juno observations, ~100 keV could also produce ~10-min pulsation according to Figure 3b. For the 3-keV electron case, there are no indications in color ratio maps in these poleward patchy structure regions, while the ~100-keV electron should affect the map. UV spectral observation of these patchy structure would constrain these uncertainties.

The observations analyzed in this paper were performed over a short time interval during one night. If it was caused by precipitating electron variations, UV emission in the same region should have pulsating features with similar (or faster) timescales. Unfortunately, our observation was not simultaneous with the UV auroral observations by HST (Hubble space telescope), or the UV and IR aurora observations by Juno. The closest HST images in time were observed on 20:56–21:40 of 24 May, which is a lag of 5.6 hr relative to our observations. For future work, it is recommended to acquire additional near IR aurora imaging observations with high temporal and spatial resolutions equivalent to this study with simultaneous UV imaging observations.

Acknowledgments

This paper is based on data collected using the Subaru telescope, which is operated by the National Astronomical Observatory of Japan (NAOJ). The observed data are opened on Subaru-Mitaka Okayama-Kiso Archive System (http://smoka.nao.ac.jp/) that is operated by NAOJ. We express our thanks to T.-S. Pyo and J. Bulger for their kind advice and helpful support on the operation of the IRCS and the AO188. This work is supported by a Grant-in-Aid for Scientific Research (26287118, 15H05209, and 15K17769) from the Japan Society for the Promotion of Science. H. K. was supported by a Grant-in-Aid for JSPS Research Fellow. H. K. and Y. K. are also thankful for the support from ISAS/JAXA for the Jovian scientific studies related to our contributions to the ESA JUICE mission studies.