1 Introduction

Lightning at Jupiter generates a strong electromagnetic impulse in the atmosphere, producing three kinds of radio signatures. The first is comprised of Jovian whistlers typically observed at frequencies below 20 kHz with several seconds long dispersed falling tones when detected in the Io torus (Gurnett et al., 1979; Kurth et al., 1985) or with fast signatures lasting from a few milliseconds to a few tens of milliseconds in the upper ionosphere (Kolmašová et al., 2018). In the strongly magnetized plasma near Jupiter, the whistler frequency is limited below the local electron plasma frequency due to the orientation of the wave propagation with respect to the local magnetic field line (Stix, 1992). These radio signals may possibly propagate up to several thousand kilometers horizontally away from lightning strokes below the ionosphere before ultimately escaping into the inner magnetosphere, but their direct vertical propagation cannot be excluded (Imai et al., 2018). The second kind consists of dispersed millisecond pulses called Jupiter dispersed pulses (JDPs), observed at frequencies below 150 kHz but above the maximum plasma frequency encountered during the wave propagation through the ionosphere. JDPs propagate directly from lightning strokes but can leak into the inner magnetosphere only at places where the ionospheric density is sufficiently low, either in localized holes or over the nightside (Imai et al., 2019). The third kind of signal is a nondispersed sferic, recorded by the Galileo Probe at frequencies <100 kHz below the Jovian ionosphere (Rinnert et al., 1998) and by Juno at 600 MHz and 1.2 GHz leaking into the inner magnetosphere (Brown et al., 2018). The latter ultrahigh frequency (UHF) sferics freely traverse the ionosphere from the lightning strokes as straight-line propagation. Note that, in the atmospheric electricity community, the term “sferic” is traditionally used for impulsive radio pulses propagating in the Earth-ionosphere waveguide, but lately, it has also been used for transionospheric radio pulses of lightning from Jupiter, Saturn, or Uranus (Aplin et al., 2020). Monitoring these three kinds of lightning-induced radio waves yields a global lightning map on both the dayside and nightside of Jupiter and puts constraints on the plasma environment near the planet.

For terrestrial lightning we measure the duration of a lightning flash from the first initial breakdown pulse to the last return stroke pulse. This lightning duration might vary from milliseconds to tens or even hundreds of milliseconds (Rakov & Uman, 2003). Different processes within the duration of a lightning flash are observed on Earth at shorter timescales decreasing down to microseconds in the case of stepped or dart leader pulses, tens of microseconds for initial breakdown pulses, or hundreds of microseconds for return strokes. Not knowing yet the details of development of lightning flashes at Jupiter, we refer to these collectively as “lightning processes” in this paper.

Another set of Jupiter lightning observations prior to Juno came from the optical detections of illuminated clouds on the nightside of the planet from Voyager 1 (Borucki & Williams, 1986; Cook et al., 1979), Voyager 2 (Borucki & Magalhães, 1992), Galileo (Dyudina et al., 2002; Gierasch et al., 2000; Little et al., 1999), Cassini (Dyudina et al., 2004), and New Horizons (Baines et al., 2007). Nightside lightning from Juno has also been detected by navigation cameras: the SRU (Becker et al., Views of lightning on the darkside of Jupiter by Juno's Stellar Reference Unit, EGU General Assembly Abstract 6393, 2019) and the Advanced Stellar Compass (ASC) (Joergensen et al., Juno ASC observations of low light phenomena on the Jovian nightside, AGU Fall Meeting Abstract P24B-04, 2018). The nightside lightning images suggested that (1) the deepest lightning flashes were situated within or below the water cloud layer (≥5-bar level), (2) the lateral extent of thunderstorms typically ranges from 100 to 2,000 km, and (3) some of the thunderstorms may have lightning activity exceeding several minutes. It is still questionable how many lightning strokes comprise the imaging detections (with ≥5-s exposure time (Baines et al., 2007)). A scanned Galileo imager frame gave average flash rates of about 0.3 flashes/s in two storm clusters, but multiple strokes may have been included in individual measured flashes because of the 1.5 s/pixel effective temporal resolution of the scanned frame (Little et al., 1999). Multifilter dayside imaging of lightning storm locations (separated by 2–4 hr from nightside lightning imaging) has demonstrated the presence of both deep clouds associated with water condensation and tall, high-opacity convective towers in the vicinity of lightning flashes (Dyudina et al., 2004; Gierasch et al., 2000).

In this paper, we present colocated observations of lightning detected at radio wavelengths and cloud features captured at visible and near-infrared wavelengths during Juno's nineteenth perijove on 6 April 2019. During this interval, unique detections of very low frequency/low-frequency (VLF/LF) radio pulses below 150 kHz were made and the magnetic footprints of these were spatially very close to the sources of UHF sferics at 600 MHz and thunderstorms. These observations with 2.67-μs time sampling allow the investigation of lightning processes at submillisecond scales.

2 Description of Instruments and Measurements

Since 5 July 2016, the Juno spacecraft has been in polar orbit around Jupiter (Bolton et al., 2017). During Juno perijoves every 53 days, synoptic observations of lightning at radio wavelengths are made by the radio and plasma wave instrument (Waves; Kurth et al., 2017) and by the Microwave Radiometer (MWR;  Janssen et al., 2017). Constraints from operational modes and geometrical considerations mean that the two instruments rarely achieve truly simultaneous coverage. The reception of lightning-induced radio waves is dependent on the topology of Jupiter's magnetic field lines for Jovian whistlers and on the orientation of the spacecraft spin plane with respect to the planet's atmosphere for UHF sferics detected by the MWR. Here we investigate quasi-simultaneous Waves and MWR data acquired around noon UTC on 6 April 2019. During this pass, Juno's onboard imaging instruments were shut down to accommodate an MWR cross-track scanning orientation (Bolton et al., Initial results of Juno's microwave imaging of Jupiter's atmosphere at multiple depths, AGU Fall Meeting Abstract P21G-3455, 2019), so we use Hubble Space Telescope (HST) imaging for atmospheric context.

The Waves instrument is capable of collecting electric field data in a frequency range of 50 Hz to 41 MHz with an electric dipole antenna and magnetic field data in a frequency range of 50 Hz to 20 kHz with a magnetic search coil sensor, through three onboard receivers. In this study, the effective length of the dipole antenna is assumed to be 0.5 m, instead of 2.4 m for the geometrical effective length (Kurth et al., 2017). One of the receivers—the Low Frequency Receiver (LFR)—is divided into two frequency bands: the low-band frequency channels of electric and magnetic fields (LFR-Lo) from 50 Hz to 20 kHz and the high-band frequency channels of electric fields (LFR-Hi) from 10 to 150 kHz. The LFR-Lo and LFR-Hi burst modes record snapshots containing 6,144 points at a cadence of one snapshot per second. The 6,144-point LFR-Lo and LFR-Hi snapshots have respective durations of 122.88 ms (at 20-μs resolution) and 16.384 ms (at 2.67-μs resolution). These frequency bands conduct interleaved observations due to the instrumental design. The LFR-Lo mode was utilized for the previous studies of Jovian low-dispersion whistlers (Imai et al., 2018; Kolmašová et al., 2018). As described in the work of Kolmašová et al. (2018) and Imai et al. (2018, 2019), we converted waveform data on the ground into spectral data by means of a 256-point fast Fourier transform for identifying Jupiter's lightning-induced wave morphology. In measuring an emission duration, we limit the time domain waveforms using the fourth-order Butterworth filter between 40 and 100 kHz. This process allows us to remove most of the dispersed signals at the low-band frequency that contain the majority of delays accumulated during the wave propagation through the ionospheric plasma.

The MWR instrument is designed to detect one linearly polarized electric field component with two patch-array antennas at 600 MHz and 1.25 GHz, three slot-array antennas at 2.6, 5.2, and 10.0 GHz and one corrugated horn antenna at 22.0 GHz. Electric field strengths are sampled with a fixed temporal resolution of 100 ms (corresponding to 1.2° of spacecraft spin per sample) and converted into antenna temperatures T_A in K. The antenna beams lie in the spacecraft spin plane, with the high-band frequency antennas coaligned on one side of the spacecraft, and the 600-MHz antenna on a separate side resulting in a 120° pointing offset. Although MWR has detected UHF sferics in the 1.2-GHz channel, we concentrate on the much more common UHF sferics at 600 MHz in this paper. The selection of the UHF sferic events is made by finding a signal higher than six standard deviation above the noise floor, after the background signal is determined via a low-pass filter (Brown et al., 2018). The −10-dB level (about 90% received power) of the 600-MHz antenna forms a 17° beam half-angle that we use to define the uncertainty of the sferic position. Using straight-line propagation, we estimate the MWR boresight of the lightning source mapped onto the 5-bar atmosphere.

Jupiter atmospheric context observations have been acquired with the HST as part of the Wide Field Coverage for Juno (WFCJ) program (Wong et al., 2020). The first observation was made during Juno's third perijove on 11 December 2016 (Tollefson et al., 2017). This program utilizes filters from ultraviolet (UV) to near-infrared wavelengths in the Wide Field Camera 3 (WFC3/UVIS) instrument, including 631-nm, 727-nm, and 889-nm narrow filters sensitive to cloud and haze particles at different altitudes. Imaging data are mapped to the 1-bar level in Jovigraphic System III coordinates as described in Wong et al. (2020). At 60°N, the spatial resolution is about 400 km in the east-west direction and 700 km in the north-south direction, with a mapping accuracy of about 80 × 160 km.

3 Observations

From 11:58:00 through 12:00:30 on 6 April 2019, the Juno Waves LFR-Hi and LFR-Lo channels independently detected a multitude of lightning-induced radio pulses below 150 kHz shown in Figures 1a and 1d. At that time, Juno's position changed in altitude above the 1-bar level from 17,630 to 14,730 km in the northern hemisphere, crossing magnetic field lines connecting to the middle radiation belt (Kollmann et al., 2017) at M = 2.34 – 2.87 (pink line of Figure 1a inset). M is defined as the magnetic equatorial radius mapped along Juno's magnetic field line using the JRM09 magnetic field model (Connerney et al., 2018) with the current sheet (Connerney et al., 1981), divided by the equatorial radius of Jupiter (1 R_J = 71,492 km). M is used in the same way as the dipole L but with nondipole field lines.

Of the 150 available snapshots each in the LFR-Hi and LFR-Lo channels (asynchronously acquired), 7 LFR-Hi snapshots and 56 LFR-Lo snapshots contained one or more radio pulses. Figure 1e shows two whistlers in a single 122-ms LFR-Lo snapshot, with one whistler appearing to extend to frequencies higher than the 20-kHz cutoff of the LFR-Lo channel. Of the 93 whistlers detected in the LFR-Lo data, 26 appeared to extend to frequencies above the 20-kHz cutoff. In the LFR-Hi data, 17 VLF/LF radio pulses were observed, extending as high as 128 kHz. A single 16.384-ms snapshot contained 10 discrete VLF/LF radio pulses (Figure 1b), while the details of the time domain waveform for one such pulse at 40–100 kHz gives a duration of 0.392 ms in Figure 1c. Six other snapshots containing VLF/LF radio pulses are illustrated in Figure 2 and listed in Table 1.

As the definitive mode of 17 VLF/LF radio pulses acquired from LFR-Hi data is unclear, they may be either JDPs or Jovian whistlers. Some of the observed pulses show signatures of a lower cutoff and dispersed features close to it. Therefore, they might not be whistlers below the plasma frequency but JDPs above the plasma frequency. Recall that the plasma frequency corresponds to the upper frequency of Jovian whistlers and the lower cutoff frequency of JDPs. However, this distinction may not be stated with certainty for all our observations, as the determination of the local plasma frequency is not always possible. In some cases, we can also examine positions of observations. For example, in Figure S1 in the supporting information, the maximum observable range (or radio horizon) overlaps with a small portion of nightside region, and this condition might be unfavorable for the JDP assumption because most of the observable region faces the dayside and the previous JDP observations were concentrated near the terminator (Imai et al., 2019).

Because the electromagnetic impulse generated from lightning is the source of whistlers and/or JDPs, these waveform durations give upper bounds on the duration of lightning processes, which generated the observed pulses. We measured the time domain LFR-Hi waveform at 40–100 kHz to identify discrete wave packets within seven samples. According to Table 1, all measured wave packet durations vary from 0.277 ms through 0.715 ms. These timescales fit well within the range of observed durations of lightning generated radio pulses on the low-Earth orbit. The whistlers at frequencies <1.25 kHz exhibit dispersion over a few tens of milliseconds above the nightside ionosphere (Santolík et al., 2009) and longer on the dayside (Santolík et al., 2008), while these time intervals become progressively shorter at VLF range (Parrot et al., 2008, 2015). On the other end of the wide frequency range of lightning radio signals, dispersed transionospheric pulse pairs generated by intracloud discharges (e.g., Jacobson & Light, 2003) have a duration between a few tens and one hundred microseconds.

The magnetic footprints of the observed VLF/LF radio pulses overlapped with detections of UHF sferics in a region shown by HST maps to host active moist convection. The MWR instrument detected nine UHF sferic signals between 12:02:14 and 12:04:16 on 6 April 2019, with T_A varying from 3.7 to 179.0 K. Recall that these times are not aligned with the VLF/LF radio pulse times because the UHF sferic times are limited to times when the MWR boresight rotates around and aligns with the direction of the thunderstorms. MWR boresight positions for these UHF sferics, together with the LFR-Hi VLF/LF radio pulse magnetic footprints, are superimposed on HST maps in Figure 3 and listed in Table 1.

Localization of radio signal source regions is not precise enough to associate the signals with specific cloud features, and in fact, we cannot rule out the possibility that all of the signals originated from a single cloud feature. However, HST maps of the general area of the radio signal source region show a particular juxtaposition of three cloud structure types that are commonly found in regions with active convection and lightning: deep water clouds, tall convective towers, and cloud clearings. Figures 3a and 3c use three HST filters probing different depths to show the presence of these three cloud structures in the vicinity of the radio signal source regions.

Deep clouds appear as red in this representation and correspond to pressure levels >4 bar at normal incidence (Banfield et al., 1998; Li et al., 2006; West et al., 2004) where only water can condense. The shallow HST viewing angle near 60°N means that these deep clouds could be at slightly lower pressures, but any cloud material at P > 2.5 bar can still only be composed of water given our knowledge of Jupiter's composition (e.g., Wong et al., 2015). Water clouds indicate the potential for moist convection, while adjacent thick and tall clouds suggest active convective towers. The water clouds may also be rendered visible by downdrafts surrounding the convective towers that clear material from the levels at P < 4 bar. HST cloud-depth color-ratio maps in Figures 3b and 3d show both very deep and very high clouds in the lightning source region.

Cloud-free clearings are the third element of the cloud structure that observationally signifies moist convection and lightning. The other two elements (convective towers and deep water clouds) have been noted in lightning source regions in the Great Red Spot turbulent wake (as observed by the Galileo Orbiter in  Gierasch et al., 2000) and in both individual discrete storms as well as cyclonic vortices (as observed by the Cassini Orbiter in  Dyudina et al., 2004, and by the HST in Wong et al., 2020). Although the deep water clouds and convective towers in these lightning source regions dominated previous attention, we now recognize that cloud-free clearings are a third element shared among these source regions. The three-way signature of lightning and active convection—convective towers, deep water clouds, and clearings—appears as a combination of white, red, and black areas in the color scheme shared by Figure 1 of Gierasch et al. (2000), Figure 9 of Dyudina et al. (2004), Figures 9 and 11 of Wong et al. (2020), and our Figure 3c. The significance of clearings as an element of the cloud structure of active convective areas is not yet known. Clearings could either be an indication of thermal conditions that promote convective outbreaks (e.g., Sugiyama et al., 2014) or an after-effect due to convective downdrafts (e.g., Lunine & Hunten, 1987; Li & Ingersoll, 2015).

4 Discussion and Conclusions

This paper reports the detections of VLF/LF lightning-induced radio pulses (10–150 kHz) from the Juno Waves instrument on 6 April 2019. These VLF/LF radio pulses may be either whistlers or JDPs. Their magnetic footprints were spatially colocated with UHF sferic locations recorded at 600 MHz by the Juno MWR instrument. In the lightning source region, HST maps isolated deep clouds (presumably water), as well as compact opaque clouds extending to high altitudes. The durations of VLF/LF radio pulses range from 0.277 through 0.715 ms after the fourth-order Butterworth filter between 40 and 100 kHz, while their upper frequencies reach as high as 128 kHz.

The lightning flash rate from optical and radio detections is still controversial at Jupiter. The optical flash average rate for all flashes stronger than the optical energy of 2 × 10⁸ J is 4 × 10⁻³ flashes/year/km² using Galileo images with an exposure time of 6.4–179.2 s (Little et al., 1999) and the flash rate from MWR UHF sferic 100-ms events is less than 0.03 flashes/year/km² (Brown et al., 2018). These values are much lower than 1–30 flashes/year/km² based on Jupiter whistlers detected by Waves (Imai et al., 2018; Kolmašová et al., 2018). The HST maps link lightning to meteorologically distinct areas with signs of both deep water clouds and high/thick clouds extending at least one scale height above them. These localized areas of intense moist convection may dominate the lightning activity, leading to systematic errors in flash rates determined from surveys that are not global in scope.

We found radio signatures of lightning processes lasting ≤0.277 ms, implying that many previous visible-imaging lightning detections (as well as MWR UHF sferic detections at timescales of 100 ms) may have been integrated over multiple events. The total durations of JDPs are also short, with the majority (95% of 445 detections) within 3.2 ms (Imai et al., 2019). Hence, the optical and microwave radio estimates of lightning flash rates are probably underestimated, in contrast to the whistler-based lightning flash rate from Waves. The cluster of VLF/LF radio pulses and UHF sferics during Juno's nineteenth perijove effectively links lightning (and thus moist convection) to particular patterns of cloud structure (from HST maps), which may be used to extend studies of the broader spatial variability of the lightning flash rate on Jupiter.