Brief Overview of Lightning Detection Technologies for Ice Giants
The easiest way to detect lightning at the ice giants is by observing their electromagnetic emissions with antennas. The ionized lightning channel itself acts as an antenna and radiates electromagnetic waves over a broad frequency range from a few Hz up to several GHz (Rakov and Uman 2003). At the lowest frequencies of a few Hz, lightning radio emissions can produce standing waves called Schumann resonances in the ionospheric cavity of a planet. However, their intensity is low and classical theory indicates that a sensitive in situ detector is needed, as the frequency is normally considered too low for the waves to escape the “ionospheric cutoff” (there is some evidence for a “leaky” ionosphere, permitting remote sensing of terrestrial Schumann resonances (Simões et al. 2011)). Schumann resonances were detected on Titan but are attributed to a non-atmospheric electricity cause (Béghin et al. 2007), so this type of data needs to be carefully interpreted. A probe delivered to an ice giant atmosphere should thus have a lightning detector in the very low frequency range (VLF, 3–30 kHz), because lightning radio emissions are much stronger at these frequencies and can propagate over several thousands of kilometres within the ionospheric cavity. For example, the Galileo probe detected VLF bursts attributed to lightning with its lightning and radio emission detector (LRD) during its descent into the Jovian atmosphere. The LRD used a ferrite-core magnetic radio frequency antenna from 100 Hz to 100 kHz (Lanzerotti et al. 1992; Rinnert et al. 1998). VLF signals from lightning can also be detected from outside the planet’s ionosphere in the form of whistlers, which are electromagnetic waves guided along magnetic field lines. Whistlers detected by the Voyager 2 plasma wave instrument around 6–12 kHz are the most important indication for lightning on Neptune (Gurnett et al. 1990), and we will discuss this observation in more detail in the next subsection.
The radio emissions from lightning called “sferics” can also be detected in the high frequency (HF) range (3–30 MHz). Such HF sferics, whose frequency is above the ionospheric cutoff, can pass directly through the ionosphere and freely propagate to orbiting spacecraft. Prominent HF sferics were detected at Saturn and (incorrectly) named “Saturn Electrostatic Discharges” (SED, Warwick et al. 1981), and at Uranus, where they were analogously called “Uranian Electrostatic Discharges” (UED, Zarka and Pedersen 1986). SED were detected by the radio instruments on Voyagers 1 and 2 (Zarka and Pedersen 1983), by Cassini (Fischer et al. 2008), and by ground-based telescopes (Konovalenko et al. 2013). UED were only detected by Voyager 2 (Zarka and Pedersen 1986); SED and UED are compared in the next subsection. The spacecraft used electric monopole or dipole antennas and corresponding receivers for radio wave reception (Warwick et al. 1977; Gurnett et al. 2004). In the HF range the receivers swept through the frequencies with step increments of a few hundred kHz and dwelled at each frequency for several tens of milliseconds.
Another interesting detection was made recently with the Juno Microwave Radiometer (MWR) which measured impulses attributed to Jovian lightning at frequencies of 600 and 1200 MHz with a receiver bandwidth of 18 MHz (Brown et al. 2018). This detection was very surprising since no HF sferics were detected at Jupiter (probably due to ionospheric absorption, as pointed out by Zarka 1985), and radio emissions of terrestrial lightning in the ultra-high frequency band (UHF, 300–3000 MHz) have rarely been studied due to the decline of intensity with increasing frequency. The MWR high frequency observations have been confirmed by parallel observations of whistlers with the Juno Waves instrument (Kolmasova et al. 2018; Imai et al. 2018). Modern receivers with low noise figures and wide bandwidth should allow good observations of impulsive radiation of lightning at microwave frequencies (Petersen and Beasley 2014). Thus, the MWR lightning detections at Jupiter have opened up a new frequency window to study planetary lightning (e.g., at ice giants). At frequencies of a few hundred MHz the flux of Jovian synchrotron radiation from electrons trapped in the radiation belts is typically much higher than the flux from Jovian lightning (Brown et al. 2018), but Juno was flying below Jupiter’s radiation belts, improving the signal to noise ratio. At Uranus there is no synchrotron radiation that could obscure potential microwave radio emissions from Uranus lightning.
Detecting optical emissions from lightning at Uranus and Neptune is probably very difficult, since the discharges might take place in the water or ammonium hydrosulphide clouds (see Sect. 2.3) deeper in the atmosphere (\(40 \times 10^{3}\) hPa or 40 bar) than at Jupiter or Saturn (Atreya and Wong 2005). While many spacecraft easily detected the optical flashes from Jupiter’s night side (Voyager 1 and 2, Galileo, Cassini, New Horizons, Juno), detecting optical flashes from Saturn’s night side with the Cassini camera turned out to be more complicated. This was due to the ring shine and the greater depth of the discharges at the \(8\mbox{--}10 \times 10^{3}\) hPa level (Fischer et al. 2008) compared to typical depths of \(5 \times 10^{3}\) hPa at Jupiter (Dyudina et al. 2004). Finally, the first optical flash detection from Saturn’s night side by Cassini was made around Saturn equinox in August 2009 when the ring shine was minimal (Dyudina et al. 2010). Interestingly, during the Great White Spot event on Saturn with its high SED rate of \(10~\mbox{s}^{-1}\) (Fischer et al. 2011), the Cassini camera also managed to image flashes on Saturn’s day side with a blue filter by subtracting two temporally close images from each other (Dyudina et al. 2013). However, this technique might only work with high flash rates, and the UED rate measured by Voyager 2 was quite low. Nevertheless, optical images of atmospheric features at the ice giants are still very important since they can give clues about the location of possible lightning flashes or if storms are present at all. At Saturn, for example, it was found that storm clouds were brighter in the images when the SED rate was high (Dyudina et al. 2007), indicating enhanced vertical convection. Optical observations of the ice giants with ground-based telescopes or the Hubble Space Telescope are also important to study atmospheric dynamics and to specify times when it is worth searching for lightning radio emissions with large ground-based radio telescopes. We note that the LRD on-board the Galileo probe also had two photodiodes to measure optical flashes, but no optical signatures were found (Rinnert et al. 1998). Optical Jupiter lightning flashes were detected recently by the Juno orbiter’s camera (JunoCam) and star tracker (Becker et al. 2019). If one does not intentionally fly into a thunderstorm (which would be very hard to realize technically at ice giants), detection of optical flashes or acoustic thunder with an in-situ probe seems improbable.
Voyager 2 Observations
The PRA (Planetary Radio Astronomy) instrument on Voyager 2 detected 140 impulsive bursts in the frequency range of 0.9 to 40 MHz (upper frequency limit of the PRA) during the January 1986 Uranus flyby (Zarka and Pedersen 1986). As was described above, these bursts were termed UED (Uranian Electrostatic Discharges) in analogy to the similar radio emissions of SED (Saturn Electrostatic Discharge). The mean UED burst duration was 120 ms, and they were detected within distances of ∼600000 km of Uranus on 24th–25th January 1986. Figure 1 shows both the UED rate as a function of distance in Uranian radii (\(1~\mathrm{R}_{\mathrm{U}}=25600\) km), and the distribution of all UED in the time-frequency plane. Due to the sweeping PRA receiver with a dwell time of 30 ms in each frequency channel, the UED are seen as short bursts over a limited frequency interval, despite the notion that they should be intrinsically broadband in reality. The low number of UED from ∼20 to 30 MHz seen in the lower panel of Fig. 1 is likely to be from an increased spacecraft noise level.
Although the UED tend to group in episodes, no periodicity corresponding to the planetary rotation (∼17.25 h) was detected, unlike SED. The intensity of UED is about an order of magnitude weaker than the intensity of SED. The average intensity normalized to the corresponding intensity that would be received at the Earth (at 1 AU) is \(6\times10^{-24}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) for the UED in the HF (high frequency) band (1.3–40 MHz) and \(2\times10^{-22}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) in the LF (low frequency) band (below 1.3 MHz). This corresponds to spectral source powers of 2 and \(60~\mbox{W}\,\mbox{Hz}^{-1}\), which is 2 to 3 orders of magnitude larger than the source power of terrestrial lightning, respectively. Neither whistlers nor optical signals of lightning or aurora were detected on the night side of Uranus by Voyager 2 (Smith et al. 1986).
During the Voyager 2 Neptune flyby on 25th August 1989, the plasma wave system (PWS) detected a series of 16 whistler-like events within ∼20 minutes at radial distances from ∼1.3 to 2 Neptune radii (\(1~\mathrm{R}_{\mathrm{N}}=24762\) km) and at magnetic latitudes from \(-7^{\circ}\) to \(33^{\circ}\) (Gurnett et al. 1990). The frequencies ranged from 6 to 12 kHz, and the large dispersions around \(26000~\mbox{sHz}^{1/2}\) fit the Eckersley law for lightning generated whistlers, for which the dispersion is frequency-independent (Rakov and Uman 2003). Eckersley (1935) had shown that the arrival time \(t\) of a terrestrial whistler is given by \(t=t_{0}+D/\mathit{sqrt}(f)\) with \(t_{0}\) as the time of the lightning flash, \(f\) as the wave frequency, and \(D\) as the dispersion constant. The dispersions are too large for a single direct path from the lightning source to the Voyager 2 spacecraft, and so the most likely propagation path involves lightning on the dayside of the planet with multiple bounces from one hemisphere to the other. Figure 2 shows a frequency time-spectrogram of Neptune whistler number 4, which lasts tens of seconds.
Farrell (1996) interpreted the highly dispersed whistler-like signals as Z-mode radiation and not as whistler mode emission. Its source could be lightning, but a magnetospheric source is also possible. A magnetoplasma is a birefringent medium in which radio waves can propagate as ordinary or extraordinary waves. The Z-mode can be seen as the low frequency branch of the extraordinary wave, whereas the whistler is the low frequency branch of the ordinary wave (see, e.g., Gurnett and Bhattacharjee 2017). The Neptune lightning hypothesis is somewhat supported by the fact that Kaiser et al. (1991) also detected four weak sferics at high frequencies (18–31 MHz) from a distance of 5–6 RN (Neptune radii) in the Voyager 2 PRA Neptune data. The average Neptune sferic intensity was \(5\times10^{-18}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) at 1 RN corresponding to an intensity of \({\sim }1.35 \times 10^{-25}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) at 1 AU, which is about 45 times weaker than the average UED intensity in the high band. The spectral source power of Neptune sferics would be \({\sim }0.04~\mbox{W}\,\mbox{Hz}^{-1}\), which is comparable to the source power of a strong terrestrial lightning flash. No optical lightning detection was made by Voyager 2 at Neptune. We do not know if lightning on the ice giants is constant, like on Jupiter, or intermittent, like on Saturn. Nevertheless, it is remarkable that Voyager 2 detected lightning at all four giant planets, albeit tentatively at Neptune. The properties of Uranus and Neptune lightning detected by Voyager 2 are summarised in Table 1.
Table 1 Characteristics of Uranus and Neptune lightning detected by Voyager 2. The average flux and source power of the HF sferics represent the values around 15 MHz Possible Origins of Lightning—Clouds and Microphysics
Uranus and Neptune have very similar atmospheric structures, inferred from remote sensing observations, radiative transfer and photochemical modelling. The most recent interpretation, broadly applying to both ice giants, in Mousis et al. (2018) has a stratosphere (0.1–30 hPa) of an extended, mainly hydrocarbon haze, generated by gravitational settling of aerosol particles from methane photolysis. In the troposphere there are expected to be ice cloud layers of methane (CH4), with their base at 1300 hPa, a physically thin but optically thick hydrogen sulphide (H2S) layer between 2000–4000 hPa, and beneath this ammonium hydrosulphide (NH4SH), followed by water (H2O) down to about \(50 \times 10^{3}\) hPa. The water-ice cloud forms the top of a massive liquid water cloud that could extend down to at least \(1000 \times 10^{3}\) hPa (Mousis et al. 2018). In a study of Neptune cloud charging, a slightly different structure was assumed by Gibbard et al. (1999). This included a region of ammonia (NH3) ice cloud at the same level as the H2S ice cloud, with the deepest liquid cloud as a mixture of H2O, NH3 and NH4SH.
Terrestrial thunderstorms are used as an analogy when considering whether these clouds could support lightning. Observations and experiments have shown that discharges are generated in mixed-phase water clouds, specifically, from collisional charge transfer between soft hail (graupel) and ice crystals, producing oppositely charged particles which are then separated by convection to generate a potential difference that eventually exceeds the breakdown voltage of air, causing a discharge (Saunders 2008). Lightning at the giant planets has been attributed to a terrestrial-like process in mixed-phase water clouds, mainly because the flash depth from visible observations at Jupiter and Saturn is consistent with the anticipated depth and temperature range of the water cloud region (Aplin and Fischer 2017). Lightning is possible in non-water clouds as long as there is adequate convection to create the clouds and sustain separation of the charged particles, and the cloud material is sufficiently polar to support charge transfer (physical properties of each of the proposed cloud layers are summarised in Table 2). Additional constraints related to the local atmospheric properties are that the breakdown voltage must be achievable by charge separation within the thundercloud. If the gas is too electrically conductive this limits particle charging through decreasing the relaxation time \(\tau \) given by \({1} / {\varepsilon _{0} \lambda }\) where \(\lambda \) is the conductivity and \(\varepsilon_{0}\) the permittivity of free space, and preventing an electric field from building up.
Table 2 Physical properties of cloud-forming materials in ice giant atmospheres Gibbard et al. (1999) simulated particle growth, charging, fall velocities and breakdown voltage for the cloud layers described above to determine which layer could support lightning, with collisional charging parametrised from laboratory experiments for terrestrial clouds (e.g. Saunders 2008). H2S and CH4 ice clouds were essentially ruled out as possible lightning generators due to their single phase and low polarisability. In the deep water cloud the limiting factor was the breakdown voltage, which is expected to be \(250~\mbox{MV}/\mbox{m}\) at \(50 \times 10^{3}\) hPa, whereas the electric fields achieved are only \(10~\mbox{MV}/\mbox{m}\). Electric fields are limited by electrostatic levitation of charged particles, which suppresses the generation of distinct areas of opposite charge within the cloud. Similar effects are expected in NH4SH clouds, but the electric field was a factor of 3 lower than the breakdown voltage. Based on this, Gibbard et al. (1999) state that lightning is very unlikely in Neptune’s water clouds, but could be possible in NH4SH. These calculations were limited by a lack of data on the physical properties of NH4SH, most likely because it is unstable at terrestrial surface conditions, hindering laboratory characterisation (Loeffler et al. 2015). Gibbard et al.’s (1999) results are consistent with the lack of optical detection of lightning from Uranus and Neptune, as lightning in the deep cloud layers would not be visible from orbit. This work also neglected the background conductivity of the gas in the cloud layers, for which no information was available (see Sect. 3.3.1).
Ground-Based Radio Observations
Lightning Detection with Ground-Based Radio Telescopes
Searching for wide-band signals like lightning from other planets with ground-based radio telescopes is not trivial given the presence of Earth lightning and other natural and artificial radio interference. So far this has only been successful for Saturn (Zakharenko et al. 2012; Konovalenko et al. 2013), and we will describe in the following paragraph how this was done with the UTR-2 radio telescope.
The Ukrainian T-shaped Radio telescope model 2 (UTR-2) was constructed near Kharkov in the early 1970s, and it is still one of the largest ground-based radio telescopes in the decametric frequency range. The telescope is split into 12 sections that form three T-shaped arms (North, South, West) each 900 m long. In total it consists of 2040 fat linear dipoles (which have a broader frequency response than thin dipoles), with a frequency range of 8 to 32 MHz. UTR-2 has a large effective area of up to \({\sim }140000~\mbox{m}^{2}\) and a high directivity, with the main beam \(0.5^{\circ}\) wide (Konovalenko et al. 2016). UTR-2 can provide simultaneous observations with up to 5 spatially separated antenna beams, and the beam can be electronically steered within a wide range of both sky coordinates (azimuth, elevation). The multi-beam capability was essential for the detection of Saturn lightning, for which two beams were used, one directed at the source, Saturn, here called the ON beam, and one directed a few degrees off target (OFF). A Saturn lightning (SED) signal should only occur in the ON beam and not in the OFF beam, whereas most interference signals come in through the side lobes of the telescope and appear in both ON and OFF beams. The known characteristics of SED (duration, intensity, wide-band signal, almost flat spectrum in decametric frequency range) and the simultaneous SED observations with the Cassini Radio and Plasma Wave Science (RPWS) instrument (Gurnett et al. 2004) in the Cassini era (2004–2017) also helped to correctly identify the signals.
After the initial ground-based detection, SED were also observed with higher time resolution, and it was found that, just like the pulsars more usually observed with radio telescopes, the signals are dispersed by the interplanetary medium (and the ionospheres of Saturn and Earth) with a characteristic frequency-dependent propagation delay. This time is typically several hundreds of microseconds over a 10–20 MHz difference in frequency (Mylostna et al. 2013). This dispersion is typical in radio astronomy, and the time delay it causes can be defined in terms of the “dispersion measure” (DM), a constant which is expressed in units of parsecs per cubic centimetre (\(\mbox{pc}\,\mbox{cm}^{-3}\)), to represent the distance and the electron concentration in the interplanetary medium (e.g. Kraus 1966). The DM is often found by empirically searching through a range of possible values to assess which gives the best overall signal-to-noise ratio. “De-dispersion” is often applied as a post-detection data analysis technique to compensate for the delay introduced by dispersion and maximise the signal-to-noise (e.g. Hankins and Rickett 1975).
Zakharenko et al. (2012) suggested that the SED intensity peaks are in short bursts that become blurred at high time resolution. The dispersion delay across the range of frequencies observed would also affect the smoothing of short broadband bursts, especially if the bursts are infrequent. This was confirmed in high spectral resolution observations (Mylostna et al. 2014), Fig. 3. An important benefit of ground-based SED observations is the discovery of several time scales in which Saturn’s lightning was especially intense. In the case of the 2010–2011 storm (Fischer et al. 2011), these were characteristic durations of (a) tens of ms, (b) 30–300 μs, and (c) 2–5 μs (Mylostna et al. 2014).
Figure 3 shows that intense bursts only occupy a small fraction (10–20%) of the total flash duration. Therefore, their peak intensity when detected with a low temporal resolution will be significantly underestimated. In addition, the dispersion delay between the lower and upper frequency limits of 16.5 to 33.0 MHz is about 300 μs. Over the same period, the average duration of the most intense sub-millisecond components of the discharge ∼70 μs. Thus, integration without eliminating dispersion delay also underestimates the lightning flux density. Figure 3 demonstrates this effect by showing the maximum flux densities obtained from the same data with and without elimination of the dispersion delay with a simple post-detector de-dispersion technique. The calculated flux density is enhanced by a factor of two if the de-dispersion is applied (Mylostna et al. 2014). The gain in sensitivity of a factor of two or three can be decisive for Uranus lightning detection, because without it the measurements are at the sensitivity threshold. In the next subsections, we will estimate this threshold in terms of the flux density of the UED (Zarka and Pedersen 1986) and the use of radiometric gain. We will also discuss the possibilities of increasing sensitivity, using the radiative properties described above, optimising the observations for the presence of short bursts and dispersive delay of signals, and potentially with the help of two or more antennas far apart on Earth’s surface.
Potential for Ground-Based Observations of Lightning from the Ice Giants
It will be shown below that the detection of Uranus lightning (UED) is within the technical capabilities of large ground-based radio telescopes (see also Zarka et al. 2004). The fluctuation \(\sigma_{\mathrm{sky}}\) of the galactic background is given by
$$ 4 \sigma _{\mathrm{sky}} = \frac{8 k_{B} T}{A_{\mathrm{eff}} \sqrt{\Delta f\Delta t}} $$
(1)
where \(k_{{B}}\) is Boltzmann’s constant, \(T\) the galactic background temperature, \(A_{\mathrm{eff}}\) the antenna effective area, \(\Delta f\) the frequency bandwidth, and \(\Delta t\) the integration time. We multiplied the sky background by a factor of 4 to account for the fact that a detectable signal should be at least a factor of 4 above the background fluctuations. The galactic background temperature is ∼30000 K at 20 MHz (see e.g., Kraus 1966). The total effective area of the UTR-2 radio telescope is \({\sim }140000~\mbox{m}^{2}\), but here we take \(A_{\mathrm{eff}} = 90000~\mbox{m}^{2}\). This arises because for non-zenith sources, the effective area is scaled by \(\cos(z)\) where \(z\) is zenith angle. For example, for a source with \(\mbox{declination} = 0^{\circ}\) and latitude of the \(\mbox{UTR-2} = 49.63^{\circ}\), \(A_{\mathrm{eff}} \sim 90000~\mbox{m}^{2}\). In Fig. 4 we have drawn \(4\sigma _{\mathrm{sky}}\) as a function of the receiver bandwidth (from 100 kHz to 10 MHz) and the integration time (20 ms or 0.1 s).
Figure 4 shows that it is necessary to use at least a bandwidth of 1 MHz with an integration time of 20 ms to get a background fluctuation that is smaller than the peak flux of Uranus lightning (UED). The average UED flux at Earth was calculated using the flux of \(6\times10^{-24}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) in the HF band at 1 AU (Zarka and Pedersen 1986), which translates to a flux of 1.7 Jy (\(1~\mbox{Jy} = 10^{-26}~\mbox{Wm}^{-2}\,\mbox{Hz}^{-1}\)) at a distance of 19 AU (average Uranus-Earth distance). The peak flux of UED at Earth might be almost 30 Jy (\(10^{-22}~\mbox{W}\,\mbox{m}^{-2}\,\mbox{Hz}^{-1}\) at 1 AU around 15 MHz in Fig. 4 of Zarka and Pedersen 1986). Since the UED rate detected by Voyager 2 was low, one should base the choice of receiver bandwidth and integration time on the average UED flux, which is not even reached with a bandwidth of 10 MHz. An integration time of 0.1 s is of the same order as the expected signal duration, which is a reasonable choice to achieve a first detection. Longer integration times would dilute the signal, and shorter integration times would need strong UED around the peak flux which should be rather rare events. It is important to note that for short signals like lightning one cannot simply enhance the detectability by using very long integration times. With the UTR-2 frequency range of 8–32 MHz (Konovalenko et al. 2013) one also cannot have a much larger bandwidth either. The integration in bandwidth can be done in the post-processing stage, so it is possible that the receiver bandwidth during the actual observation is smaller. The same holds for the integration time.
We conclude that UED detection should be possible with the UTR-2 radio telescope, but we are close to its sensitivity threshold. In contrast to the UED, the SED intensity at Earth are a few hundred Jy on average with a peak intensity as high as 45000 Jy. This has enabled study of the fine structure of SED down to the microsecond range (Mylostna et al. 2014). Finally, we note that the expected average flux of Neptune lightning at Earth would only be around 15 mJy (Kaiser et al. 1991), which would need a radio telescope more than 100 times larger than UTR-2 for a detection in the decametric frequency range.
First Ground-Based Attempts at Uranus Lightning Detection
In summer 2014 ground-based infrared images made with the W.M. Keck observatory showed several storms in the atmosphere of Uranus. In spite of the expected decline in convective activity following the 2007 equinox, eight storms were detected on the planet’s northern hemisphere on August 5–6 2014 (de Pater et al. 2015). One of them was the brightest storm ever seen on Uranus, located around a planetocentric latitude of \({\sim }15^{\circ}\mbox{N}\) and reaching altitudes of ∼330 hPa, well above the uppermost methane-ice cloud layer. The brightness of this feature had already decreased substantially by August 17, and it might have been formed by strong updrafts. Another, deeper, cloud feature (at about 2000 hPa) was seen later (October 2014) by amateur astronomers and by the Hubble Space Telescope at a latitude of \(32^{\circ}\)N, but overall the storm activity was significantly decreased by October 2014.
Based on initial information from infrared and optical observations, two campaigns were conducted at UTR-2 in 2014: August 18–25 and October 6–12. As previous work had indicated that the source of lightning may not be tied to the exact position of the storm, observations were made during the entire period when the planet was above the horizon and the effective antenna area did not drop very much. Observations with time interval \(+/{-} 3\) hours from culmination give a zenith angle of Uranus in culmination \({\sim }45^{\circ}\) (declination of the planet in August-October 2014 was about \(5^{\circ}\)), and near \(70^{\circ}\) at the start and end of a measurement sequence. The observation technique was as follows: three receivers in correlation mode (Zakharenko et al. 2016) of antennas North-South and West-East (which provides the maximum set of analyzed parameters: module and phase of antenna signal cross-spectra and their individual power spectra) were connected to beams 1, 3 and 5 of the radio telescope. Beam 3 was directed at the source (ON), and beams 1 and 5 (both OFF) were turned away from the source by \(1^{\circ}\) along the meridian: beam 1, to the south and beam 5 to the north. The height of the source above the horizon varied from 20 to 45 degrees, while the effective area of the radio telescope was \(50000\mbox{--}100000~\mbox{m}^{2}\). With a bandwidth of about 10 MHz and an integration time of 20 ms, the sensitivity of the UTR-2 was sufficient to detect the maximum lightning flux (see Fig. 3). However, over 15 days of observation, there were no events that were clearly visible in the ON beam and absent in the OFF beams.
Subsequently, one week of similar Uranus observations have been carried out each September-October since 2015, when the culmination of Uranus in the middle of the night provided the minimum radio frequency interference and therefore the best conditions for scanning observations. No lightning signals from Uranus have yet been recorded.