Drifting discrete Jovian radio bursts reveal acceleration processes related to Ganymede and the main aurora

Radio detection at high time-frequency resolutions is a powerful means of remotely studying electron acceleration processes. Radio bursts have characteristics (polarization, drift, periodicity) making them easier to detect than slowly variable emissions. They are not uncommon in solar system planetary magnetospheres, the powerful Jovian “short bursts (S-bursts)" induced by the Io-Jupiter interaction being especially well-documented. Here we present a detection method of drifting radio bursts in terabytes of high resolution time-frequency data, applied to one month of ground-based Jupiter observations. Beyond the expected Io-Jupiter S-bursts, we find decameter S-bursts related to the Ganymede-Jupiter interaction and the main Jovian aurora, revealing ubiquitous Alfvénic electron acceleration in Jupiter’s high-latitude regions. Our observations show accelerated electron energies are distributed in two populations, kilo-electron-Volts and hundreds of electron-Volts. This detection technique may help characterizing inaccessible astrophysical sources such as exoplanets.

For a significant fraction of the processed dynamic spectra (5 to 10%, see Table 1 in the article), positive slopes are found.We have checked visually all the corresponding cases with SN R ≥ 5, and we have found that none of them actually corresponds to real positively drifting bursts.All consist of dynamic spectra devoid of Jupiter signal and entirely dominated by RFI.These dynamic spectra are not completely cleaned by the RFI mitigation steps, and some of them accidentally result in detections with positive slopes very close to the vertical or horizontal axis ±15 • .Figure 1 displays a representative example of such a positive slope detected by the pipeline.

SNR distribution per type of emission and slope interval
Figure 2 shows, for each type of emission (Io-induced, Ganymede-induced, Main aurora) detected in LH and RH circular polarization, the distribution of SNR for the three intervals of drift-rates listed in Table 1 and displayed in Figure 3  Supplementary Figure 2: Histograms of SNR values for the three types of emissions identified and two polarizations.The same intervals of drift-rate values as in Table 1 in the article are used, and the same color codes as in Figure 3 in the article.See text for details. and Main aurora).The data were recorded by the NewRoutine digital receiver also connected to the NDA.This receiver observes Jupiter for longer periods of time than JunoN (8 hours per day) and provides thus complementary observations with high SNR but lower resolutions than JunoN [2,1].These context dynamic spectra, plotted with a compressed timescale (1-3 hours on Figure 3a-c), reveal the overall arc-shaped structures in which we detected S-bursts with JunoN.Their sense of curvature, polarization, spectral range, and time of occurrence can be compared to simulation results for identifying their origin [3].The geometry at the time of each observation is indicated by the color lines on the CML-ϕ Io and CML-ϕ Ganymede maps of Figure 3d,e (CML = observer's longitude, ϕ Io = orbital phase of Io, ϕ Ganymede = orbital phase of Ganymede).Both approaches consistently indicate that panels a,b and c of Figure 3 correspond respectively to Io-C, Ganymede-C, and non-Io/non-Ganymede (i.e.main aurora) Jovian decameter emissions.

Presence of O mode emission ?
Figure 4 displays the distribution of drift rates per circular polarization only for the bursts detected in the Ganymede "C" box of Figure 1b in main text.These are the only bursts unambiguously related to Ganymede (bursts in box "B" are actually related to Io, and box "D" is defined with less well confidence [4]).Ganymede "C" bursts come from Jupiter's southern hemisphere, and should thus have LH circular polarization if emitted on the X mode.The left panel of Figure 4 indeed contains the largest number of bursts, of LH polarization, with both fast and slow negative drifts, thus slow drifting bursts cannot be explained by O mode only.But the right panel shows a dominant peak for slow drifts, in RH polarization, consistent with O mode emission produced by <1 keV electrons.It suggests that O mode bursts may indeed be present.Further detailed statistics on the polarization and slopes of a larger number of bursts sorted by emission type (Ioor Ganymede-A, B, C, D) should allow to study in details the emission mode and its relation with the bursts drifts.This provides a powerful means to characterize the electron cyclotron maser efficiency for various electron energies.

Ganymede C (RH)
Supplementary Figure 4: Histograms of slopes (drift-rates) for Ganymede-C emissions in both polarizations.These histograms shows the propotion of drift-rates in LH polarization (left) and RH polarization (right), for the bursts detected in the Ganymede "C" box of Figure 1b in the paper.The color code is the same as in Figure 2.

Supplementary Figure 1 :
Example of a spurious detection of signals with a positive slope.The dynamic spectrum (left panel) is dominated with weak vertical RFI that could not be completely removed by the RFI mitigation step of the processing.This results in a very intense horizontal line in the 2D-FFT (middle panel), which is not completely eliminated by the contrast enhancement procedure.As a result, a peak is detected with an SNR ≥ 7 in the Radon transform (rightside panel) very near 105 • , i.e. close to the horizontal axis but on the side of positive slopes.
Figure2shows, for each type of emission (Io-induced, Ganymede-induced, Main aurora) detected in LH and RH circular polarization, the distribution of SNR for the three intervals of drift-rates listed in Table1and displayed in Figure3in the article.Positive drifts are clearly associated with lower SNR in all cases, consistent with weak residual RFI.They are much less numerous than negative (real) drifts in Fig 2a,b (Io-induced case), where the largest SNR values are associated with faster drift-rates <-10 MHz/s.Positive drifts are less numerous in Fig 2c (Ganymede-induced, LH), but dominant in Fig 2d (Ganymede-induced, RH).This can be explained by the fact that for the time intervals that we analyzed, most of the Ganymede-induced emissions were LH polarized.For Ganymede, the faster negative drifts (df /dt < −10 MHz/s) are less numerous than the slower ones (−10 ≤ df /dt < 0 MHz/s), but they are detected with slightly larger SNR.Drifting bursts associated with the main aurora (Fig2e,f) display the same general behaviour as Ganymedeinduced emissions, except that in this case most of the detected emissions were RH polarized.