The injections within Earth's magnetosphere (Fig. 1) involve particles with kilo-electron volt (keV) to mega-electron volt (MeV) energies4. Often occurring at radial distances of 6 to 10 Earth radii, injections are one component of global dynamic events called ‘magnetospheric substorms’. Substorms represent, in part, the transient release of energy stored in the magnetosphere with stressed magnetic fields13. The energy source for Earth's substorms is the solar wind of charged gases, or plasmas, emanating from the Sun. Substorms create dramatic brightening of the aurora at high geographic latitudes and a substantial expansion of the regions where auroral emissions occur.

Figure 1: Energetic electron injection measurements within Earth's space environment.
figure 1

The response of three different electron energy channels is shown as measured from the Earth's geosynchronous orbit (6.7 Earth radii circular, near-equatorial). We note the energy-dispersed nature of the channels, with different energies arriving at the spacecraft at different times. Plotted after ref. 5.

The recent discovery of Earth-like charged particle injections within Jupiter's magnetosphere14,15 is surprising because Jupiter's magnetosphere is powered mostly from the inside by the rapid but steady planetary rotation rather than from the outside by the variable solar wind. The role of injections in generating auroral emissions at Jupiter has been heretofore unknown, to our knowledge.

A unique opportunity to address dynamics in Jupiter's space environment was made available by a Jupiter joint observation campaign in late 2000 and early 2001. It involved the fly-by of the Cassini spacecraft, headed toward Saturn, the Galileo spacecraft orbiting Jupiter, and remote imaging by the Hubble Space Telescope (HST). During the campaign, Galileo recorded energetic electron injection signatures at radial distances of 10 to 13 Jupiter radii (Fig. 2). A simple model (Fig. 3) explains the energy-dispersed character of these signatures (different particle energies arrived at Galileo at different times). The model is closely analogous to models derived from injections at Earth16,17,18. Quantitative analysis (Fig. 4) reveals the temporal relationship between the injections and the signatures. At the times of the injections, around 15 h before the dispersed signatures were observed, Galileo was at a radial distance of about 20 Jupiter radii and recorded no obvious signature of the injections occurring closer to Jupiter.

Figure 2: Energetic electron injection measurements within Jupiter's magnetosphere.
figure 2

Log [energy (keV)] versus time (hours of day 363, 2000; top scale) versus particle log [intensity (cm-1 s-1 sr-1 keV-1], shown as a colour scale, display of ion (top) and electron (bottom) measurements from the energetic particle detector on the Galileo spacecraft for the radial range of 19 to 8 Jupiter radii (bottom scale). The energy-dispersed injections are visible in the right-hand portion of the electron display beginning at about hour 13. The electron sensor is nearly saturated at the lower energies (top of the electron display) and so the relative variations at low energies is underrepresented here.

Figure 3: Schematic for the generation of injections within Jupiter's magnetosphere.
figure 3

The hot plasma injection (left side) occurs quickly, and then the dispersive drift, driven by Jupiter's rotation and magnetic field inhomogeneities, occurs slowly and generates energy-dispersed particle signatures at Galileo. The phase space density (PSD) is a transformation of the particle intensities into a form that is invariant for the kind of transport thought to occur here. Plotted after refs 14 and 15.

Figure 4: Quantitative analysis of Jupiter's energy-dispersed electron injection signatures.
figure 4

Plotted points are the times (hours in day 363 of 2000) versus energy of the peaks in the electron channel responses for the three main electron injection signatures shown in Fig. 2. The theoretical fits use the best estimates of the energy-dependent drifts within Jupiter's magnetosphere15 to reconstruct the times (shown in the figure) when the fast injections (Fig. 3) occurred. These estimates make use of a magnetic field model that incorporates electric currents internal to Jupiter and the latest estimates of typical currents external to Jupiter within the magnetosphere19.

Ultraviolet HST auroral images, similar to those in previous reports7, were taken during the Galileo operations and remapped to polar coordinates (Fig. 5). The images reveal a distinct auroral emission patch in eight consecutive images (100-s exposures) obtained over a 36-min HST operation period (Fig. 5, extreme right and close to the yellow square labelled ‘15 h’). The auroral patch moved with Jupiter as Jupiter rotated, and thus maintained its latitude–longitude position. The last image of the set was taken at about 1230 UT, about 30 min before the beginning of the first injection signature in Fig. 2.

Figure 5: Hubble Space Telescope (HST) ultraviolet image of Jupiter's northern hemisphere aurora transformed to a Jupiter system-III polar coordinate system.
figure 5

Zero and 90 degree longitudes are straight up and horizontal to the right, respectively. The auroral emission patch of interest is to the extreme right, adjacent to the yellow square labelled ‘15 h’. The yellow squares and yellow dashed line show magnetic projections of the Galileo spacecraft onto Jupiter's upper atmosphere using two models of Jupiter's magnetic field (see text for details). The model of Khurana19, that includes currents external to Jupiter, was updated by replacing the internal field model, O6, with the latest internal field model, VIP421. We note that as Galileo moved closer to Jupiter, the trajectory appears to move closer and closer to the strong emissions of the bright, poleward ring of the auroral oval. Although this is not the subject of this Letter, this characteristic is contrary to what one might expect if this ring of emission has a source that predominates steadily at some radial distance. However, as verified with previous work (see Fig. 1 of ref. 9), present field models do not predict the poleward kink in the global auroral distribution of the brightest aurora, revealed here aligned along the 140–150° longitude meridian and in previous work7. The structure of the bright auroral oval is clearly different in the kink region from its structure elsewhere, and so more than just a refinement in magnetic mapping at high latitudes will be needed to understand it. Local time effects6,9 and perhaps even the effects of dynamics may be involved here. The brightest aurora is thought to map to distances as large as 20–30 Jupiter radii, and given the mapping sensitivities9, our results are consistent with that hypothesis.

We believe the patch was generated by the first of the injections observed by Galileo (Fig. 4). In Fig. 5, the yellow squares and the yellow dashed line show two different calculations of the motion of the Galileo spacecraft trajectory when mapped along magnetic field lines from the spacecraft to Jupiter's upper atmosphere. The yellow dashed line is the better estimate because the magnetic field model used to map Galileo's position includes contributions both from currents internal to Jupiter and from up-to-date estimates of average currents flowing external to Jupiter within its magnetosphere19. Adequate approximation to the magnetic field requires that external currents be included because the actual magnetic field configuration beyond about 9 Jupiter radii is distorted away from the nearly dipolar configuration expected from internal currents alone19,20. Projections derived using a field model with only internal currents (yellow squares in Fig. 5; VIP421) are shown to provide a sense of the uncertainties involved with mapping. With the better estimate, the Galileo trajectory appears to have crossed into the isolated patch between 14.7 and 15.3 h, centred on the maximum of the higher-energy portion of the injection feature (Fig. 4). This result does not depend on detailed timing because, as mentioned, the auroral patch maintained its latitude–longitude position as Jupiter rotated.

The auroral patch characteristic of rotating with Jupiter is expected because the injected electrons also rotate with Jupiter (within several per cent at 12 Jupiter radii15). Although the auroral patch of interest here was the brightest observed during the joint observation campaign, such patches are common within HST images. Likewise, the occurrence of jovian electron injections is also common15. The other energetic electron injections revealed in the Galileo data (Figs 2 and 4) map to regions (Fig. 5) that also show measurable auroral emissions. However, those emissions do not appear patch-like and may have been active even in the absence of injections.

Studies of Earth's magnetosphere13 suggest two different ways that injected energetic particles can generate auroral emissions. (1) The particle energy distributions are modified during injection and become unstable to exchanges of energy with magnetospheric wave modes. The waves scatter particles so that some travel narrowly along the magnetic field lines until they encounter the atmosphere. (2) The injected particle cloud is a high-pressure region and so electric current flows along its boundary. This pressure-driven (diamagnetic) current diverges along the leading and trailing edges of the rotating cloud because the magnetic field strength changes with radial distance. Currents are driven along the magnetic field lines towards and away from the planet and can interact strongly with plasmas close to the planet. At Jupiter, that interaction would yield downward accelerated electrons, and atmospheric auroral emissions, at the trailing edge of the rotating plasma cloud. Although there is substantial uncertainty in the magnetic mapping, the position of the auroral patch does match best with either the trailing edge of the electron cloud (the second mechanism) or the centre of that portion of the cloud that contained the higher-energy electrons measured (the first mechanism).

For an aurora resulting from the scattering process, the maximum power density that the measured (>20 keV) electron cloud can provide to the aurora is 60 ± 30 erg cm-2 s-1. Higher power densities are possible with the electric current generation mechanism. Models of interaction between electrons and Jupiter's atmosphere22, recalculated with the energy distribution shapes measured in Fig. 2, yield about 3 erg cm-2 s-1 for the electron input needed to explain the auroral emissions. Thus, measured electrons can supply the requisite energy with scattering efficiencies of only 3% to 10% of the maximum. Auroral optical emission spectra were not available for this event to estimate independently the electron energies involved. However, recent Galileo auroral observations measured the tangent altitude (above the 1-bar atmospheric pressure level) of peak auroral emissions at 245 ± 30 km, with some emissions extending to an altitude of 120 ± 40 km (ref. 23). Atmospheric penetration of electrons modelled for diffuse aurorae require the involvement of over 48 keV electrons to explain even the peak auroral emissions24,22.