Letter

# Evidence of a plume on Europa from Galileo magnetic and plasma wave signatures

Accepted:
Published:

## Abstract

The icy surface of Jupiter’s moon, Europa, is thought to lie on top of a global ocean1,2,3,4. Signatures in some Hubble Space Telescope images have been associated with putative water plumes rising above Europa’s surface5,6, providing support for the ocean theory. However, all telescopic detections reported were made at the limit of sensitivity of the data5,6,7, thereby calling for a search for plume signatures in in-situ measurements. Here, we report in-situ evidence of a plume on Europa from the magnetic field and plasma wave observations acquired on Galileo’s closest encounter with the moon. During this flyby, which dropped below 400 km altitude, the magnetometer8 recorded an approximately 1,000-kilometre-scale field rotation and a decrease of over 200 nT in field magnitude, and the Plasma Wave Spectrometer9 registered intense localized wave emissions indicative of a brief but substantial increase in plasma density. We show that the location, duration and variations of the magnetic field and plasma wave measurements are consistent with the interaction of Jupiter’s corotating plasma with Europa if a plume with characteristics inferred from Hubble images were erupting from the region of Europa’s thermal anomalies. These results provide strong independent evidence of the presence of plumes at Europa.

## Main

Plumes thus far identified from Hubble images are similar in spatial scale, appearing to rise ~200 km above the disk of Europa’s solid body5,6,7,10. The ones near the equator are located on Europa’s trailing hemisphere south of the equator in a region of comparatively high surface temperature11. Additional indirect evidence of a ‘unique feature’, possibly associated with a plume on the trailing edge of Europa, has been reported12. Magnetometer (MAG) data8 were acquired on eight targeted passes by Europa during Galileo’s eight years in orbit around Jupiter, but only two passes (E12 and E26) came closer to the surface than 400 km—a height at which the reported plumes5,6,7 might impose a plasma and field signature. Both passes crossed the trailing hemisphere of Europa (E12 near the equator and E26 at high southern latitude) and recorded short time-duration, large-amplitude perturbations accompanied by a sharp decrease of the field magnitude near closest approach. A previous study of the potential effect of plumes on Europa’s plasma interaction suggested that the magnetic perturbations during the E26 flyby could be associated with atmospheric inhomogeneity resulting from a plume13. Here, we show that a localized signature in the MAG data acquired on Galileo’s closest encounter with Europa (E12 flyby) is fully consistent with the perturbations expected if the spacecraft crossed a plume rising above the nearby surface.

Figure 1 shows the MAG data from E12, which remained below 400 km altitude between 12:00:59 and 12:05:37 ut on 16 December 1997. Both the magnetic field magnitude and the plasma density were exceptionally high upstream of Europa14,15, and the field fluctuated on time scales of minutes. Closest approach (altitude: 206 km) was at 12:03:20 ut. Between ~12:00 and 12:03 ut, significant changes in all 3 components of the magnetic field were observed for ~3 min. About 1 min before closest approach, the magnetic field changed by hundreds of nT in 16 s. It is this short time-scale fluctuation appearing in the context of slower background fluctuations that we regard as marking the passage through a plume of the characteristics extracted from Hubble images. Given Galileo’s 6 km s−1 speed relative to Europa, 3 min corresponds to a transverse spatial scale of ~1,000 km, comparable to the characteristic scale of imaged plumes at the spacecraft altitude of ~400 km.

The plasma wave spectrum obtained by Galileo’s Plasma Wave Spectrometer (PWS)9 reveals an isolated change concurrent with the short time-scale magnetic field perturbations. The sudden, short-duration jump in the frequency of intense emissions can be interpreted as consistent with a highly localized source of plasma, thereby supporting the hypothesis that the magnetic perturbations arise from passage through a localized plume. In the electric field spectra (Fig. 2a), a narrow-banded enhancement of power evident through the entire pass is interpreted as the upper hybrid resonance (UHR) emissions. In conjunction with the known electron cyclotron frequency, this reveals that the electron plasma density remained above or near 600 cm−3 on the inbound leg of this pass and fell abruptly on the outbound leg (Fig. 2b). This upstream density is of order three times greater than observed during all other Europa flybys14, immediately highlighting this flyby as remarkable. A pink arrow in Fig. 2 indicates the time (~12:02 ut) of the rapid field rotation identified in the MAG data. At this time, just before closest approach, the spectrum changes abruptly; there is a brief burst of electron cyclotron harmonics, including one band above the upper hybrid band previously identified14. The authors of the original study14 were reluctant to associate this feature specifically with Europa since such waves are also a characteristic of Jupiter’s magnetic equator. However, the brief excursion of the upper hybrid band to about 400 kHz might be interpreted as a local spike in the plasma density. Given the recent evidence for plumes, the changes would be consistent with entry into a different plasma regime and, if the emissions near 400 kHz are UHR waves, the electron density exceeded 2,000 cm−3 (Fig. 2b).

To test the hypothesis that the signatures seen on E12 are imposed by a plume, we modelled the effect of a plume on plasma and field properties near Europa using three-dimensional (3D) multi-fluid magnetohydrodynamic (MHD) simulations16,17. The simulation model tracks O+ (representative of magnetospheric plasma), O2+ (representative of ions originating from Europa) and electron fluids separately, and includes ionization, charge-exchange and recombination processes occurring in Europa’s atmosphere18. We placed a plume in the region of Europa’s thermal anomaly11 using the structural and density parameters consistent with plume properties inferred from the previous telescopic observations5,6,7. Our simulation assumes that upstream conditions are steady, with a background plasma density of ~600 cm−3. The strong upper hybrid frequency emissions in Fig. 2a indicate that the background density remained quite steady at this value until 12:06 ut, after which it began to decrease, dropping to ~100 cm−3 by the end of the pass. It is not clear whether the change of background density was a temporal or spatial feature, such as exit from a possible region of anomalously cold, high-density plasma referred to as a cold dense blob19. In either case, the density decrease should not have affected the analysis of the plume because it occurred only after the perturbations we were focusing on had diminished. The region of abrupt large-amplitude fluctuations that we link to a plume lasted <3 min, ending by ~12:03 ut, after which high-frequency fluctuations of the magnetic field are consistent with nominal background noise. For the purpose of this investigation, changes in the background conditions that were encountered about 0.7 Europa’s radius (RE) beyond the signature of the plume and some 3 min after the spacecraft exited the region perturbed by its presence should not affect the results. As shown below, the simulation reproduced the rapidly changing magnetic field and plasma density signatures identified in the Galileo E12 data.

Hubble observations6,7 imply that plumes may erupt over a range of Europa latitudes and longitudes within the region of elevated nighttime temperatures11. Thus, we regard the location of the base of the plume as weakly constrained and take the location of the base as parameters. Images from Saturn’s moon Enceladus20 show that plumes have complex structures. Their axes diverge from radial and there can be localized structures within a single plume. Assuming that Europa plumes may exhibit similar properties, we allow the central axis of our simulated plume to be inclined relative to the radial direction. The plume is given a conical structure and the opening angle and scale height are taken as free parameters (see Methods).

Based on the constraint on the approximate location of the plume imposed by the MAG and PWS data, a number of different combinations of plume parameters were tested. The results presented here were extracted from the run in which the plume emerged from a longitude of 245° W and a latitude of 5° S, somewhat east of the plume imaged6 in a region with elevated nighttime temperatures (90–110° on Europa’s thermal maps11). The plume is tilted in the azimuthal direction by 15° and in the latitudinal direction by 25°. The location of the plume used is shown on maps of Europa’s surface features in Fig. 3. The central column density of the modelled plume is 3 × 1020 m−2, which falls within the range of column densities (1.5 × 1020–2.3 × 1021 m−2) inferred from spectroscopic observations5,6.

The modelled magnetic fields extracted from the MHD simulations with and without a plume were compared with the measured field in Fig. 1. The measured and modelled plasma densities are plotted in Fig. 2b. Overall, the simulation corresponds well with the smoothed data, although the shocklets (very small amplitude shocks15) present in the upstream portion of the pass are not reproduced. Some discrepancies in the smoothed data are readily accounted for. On the E12 pass, the spacecraft crossed Jupiter’s equatorial plane at 11:59 ut, shortly before reaching the region of anomalous variations. The background B y component reverses sign across the equator, but the simulation was carried out in a fixed background field with B y  = 0. This explains why in Fig. 1 the measured B y is positive at the start of the interval and negative at the end of the interval, but the simulated field starts and ends with B y  = 0. The plasma conditions assumed are close to the upstream conditions measured, but the gradient of background density after closest approach is not part of the model, which is why the model field diverges from measurements after 12:08 ut. The high upstream density leads to pile-up of the field close to Europa, which is well reproduced by the model. Shocklets in the upstream flow may account for the changes in field magnitude between 11:58 and 12:00 ut.

Overlooking these discrepancies between the model and measurements on the large scale, the effect of a plume can be established by comparing the two simulation results within the region marked by the vertical lines in Fig. 1. Flow diversion around an obstacle twists the magnetic field. The rapid changes in B y near the centre of the plume, where the y-component of the field abruptly turns positive (125 nT) and then negative (−219 nT), result from flow diversion around a plume. The simulation reproduces the rotation of B y arising through a localized diversion of the flow both towards and away from Jupiter as the trajectory crosses the plume, albeit at a reduced amplitude.

The change of B x imposed by the plume can be interpreted in terms of an Alfvén wing structure21. The slowing of plasma flow on approach to Europa (Supplementary Fig. 3d) generates Alfvénic perturbations that bend the field in the direction of the flow, resulting in negative B x perturbations above the moon and positive B x perturbations below it. The B x signature at ~12:02 ut can be understood as characteristic of a mini Alfvén wing formed below a confined plume (Supplementary Fig. 5).

Starting at 12:00 ut, the variations of B z and |B| arise from a local flow stagnation and diversion. The field magnitude becomes large as a result of field pile-up where the high-density ambient plasma14 is substantially slowed as it approaches the plume (Supplementary Fig. 4f). The strong decrease in field strength then occurs where the flow is re-accelerated in the wake of the plume. The peak plasma density at the centre of the plume extracted from the simulation is ~2,000 cm−3 (Fig. 2b), very close to the enhanced electron density inferred from PWS if the anomalous emissions at 12:02 ut correspond to UHR waves.

In addition to E12, the only other Galileo pass that includes an interval at an altitude below 400 km is E26 (closest approach at 17:59:43 on 3 January 2000). Again, the spacecraft speed was ~6 km s−1, but given the relatively high altitude of this pass, a plume signature might have been encountered for only ~1 min. In the low-altitude region, no field perturbations last for even 1 min, although there is a diamagnetic decrease lasting for ~5 s. As this signature does not approximate the characteristics we have previously identified, we suggest that the magnetic perturbation observed near closest approach on the E26 pass is unlikely to be produced by a plume.

The consistency of the particles and field perturbations during Galileo’s E12 flyby with the expected signature of a plume of the scale and density inferred from analysis of Hubble images underscores the value of acquiring in-situ data at low altitude on upcoming missions carrying appropriate instrumentation. The European Space Agency’s JUICE mission22 to Ganymede plans two passes with its closest approach 400 km above Europa’s surface. The National Aeronautics and Space Administration (NASA)’s Europa Clipper mission23 will make ~40 passes at altitudes <400 km. In-situ and remote-sensing measurements from these missions—especially those in regions close to the thermal anomaly that may be a preferred location of plumes—will be highly desirable to obtain detailed characterization of plume activity at Europa.

## Methods

### Galileo trajectories for all Europa encounters

During its entire mission, the Galileo spacecraft encountered Europa on 11 different orbits. Supplementary Fig. 1 shows the Galileo trajectories for all Europa encounters in EphiO coordinates, where x is parallel to Europa’s orbital velocity, y is directed towards Jupiter and z completes the right-handed system. MAG data were acquired on eight passes15, whereas PWS data were acquired on nine passes14. Recorded MAG data were lost on E6, E16 and E18, and recorded PWS data were lost on E16 and E18. Among the 11 flybys, only the E12 and E26 flybys came within 400 km of the surface with closest approach altitudes of 196 km for E12 and 348 km for E26.

### Multi-fluid MHD model and simulation results

#### Model basics

We modelled the plasma interaction with Europa and its exosphere using a well-established MHD code, BATSRUS (Block-Adaptive-Tree-Solarwind-Roe-Upwind-Scheme)16,24,25. The BATSRUS code solves the governing MHD equations in finite-volume form. It has been widely applied to simulations of planetary magnetospheres and plasma interactions with planetary moons, including Europa. Previous modelling studies on Europa using BATSRUS include single-fluid26, multi-species27 and multi-fluid17 MHD simulations.

In this work, we employed the multi-fluid version of BATSRUS16,17 to separately track the plasma species originating from different sources at Europa. In particular, the multi-fluid MHD model includes separate continuity, momentum and energy equations for two ion fluids: an O2+ fluid representing the main ion species produced from Europa’s O2-dominated exosphere18,28,29 and an O+ fluid consisting of Jupiter’s ambient magnetospheric plasma and O+ ions produced as a daughter species from Europa’s O2 exosphere. In our model, each of the ion fluids is simulated with its own density, bulk velocity and temperature. Electrons in the system are treated as a charge-neutralizing fluid with a separate pressure equation that describes the evolution and spatial distribution of the electron temperature. Source terms for the electron pressure in our model include the thermal energy contributed by the newly implanted electrons through impact ionization and the excess energy of photoelectrons, as well as the energies transferred through elastic collisions with the ions and neutrals. Loss terms for the electron pressure account for the reduction due to ion–electron recombination and the ionization energy provided by the ionizing electrons.

The key to modelling the plasma interaction with Europa and its exosphere is to include the important mass-loading processes present in the vicinity of the moon, such as ionization due to photo- and electron impact ionization, charge-exchange and dissociative recombination. Our multi-fluid MHD model incorporates these mass-loading processes by adding appropriate source and loss terms in the set of MHD equations17.

The mass-loading model in our simulations assumes a prescribed density distribution for Europa’s exosphere, which is composed predominantly of molecular oxygen18,29. The assumed neutral density (Nn) distribution is described in a form that contains two exponential functions with different surface densities and scale heights representing the sublimated and sputtered components of the exosphere30,31,32,33.

$N n = 1 + 2 cos ϕ ⋅ n 10 e r - R E h s 1 + n 20 e r - R E h s 2 , if ϕ ≤ 90 ∘ n 10 e - r - R E h s 1 + n 20 e - r - R E h s 2 , if ϕ > 90 ∘$
(1)

Here, r is the radial distance in units of RE (RE = 1,570 km is Europa’s radius), n10 = 4 × 107 cm−3 and hs1 = 100 km are the surface density and scale height for the relatively confined thermal component of the exosphere34, whereas n20 = 1 × 106 cm−3 and hs1 = 500 km are used to represent the relatively extended sputtered component of the exosphere32,33. For the upstream (or trailing) hemisphere, the sum of the two exponential functions is multiplied by the coefficient dependent on ϕ, which is the azimuthal angle measured from the –x axis in EphiO coordinates. The coefficient involving ϕ results in higher neutral densities on the trailing hemisphere than on the leading hemisphere, consistent with the idea that sputtering by Jovian magnetospheric particles is a dominant process in producing Europa’s exosphere35. The resulting neutral distribution for the exosphere is shown in Supplementary Fig. 2a. The total column density arising from the sum of the two components ranges from 4.5 × 1018 m−2 on the leading hemisphere to 1.3 × 1019 m−2 on the trailing hemisphere. The range of exospheric column densities used in our simulations is consistent with those obtained from observations (2.4 × 1018–1.4 × 1019 m−2 (refs 18,28)) and those used in previous simulations of Europa’s plasma–exosphere interaction13,1736,37. The exosphere model described in equation (1) is used as the background neutral exosphere model in simulations with and without a plume included. For simulations with a plume, a localized plume-related neutral distribution, which is detailed below, is added to this background neutral model.

Based on the prescribed neutral exosphere, the MHD model calculates the corresponding ion production rate by multiplying the neutral density by a given ionization rate. Here, we adopt a constant ionization rate of 1 × 10−6 s−1 for O2+ production and 1 × 10−7 s−1 for O+ production according to the mean ionization rates derived previously17. Although the ionization rates are constants in our model, the ion production rates are asymmetric, with higher rates on the upstream than on the downstream side because the neutral exosphere density is asymmetric between the leading and trailing hemispheres. The total ion production rate in our simulation without plume amounts to ~14.7 kg s−1 or 2.9 × 1026 ions s−1.

The charge-exchange interaction between O2 and O2+, which does not yield additional mass input into the system but contributes to the momentum loading of the flow, is also included in our model through ion-neutral friction. The rate for the ion-neutral friction is given by38:

$k in =2.59×1 0 - 17 × N n T ̄ 1 - 0.073 × log 10 T ̄ 2$
(2)

where N n is the exospheric neutral density and $T ̄ = 1 2 T O 2 + T O 2 +$ is the mean temperature between O2 and O2+. The resulting total charge-exchange rate integrated over the entire interaction region is ~27 kg s−1 or 5 × 1026 ions s−1.

Plasma generated from Europa’s exosphere may be lost by dissociative recombination between ions and electrons. The recombination rates for O+ ($α O +$) and O2+ ($α O 2 +$) are prescribed according to the following functions given by38:

$α O + =3.7×1 0 - 18 250 T e 0.7 , α O 2 + =2.4×1 0 - 13 300 T e 0.7$
(3)

where Te is the electron temperature calculated from the electron energy equation. The total recombination rate in our model is ~8 × 10−4 kg s−1—several orders of magnitude smaller than the ion production rate. This result is consistent with the previous finding that recombination appears to be an insignificant loss process for Europa’s ionosphere17,37.

#### Plume model

To model the effect of a plume on the plasma interaction, we incorporated in our MHD model an analytical form for the plume neutral density distribution similar to that derived based on the Hubble observations5. As described in equation (4), the density distribution (NP) has a conical structure governed by several parameters, including the central surface density (NP0), scale height (HP) and opening angle (θP).

$N P = N P 0 ⋅ e - r - R E H P ⋅ e - θ θ P 2$
(4)

Here, r is the radial distance in units of RE and θ is the polar angle measured relative to the central axis of the plume.

The location of the plume on Europa’s surface is also a free parameter in our model, but the Galileo MAG and PWS observations of the anomalous variations in particles and field conditions provide a key constraint on the general region where the plume should be located. Specifically, the signatures seen around 12:02 ut in the MAG and PWS data that we identify to be associated with a plume were encountered at a longitude of 235° W and a latitude of 9° S. Images from Saturn’s moon Enceladus20 show that plumes possess a complicated structure in that their axes diverge from the radial direction and there can be localized structures within a single plume. Assuming that Europa plumes may exhibit similar properties, we allow the central axis of our simulated plume to bend relative to the radial direction. Guided by this information, we tested a number of different combinations of the parameters controlling the location and spatial distribution of the plume. The results shown in this paper were extracted from the run in which the base of the plume is located at a longitude of 245° W and a latitude of 5° S, and the plume’s central axis is tilted relative to the radial direction by 15° in the azimuthal direction towards the east and by 25° in the latitudinal direction towards the south. Other parameters used are NP0 = 2 × 109 cm−3, HP = 150 km and θP = 15°. These parameters yield a central column density of Ncol = 3 × 1020 m−2, which falls right within the range of plume column densities inferred from spectroscopic observations: 1.5 × 1020 m−2 (ref. 5) to 2.3 × 1021 m−2 (ref. 6). Supplementary Fig. 2b shows the distribution of the total neutral density, which is the sum of the background exosphere density and plume density. The net ion production rate associated with the modelled plume is estimated to be ~0.3 kg s−1.

#### Numerical grid

The MHD simulations were conducted on a non-uniform, high-resolution spherical mesh that ensures high grid resolution of the near-Europa interaction region and an accurate prescription of boundary conditions at the moon. The outer boundary of the simulation corresponds to a sphere of radius of 56 RE, and the inner boundary corresponds to Europa’s surface of radius of 1 RE. Using the adaptive mesh refinement capability of BATSRUS16, we generated a spherical mesh with the multi-level refined grid structure shown in Supplementary Fig. 2c,d. The grid resolution at r = 2 RE is ~0.02 RE (or 31 km). Given the small-scale size of the plume of interest, we refined the grid resolution even further in the vicinity of the plume, which was essential for resolving the fine structure in the magnetic and density perturbations associated with the plume. The smallest grid size around the plume is ~0.004 RE or 6 km.

#### Boundary conditions

At the upstream outer boundary, we specify the magnetic field according to the Galileo MAG observations obtained around closest approach. The vector components (in EphiO coordinates) used in the simulation were (B x , B y , B z ) = (78, 0, −395) nT. Based on the Galileo PWS and PLS measurements14,39, a uniform electron number density of 500 cm−3, a flow velocity of 100 km s–1 in the corotation direction and a temperature of 100 eV were specified for the upstream plasma flowing into the simulation domain. On the downstream outer boundary, floating boundary conditions were applied to allow the plasma to freely leave the simulation domain.

At the inner boundary (Europa’s surface), we included the induced dipole field arising from the inductive response of a subsurface ocean to the time-varying external magnetic field1,4. Assuming a 100% induction efficiency40, an equatorial dipole with its axis parallel to the –x axis (in EphiO coordinates) and an equatorial surface strength of 39 nT were prescribed as initial conditions and fixed in our simulations to represent the induced field. In regions at the surface where the plasma flow in the computational cell next to the surface had an inward radial component, floating boundary conditions were applied to the plasma parameters (that is, zero gradients in density, pressure and velocity), such that Europa’s surface absorbed incoming flows. In regions where the plasma flow had an outward radial flow near the surface, fixed densities of 1 cm−3 for O+ and 10 cm−3 for O2+ and a temperature of 600 K for both ion fluids were specified to represent the situation where Europa’s surface contributes only a minimal amount of cold plasma to the ionosphere.

#### Comparison of model results between simulations with and without a plume

With the input parameters described above, we performed two sets of MHD simulations: one with only the background exosphere and another with the plume (described in the text above) added to the existing exosphere. In all simulations, the upstream parameters and inner boundary conditions were kept the same and the MHD model was iterated until the system had reached a quasi-steady state, from which the simulation results presented here were extracted.

Supplementary Figs. 35 present a series of comparisons of the model results between the two cases to reveal the perturbations to plasma and magnetic field conditions caused by the plume. In particular, Supplementary Fig. 3a,b compares the density distribution for the dominant ionospheric species, O2+, in the xy plane (at z = −0.2 RE) that contained the Galileo trajectory during the E12 flyby. In the simulation with just the background exosphere, during the interval of interest (12:00–12:03 ut) the Galileo spacecraft would have encountered a plasma population with a roughly constant density of ~700 cm−3. In contrast, in the simulation with the plume included, the ionization of the plume particles produced localized density enhancements with a peak density of ~2,000 cm−3 around 12:02 ut at the spacecraft altitude, consistent with the enhanced electron density inferred from PWS at this time. Without a plume, as the ambient Jovian plasma flow approaches Europa from upstream, it is slowed by ion pickup and by Europa’s ionosphere21; some of the flow then diverts around the moon and is re-accelerated on the flanks (Supplementary Fig. 3c). The flow speeds during the interval of interest range from 40 to 60 km s−1. In the presence of the plume (Supplementary Fig. 3d), the flow that is originally diverted around Europa is further slowed down locally due to the pickup processes occurring in the vicinity of the plume. The flow speeds in the same region are reduced to 25–40 km s−1.

Supplementary Fig. 4 compares the magnetic perturbations in the xy plane that contains the Galileo trajectory for the two cases. The left column applies to the simulation without a plume and illustrates the large-scale structure of the magnetic perturbations arising from the interaction between the ambient Jovian plasma and Europa’s exosphere. The slowing down of the ambient flow near Europa causes the magnetic field bendback above and below the equatorial plane21 with positive B x perturbations below Europa’s equator in the plane of Galileo’s E12 pass (Supplementary Fig. 4a). The slowing down of the flow also produces compressional waves that result in enhancement of the magnitude of the dominant component B z (Supplementary Fig. 4e) and of the field magnitude upstream of Europa. The ambient field for the E12 pass is approximately aligned with the z axis near closest approach, and the plasma interaction produces minimal perturbations in the B y component upstream of Europa. Shown in the right column of Supplementary Fig. 4 are the results from the case with the plume included. The magnetic perturbations remain similar to those found in the case without a plume, except in the immediate vicinity of the plume where the perturbed Jovian plasma interacts with plume particles to generate strong local magnetic and plasma perturbations. The base of the plume introduced in our model lies above the Galileo trajectory, and consequently the Alfvénic perturbations caused by the locally modified flow cause additional bendback of the field lines, seen as locally enhanced positive B x perturbations in Supplementary Fig. 4b. The localized positive and negative B y perturbations (Supplementary Fig. 4d) near the plume arise from the diversion of flow around the plume. Upstream of the plume, the slowing down of the flow leads to pile-up of the magnetic field producing negative perturbations of B z (Supplementary Fig. 4f), analogous to those that develop where the ambient Jovian flow interacts with Europa’s global exosphere upstream of the moon. The effect appears on a much smaller spatial scale because of the small spatial size of the plume. Downstream of the plume, the diverted flow around the plume centre is re-accelerated, producing a decrease in the field magnitude. The variations of the simulated magnetic field components and strength associated with the plume discussed above are entirely consistent with the Galileo MAG observations.

Supplementary Fig. 5 compares the 3D configuration of the interaction region for cases with and without the plume. An iso-surface with a flow speed of 25 km s−1 is chosen to illustrate, as an approximation, the morphology of the interaction region. Colour contours of B x are plotted on the iso-surface to show the bendback of the field, which leads to negative B x perturbations above and positive B x perturbations below the central plane where the peak mass-loading occurs. Supplementary Fig. 5a,b reveals the 3D structure of the plume-generated perturbations. Supplementary Fig. 5b shows that a ridge-like structure forms surrounding the plume, with negative B x perturbations above and positive B x perturbations below the central axis of the plume. These perturbations are characteristic of those expected from an Alfvén wing-like structure21. The figure also shows the trajectory geometry during the E12 flyby relative to the plume-generated mini Alfvén wing.

### Data availability

The data that support the plots within this paper and other findings of this study are available from the corresponding author upon reasonable request.

Publisher’s note: Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

## References

1. 1.

Khurana, K. K. et al. Induced magnetic fields as evidence for subsurface oceans in Europa and Callisto. Nature 395, 777–780 (1998).

2. 2.

Carr, M. H. et al. Evidence for a subsurface ocean on Europa. Nature 391, 363–365 (1998).

3. 3.

Pappalardo, R. T. et al. Does Europa have a subsurface ocean? Evaluation of the geological evidence. J. Geophys. Res. 104, 24105–24055 (1999).

4. 4.

Kivelson, M. G. et al. Galileo magnetometer measurements strengthen the case for a subsurface ocean at Europa. Science 289, 1340–1343 (2000).

5. 5.

Roth, L. et al. Transient water vapor at Europa’s south pole. Science 343, 171–174 (2014).

6. 6.

Sparks, W. B. et al. Probing for evidence of plumes on Europa with HST/STIS. Astrophys. J. 829, 121 (2016).

7. 7.

Sparks, W. B. et al. Active cryovolcanism on Europa? Astrophys. J. Lett. 839, L18 (2017).

8. 8.

Kivelson, M. G., Khurana, K. K., Means, J. D., Russell, C. T. & Snare, R. C. The Galileo magnetic field investigation. Space Sci. Rev. 60, 357–383 (1992).

9. 9.

Gurnett, D. A. et al. The Galileo plasma wave investigation. Space Sci. Rev. 60, 341–355 (1992).

10. 10.

Roth, L. et al. Orbital apocenter is not a sufficient condition for HST/STIS detection of Europa’s water vapor aurora. Proc. Natl Acad. Sci. USA 111, E5123–E5132 (2014).

11. 11.

Spencer, J. R., Tamppari, L. K., Martin, T. Z. & Travis, L. D. Temperatures on Europa from Galileo photopolarimeter-radiometer: nighttime thermal anomalies. Science 284, 1514–1516 (1999).

12. 12.

McGrath, M. & Sparks, W. B. Galileo ionosphere profile coincident with repeat plume detection location at Europa. Res. Notes AAS 1, 14 (2017).

13. 13.

Blöcker, A., Saur, J. & Roth, L. Europa’s plasma interaction with an inhomogeneous atmosphere: development of Alfvén winglets within the Alfvén wings. J. Geophys. Res. 121, 9794–9828 (2016).

14. 14.

Kurth, W. S. et al. The plasma wave environment of Europa. Planet. Space Sci. 49, 345–363 (2001).

15. 15.

Kivelson, M. G., Khurana, K. K. & Volwerk, M. in Europa (eds Pappalardo, R. T., McKinnon, W. B. & Khurana, K. K.) 545–570 (Univ. Arizona Press, Tucson, AZ, 2009).

16. 16.

Toth, G. et al. Adaptive numerical algorithms in space weather modeling. J. Comp. Phys. 231, 870–903 (2012).

17. 17.

Rubin, M. et al. Self-consistent multi-fluid MHD simulations of Europa’s exospheric interaction with Jupiter’s magnetosphere. J. Geophys. Res. 120, 3503–3524 (2015).

18. 18.

Hall, D. T., Strobel, D. F. & Feldman, P. D., McGrath, M. A. & Weaver, H. A. Detection of an oxygen atmosphere on Jupiter’s moon Europa. Nature 373, 677–679 (1995).

19. 19.

Bagenal, F., Dougherty, L. P., Bodish, K. M., Richardson, J. D. & Belcher, J. M. Survey of Voyager plasma science ions at Jupiter: 1. Analysis method. J. Geophys. Res. 122, 8241–8256 (2017).

20. 20.

Porco, C. C. et al. Cassini observes the active south pole of Enceladus. Science 311, 1393–1401 (2006).

21. 21.

Neubauer, F. M. The sub-Alfvénic interaction of the Galilean satellites with the Jovian magnetosphere. J. Geophys. Res. 103, 19843–19866 (1998).

22. 22.

Grasset, O. et al. JUpiter ICy moons Explorer (JUICE): an ESA mission to orbit Ganymede and to characterise the Jupiter system. Planet. Space Sci. 78, 1–21 (2013).

23. 23.

Pappalardo, R. et al. Science objectives and capabilities of the NASA Europa mission. In Proc. 47th Lunar and Planetary Science Conf. 3058 (Lunar and Planetary Institute, 2016).

24. 24.

Powell, K. G., Roe, P. L., Linde, T. J., Gombosi, T. I. & De Zeeuw, D. L. A solution-adaptive upwind scheme for ideal magnetohydrodynamics. J. Comp. Phys. 154, 284–309 (1999).

25. 25.

Gombosi, T. I. et al. Semirelativistic magnetohydrodynamics and physics-based convergence acceleration. J. Comput. Phys. 177, 176–205 (2002).

26. 26.

Kabin, K. et al. On Europa’s magnetospheric interaction: a MHD simulation of the E4 flyby. J. Geophys. Res. 104, 19983–19992 (1999).

27. 27.

Liu, Y. et al. Two-species, 3D MHD simulation of Europa’s interaction with Jupiter’s magnetosphere. Geophys. Res. Lett. 27, 1791–1794 (2000).

28. 28.

Hall, D. T., Feldman, P. D., McGrath, M. A. & Strobel, D. F. The far-ultraviolet oxygen airglow of Europa and Ganymede. Astrophys. J. 499, 475–481 (1998).

29. 29.

McGrath, M. A., Hansen, C. J. & Hendrix, A. R. in Europa (eds Pappalardo, R. T., McKinnon, W. B. & Khurana, K. K.) 485–505 (Univ. Arizona Press, Tucson, AZ, 2009).

30. 30.

Shematovich, V. I., Johnson, R. E., Cooper, J. F. & Wong, M. C. Surface-bounded atmosphere of Europa. Icarus 173, 480–498 (2005).

31. 31.

Smyth, W. H. & Marconi, M. L. Europa’s atmosphere, gas tori, and magnetospheric implications. Icarus 181, 510–526 (2006).

32. 32.

Cassidy, T. A., Johnson, R. E., McGrath, M. A., Wong, M. C. & Cooper, J. F. The spatial morphology of Europa’s near-surface O2 atmosphere. Icarus 191, 755–764 (2007).

33. 33.

Plainaki, C. et al. The role of sputtering and radiolysis in the generation of Europa exosphere. Icarus 218, 956–966 (2012).

34. 34.

Saur, J., Strobel, D. F. & Neubauer, F. M. Interaction of the Jovian magnetosphere with Europa: constraints on the neutral atmosphere. J. Geophys. Res. 103, 19947–19962 (1998).

35. 35.

Johnson, R. E. et al. in Europa (eds Pappalardo, R. T., McKinnon, W. B. & Khurana, K. K.) 507–528 (Univ. Arizona Press, Tucson, AZ, 2009).

36. 36.

Schilling, N., Neubauer, F. M. & Saur, J. Influence of the internally induced magnetic field on the plasma interaction of Europa. J. Geophys. Res. 113, A03203 (2008).

37. 37.

Dols, V., Bagenal, F., Cassidy, T., Crary, F. J. & Delamere, P. A. Europa’s atmospheric neutral escape: importance of symmetrical O2 charge exchange. Icarus 264, 387–397 (2016).

38. 38.

Schunk, R. & Nagy, A. Ionospheres: Physics, Plasma Physics, and Chemistry (Cambridge Univ. Press, Cambridge, UK, 2009).

39. 39.

Bagenal, F. et al. Plasma conditions at Europa’s orbit. Icarus 261, 1–13 (2015).

40. 40.

Zimmer, C., Khurana, K. K. & Kivelson, M. G. Subsurface oceans on Europa and Callisto: constraints from Galileo magnetometer observations. Icarus 147, 329–347 (2000).

## Acknowledgements

We thank M. McGrath for an illuminating presentation at a Europa Clipper Project Science Group meeting on observations of Europa’s plumes, which led us to re-examine the Galileo MAG data on which this paper is largely based. The work at the University of Michigan was supported by NASA through grants #NNX12AM74G and #NNX15AH28G, contract #1532308 through the Jet Propulsion Laboratory and contract #143448 through the Applied Physics Laboratory at Johns Hopkins University. The research at the University of Iowa is supported by NASA through contract UTA16-001080 through the University of Texas at Austin. Additional funding for work at UCLA was provided by NASA grants #NNX13AL05G:000002 and #NNX14AO24G.

## Author information

### Affiliations

1. #### Department of Climate and Space Sciences and Engineering, University of Michigan, Ann Arbor, MI, USA

• Xianzhe Jia
•  & Margaret G. Kivelson
2. #### Department of Earth, Planetary and Space Sciences, UCLA, Los Angeles, CA, USA

• Margaret G. Kivelson
•  & Krishan K. Khurana
3. #### Department of Physics and Astronomy, University of Iowa, Iowa City, IA, USA

• William S. Kurth

### Contributions

X.J., M.G.K. and K.K.K. contributed to the analysis of the Galileo magnetic field data. W.S.K. analysed the Galileo plasma wave data. X.J. performed the simulations and led the interpretation of the model results. All authors discussed the results and contributed to writing the manuscript.

### Competing interests

The authors declare no competing interests.

### Corresponding author

Correspondence to Xianzhe Jia.

## Supplementary information

1. ### Supplementary Information

Supplementary Figures 1–5.