Magnetosphere‐Ionosphere‐Thermosphere Coupling Study at Jupiter Based on Juno's First 30 Orbits and Modeling Tools

Abstract The dynamics of the Jovian magnetosphere is controlled by the interplay of the planet's fast rotation, its solar‐wind interaction and its main plasma source at the Io torus, mediated by coupling processes involving its magnetosphere, ionosphere, and thermosphere. At the ionospheric level, these processes can be characterized by a set of parameters including conductances, field‐aligned currents, horizontal currents, electric fields, transport of charged particles along field lines including the fluxes of electrons precipitating into the upper atmosphere which trigger auroral emissions, and the particle and Joule heating power dissipation rates into the upper atmosphere. Determination of these key parameters makes it possible to estimate the net transfer of momentum and energy between Jovian upper atmosphere and equatorial magnetosphere. A method based on a combined use of Juno multi‐instrument data and three modeling tools was developed by Wang et al. (2021, https://doi.org/10.1029/2021ja029469) and applied to an analysis of the first nine orbits to retrieve these parameters along Juno's magnetic footprint. We extend this method to the first 30 Juno science orbits and to both hemispheres. Our results reveal a large variability of these parameters from orbit to orbit and between the two hemispheres. They also show dominant trends. Southern current systems are consistent with the generation of a region of sub‐corotating ionospheric plasma flows, while both super‐corotating and sub‐corotating plasma flows are found in the north. These results are discussed in light of the previous space and ground‐based observations and currently available models of plasma convection and current systems, and their implications are assessed.


25
In this Supplementary Information document, we provide more details about the 26 numerical methods used in the study described in the main paper. We discuss first in 27 section 1 a correction in the electrodynamics model which allows us to take into consid-28 eration the uncertainties in the data to compute the MIT coupling parameters. Secondly, 29 in section 2, we describe the method developed for this study to compute the residual 30 magnetic field from the Juno MAG instrument and the JRM09 magnetic model using 31 Fast Fourier Transform filtering methods. Thirdly, in section 3 we describe briefly the 32 method used to determine the amount of time by which we shift the UVS data. Finally, 33 in section 4, we describe the method used to compute the uncertainties over the calcu- In our calculation scheme, we perform a geometric mapping of the parameters that 38 are measured by Juno in the 3-D space corresponding to the "high latitude, auroral and 39 polar magnetosphere" region, referred to as Region II in Wang et al. (2021), to the con-40 ducting layer of the ionosphere/thermosphere, referred to as Region III, along magnetic 41 field lines. This mapping is done to calculate the key parameters of MIT coupling in this 42 layer, which is modelled as a 2-D infinitely thin shell surrounding the planet. Indeed, its 43 vertical thickness, on the order of 3.10 −3 times the Jovian radius R J , is very small com-  Figure 1: Northern hemisphere illustration of the geometric elements and reference frames used in our calculations of MIT coupling parameters from Juno data. The magnetic field lines mapping to the mean latitude of the main oval (shown as the lower red curve on the ionospheric sphere) are shown in purple. They make it possible to map this main oval location to the altitude of Juno (upper red curve). Juno's trajectory as well as the trajectory of its magnetic footprint are in sky-blue and white, respectively. Two local reference frames are used for our differential calculations. At the altitude of Juno and everywhere along magnetic field lines above the ionosphere, unit vectorẑ points along the magnetic field,ŷ is positive eastward along the tangent to the cone of field lines connected to the main oval, andx complements the frame, positive towards the equator in both hemispheres. At the ionospheric altitude, unit vectorsx ′ andŷ ′ are horizontal and tangent to the conducting layer of the ionosphere which is assumed to be an infinitely thin surface surrounding the plane.x ′ is oriented towards the equator orthogonal to the auroral oval (red curve),ŷ ′ is oriented eastward along the oval, andẑ ′ is vertical, positive upward in the northern hemisphere and downward in the southern hemisphere.x ′ andŷ ′ taken together provide a local 2-D reference frame for horizontal vectors and differential expressions defined in the plane tangent to the ionospheric conducting layer. For the description of the 3-D space of Region II, we introduce a set of orthogo-48 nal curvilinear coordinates (x, y, z) such that x and y are constant along magnetic field 49 lines, to facilitate the mapping of quantities between Juno's location and its magnetic 50 footprint in the ionosphere/thermosphere. Each point in (x, y, z) space is associated with 51 a local orthogonal reference frame with vector units (x,ŷ,ẑ).ẑ is along the magnetic field 52 direction,ŷ is orthogonal toẑ in the plane containing the local magnetic field vector and 53 the local tangent to the oval, andx =ŷ ×ẑ . In this reference system, distances are 54 given by the metric tensor ds 2 = h 2 x dx 2 +h 2 y dy 2 +h 2 z dz 2 , in which h x , h y and h z are 55 the metric coefficients in the three directions. In the dipolar approximation, where we 56 assume that the magnetic field is radial, we can choose to work in the spherical frame 57 given by (x, y, z) = (θ, ϕ, r). We then have explicitly h x = h θ = r , h y = h ϕ = r sin θ 58 and h z = h r = 1.

59
-2- tensor. In this reference system, distances are given by the metric tensor (ds ′ ) where µ 0 is the vacuum permittivity. The magnetic perturbation δB y is obtained from one can write a direct relationship between the magnetic perturbation generated by field-92 aligned currents at the altitude of Juno and the intensity of these same currents at the 1.4 Calculation of J x from J ∥ at the ionospheric altitude In the specific case where, at least locally, the oval runs along a circle of constant 103 magnetic latitude, one can introduce the global magnetic coordinates such that (x ′ , y ′ ) = 104 (θ, ϕ) and extend this coordinate system along each field line to generate the (x,ŷ,ẑ) co-105 ordinate system of region II. In this particular case, h y ′ = r sin θ, h x ′ = r, where r is 106 the radial distance to the planet's center and equations 3 and 4 write respectively: The global reference frame considered in this study is Jupiter's magnetic dipole co- Residual azimuthal magnetic field δB ϕ . Analysis of this figure shows that the amplitude of the residual magnetic field remains close to zero when Juno is far from Jupiter (r Juno > 1.5R j ). But when Juno approaches Jupiter, the residual magnetic is modulated by oscillations of very large amplitude (tens of mT for δB r , δB θ , and few mT for δB ϕ ) and very large period (> 20 min) with respect to the expected amplitude and period of the signal of interest in this study (hundreds of nT for the amplitudes, and few minutes for the period). This observation suggests that filtering away this large amplitude/low frequency component of the signal should give back the signal of interest. period are much larger than the ones expected for the signal of interest in this study, which 157 reach maximal amplitudes of hundreds of nT, and whose period is of a few minutes at 158 most. We thus need to extract from this data the data which interests us, and whose am-

181
We thus apply a bandpass FFT filter on the residual magnetic field signal, of upper cut-182 off frequency f u and lower cutoff frequency f l such that 10min ≫ 1/f u ≫ 30s and 1h ≫ 183 1/f l ≫ 10min.

184
However, such a filtering method intrinsically brings error in the resulting signal, 185 which amounts to introducing "artificial" currents in the system we are considering. These 186 artificial currents can be understood visually from figure 3. We thus needed to quantify 187 the amount of artificial currents we are introducing for each crossing. To do so, the method 188 proposed is the following: we assume first that the azimuthal component of the resid-189 ual residual magnetic field δB ϕ is entirely generated by field aligned currents J ∥ , accord-190 ing to equation 3. We thus consider a "dummy" profile of field aligned currents J ∥,0 , and 191 from this profile, we compute the resulting δB ϕ0 . We apply the filtering method of this  197 Linearity is thus very helpful for this analysis. Response calculated for an impulsive input, showing a characteristic timescale T J corresponding to Juno's rotation around its axis. Panel b: Response to a square input, where we denoted by A 1 the amplitude and by T 1 the timescale of the input, by T 0 the high characteristic timescale of the applied filter, A ′ 1 the reduced amplitude of the output and A 2 the amplitude of the side currents appearing as a result of the filtering. The aim is to characterize A ′ 1 /A 1 and A 2 /A 1 as a function of T 1 /T 0 to be able to estimate the error generated by such filtering method for each crossing. an adequate value of f l with respect to the data was possible, but in some extreme cases, Figure 4: Diagram describing the amplitude reduction factor A ′ 1 /A 1 (panel a) and the relative amplitude of the secondary peak A 2 /A 1 (panel b) as a function of the ratio T 1 /T 0 . This diagram allows for a systematic estimation of the error generated by the filtering method on the resulting MIT coupling parameters.
we had to choose larger values of the ratio T 1 /T 0 = T 1 f l , so that considering these ex-221 treme cases only, we can estimate the error on the calculated field aligned currents to 222 be at most 20%. 223 3 UVS data time shift 224 We consider in this section of the Supplementary Information the post processing 225 steps applied on the time series obtained from the UVS instrument data.

226
During this study, we worked with two kinds of data coming from the UVS instru-227 ment: imagery data, and temporal data. The imagery data were used to study visually  In this paper, the temporal data were compared to the data coming from other Juno 236 instruments, mainly the particle data coming from the JADE and JEDI instruments, and 237 the magnetic data coming from the MAG instrument. We observed on almost all cross-  To quantify systematically this apparent delay, we assumed that the precipitating 251 particle heating rate computed directly for the JADE and JEDI instruments had to be 252 correlated to the brightness profile from the UVS data. Our method for each crossing 253 then consisted in shifting progressively the UVS data in time by steps of one second, and 254 computing at each step the correlation between the particle heating rate corresponding 255 to the downward electron energy flux in the loss cone measured by the particle detec-256 tors and the brightness profile. Then, the shift which maximizes the correlation is re-257 tained as the value of the delay for this crossing, and the results are shown in table 1 for 258 the crossings studied in details in the main paper for which UVS data were available.

259
These results show value of delays which are consistent with visual observations. In ta-260 ble 1, we also showed the maximal value of the correlation for its informative value. To discuss the uncertainties associated with the derivation of the MIT coupling pa-270 rameters, we are guided by the order at which the calculations are made.

271
The first parameters derived are the ionospheric conductances Σ H and Σ H and the 272 particle heating rate P e using JADE and JEDI data. The uncertainty here is directly 273 related to the accuracy of the Juno instruments data, which we consider as exact. This means that if Y is a quantity measured by JADE or JEDI, then the relative uncertainty 275 over this quantity is taken to be zero: ∆Y /Y = 0. The relative uncertainty over the 276 conductances is also directly related to the atmospheric and ionospheric models described 277 in figure 1 of the main paper. A previous analysis conducted by Wang et al. (2021)  sen and Hall conductances Σ P,H , and the particle heating rate P e . This sets their rel-283 ative uncertainty as ∆Σ P /Σ P = ∆Σ H /Σ H = ∆P e /P e = 0.2. Thus, this uncertainty 284 originates from the models.

285
In parallel, the field aligned currents J ∥ are computed directly from the residual 286 magnetic field δB ϕ using the method described in section 2 of the Supplementary Infor-287 mation document. The analysis conducted there allowed us to bound the relative un-288 certainty associated to J ∥ to 20% in the worst case, which means that we set ∆J ∥ /J ∥ = 289 0.2. This value is set to take into account the worst cases, which concern only a few cross-