betaFeedback

WorksPhysics and Astronomy

Statistical correlations between turbulent structures and dissipation proxies in near-Earth space: the MMS 2023-2024 unbiased magnetosheath campaign

Abstract

The terrestrial magnetosheath provides excellent opportunities for the experimental study of space plasma physics. Turbulence and kinetic-scale dissipation processes have been extensively studied separately. However, examining their cross-scale interplay may shed light on unsolved long-standing questions: which kinetic processes specifically dissipate the plasma turbulent fluctuations, and what drives them? In this study, we perform a statistical study of high-resolution data from the 2023-2024 unbiased campaign of the Magnetospheric Multiscale (MMS) mission to examine correlations between local small-scale accumulation of turbulent energy and peaks of dissipation proxies. The results highlight that clear regional correlations between a vast ensemble of dissipation proxies emerge on a scale of a few ion inertial lengths. Furthermore, intermittent turbulent structures dissipate energy in a hierarchical ordering

Introduction

One of the challenges in the physics of space plasmas such as the solar wind and planetary magnetospheres is to explain the energy conversion and dissipation mechanisms responsible for plasma heating and particle energization [45]. Observations show that space plasmas remain significantly hotter than predicted from adiabatic expansion or collisional heating models [6, 28]. Since heliospheric plasmas are weakly collisional, turbulent fluctuations are thought to transfer energy from magnetohydrodynamic (MHD) scales toward kinetic scales, where it is then converted into internal energy by kinetic processes. To investigate these processes, we focus on the terrestrial magnetosheath because it hosts strong and fully developed turbulence comparable to those of the solar wind.

Plasma fluctuations follow power-law spectra in the turbulent inertial range, then transition to kinetic turbulence with a spectral break near ion scales [6]. Below ion scales, the cascade continues toward electron scales, where dissipation becomes significant [48] (O. Alexandrovа et al. 2009). This broadband cascade provides a pathway where energy conversion and, potentially, dissipation occur through a variety of mechanisms [30, 39], such as wave–particle interactions like Landau damping [22] and cyclotron resonance [28], as well as nonlinear processes like stochastic heating [7]. From both fluid and kinetic simulations, it is known that turbulent energy conversion is highly intermittent and tends to concentrate near coherent structures such as current sheets and reconnection sites [23, 49]. While these processes are well established in theory and simulations, their relative importance in observations still remains difficult to quantify.

Many studies have begun to link these processes to localized structures in turbulence. In the solar wind, current sheets are correlated with enhanced proton temperature and temperature anisotropy [32, 17], and coherent structures have also been associated with enhanced suprathermal and energetic particle populations [58, 59]. Analysis tools for the identification of strong turbulent energy transfer also reveal that intermittent events are associated with localized and regional proton heating [52, 66]. In the terrestrial magnetosheath, measurements reveal enhanced particle energization, and non-Maxwellian velocity distribution functions (VDF) near coherent structures [9, 53, 4]. MHD simulations show that turbulence generates reconnecting current sheets that efficiently convert energy and accelerate particles [61, 14], while kinetic simulations further show that electromagnetic work and pressure–strain interactions cluster around coherent structures [64].

Despite this progress, the causal relationship between turbulence and kinetic-scale energy conversion and dissipation remains unclear. Some studies interpret turbulence as the large-scale driver that forms strong gradients, current sheets, and reconnection sites where energy conversion is localized [23, 30], while others emphasize more distributed wave–particle interactions [22, 21]. Clarifying this relationship requires a statistical approach that directly connect turbulent energy transfer with dissipation signatures across varying plasma conditions. Such approach has so far been limited. Many observational studies focus on intervals pre-selected for strong, visually identifiable events [32, 9]. Such targeted selections can enhance apparent correlations but are known to introduce selection biases [34]. Moreover, most analyses rely on a limited subset of diagnostics, even though different proxies capture distinct channels of energy conversion [8, 40]. While recent works have begun combining multiple complementary diagnostics [63, 53, 39], systematic comparisons across several dissipation proxies under consistent observational conditions remain limited. Most importantly, these analyses did not account for the spatial and temporal displacement between turbulent structures and dissipation sites, resulting in weak correlations in direct point-to-point comparisons [63, 30]. This behavior arises because dissipation effectively occurs in the vicinity of turbulent structures rather than exactly at their locations [17, 58, 52].

In this study, we address these limitations by analyzing high-resolution data from the 2023–2024 Magnetospheric Multiscale (MMS) magnetosheath unbiased campaigns. Using this dataset, we systematically compare multiple complementary dissipation diagnostics and explicitly quantify their spatial displacement relative to turbulent energy-transfer events.

The 2023-2024 Magnetospheric MultiScale unbiased campaign

Burst-mode measurements from the four MMS spacecraft are used to investigate turbulence and dissipation proxies in the terrestrial magnetosheath. The mean correlation length estimated from the magnetic-field autocorrelation was approximately 806 km, significantly larger than the mean spacecraft separation of 33 km. This implies that all four spacecraft were sampling the same correlated plasma structure and supports the validity of the multi-spacecraft gradient estimates.

The magnetic field B\mathbf{B} is provided by the Fluxgate Magnetometer [47] and sampled at 128 Hz, while the electric field E\mathbf{E} is provided by the Electric Double Probe at 8192 Hz [12]. Ion velocity distribution functions (fif_i) and corresponding ion moments, such as ion number density (nin_i), bulk ion velocity (vi\mathbf{v}_i), ion pressure tensor (Pi\mathbf{P}_i), and ion temperature tensor (Ti\mathbf{T}_i), are obtained from the Fast Plasma Investigation (FPI) at 150 ms [42], with bulk velocities corrected using FPI spintone data. For electrons, only the number density (nen_e) derived from the FPI is used.

In standard MMS operations, burst-mode intervals are selected through a Scientist-In-The-Loop process [15], which biases observations toward specific conditions. The MMS 2023–2024 unbiased campaigns was designed to provide systematically sampled intervals of magnetosheath turbulence without event preselection. The resulting dataset contains 615 intervals1 collected during 2 February–27 March 2023 [35] and 13 March–27 April 2024. The campaign used repeating cycles of 2.5–3 min of burst data, followed by 7–9 min of fast data. These campaigns have supported statistical surveys of whistler waves (I. Svenningsson et al. [55]), automated classification of plasma regimes (I. Svenningsson et al. [54]), analyses of non-Maxwellian ion distributions (L. Richard et al. [46]), and systematic studies of plasma and turbulence parameters (F. Pecora et al. [36]). The left panels of Figure 1 show a representative interval from the 2024 campaign, characterized by steady conditions and strong turbulent fluctuations across all fields. The right panels summarize the spatial coverage of the 615 intervals. The broad range of plasma β=2μ0nikBT/∣B∣2\beta= 2\mu_0 n_i k_{\mathrm{B}} T / \lvert B\rvert^2 indicates that the dataset samples diverse turbulent environments throughout the magnetosheath.

Time series of plasma and field quantities and spatial coverage of MMS intervals

Figure 1. Left: Example time series of plasma and field quantities parameters extracted from MMS1. Right: Position of each interval in the GSE plane, color-coded by the plasma β\beta, for the 2023 (top) and 2024 (bottom) campaigns. The large blue circle represents the Earth. Magnetopause and bow shock are computed from M. H. Farris & C. T. Russell [13], J.-H. Shue et al. [50].

To verify that the studied intervals contain fully developed turbulence, we compute standard diagnostics, summarized in Figure 2. The inertial range is bounded by the correlation frequency fcf_c and the ion cyclotron frequency fci=e∣B∣/2πmif_{ci} = e\lvert B\rvert/ 2\pi m_i. The correlation frequency is obtained from the 1/e1/e decay of the magnetic-field autocorrelation function (W. H. Matthaeus et al. [29]). The kinetic range extends toward the ion inertial scale di=c/nie2/(ε0mi)d_i = c / \sqrt{n_i e^2 / (\varepsilon_0 m_i)}, with mim_i the ion mass. The computed root-mean-square velocity fluctuations satisfy ⟨∣v−⟨v⟩∣2⟩t/∣⟨v⟩∣<0.2\sqrt{\langle\lvert v-\langle v\rangle\rvert^2\rangle_t}/\lvert\langle v\rangle\rvert< 0.2 and supports the validity of the Taylor hypothesis. This scale is converted to the frequency fif_i through the mean plasma bulk flow velocity of each interval under the Taylor hypothesis (G. I. Taylor [57]). The trace magnetic power spectral density (PSD) PB(f)P_B(f) is computed for each interval using the Welch method (P. Welch [62]), and then ensemble-averaged over all intervals, after normalization to the lowest mode power. Panel (a) shows that the PSD exhibits a persistent but very limited inertial range followed by a kinetic-scale spectrum, with exponents consistent with typical magnetosheath turbulence ([65, 36]).

Turbulence diagnostics from the unbiased campaigns

Figure 2. Turbulence diagnostics from the unbiased campaigns. Each power spectral density is normalized to the power at the lowest frequency. Grey curves represent individual intervals, solid black curves for the ensemble averages, and dashed lines for the power-fits. The inertial range is bounded by fcf_c and fcif_{ci}, while the kinetic range extends beyond fif_i, all indicated by colored vertical dashed lines. Converted time scales are respectively τcorr\tau_{\mathrm{corr}}, τρi\tau_{\rho_i}, and τi\tau_i (same style).

Intermittency is quantified using the scaling of the magnetic-field flatness KB(τ)=⟨∣δB(τ)∣4⟩t/⟨∣δB(τ)∣2⟩t2K_B(\tau) = \langle\lvert\delta B(\tau)\rvert^4\rangle_t / \langle\lvert\delta B(\tau)\rvert^2\rangle_t^2, where δB(τ)=B(t+τ)−B(t)\delta B(\tau) = B(t+\tau)-B(t), τ\tau is the time lag, ⟨⋅⟩t\langle\cdot\rangle_t denotes a time average within each interval, and the result is averaged over the three magnetic-field components for visualization purposes (the three components showing similar scaling). Panel (b) shows that the flatness increases toward small scales and follows a power law, indicating non-Gaussian fluctuations and developed intermittency. The observed exponent is comparable to hydrodynamic turbulence [2] and quasi-perpendicular magnetosheath [18], but generally smaller than typical solar-wind and more turbulent magnetosheath values [51, 44], although significant interval-to-interval variability is present. Finally, the MHD turbulent energy cascade is estimated using the Politano–Pouquet (PP98) third-order law [41, 27], formulated in terms of the magnetic fluctuations in Alfvén units b=B/μ0nimi\mathbf{b} = \mathbf{B}/\sqrt{\mu_0 n_i m_i}. The mixed third-order structure function is defined as Y(τ)=⟨δv∥(τ)(∣δv(τ)∣2+∣δb(τ)∣2)−2δb∥(τ)δv(τ)⋅δb(τ)⟩tY(\tau) = \left\langle\delta v_{\parallel}(\tau)\left(|\delta\mathbf{v}(\tau)|^2 + |\delta\mathbf{b}(\tau)|^2\right) - 2\delta b_{\parallel}(\tau)\delta\mathbf{v}(\tau)\cdot\delta\mathbf{b}(\tau)\right\rangle_t where longitudinal increments δv∥(τ)\delta v_{\parallel}(\tau) and δb∥(τ)\delta b_{\parallel}(\tau) are taken along each interval mean flow direction. Panel (c) shows that the ensemble-averaged third-order moment follows a clear linear scaling across the inertial-range scales [19, 44].

Overall, the unbiased campaigns systematically exhibit clear signatures of developed turbulence, with an identifiable inertial range, scale-dependent intermittency, and a scaling consistent with the PP law prediction. For a more in-depth study of turbulent scales [36], see also analyses of the PP law validity and energy transfer rates [31].

Proxies for turbulent structures and kinetic dissipation

A convenient way to quantify the scale-dependent transfer of energy in turbulent plasmas is through the Local Energy Transfer (LET) proxy, which originates from the MHD PP98 law [41]. By expressing this law in a point-wise form, we obtain an estimator of the instantaneous nonlinear energy flux at scale ℓ\ell. The LET can be decomposed into physically interpretable contributions

LET=∣δv∣2δv∥+∣δb∣2δv∥−2(δv⋅δb)δb∥ℓ.(1)\mathrm{LET} = \frac{|\delta\mathbf{v}|^2 \delta v_{\parallel} + |\delta\mathbf{b}|^2 \delta v_{\parallel} - 2(\delta\mathbf{v}\cdot\delta\mathbf{b})\delta b_{\parallel}}{\ell}. \tag*{(1)}

In this study, the LET is evaluated at the scale ℓ≃di\ell\simeq d_i, which marks the transition between MHD turbulence and ion-kinetic physics. Studies have shown that intense LET events are correlated with enhanced nonlinear activity, often observed near current sheets and associated with kinetic distortions and plasma heating [52, 63, 53]. On the other hand, dissipation proxies can be estimated using the measured moments over the whole ion VDF. The energy conversion between fields and particles is quantified through the electromagnetic work [67],

J⋅E′=J⋅(E+v×B)\mathbf{J}\cdot\mathbf{E}' = \mathbf{J}\cdot(\mathbf{E}+\mathbf{v}\times\mathbf{B})

where E′\mathbf{E}' denote the electric field measured in the co-moving frame of ions. This proxy isolates non-ideal field–particle interactions and identifies localized energization in kinetic turbulence [61, 63].

Complementary information on how this energy is redistributed is provided by the pressure–strain interaction [11],

(P⋅∇)⋅v=Pθ+Π:D,(\mathbf{P}\cdot\nabla)\cdot\mathbf{v} = P\theta+ \mathbf{\Pi}:\mathbf{D},

where θ=∇⋅v\theta= \nabla\cdot\mathbf{v}, PP is the isotropic pressure, D\mathbf{D} is the traceless strain tensor, and Π\mathbf{\Pi} is the deviatoric pressure tensor. The PθP\theta term represents compressive work, while the Π−D\mathbf{\Pi}-\mathbf{D} term captures shear-driven energy conversion. Additional filtering is applied to suppress residual spintone (see Appendix § 9). These proxies describe the transfer of bulk flow energy into internal energy and provide a measure of irreversible heating processes [64, 38].

Example time series of all computed proxies

Figure 3. Example time series of all computed proxies. Detected events of turbulent or dissipation activity are marked with red stars. An example of event displacement matching analysis is performed between the LET and j′⋅E′\mathbf{j}' \cdot\mathbf{E}', following the methodology in Appendix § 10. Identical green shapes mark a unique pair of matches, in which their amplitude and spatial distances are recorded for statistical analysis.

Additional insights into the kinetic processes can also be obtained from the structure of the particle distribution. Departures from gyrotropy are quantified using the pressure agyrotropy parameter Q\sqrt{Q} [56], defined in a field-aligned coordinate with P=(P∥P12P13P12P⊥P23P13P23P⊥)\mathbf{P} = \begin{pmatrix} P_{\parallel} & P_{12} & P_{13} \\ P_{12} & P_{\perp} & P_{23} \\ P_{13} & P_{23} & P_{\perp} \end{pmatrix} as

Q=P122+P132+P232P⊥2+2P∥P⊥.\sqrt{Q} = \sqrt{\frac{P_{12}^{2} + P_{13}^{2} + P_{23}^{2}}{P_{\perp}^{2} + 2P_{\parallel}P_{\perp}}}.

A value Q=0\sqrt{Q}=0 corresponds to a gyrotropic distribution, while positive values indicate increasing departures from gyromotion around the magnetic-field direction and captures partial demagnetization ([10] (Z. Vörös et al. 2017)).

Departures from local thermal equilibrium (LTE) are quantified using the normalized non-Maxwellianity ([17, 39]), which quantifies the deviation between the measured VDF fif_i and the corresponding Maxwellian gig_i,

ξ=vth,i3/2ni∫R3(fi(v)−gi(v))2 d3v,\xi= \frac{v_{\mathrm{th},i}^{3/2}}{n_i}\sqrt{\int_{\mathbb{R}^3} \left(f_i(\mathbf{v})-g_i(\mathbf{v})\right)^2 \,\mathrm{d}^3\mathbf{v}},

where vth,i=2kBT/miv_{\mathrm{th},i}=\sqrt{2k_{\mathrm{B}}T/m_i} is the ion thermal speed. Larger values indicate possible velocity-space distortions, such as beams, heating ([53]), and other non-equilibrium kinetic features ([16]).

A complementary measure of departure from equilibrium is provided by the entropy deficit ([24, 26]). The kinetic entropy s=−kB∫R3f(v)ln⁡f(v) d3vs=-k_{\mathrm{B}}\int_{\mathbb{R}^3}f(\mathbf{v})\ln f(\mathbf{v})\,\mathrm{d}^3\mathbf{v} is compared with that of the corresponding Maxwellian distribution defined as sM=(3/2)kBni[1+ln⁡(2πkBT/mini2/3)]s_M=(3/2)k_{\mathrm{B}}n_i\left[1+\ln\left(2\pi k_{\mathrm{B}}T/m_i n_i^{2/3}\right)\right], to define the normalized entropy deficit

MKP=sM−s(3/2)kBni.M_{\mathrm{KP}}=\frac{s_M-s}{(3/2)k_{\mathrm{B}}n_i}.

Positive values indicate increasing departures from equilibrium with the presence of non-thermal velocity-space structure ([3, 25]).

Finally, the HORNET proxy ([5]) characterizes the temporal evolution of velocity-space entropy through a Lagrangian derivative,

PHORNET=−niTddt(sv,relni),P_{\mathrm{HORNET}}=-n_iT\frac{\mathrm{d}}{\mathrm{d}t}\left(\frac{s_{\mathrm{v,rel}}}{n_i}\right),

with sv,rel=−kB∫R3fi(v)ln⁡(fi(v)/gi(v)) d3vs_{\mathrm{v,rel}}=-k_{\mathrm{B}}\int_{\mathbb{R}^3}f_i(\mathbf{v})\ln\left(f_i(\mathbf{v})/g_i(\mathbf{v})\right)\,\mathrm{d}^3\mathbf{v}. Positive values indicate an ongoing evolution away from LTE, while negative values correspond to relaxation toward equilibrium.

Statistical relationship between turbulent structures and kinetic signatures

To investigate whether the most intense kinetic dissipation occurs near intermittent turbulent structures, we initially quantify the spatial separation between each peak of the LET and the nearest peak of each dissipation proxy, using a peak-matching procedure (see Appendix § 10). Figure 4 shows the histograms of such LET-dissipation proxy distances, obtained from the whole unbiased campaign (red), compared to the distribution of distances between two successive peaks of the LET (blue). Note that all the dissipation proxies show a similar peak-to-peak distribution (not shown). In each panel, we also report the median of the distributions.

Distributions of spatial separations between peaks of proxies

Figure 4. Distributions of spatial separations between peaks of proxies. Red histograms show the separations between successive LET peaks with dpeaksLETd_{\mathrm{peaks}}^{\mathrm{LET}} the median of the distribution. Blue histograms show the separations between each LET peak and the nearest peak of the kinetic proxy, with dpeaksProxyd_{\mathrm{peaks}}^{\mathrm{Proxy}} the median of the distribution. All spatial distances are normalized by the ion inertial length did_i of the corresponding interval.

Figure 4 shows that successive LET peaks are separated by a characteristic distance dpeakLET≃11did_{\mathrm{peak}}^{\mathrm{LET}} \simeq11d_i and indicates that intermittent turbulent structures are sparsely distributed at ion scales, typical of intermittent turbulence. In contrast, all kinetic proxies exhibit significantly smaller separations relative to LET peaks, suggesting that sites of dissipation might be correlated with the intermittent structures. A systematic ordering is observed in the distributions. Energy-based proxies (J⋅E\mathbf{J} \cdot\mathbf{E}, (P⋅∇)⋅v(\mathbf{P} \cdot\nabla) \cdot\mathbf{v}, P−θP-\theta, Π−D\Pi-\mathrm{D}) generally show median separations of order ∼2di\sim2d_i, while velocity-space proxies (Q\sqrt{Q}, ξ\xi, MKPM_{KP}) display broader distributions with median separations of order ∼3−4di\sim3-4d_i. This suggests a more local effect of turbulent structures on the former, and slightly more regional on the latter. Interestingly, despite the HORNET PHORNETP_{\mathrm{HORNET}} being derived from the VDF, it displays characteristic separations comparable to those of the energy-based proxies. To ensure that the methodology of our analysis is valid, we perform several robustness checks (see Appendix § 11).

A complementary cross-correlation analysis confirms the relative ordering of the separation distances and provides an independent validation of the peak-matching approach. In addition, it quantifies how the spatial relationships depend on the detection threshold, thereby assessing the robustness of the identified coupling between LET and the kinetic proxies.

These results indicate that peaks of all kinetic dissipation proxies are preferentially located in the vicinity of LET peaks, with distinct characteristic distances depending on the nature of the proxy considered.

To further investigate the relationship between turbulent activity and kinetic signatures, we examine the correlations between the amplitudes of LET and the corresponding matched kinetic events. For each pair of matched peaks, the amplitudes of LET and the dissipation proxy are compared after mean detrending and taking absolute values. Figure 5 shows the joint distribution of LET and each dissipation proxy. Each panel reports the Pearson and Spearman correlation coefficients. Pearson measures linear relationships, while Spearman captures monotonic but potentially nonlinear relationships. Strong correlations are observed between LET and electromagnetic work J⋅E\mathbf{J} \cdot\mathbf{E}, the shear-driven term Π−D\Pi-\mathrm{D}, and non-Maxwellianity ξ\xi. This is followed by moderate correlations for agyrotropy Q\sqrt{Q} and entropy deficit MKPM_{KP}. Finally, weaker correlations are obtained for the compressive term P−θP-\theta and the PHORNETP_{\mathrm{HORNET}} proxy.

Joint probability density functions between LET and kinetic proxies

Figure 5. Joint probability density functions between LET and kinetic proxies. Pearson (ρP\rho_P) and Spearman (ρS\rho_S) correlation coefficients are computed. The color bar shows the probability density estimated via kernel density estimation.

We also tested the correlation strengths of different detection threshold choices and found that their strengths remain relatively stable with the exception of VDF-based proxies (see Appendix § 11). Additional tests performed across different plasma β\beta intervals show that the correlations tend to increase with β\beta, while remaining well defined within the most populated range of typical magnetosheath conditions (see Appendix D).

Overall, the statistical analysis reveals a systematic spatial clustering of kinetic proxy peaks around LET peaks with characteristic separations at ion scales, and a range of correlation strengths between LET and the different kinetic diagnostics.

Discussion and Conclusions

The results presented in this study provide new statistical constraints on how turbulent energy is coupled to kinetic processes in the terrestrial magnetosheath. Combining an unbiased dataset with multiple complementary diagnostics, we identify a coherent spatial organization of kinetic activity around intermittent turbulent structures at ion scales.

A central result is the emergence of a scale-dependent ordering between different classes of diagnostics. Energy-based proxies show the closest spatial correspondence with LET, with small separations showing that energy conversion occurs in the vicinity of intermittent structures. This supports a picture in which electromagnetic work and pressure–strain interactions act as immediate channels for converting bulk and field energy into particle internal energy at ion scales.

In contrast, velocity-space diagnostics exhibit broader spatial distributions which extend beyond the regions of strongest energy conversion. This indicates that the particle response is not confined to these localized sites but instead spreads over larger spatial and velocity-space scales. The additional threshold analysis presented in Appendix § 11 further indicates that the strongest kinetic signatures are more closely associated with intense intermittent events. Such behavior is consistent with kinetic processes that redistribute energy in phase space, including phase mixing, finite Larmor radius effects, and wave–particle interactions.

Taken together, these observations support a multi-stage scenario. Turbulent interactions first concentrate energy at intermittent structures, energy is then converted locally into particle internal energy, and the resulting particle distributions evolve over larger scales through kinetic processes. This separation between energy transfer, conversion, and particle response is reflected in the finite spatial offsets observed between the corresponding diagnostics.

The coexistence of strong but imperfect correlations further suggests that no single diagnostic fully captures the complexity of kinetic dissipation. Some proxies may include contributions from reversible energy exchanges or depend sensitively on local plasma conditions ([20]). Differences between these proxies likely also reflect their sensitivity to distinct physical processes.

Several limitations should be noted. First, the peak-based approach simplifies the spatial structure of turbulence. Second, the LET is derived under assumptions of isotropy and incompressibility, which are only approximately satisfied in the magnetosheath (Pecora, Francesco et al. 2026). Finally, the analysis is statistical and does not establish causality between turbulent structures and kinetic processes.

Despite these limitations, the present results provide robust observational evidence for a structured coupling between turbulence and kinetic processes. Overall, they support a picture in which turbulent energy transfer, kinetic activity, and particle energization are tightly coupled but spatially offset processes. Future work should aim to distinguish between different channels of energy conversion and to separate reversible from irreversible contributions using sign-resolved proxies.

Acknowledgments

LSV was supported by the Swedish National Space Agency (SNSA) Research Grant N. 2025-00229. LSV and OP were supported by the Swedish Research Council (VR) Research Grant N. 2022-03352 and by the PRIN 2022 project “2022K2L38BK - The ULti-mate fate of TuRb ulence from space to laboratory plasmas (ULTRAS)” (Master CUP B53D2304850006) by the Italian Ministry of University and Research, funded under the National Recovery and Resilience Plan (NRRP), Mission 4 – Component C2 – Investment 1.1, “Fondo per il Programma Nazionale di Ricerca e Progetti di Rilevante Interesse Nazionale (PRIN 2022)” (PE9) by the European Union – NextGenerationEU. OP is supported by the FIS2 Starting Grant FIS-2023-00246 “PhAse-sPAce cOmplexity in turbulent nearly-reversible plasmas (PAPAAO)” (CUP B53C24009610001) funded by the Italian Ministry of University and Research. AT was supported by the ... ZV was funded by the Austrian Science Fund (FWF) under 10.55776/PAT932923. FP is partially supported by NSF SHINE 2501387 at the University of Delaware. TV is supported by NASA-FINESST grant No. 80NSSC23K1630. SS is supported by “Space It Up!”, funded by the Italian Space Agency (ASI) and the Ministry of University and Research (MUR), under Contract Grant Nos. 2024-5-E.0-CUP and I53D240000600005. VMC is supported by the Vicenç Mut grant POSTDOC2024-58 funded by Conselleria d’Educació i Universitats del Govern de les Illes Balears i pel Fons Social Europeu +.

Author Contributions

LVAD, LS and PR contributed equally to all the elements of the work. VMC and AS contributed to the development of data analysis codes and to the interpretation of results. All other authors contributed to the data analysis, interpretation of results, and manuscript preparation.

Software: Pyrfu ([43])

Appendix

Residual spintone filtering

During the computation of gradient-related quantities such as Π−D\Pi- D and pθp\theta, significant oscillations were observed and are attributed to residual spintone effects (see Figure 6). These oscillations exhibit regular patterns, and their dominant frequency was identified using a Welch power spectral density estimate. To mitigate their impact, a high-pass Fourier filter was applied and targeted frequencies above the dominant frequency with a 10% margin. This margin accounts for numerical precision issues, as filtering at the exact dominant frequency sometimes fails to suppress the oscillations effectively.

High-pass filtering applied to the $p\theta$ proxy

Figure 6. High-pass filtering applied to the pθp\theta proxy. After applying the filter with a cutoff frequency of 1.05×fdominant1.05 \times f_{\mathrm{dominant}} to the original signal (blue), the filtered signal (red) effectively suppresses these oscillations while preserving the physical trends.

Methodology for the spatial displacement matching between peaks

Peak detection is performed using a threshold defined relative to the local statistical variability of the signal. For each diagnostic time series x(t)x(t), the mean μ\mu and standard deviation σ\sigma are computed over the considered interval. A detection threshold nn is then applied to identify peak times tpeakt_{\mathrm{peak}} satisfying

∣x(tpeak)−μ∣>nσ.\left|x(t_{\mathrm{peak}})-\mu\right| > n\sigma.

Only peaks exceeding this threshold are considered candidate events. To prevent clusters of nearby local maxima from being counted as independent structures, we additionally apply prominence filtering from Python’s scipy library.

To quantify spatial separations between turbulent structures and kinetic signatures, we match peaks detected in the LET signal with peaks detected in each kinetic proxy. Since multiple peaks may occur within a short interval, a one-to-one matching procedure is required to avoid ambiguous associations. Let the detected peak times of signals AA and BB be respectively PA={tA,i}i=1nP_A = \{t_{A,i}\}_{i=1}^{n} and PB={tB,j}j=1mP_B = \{t_{B,j}\}_{j=1}^{m}, with tA,i,tB,j∈Rt_{A,i}, t_{B,j} \in\mathbb{R}. We construct a one-to-one matching using a mutual nearest-neighbor criterion. The algorithm proceeds as follows:

  1. Nearest neighbor from AA to BB: For each peak tA,i∈PAt_{A,i} \in P_A, define its nearest neighbor in PBP_B by

π(i)=arg min⁡1≤j≤m ∣tA,i−tB,j∣.\pi(i) = \underset{1 \le j \le m}{\operatorname{arg\,min}}\ \left|t_{A,i}-t_{B,j}\right|.

This yields candidate pairs (tA,i,tB,π(i))(t_{A,i}, t_{B,\pi(i)}). At this stage, multiple peaks in A may map to the same peak in B. We note that in the case of a tie-breaker distance, we choose the peak tB,π(i)t_{B,\pi(i)} with the largest amplitude. if still tied, choose the smallest index.

  1. Nearest neighbor from B to A: Similarly, for each peak tB,j∈PBt_{B,j} \in P_B, define its nearest neighbor in PAP_A by

σ(j)=arg min⁡1≤i≤n∣tA,i−tB,j∣.\sigma(j) = \underset{1 \leq i \leq n}{\operatorname{arg\,min}} \lvert t_{A,i} - t_{B,j} \rvert.

The same tie-breaker rule is applied if necessary.

  1. Mutual matching: A pair (tA,i,tB,j)(t_{A,i}, t_{B,j}) is retained if and only if

π(i)=jandσ(j)=i.\pi(i) = j \quad\text{and} \quad\sigma(j) = i.

The final matched set is

M={(tA,i,tB,j)∈PA×PB:π(i)=j and σ(j)=i}.M = \left\{(t_{A,i}, t_{B,j}) \in P_A \times P_B : \pi(i) = j \text{ and } \sigma(j) = i\right\}.

This process ensures a one-to-one correspondence between peaks and prevents spurious matches caused by asymmetric or distant assignments. All peaks that do not satisfy the mutual matching step are discarded from the analysis.

To assess the significance of the observed spatial associations between LET peaks and kinetic proxy peaks, we perform a null hypothesis test by randomly shuffling the kinetic proxy signals and repeating the peak detection and matching procedure (see Figure 7). At low thresholds, both the shuffled dataset (right) and the real data (left) show similar trends due to the high density of available peaks. In contrast, at higher thresholds, the shuffled dataset systematically yields larger peak distances and a wider spread than the real data. This demonstrates that the closer spatial alignment observed in the real dataset is not reproduced by random shuffling. We note that correlations persist at all thresholds in the shuffled data. This is expected as the thresholding step ensures similar peak amplitudes are detected regardless of signal shuffling.

Mean spatial peak distance between LET and kinetic proxies for observed and shuffled data

Figure 7. Left: Mean spatial peak distance between LET and kinetic proxies as a function of threshold for the unbiased campaign dataset. Right: Same analysis but by shuffling the dataset of each kinetic proxy within each interval and preserving LET.

Influence of the thresholding on the detected structures

The detection threshold nn, expressed in units of the local standard deviation σ\sigma, controls the amplitude of the selected events and therefore probes different levels of intermittency. Lower thresholds include weaker fluctuations, while higher thresholds isolate the most intense structures.

The left panel of Figure 6 shows that the correlations between LET and the dissipation proxies remain largely stable across the explored range of thresholds. This indicates that the coupling between turbulent energy transfer and kinetic activity is not dominated by rare extreme events, but instead reflects a persistent property of the turbulent fluctuations. The relative ordering between proxies is preserved and demonstrates that the hierarchy identified in Section 4 is robust to event selection.

A distinct behavior is observed for the velocity-space diagnostics ξ\xi and MKPM_{KP}, whose correlations with LET increase with the detection threshold. This indicates that strong deviations from equilibrium are preferentially associated with the most intense turbulent structures.

To assess whether this trend could be influenced by measurement noise in the VDFs, we restrict the velocity-space integration to 0.5vth<v<6.8vth0.5v_{th} < v < 6.8v_{th}. As shown in the right panel of Figure 8, this restriction systematically reduces the correlations with LET. If the excluded regions were dominated by counting noise, then their removal would reduce variance and increase correlations. The observed opposite behavior therefore indicates that these regions contain physically meaningful distortions associated with turbulent structures. This supports the interpretation that the enhanced correlations at higher thresholds reflect genuine kinetic effects rather than instrumental artifacts, and justifies the use of the full VDF in the main analysis.

Maximum correlations between LET and dissipation proxies across detection thresholds

Figure 8. Left: Maximum correlation between LET and the each kinetic proxy as a function of the detection threshold expressed in units of the standard deviation σ\sigma. Right: Maximum correlation for ξ\xi and MKPM_{KP} using different velocity-space integration during their computation.

Finally, Figure 9 shows that increasing the threshold reduces the number of detected events but does not significantly modify their spatial correspondence with LET structures. The persistence of clustering across thresholds indicates that the localization of kinetic activity near turbulent structures is not an artifact of event selection. Based on this analysis, n=2n = 2 provides a compromise between statistical convergence and the identification of sufficiently intermittent events.

Influence of the detection threshold on the pairing statistics between LET and the kinetic proxies

Figure 9. Influence of the detection threshold on the pairing statistics between LET and the kinetic proxies. The left panel shows the number of matched structures as a function of the detection threshold, while the right panel shows the ratio between the median separation distance of proxy events dpeakProxyd^{Proxy}_{peak} and the median distance of successive LET events dpeakLETd^{LET}_{peak}.

Turbulent-kinetic correlations across plasma β\beta regimes

To assess the dependence of the results on plasma conditions, we repeat the analysis across different ranges of the plasma β\beta, which is known to influence the nature of turbulence and kinetic processes in the magnetosheath.

Figure 9 shows that the correlations between LET and the kinetic proxies tend to increase with plasma β\beta. The binning is constructed to account for the bin width to ensure that the reported values correspond to properly normalized probability densities and allowing for a consistent comparison across β\beta regimes.

The observed increase in correlation at higher β\beta indicates a stronger coupling between turbulent fluctuations and kinetic processes in these regimes. High-β\beta plasmas are commonly associated with enhanced compressibility and stronger pressure fluctuations. Consequently, it can promote more efficient energy conversion through pressure-strain interactions and facilitate the development of velocity-space distortions. This behavior is consistent with previous studies showing that the partition of energy conversion channels depends on plasma β\beta [33].

Turbulent–kinetic correlations across plasma $\beta$ regimes

Figure 10. Turbulent–kinetic correlations across plasma β\beta regimes. The left panel shows the probability density function of intervals as a function of β\beta, normalized by the total number of intervals. The right panel shows the corresponding maximum correlations (Pearson or Spearman) between LET and each kinetic diagnostics as a function of β\beta.

At the same time, non-negligible correlations are observed across all β\beta intervals and indicate that the coupling between turbulent structures and kinetic activity is not restricted to specific plasma conditions. Although the dataset is predominantly composed of intervals in the range 1≲β≲21 \lesssim\beta\lesssim2, the persistence of correlations outside this regime demonstrates that the observed trends are not driven by a particular subset of the data.

Overall, these results indicate that the coupling between turbulence and kinetic processes is a robust feature of magnetosheath plasma, with an efficiency that increases with β\beta.

References

  1. [2]Anselmet, F., Gagne, Y., Hopfinger, E. J., & Antonia, R. A. 1984 Journal of Fluid Mechanics, 140, 63–89, doi: 10.1017/S0022112084000513DOI
  2. [3]Argall, M. R., Barbhuiya, M. H., Cassak, P. A., et al. 2022 Physics of Plasmas, 29, 022902, doi: 10.1063/5.0073248
  3. [4]Bandyopadhyay, R., Matthaeus, W. H., Parashar, T. N., et al. 2020 Physical Review Letters, 124, doi: 10.1103/physrevlett.124.255101
  4. [5]Barbhuiya, M. H., Cassak, P. A., Adhikari, S., et al. 2024 Physical Review E, 109, doi: 10.1103/physreve.109.015205
  5. [6]Bruno, R., & Carbone, V. 2005 Living Reviews in Solar Physics, 2, 4, doi: 10.12942/lrsp-2005-4
  6. [7]Chandran, B. D. G., Li, B., Rogers, B. N., Quataert, E., & Germaschewski, K. 2010 The Astrophysical Journal, 720, 503, doi: 10.1088/0004-637X/720/1/503
  7. [8]Chasapis, A., Matthaeus, W., Parashar, T., et al. 2017 The Astrophysical Journal, 836, 247, doi: 10.3847/1538-4357/836/2/247
  8. [9]Chasapis, A., Matthaeus, W. H., Parashar, T. N., et al. 2018 The Astrophysical Journal Letters, 856, L19, doi: 10.3847/2041-8213/aaadf8
  9. [10]Che, H., Schiff, C., Le, G., et al. 2018 Physics of Plasmas, 25, doi: 10.1063/1.5016853
  10. [11]Del Sarto, D., Pegoraro, F., & Califano, F. 2016 Phys. Rev. E, 93, 053203, doi: 10.1103/PhysRevE.93.053203
  11. [12]Ergun, R. E., Tucker, S., Westfall, J., et al. 2016 Space Science Reviews, 199, 167, doi: 10.1007/s11214-014-0115-x
  12. [13]Farris, M. H., & Russell, C. T. 1994 Journal of Geophysical Research: Space Physics, 99, 17681, doi: https://doi.org/10.1029/94JA01020
  13. [14]Franci, L., Cerri, S. S., Califano, F., et al. 2017 The Astrophysical Journal Letters, 850, L16, doi: 10.3847/2041-8213/aa93fb
  14. [15]Fuselier, S. A., Lewis, W. S., Schiff, C., et al. 2016 Space Science Reviews, 199, 77, doi: 10.1007/s11214-014-0087-x
  15. [16]Graham, D. B., Khotyaintsev, Y. V., André, M., et al. 2021 Journal of Geophysical Research: Space Physics, 126, e2021JA029260, doi: https://doi.org/10.1029/2021JA029260
  16. [17]Greco, A., Valentini, F., Servidio, S., & Matthaeus, W. H. 2012 Phys. Rev. E, 86, 066405, doi: 10.1103/PhysRevE.86.066405
  17. [18]Gurchumelia, A., Sorriso-Valvo, L., Burgess, D., et al. 2022 Frontiers in Physics, Volume 10 - 2022, doi: 10.3389/fphy.2022.903632
  18. [19]Hadid, L. Z., Sahraoui, F., Galtier, S., & Huang, S. Y. 2018 Phys. Rev. Lett., 120, 055102, doi: 10.1103/PhysRevLett.120.055102
  19. [20]Hellinger, P., Verdini, A., Montagud-Camps, V., et al. 2024 Astronomy & Astrophysics, 684, A120, doi: 10.1051/0004-6361/202348547
  20. [21]Howes, G. G., Klein, K. G., & Li, T. C. 2017 Journal of Plasma Physics, 83, 705830102, doi: 10.1017/S0022377816001197
  21. [22]Howes, G. G., TenBarge, J. M., Dorland, W., et al. 2011 Physical Review Letters, 107, doi: 10.1103/physrevlett.107.035004
  22. [23]Karimabadi, H., Roytershteyn, V., Wan, M., et al. 2013 Physics of Plasmas, 20, 012303, doi: 10.1063/1.4773205
  23. [24]Kaufmann, R. L., & Paterson, W. R. 2009 Journal of Geophysical Research: Space Physics, 114, doi: https://doi.org/10.1029/2008JA014030
  24. [25]Liang, H., Barbhuiya, M. H., Cassak, P. A., et al. 2020 Journal of Plasma Physics, 86, doi: 10.1017/s0022377820001270
  25. [26]Liang, H., Cassak, P. A., Servidio, S., et al. 2019 Physics of Plasmas, 26, doi: 10.1063/1.5098888
  26. [27]Marino, R., & Sorriso-Valvo, L. 2023 Physics Reports, 1006, 1, doi: https://doi.org/10.1016/j.physrep.2022.12.001
  27. [28]Marsch, E. 2006 Living Reviews in Solar Physics, 3, 1, doi: 10.12942/lrsp-2006-1
  28. [29]Matthaeus, W. H., Smith, C. W., & Bieber, J. W. 1999 AIP Conference Proceedings, 471, 511, doi: 10.1063/1.58686
  29. [30]Matthaeus, W. H., Yang, Y., Wan, M., et al. 2020 The Astrophysical Journal, 891, 101, doi: 10.3847/1538-4357/ab6d6a
  30. [31]Montagud-Camps, V., Grappin, R., & Verdini, A. 2026 The Astrophysical Journal, 853, 153, doi: 10.3847/1538-4357/aaa1ea
  31. [32]Osman, K. T., Matthaeus, W. H., Wan, M., & Rappazzo, A. F. 2012 Physical Review Letters, 108, doi: 10.1103/physrevlett.108.261102
  32. [33]Parashar, T. N., Matthaeus, W. H., & Shay, M. A. 2018 The Astrophysical Journal Letters, 864, L21, doi: 10.3847/2041-8213/aadb8b
  33. [34]Pecora, F., Yang, Y., Chasapis, A., et al. 2023 Monthly Notices of the Royal Astronomical Society, 525, 67, doi: 10.1093/mnras/stad2232
  34. [35]Pecora, F., Chhiber, R., Roy, S., et al. 2026a, 2023 MMS unbiased magnetosheath campaign intervals, 1.0 Zenodo, doi: 10.5281/zenodo.1863220DOI
  35. [36]Pecora, F., Chhiber, R., Roy, S., et al. 2026b Journal of Geophysical Research: Space Physics, 131, e2025JA034919, doi: https://doi.org/10.1029/2025JA034919
  36. [38]Pezzi, O., Perrone, D., Servidio, S., et al. 2019 The Astrophysical Journal, 887, 208, doi: 10.3847/1538-4357/ab5285
  37. [39]Pezzi, O., Liang, H., Juno, J. L., et al. 2021 Monthly Notices of the Royal Astronomical Society, 505, 4857–4873, doi: 10.1093/mnras/stab1516
  38. [40]Phillips, C., Bandyopadhyay, R., McComas, D. J., & Bale, S. D. 2022 Monthly Notices of the Royal Astronomical Society: Letters, 519, L1, doi: 10.1093/mnrasl/slac143
  39. [41]Politano, H., & Pouquet, A. 1998 Geophysical Research Letters, 25, 273, doi: https://doi.org/10.1029/97GL03642
  40. [42]Pollock, C., Moore, T., Jacques, A., et al. 2016 Space Science Reviews, 199, 331, doi: 10.1007/s11214-016-0245-4
  41. [43]PyRFU team, Richard, L., Khotyaintsev, Y., et al. 2024, pyrfu, 1.2.0 doi: 10.5281/zenodo.10678695
  42. [44]Quijia, P., Fraternale, F., Stawarz, J. E., et al. 2021 Monthly Notices of the Royal Astronomical Society, 503, 4815, doi: 10.1093/mnras/stab319
  43. [45]Retinò, A., Khotyaintsev, Y., Le Contel, O., et al. 2022 Experimental Astronomy, 54, 427, doi: 10.1007/s10686-021-09797-7DOI
  44. [46]Richard, L., Servidio, S., Svenningsson, I., et al. 2025, Non-Maxwellianity of Ion Velocity Distributions in the Earth’s Magnetosheath, https://arxiv.org/abs/2504.05502
  45. [47]Russell, C. T., Anderson, B. J., Baumjohann, W., et al. 2016 Space Science Reviews, 199, 189, doi: 10.1007/s11214-014-0057-3
  46. [48]Sahraoui, F., Goldstein, M. L., Robert, P., & Khotyaintsev, Y. V. 2009 Phys. Rev. Lett., 102, 231102, doi: 10.1103/PhysRevLett.102.231102DOI
  47. [49]Servidio, S., Valentini, F., Perrone, D., et al. 2015 Journal of Plasma Physics, 81, 325810107, doi: 10.1017/S0022377814000841DOI
  48. [50]Shue, J.-H., Song, P., Russell, C. T., et al. 1998 Journal of Geophysical Research: Space Physics, 103, 17691, doi: https://doi.org/10.1029/98JA01103
  49. [51]Sorriso-Valvo, L., Marino, R., Foldes, R., et al. 2023 A&A, 672, A13, doi: 10.1051/0004-6361/202244889
  50. [52]Sorriso-Valvo, L., Perrone, D., Pezzi, O., et al. 2018 Journal of Plasma Physics, 84, doi: 10.1017/s0022377818000302
  51. [53]Sorriso-Valvo, L., Catapano, F., Retinò, A., et al. 2019 Physical Review Letters, 122, doi: 10.1103/physrevlett.122.035102
  52. [54]Svenningsson, I., Yordanova, E., Khotyaintsev, Y. V., André, M., & Cozzani, G. 2025 Journal of Geophysical Research: Space Physics, 130, e2024JA033272, doi: https://doi.org/10.1029/2024JA033272
  53. [55]Svenningsson, I., Yordanova, E., Khotyaintsev, Y. V., et al. 2024 Journal of Geophysical Research: Space Physics, 129, e2024JA032661, doi: https://doi.org/10.1029/2024JA032661
  54. [56]Swisdak, M. 2016 Geophysical Research Letters, 43, 43–49, doi: 10.1002/2015gl066980DOI
  55. [57]Taylor, G. I. 1938 Proceedings of the Royal Society of London. A. Mathematical and Physical Sciences, 164, 476, doi: 10.1098/rspa.1938.0032DOI
  56. [58]Tessein, J. A., Matthaeus, W. H., Wan, M., et al. 2013 The Astrophysical Journal Letters, 776, L8, doi: 10.1088/2041-8205/776/1/L8DOI
  57. [59]Tessein, J. A., Ruffolo, D., Matthaeus, W. H., et al. 2015 The Astrophysical Journal, 812, 68, doi: 10.1088/0004-637X/812/1/68DOI
  58. [61]Wan, M., Matthaeus, W. H., Roytershteyn, V., et al. 2016 Physics of Plasmas, 23, 042307, doi: 10.1063/1.4945631DOI
  59. [62]Welch, P. 1967 IEEE Trans. Audio Electroacoust., 15, 70, doi: 10.1109/TAU.1967.1161901DOI
  60. [63]Yang, Y., Wan, M., Matthaeus, W. H., et al. 2018 Monthly Notices of the Royal Astronomical Society, 482, 4933–4940, doi: 10.1093/mnras/sty2977DOI
  61. [64]Yang, Y., Matthaeus, W. H., Parashar, T. N., et al. 2017 Phys. Rev. E, 95, 061201, doi: 10.1103/PhysRevE.95.061201DOI
  62. [65]Yordanova, E., Vaivads, A., André, M., Buchert, S. C., & Vörös, Z. 2008 Phys. Rev. Lett., 100, 205003, doi: 10.1103/PhysRevLett.100.205003DOI
  63. [66]Yordanova, E., Vörös, Z., Sorriso-Valvo, L., Dimmock, A. P., & Kilpua, E. 2021 The Astrophysical Journal, 921, 65, doi: 10.3847/1538-4357/ac1942DOI
  64. [67]Zenitani, S., Hesse, M., Klimas, A., & Kuznetsova, M. 2011 Physical Review Letters, 106, doi: 10.1103/physrevlett.106.195003DOI

Paper details

vlnc:W_39tt2hch
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 2

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

Not recorded for this paper

View statement
Looking for related papers…
Contents