Geophysical Research Letters

Firehose driven magnetic fluctuations in the magnetosphere



[1] The nonlinear saturation of the firehose instability in the high plasma pressure central plasma sheet is shown to produce a wide spectrum of Alfvénic fluctuations in the range of Pi-2 geomagnetic pulsations. The wave energy sources are the small p/p > 1 + B20p anisotropies which are created by Earthward ion convection at constant first and second adiabatic invariants. In the nonlinear state, the field-line curvature force is weaker than the linear force. This weakening of the driving force limits the amplitude of the Alfvénic fluctuations. Away from the equatorial plane, the plasma is firehose stable, but carries large magnetic fluctuations.

1. Introduction

[2] Geomagnetic pulsations of the Pi-2 type are defined as magnetic fluctuations with periods in the range 40–200 s with irregular wave forms. The geomagnetic Pi-2 oscillations are known to be intimately correlated with substorm growth and onsets [Sigsbee et al., 2002] and bursty bulk flows [Kepko and Kivelson, 2001].

[3] During the time of enhanced Earthward plasma convection, there are numerous nonequilibrium features that develop in the ion phase space density distributions, including currents and anisotropies, that may serve to trigger low and ultra-low frequency instabilities. Here we argue that ion pressure anisotropies p > p associated with convection or bursty bulk flows drive nonlinear Pi-2 type fluctuations in the geomagnetic tail. We find that the nonlinear firehose driven fluctuations have a rich kω-spectrum and that the local nonlinear firehose stability parameter

equation image

fluctuates with only short-lived excursions into the unstable domain σ(t) < 0 occurring at the equatorial plane where the curvature equation image = (b · ∇)b vector has its maximum value.

[4] Kaufmann et al. [2000] use one year of one-minute Geotail data to analyze the firehose parameter A = μ0(pp)/B2 as a function of β. Kaufmann et al. [2000] report A ranging from −0.1 to +0.4 with the values closer to zero occurring in the high-β bins. The number of data samples in the high-beta bins, β > 30 and 10 < β < 30, is, however, low. Kaufmann et al. [2000] conclude that while A from the data is a few tenths, that the methodology used underestimates the value of A.

[5] Chen and Wolf [1999] develop a model for bursty bulk flows that follows an Earthward accelerated flux tube that develops a firehose instability due to the faster increase of p from the shortening of the magnetic field line lengths than of p from B. Ji and Wolf [2003a] follow up the Chen and Wolf model with a detailed Lagrangian simulation model that couples the firehose dynamics to the magnetoacoustic wave dynamics. They find a shock front propagating to the Earth and the development of the firehose instability at the largest k = π/Δ in the simulation. They pose the problem of finding the kinetic theory physics that limits the ∣k∣ for the growth rate and finding the nonlinear saturation level.

2. Nonlinear Firehose Model

[6] In this letter we present a kinetically-modified Eulerian ion fluid description of the nonlinear firehose instability for the magnetotail problem. We use a simple field-line model with an anisotropic pressure.

[7] At latitude θ, the σ(t, θ) → 1 with the fluctuation energy wB = (δBx2/2μ0) and wK = equation imageρ(Ey2/B2) flowing into the region from the magnetic equatorial plane. Outside the high plasma pressure region β = 2μ0p/B2 ≫ 1 the oscillations are large amplitude Alfvén waves at finite kρi where ρi = (miTi)1/2/eB is the ion gyroradius. The finite ion gyroradius effects and the finite ion inertial scale cpi effects determine the maximum growth rate γmax = equation image[γ(kz)] of the firehose instability. The calculation of these dispersion terms is well known and gives rise to the following linear dispersion relation

equation image

where we neglect the Landau damping contributions from the warm plasma kinetic dispersion relation [Stix, 1992; Gary, 1993]. For kρi = (kcpi)(βi/2)1/2 ≪ 1 the linear dispersion term in ω/ωci is negligible and equation (2) yields the MHD firehose instability ω2 = k2vA2σ with σ defined in equation (1). For p/p > 1 + 2/βi the MHD growth rate is image = ∣kvAequation image increasing monotonically with k. From the kinetic dispersion relation there is a well-defined maximum growth rate at kρi = equation image(1 − T/T − 2/βi)1/2 where the maximum growth rate

equation image

We have written an initial value code using the dipole magnetic field lines and typically taking the mass density ρ(s) equation imageB(s) to study the linear and nonlinear kinetically-modified Alfvénic-Firehose turbulence occurring in the night-side magnetotail. The nonlinear equation for the complex displacement field ξ(z, t) is

equation image

where σn is given in equation (1) with B2 = Bn2(1 + ∣∂zξ∣2) and ν = ωciρi2 = cTequation image/eB follows from equation (2). The complex valued fields are ξ = ξx + iξy and δB/Bn = ∂zξx + izξy with the real δB2 = Bn2 + δBB used in equation (1). The last term in equation (4) describes the Faraday rotation of the polarization vector of the Alfvén wave. The phase rotation δϕ in time Δt is δequation image = νk2Δt = k2(cTi/eBt.

[8] The derivation of equation (4) follows from the acceleration equation

equation image

where P = (pp)(BB/B2) + pI, B = Bnequation imagex + δBxequation imagex, jy = equation image = equation image and δBx = Bzzξ from Faraday's law ∂zEy = ∂Bx/∂t and Ohm's law EyvxBz = EyBz(∂ξ/∂t) = 0. The inertial acceleration from the jyBz force gives ∂t2ξ = vA2z2ξ. The nonlinear pressure gradient force follows from the calculation of

equation image

which reduces to

equation image

after using B = Bn(1 + ∣∂zξ∣2)1/2 and Bn = Bz = constant.

[9] The nonlinear driving term equation image at small δBx/B = ∂zξ ≪ 1 is proportional to the field-line displacement ξ. At large amplitudes δBx/Bequation image 1 the force is greatly weakened, eventually decreasing with ξ as ∂z2ξ/∣∂zξ∣2. The effect is familiar as the decrease of curvature for a string y = y(x) given by the curvature formula y″/(1+y2)3/2. This may be called the magnetic field line crinkling.

[10] The nonlinear partial differential equation (4) develops mode coupling from the nonlinear curvature force. The nonlinear partial differential equation for U(equation image, τ) = Bzξ(z/L, tvA0/L) with τ = tBz/(μ0ρ0)1/2L where L is the scale length (∼RE) of the field line variation is

equation image

with boundary conditions U(z = ±L, τ) = 0. Here ν = Ti/eBLvA0 = ρixvi/LvA.

[11] The total energy is

equation image

where K = equation imagedθequation imageρ(∂ξ/∂t)2, WB = equation imagedθδequation image2/2μ0 and WP = −A/2equation imageequation imagenequation imagedθ. We use the energy invariant to test the accuracy of the numerical solutions.

[12] The nonlinear potential is of the type that occurs in the derivative nonlinear Schrödinger equation (DNLS) which governs dispersive Alfvén waves [Horton and Ichikawa, 1996]. In perturbative theory the nonlinear force weakens as equation imageequation image [A − 1 − A∣∂zξ∣2](∂z2ξ). Here we show a typical example of the solution of equation (8). In a future work, we will develop solutions for the magnetotail that include parametric decays of these fluctuations into Alfvén waves and acoustic waves. A single k-mode calculation shows that the saturated amplitude varies as ξmax = k−1(A1/2 − 1)1/2 and has the nonlinear frequency ωA(A − 1)1/2.

[13] In addition to the intrinsic nonlinear force derived in equation (8) that saturates the instability and produces the nonlinear oscillations shown in Figures 14 there is the quasilinear change of the pressure anisotropy due to reaction of the magnetic fluctuations on the background ion phase space distribution function [Gary, 1993; Sagdeev and Galeev, 1969]. Here we ignore the quasilinear change of the pressures arguing that on the background evolution time scale there are processes building up the anisotropy competing with the quasilinear relaxation of the anisotropy. Thus, the turbulence shown here for a fixed value of A = μ0(pp)/Bn2 given here overestimates the amplitude obtained with addition of the quasilinear background transport. In the simulation we model this effect rather crudely by setting p = p after some number of nonlinear oscillations. There follows a period of large amplitude Alfvén wave oscillations which may be subject to parametric instabilities for A above a critical value.

Figure 1.

(a) Fastest growing eigenfunction (solid line real part, dashed imaginary part) for A = 1.1 and β(0) = 5, (b) the eigenfrequency (solid line) and growth rate (dashed line) in units of vA0/R0 where vA0 and R0 are the equatorial Alfvén speed and mirror field scale length, (c) the k-spectrum of the eigenmode and (d) the initial profile of σ(θ) in equation (1).

Figure 2.

One-hour interval of steady-state turbulence driven by fixed p/p = 1.44. (a) The magnetic fluctuations δBx(t), (b) the magnetic power spectrum ∣δBx(ω)∣2. (c) the electric field fluctuation Ey(t) and (d) the displacement field ξ = ξx + iξy fluctuations (ξx solid, ξy dashed).

Figure 3.

The three energy components in equation (9) for the solution shown in Figure 2. The total energy Wtotal = Wtotal(t = 0) = 10−30.

Figure 4.

The increase of the mean value of the magnetic fluctuation energy WB as a function of the anisotropy parameter A for fixed ν = 0.01. The small error bars at low WB are due to the small oscillations of WB(t) about the mean value. The actual errors are larger.

[14] The simulations are performed as an initial value problem with equatorial field-line crossing at x ∼ 8 RE with β(0) = 2μ0p/Bn2 = 5 and p/p = 1.44 so that A = 1.1 or σ(t = 0) = −0.1 and ν = 0.01. We use a 4th–5th over adaptive RK integrator to advance the finite differential equations on 129 grid points along the magnetic field line. The local equatorial field strength is Bn ≃ 100 nT. The local ion gyroradius ρi ≅ 200 km.

3. Simulation Results

[15] Figure 1a shows the fastest growing eigenfunction at 10tA ≃ 820 s as a function position along the field line. Figure 1b shows that there is a well-defined eigenmode frequency (ω = 2πf ∼ 8/tA ∼ 5–10 mHz), Figure 1c shows the spectrum of k = 2πn/L values in the eigenmode. Figure 1d gives the profile of σ(θ, t = 10tA). We have not calculated analytically the eigenmodes. The local equatorial plane frequency and growth rate for the most unstable k are ω ≃ γmax = 10(vA0/L), but the observed eigenmode growth rate is much smaller than γmax due to the localization of σ(θ).

[16] Figures 2a and 2b show the magnetic fluctuations δBx(t) and its frequency spectrum δBx2(ω). Figure 2c gives the electric field fluctuations Ey(t). Figure 2d shows the nonlinear state of the plasma defined by ξ(t) = ξx(t) + iξy(t) for a time interval Δt = 15tA ≃ 20 min in the saturated state. There are several characteristic frequencies in the dynamics. In this example, there is a short period signal T1 ≃ 100 s and a long nonlinear period T0 ≃ 300 s.

[17] The time series in Figure 2 for the electric and magnetic fields show a chaotic structure even though the power is concentrated in two frequencies. Such spectra are typical of lower-order dynamical systems suggesting a search for a low-order model derived from the pde. There are bursts of energy releases from the nonlinear dynamics that are one aspect of a self-organized criticality (SOC) system. The second aspect of space-scale invariance of an SOC system is not satisfied due to the key role of the ion gyroradius in defining the fast-growing linear mode at k∥,max. The turbulent fluctuations mode-couple to both shorter and longer wavelength fluctuations. For σ = 1 there are soliton solutions to the associated DNLS equation [Horton and Ichikawa, 1996] derived by factoring the Alfvén equation into uncoupled right and left propagating waves. In the present problem the driving by the pressure anisotropy disrupts those solitons and couples the parallel (right) and antiparallel (left) propagating Alfvén waves. For equation (8) we find coherent, localized solutions not unlike solitons for ν = 0.03.

[18] Figure 3 shows the three energy components in equation (9) for the solutions in Figure 2. After the exponential growth the turbulence saturates with the magnetic fluctuation energy WB ≃ −WP the source of instability and both large compared with kinetic energy K. In the simulation the total energy is well conserved as follows analytically from the model. Thus, there is an efficient conversion from the anisotropy thermal energy reservoir into magnetic energy. The increase of the magnetic turbulence with the anisotropy parameter A is shown in Figure 4. At small A − 1 values the system shows coherent structures and the error bars may underestimate the actual error.

4. Conclusions

[19] In conclusion, the presence of small fractional pressure anisotropies in the high-beta magnetotail plasma produce an intermittent spectrum of Alfvénic fluctuations that are in the range of the Pi-2 signals commonly associated with bursty bulk flows [Kepko and Kivelson, 1999, 2001] and substorm dynamics. The kinetic ballooning pressure gradient instability [Roux et al., 1991] remains a possible cause of these fluctuations. The nonlinear firehose model also produces candidate magnetic fluctuations and requires the kinetic dispersion term to give the maximum growth rate, the wave dispersion and polarization. The nonlinear model is integrated and shows irregular wave forms with intermittent turbulent energy releases. Further research that includes the driving sources of the anisotropy and the quasilinear velocity scattering of the ions from the magnetic fluctuations is required for a more detailed picture of the long-time evolution of the system.

[20] In a related work Ji and Wolf [2003b] argue that the appearance of the anisotropy is a product of the near-Earth neutral line formation. Consequently, the observations of Pi-2 turbulence described here along the auroral field lines may be a signature of the flow braking from fast Earthward flows produced by a near-Earth neutral line. It remains for future 2D theory and simulations to find the balance of the quasilinear reduction of the anisotropy and its production by Earthward flows.

[21] Finally, the problem of correlating the observed Pi2 oscillations with the various theoretical models remains.


[22] The authors thank Richard Wolf and Shuo Ji for numerous useful discussions along the path to this solution. The work was supported by the National Science Foundation ATM-0229863.