Magnetic Turbulence, Fast Magnetic Field line Diffusion and Small Magnetic Structures in the Solar Wind G. Zimbardo , P. Pommoisy and P. Veltri y Center Dipartimento di fisica, Università della Calabria (Italy) for High Performance Computing, Università della Calabria (Italy). Abstract. The influence of magnetic turbulence on magnetic field line diffusion has been known since the early days of space and plasma physics. However, the importance of “stochastic diffusion" for energetic particles has been challenged on the basis of the fact that sharp gradients of either energetic particles or ion composition are often observed in the solar wind. Here we show that fast transverse field line and particle diffusion can coexist with small magnetic structures, sharp gradients, and with long lived magnetic flux tubes. We show, by means of a numerical realization of three dimensional magnetic turbulence and by use of the concepts of deterministic chaos and turbulent transport, that turbulent diffusion is different from Gaussian diffusion, and that transport can be inhomogeneous even if turbulence homogeneously fills the heliosphere. Several diagnostics of field line transport and flux tube evolution are shown, and the size of small magnetic structures in the solar wind, like gradient scales and flux tube thickness, are estimated and compared to the observations. INTRODUCTION TURBULENT VERSUS GAUSSIAN DIFFUSION The magnetic turbulence found in many plasmas causes a magnetic field line random walk which “destroys" the magnetic surfaces [1, 2, 3, 4, 5, 6], and causes a fast plasma transport across the magnetic field structure, a phenomenon which is sometimes called stochastic diffusion. Recently it was shown that magnetic field line diffusion gives a natural explanation for the Ulysses observations of energetic particles at high southern heliographic latitudes [7]. Nevertheless, the relevance of stochastic diffusion in space plasmas has been challenged on the basis of the fact that sharp gradients of particular ion composition are often observed in space (see e.g. Ref. [8]). Also, sharp intensity variations in impulsive energetic particle events seen by the SWICS instrument on ACE may suggest that turbulent diffusion is not smoothing out the gradients [9]. Yet, it was shown by Giacalone et al. [10], by means of a numerical simulation, that magnetic field line random walk is consistent with both fast diffusion and small-scale gradients in energetic particle intensity. In this paper we argue that this is a general feature of the transport of a passive tracer by a turbulent field, that turbulent diffusion is different from Gaussian diffusion, and show by a numerical simulation that fast magnetic field line and plasma transport can be found simultaneously to small magnetic flux tube structure and field line mixing. Moreover, we argue on the type of anisotropy present in the solar wind turbulence. In a normal, Gaussian diffusion process, the diffusing particles move randomly in all directions, because of either molecular diffusivity, in an ordinary fluid, or collisions, in a plasma. Such random motion tends to smooth out a density gradient, and a point like initial concentration of, say, blue dye, evolves with the well known Gaussian smooth profile. Also, a Gaussian random walk is characterised by a finite mean square step length, and by random, uncorrelated directions of motion. In such a case normal diffusion results, h x2i i Di t, where Di is the diffusion coefficient and t the time, and, in particular, the diffusing particles moves in all direction with equal probability. This leads to the smoothing of density or concentration gradients. We consider here turbulent diffusion as the result of the advection of a passive tracer by a turbulent velocity (or magnetic) field. The transport of the passive tracer (or of the test particles in a turbulent magnetic field) is due to the chaotic evolution of the non linear dynamical system. Such evolution can be very complex and “unpredictable", but still deterministic. This means that each passive tracer does not move randomly, but follows a flow (or field) line for long distances regardless of the density or concentration gradients. Basically, this is what happens when you are mixing a blue dye with a white paint: at least for some time (until molecular diffusion becomes important) blue stripes stand out on a white CP679, Solar Wind Ten: Proceedings of the Tenth International Solar Wind Conference, edited by M. Velli, R. Bruno, and F. Malara © 2003 American Institute of Physics 0-7354-0148-9/03/$20.00 409 =2 background. In a turbulent medium, long range correlations among the fluctuation occurr, and, indeed, anomalous transport regime can be found, h x 2i i Di ti (i x;y ), with i > (see Pommois et al. [5, 6]). The main features of such chaotic (not random) transport can be grasped by considering the evolution of a magnetic flux tube in the presence of magnetic turbulence. In a uniform background field, a flux tube with an initially circular cross section will be elongated in one direction and squeezed in the other (because of r , this is the typical behaviour). Then the flux tube cross section will be distorted and further on it will develop branches (see figures by Rechester and Rosenbluth [3]; and Isichenko [11]). A very complex and ramified structure can be obtained, with the outer size of the flux tube growing fast, but with many “voids" in the cross section. A numerical simulation of the flux tube evolution is reported here. The magnetic field model is = 1 =2 B=0 B(r) = B e^z + B(r) 0 that is an uniform background field plus zero-average fluctuations. Here the magnetic fluctuations are constructed in a parallelepipedal box with sides Li ' li , where li is the correlation length in the i-th direction (i x;y;z ). The magnetic fluctuations are represented as ( ) B i (k) k; 4 = X B(r) = (k)e (k)exp [k r + r=lx r=lx/5 ] FIGURE 1. Evolution of a magnetic flux tube cross section in the case of quasi 2-D spectrum with lx =lz = ly =lz = 0:1 (from bottom to top). The magnetic flux tube is starting in the circle of radius r at bottom of the figure and is traced with 5000 points. Each adjacent line is integrated, and we plotted here the evolution for interval Z = 0:25Lz lz (where Lz is the box simulation dimension, and lz is the correlation length in z direction). The turbulence level is B=B0 = 0:4. Flux tube on the left, r = 0:05Lz 0:2lz ; flux tube on the right, r = 0:25Lz lz . Magnetic turbulence appears to be anisotropic both in the solar wind [12, 13] and in the interstellar medium [14]. Therefore, the Fourier intensity is given by B (k) = (kx 2 lx 2 + ky 2 ly C 2 + kz 2 lz 2 )=4+1=2 where C is a normalization constant. Note that with this choice of B , the Fourier amplitude depends on all of lx , ly , lz , and that the isolevels of B are ellipsoids in space (see, Pommois et al. [5, 6], for more details). Once the magnetic field model is set up, we can trace the magnetic field structure by integrating the magnetic field line equations d =ds =j j. It is usually found that the level of stochasticity increases with the turbulence level B=B0 . Here, B is the rms value of fluctuations. In a similar way the level of stochasticity increase with the ratio of correlation lengths lz =lx [6, 15, 16]. A suitable parameter to caracterize the level of stochasticity and the transport regime was shown to be the Kubo number R Bl z =B0 lx [15, 16]. This expression of R was found to be valid even when l x 6 ly [16]. Besides, the study of magnetic field line transport carried out in [15, 16] shows that anomalous diffusion is found for R , quasilinear diffusion, i.e. D i / B 2 , is found for : < R< , and the percolation scaling of the . diffusion coefficient, Di / B 0:7 , is found for R > (k) (k) k r = B(r) B(r) = 02 1 1 NUMERICAL RESULTS Figures 1, 2, and 3 show the evolution of a magnetic flux tube for different realizations of the turbulent magnetic field. All of them have the same fluctuation level B=B0 : , and different degrees of anisotropy. In each case, we choose the starting positions of the field lines on a path representing the boundary of a circular flux tube cross section. Then, integrating up to 5000 adjacent field lines, we can devise the evolution of the shape of a flux tube cross section [3, 11, 17]. We argue that the spatial distribution of not too energetic particles injected in a magnetic flux tube will follow the shape of the flux tube, except for finite Larmor radius effects. In Figure 1 a quasi-2D anisotropy is considered, with lz lx ly . In such a case we have relatively large =04 = = 10 = 10 10 410 r=lx/5 r=lx r=lz/5 FIGURE 2. Same Figure as Figure 1. We use here a Slab-like spectrum with lx =lz = ly =lz = 10 for the magnetic turbulence, and the flux tube cross section from bottom to top are plotted at intervals Z = 2Lz (which correspond to 8Lz ). Flux tube on the left, r = 0:05Lx lx =5; flux tube on the right, r = 0:25Lx lx . =( )( FIGURE 3. Same Figure as Figure 1. We use here an anisotropic spectrum with lx =lz = 8 and ly =lz = 1 for the magnetic turbulence, and the flux tube cross section from bottom to top are plotted at intervals Z = 2Lz (which correspond to 8lz ). Flux tube on the left, r = 0:05Lz lz =5; flux tube on the right, r = 0:25Lz lz . )=4 ly = 1, and lz = 1 was Kubo number, R B=B0 lz =lx , and, indeed, a very fast evolution of the magnetic flux tube is obtained. In the left “column”, from bottom top, the initial size of the flux tube, that is the radius r, is much smaller than the turbulence correlation length, r : Lz : lz lz . On the right side of Figure 1, the initial size of the flux tube is comparable to the turbulence correlation length, r : Lz lz . It can be seen that very elongated and ramified structures quickly form. This is in agreement with the magnetic field line evolution reported by Zank et al. [18]. In Figure 2 a quasi-slab anisotropy is considered, with lz lx = ly = . In such a case we have a relatively small Kubo number, R : , and the evolution of the flux tube is slower. Indeed, it can be seen that the flux tube is distorted, but this process is much slower than in the previous case. It appears that the Kubo number also quantifies the speed of ramification of the flux tubes. In Figure 3, an anisotropic spectrum with l x , = 0 05 used, as appropriate to the solar wind case (see [13, 7]). In this case the Kubo number : . In Figure 3, the flux tube has been followed is R for larger distances, up to 32l z , in order to make a direct comparison with the observations in the solar wind. Indeed, considering that at 1 AU from the Sun, the correlation length is about 0.03 AU [7], we can argue that to travel for about 1 AU in the heliosphere (the typical distance where most of the spacescraft take measures) we need to cover about 30 correlation lengths l z Each plot, from bottom to top, is spaced by a distance in the z direction which corresponds to 8lz . At the top of the flux tube cross section of left panel of Figure 3, we can observe filamentation of the curves, which corresponds to small structures in the magnetic field. If we consider that a particular ion composition or energetic particles are injected at the base of the magnetic flux tube, we can understand how sharp gradients can be found in the solar wind. Here two filaments are separed by 2–3lz , which = 0 05 02 = 0 25 = 10 = 10 r=lz = 0 04 =8 411 are the largest structures observable in the plot, even though smaller structures are present. Converted in the solar wind dimension this corresponds to structures of about 0.06–0.09 AU and is consistent with the findings of Giacalone et al. [10]. Clearly the smaller structures will give rise to shorter time variations. On the other hand, the flux tube evolution represented in Figures 1, 2, and 3 shows that this is very sensitive to the anisotropy of the turbulence spectrum. Different kinds of anisotropy give rise to rather different evolution (see also [4, 18]). In particular, the details of the flux tube fine structure can be different for different kind of anisotropies. Hence, we argue that an accurate study of the morphology of impulsive energetic particle events when compared to the simulation results, can give information on the anisotropy of solar wind magnetic turbulence. Indeed, when the Kubo number is large (like for quasi 2-D anisotropy) the evolution of the flux tube is very fast, with the formation of fine scale structures. Consequently, the energetic particle intensity time profile will exhibit many short time peaks, possibly with a hierarchy of durations. (Indeed, a fractal structure is often the result of the evolution of a strongly nonlinear system). On the other hand, when the Kubo number is small (like for slab turbulence, or when l x lz ), the evolution of the flux tube is slow, and only a few, well defined periods of high energetic particle fluxes will be found in the intensity profile. dients. At the same time, fast spreading out, that is fast perpendicular transport, is obtained with respect to the average magnetic field. Further study may help to use the observation of the fine structure in the ion composition or energetic particle fluxes to infer information on the anisotropy of turbulence. ACKNOWLEDGMENTS This work is part of a research programme which is financially supported by the Ministero dell’Università e della Ricerca Scientifica e Tecnologica (MURST), the Agenzia Spaziale Italiana (ASI), contract no. I/R122/01, and the High Performance computing center at the University of Calabria. REFERENCES 1. Rosenbluth, M. N., Sagdeev, R. Z., Taylor, G. B., and Zaslavsky, G. M., Nucl. Fusion, 6, 297 (1966). 2. Jokipii, J. R., and Parker, E. N., Phys. Rev. Lett., 21, 44 (1968). 3. Rechester, A. B., and Rosenbluth, M. N., Phys. Rev. Lett., 40, 38 (1978). 4. Matthaeus, W. H., Gray, P. C., Pontius, Jr., D. H., and Bieber J. W., Phys. Rev. Lett., 75, 2136 (1995). 5. Pommois, P., Zimbardo, G., Veltri, P., Phys. Plasmas, 5, 1288 (1998). 6. Pommois, P., Veltri, P., Zimbardo, G., Phys. Rev. E, 59, 2244 (1999). 7. Pommois, P., Veltri, P., Zimbardo, G., J. Geophys. Res., 106 (A11), 24,965 (2001). 8. Zurbuchen, T. H., Hefti, S., Fisk, L A., Gloecker, G., and Schwadron, N. A., J. Geophys. Res., 105, 18,327 (2000). 9. Mazur, J. E., Mason, G. M., Dwyer, J. R., Giacalone, J., Jokipii, J. R., Stone, E. C., Astrophys. J., 532, L79 (2000). 10. Giacalone, J., Jokipii, J., Mazur, J. E., Astrophys. J., 532, L75 (2000). 11. Isichenko, M. B., Rev. Modern Phys., 64, 961 (1992). 12. Klein, L., Bruno, R., Bavassano, B., Rosenbauer, H., J. Geophys. Res., 98 (A5), 7837 (1993). 13. Carbone, V., Malara, F., Veltri, P., J. Geophys. Res., 100 (A2), 1763 (1995). 14. Bhattacharjee, A., and Ng, C. S., Astrophys. J., 548, 318 (2001). 15. Zimbardo, G., Veltri, P., Pommois, P., Phys. Rev. E, 61, 1940 (2000). 16. Pommois, P., Veltri, P., Zimbardo, G., Phys. Rev. E, 63, 066405 (2001). 17. Isichenko, M. B., Plasma Phys. Control. Fusion, 33, 809 (1991). 18. Zank, G., Matthaeus, W., and Zhou, Y., Nearly Incompressible Hydrodynamics in the Inhomogeneous Solar Wind, this volume. CONCLUSIONS In this paper we have pointed out the different features of normal diffusion due to collisions and of turbulent diffusion due to the convection of a passive tracer by turbulence. We have shown by means of a numerical simulation of the evolution of a magnetic flux tube that fast magnetic field line transport and the formation of small structures in the flux tube cross section go together. We also have found that the rate of small structure formation depends on the anisotropy of turbulence, and in particular on the Kubo number. If we assume that either energetic particles accelerated by a flare, or particular ion compositions are injected from the corona in the solar wind in a given flux tube, we can imagine that the charged particles in their propagation will follow to a good extend the magnetic field lines. Hence, the spatial distribution of particles will be organized as the cross sections of the magnetic flux tubes shown in Figures 1 to 3. As a consequence, when such structure is convected over the spacecraft from the solar wind flow, sudden variations of energetic particles or ion concentration will be seen, corresponding to sharp gra- 412
© Copyright 2026 Paperzz