Presidential Address Turbulent magnet Nigel Weiss recounts his Presidential Address 2001, given to the RAS A&G Ordinary Meeting on 9 February 2001. T he resurgence of interest in cosmical magnetic fields has been driven by a combination of observational advances with rapid progress in theory and interpretation. My chosen subject of magnetic fields in stars like the Sun not only reflects my own interests but also provides an opportunity to show how numerical models, coupled with a deeper understanding of nonlinear processes, help us to explain the features that are being seen. Here I shall focus on small-scale local structures near the surface of the Sun, leaving global magnetic fields – activity cycles, dynamos and so forth – for next year. Fortunately, these topics can largely be decoupled. I shall start by recalling the key properties of our local star, the Sun, and go on to describe some recent observations from space, with Yohkoh, SOHO and TRACE, and from the ground, at the Swedish Vacuum Solar Telescope (SVST) on La Palma. Then I shall discuss theoretical results, starting with simple kinematic models, proceeding to fully nonlinear dynamic models of stratified magnetoconvection and continuing to small-scale turbulent dynamos. Finally, I shall confront these model calculations with the actual observations. The Sun as our star The Sun is the only star on which fine-scale structures can actually be resolved. The image Early sunspot observations Modern observational astronomy began with a technological advance: the invention of the telescope in about 1608. The news spread rapidly and reached Galileo at Padua in June 1609. He used his improved instrument to observe the Moon and the Milky Way and he discovered the satellites of Jupiter. There is no reason to doubt his statement in May 1612 that he had been observing sunspots for 18 months, i.e. since November 1610, or his later claim that he had noticed them earlier that year, before moving from Padua to Florence. Fortunately solar activity was near a maximum and there were plenty of spots to 3.10 ecent high-resolution observations, from the ground and from space, have revealed the fine structure of magnetic features at the surface of the Sun. At the same time, advances in computing power have at last made it possible to develop models of turbulent magnetoconvection that can be related to these observations. The key features of flux emergence and annihilation, as observed by the MDI experiment on SOHO, are reproduced in kinematic calculations, while three-dimensional numerical experiments reveal the dynamical processes that are involved. The pattern of convection depends on the strength of the magnetic field: as the mean field decreases, slender rising plumes give way to a regime where magnetic flux is separated from the motion and then to one where locally intense magnetic fields nestle between broad and vigorously convecting plumes. Moreover, turbulent convection is itself able to act as a small-scale dynamo, generating disordered fields near the solar surface. R from TRACE (the Transition Region and Coronal Explorer) in figure 1 shows slender loops, only a few hundred km across, that follow field lines in the solar atmosphere and are linked to magnetic flux emerging through the solar photosphere. Such features make the Sun a unique laboratory for plasma physics, in parameter regimes that cannot be matched in the laboratory. Observations keep throwing up phenomena that would never have been anticipated had they not been seen. Sunspots themselves, which are dark because they are the sites of strong magnetic fields, remain a classic example. The first telescopic observations were made almost 400 years ago and led immediately to unseemly arguments about priority (see “Early sunspot observations” below). Between then and now there have been enormous increases in resolution, as can be seen by comparing the images in figures 2 and 3, yet we still do not properly understand the origin and stability of isolated flux tubes, or the nature and cause of a sunspot’s filamentary penumbra (Thomas and Weiss 1992). As Eugene Parker puts it, “Nature is cleverer than we are” or, in the words of Blaise Pascal, who was born in Galileo’s lifetime: “Imagination tires sooner of conceiving than Nature does of providing”, (“L’imagination se lassera plutôt de concevoir que la nature de fournir”). At this point we should recall the structure of the Sun (Stix 1989). Energy is generated by thermonuclear processes in its central core, with a radius of 0.2 R , where the solar radius R = 700 000 km. This energy is transported outwards by photons through the radiative zone, to a radius of 0.7 R , where radiative transport gives way to convection. be seen, so other independent discoveries followed rapidly. The first recorded observations are due to Thomas Harriot in England – but his manuscripts, with drawings made in December 1610, were only rediscovered at Alnwick Castle in 1786. So the credit for the first published account (printed in the autumn of 1611) goes to Johann Fabricius (or Goudsmid) from East Friesland, who observed several dark spots in March 1611 and noted that they appeared to rotate with the Sun. Christoph Scheiner, at the Jesuit university in Ingolstadt, had noticed a similar spot just three days earlier, and observed continuously from October to December 1611. His account, published anonymously in January 1612, drew a response from Galileo, who carried out a systematic series of observations from June to August 1612 and published his results the following year, arguing conclusively that the spots were on the Sun and that they rotated with it. From then on there was a fierce dispute between Galileo and Scheiner (neither of whom ever mentioned Fabricius) as to who had first discovered sunspots. Their arguments were aggravated by disagreement over Aristotelian cosmology and the Copernican world system. Scheiner later moved to Rome, where he continued his observations and eventually described them in a sumptuous volume, Rosa Ursina sive Sol, which appeared in June 2001 Vol 42 Presidential Address ic fields in the Sun 1: Fine loops above an active region on the Sun are revealed in this image from TRACE, taken in the extreme ultraviolet. These loops, only a few hundred kilometres across, follow magnetic field lines that emerge from the photosphere and extend upwards into the corona before returning to the solar surface. What we see at the solar photosphere is the upper boundary of this convection zone. We have an enormous wealth of observational results relating to the Sun’s atmosphere (which contains only 10–10 of the solar mass). By contrast, there is still relatively little direct information about the solar interior, though we can exploit helioseismology to probe deeper into the convection zone. Yet the structure of the atmosphere is determined by magnetic fields that are generated within the convection zone and emerge from it through the photosphere. Turbulent convection near the photosphere gives rise to various cellular patterns (Schrijver and Zwaan 2000). The remarkable image in figure 2, obtained at the SVST, shows not only sunspots but also the photospheric granulation, caused by small-scale convective plumes, about 1000 km across, rising and 1630. His dispute with Galileo probably affected the subsequent trial. It took 300 years before a scientific advance revealed the origin of sunspots. The development of spectroscopy and the discovery of the Zeeman effect in 1894 made it possible for George Ellery Hale to measure magnetic fields of several thousand gauss in sunspots. Thus began the study of solar magnetism, which was eventually extended to other stars by Harold and Horace Babcock in the 1950s. The explanation of the darkness of sunspots as being caused by magnetic suppression of convection was put forward by Ludwig Biermann and Thomas Cowling in the 1940s. June 2001 Vol 42 2: This remarkable image of the solar photosphere shows two sunspots and several dark pores. The solar granulation, caused by cellular convection just below the photosphere, is clearly visible and the tiny, bright points are sites of intense magnetic fields. The field shown is about 120 000 km2 and the smallest features are a few hundred km across. The image, in the CH G-band at 4305 Å, was obtained on the Swedish Vacuum Solar Telescope at La Palma by T Berger and G Scharmer on 12 May 1998 and processed at the Lockheed-Martin Advanced Technology Center in Palo Alto. (Courtesy of T Berger.) 3: Early sunspot observations. This figure, from Scheiner’s Rosa Ursina, completed in 1630, shows several sunspots on successive days, as they are carried round by the Sun’s rotation. The distinction between the umbra and penumbra of a sunspot had already been recognized. (Courtesy of the RAS.) 3.11 Presidential Address 4: This image of the Sun, from the Michelson Doppler Imager on SOHO, shows motion towards and away from the Earth as blue- and red-shifted patches, after the axisymmetric differential rotation has been removed. The patches are produced by horizontal motion at the solar surface, corresponding to supergranular cells with characteristic diameters of about 30 000 km. (Courtesy of D Hathaway.) (a) (b) 5: X-ray emission from the hot corona of the Sun maps magnetic fields at the solar surface. (a) A soft X-ray image of the Sun, obtained with the Normal Incidence X-ray Telescope on a rocket flight on 11 September 1989. (b) The corresponding magnetogram, from Kitt Peak National Observatory, shows oppositely directed fields as blue and red against a lighter neutral background. (Courtesy of L Golub.) 3.12 impinging on a stable layer, so that bright, hot features are enclosed by a network of cooler, sinking gas (Berger et al. 1998). The pattern is constantly changing and individual cells have lifetimes of about 10 minutes. Superimposed on this is the supergranulation, a much larger pattern of cells with diameters between 20 000 and 30 000 km (Hagenaar et al. 1997). This is demonstrated by the Doppler image in figure 4, from the MDI instrument on SOHO (the Solar and Heliospheric Orbiter), which shows gas spreading outwards from cell centres and moving towards or away from the observer (Hathaway et al. 2000). Associated with this motion is a magnetic network surrounding the supergranules and made visible by Ca+ emission. There is also evidence for an intermediate cellular scale, the mesogranulation (Shine et al. 2000), which is probably produced by collective interactions between the granular convection cells (Cattaneo et al. 2001a). Magnetic fields on the Sun The outermost part of the solar atmosphere, the corona, is heated magnetically to temperatures of several million degrees and therefore radiates in the ultraviolet and X-ray regions of the spectrum. Its rich and complicated structure is determined by the fields that emerge from the photosphere, as can be seen by comparing the soft X-ray image in figure 5 with the accompanying magnetogram. The magnetogram shows features on several different scales. First, there are the prominent active regions within which sunspots are located, which correspond to bright loops and flaring activity in the corona. Then there is the irregular magnetic network that outlines the supergranules and, finally, there are tiny features distributed across the face of the Sun. The close-up image of magnetic fields in figure 6, obtained from the MDI instrument on SOHO, shows many such features, scattered like pepper and salt across the field of view. The evolution of these fields can be followed from time-sequences of the MDI results (Schrijver et al. 1997, 1998). Small-scale magnetic flux emerges within supergranules as bipolar features (ephemeral active regions), whose footpoints spread apart, while the fields are fragmented to yet finer scales by the granulation. These small flux elements migrate towards the boundaries of the supergranules and merge into the magnetic network, where fields with opposite polarities collide, reconnect and cancel out. The total rate at which unsigned magnetic flux emerges on these scales is significantly greater than that associated with solar activity and sunspots. The ratio of the total unsigned flux in the network to the rate at which it emerges implies that individual flux elements have a June 2001 Vol 42 Presidential Address lifetime of less than a day before they are annihilated (Hagenaar 2001). The interaction between magnetic flux elements and granular convection can be followed in images obtained with the SVST on La Palma and processed digitally at the Lockheed-Martin Solar and Astrophysics Laboratory in Palo Alto. When the solar p-modes – which are so important for helioseismology – are filtered out, using techniques originally developed for observations from space, the individual granules are plainly visible, as can be seen in figures 3 and 7. These images were both taken in the CH G-band at 4305 Å and the tiny, bright features are the sites of intense magnetic fields, with strengths of up to 1500 gauss, so that the magnetic pressure in a flux tube would be close to the external gas pressure, and the flux tube should therefore be almost completely evacuated. It is apparent that these flux elements nestle between the individual granules and then move along the dark intergranular lanes as the cellular pattern evolves (Berger et al. 1995, 1998; Berger and Title 1996). 6: A composite image of intense small-scale magnetic fields that form a “magnetic carpet” at the solar photosphere. Bipolar regions constantly emerge and split into tiny flux elements that are swept into the magnetic network that partially encloses supergranules. Black and white pixels denote oppositely directed fields at high resolution. Extreme ultraviolet emission is represented by the green-shaded background. The field lines correspond to a potential field in the solar atmosphere. (N E Hurlburt.) 7: Close-up of bright points between granules in the photosphere. They correspond to intense magnetic fields in the intergranular lanes, where cooler gas is sinking (but not all strong fields are bright). This G-band image of a region about 60 000 km across was obtained by Göran Scharmer at the SVST on 10 July 1997. (G Scharmer and D Hathaway.) Kinematic modelling of emerging flux The simplest numerical models are twodimensional and purely kinematic, without any dynamical effects. In a perfectly conducting plasma the magnetic field B satisfies the induction equation ∂B = curl u × B (1) ∂t where u is the fluid velocity, and field lines are frozen into the plasma. So if B is vertical at the surface, where the vertical velocity vanishes, ∂B z = –uH . ∇Bz – Bz∇ . uH (2) ∂t where uH is the horizontal velocity and the z-axis points upwards. Thus individual flux elements will move with the flow like passive test particles (or “corks”) and accumulate in regions where the horizontal velocity converges. To model the emergence and fragmentation of ephemeral active regions we need to introduce two families of corks (labelled + and –) to represent oppositely directed fields and then to arrange for these corks to annihilate each other when they collide. Then we can represent the supergranules by axisymmetric sources, thereby generating a Voronoi tessellation of the plane with a network that encloses every source, and allow each supergranule to have a finite lifetime, after which a new source appears (Simon et al. 1995). The results in figure 8 show that such a simple calculation already reproduces the essential features of a statistically steady network, with a balance between flux emergence and cancellation, and a lifetime for flux elements that is compatible with that derived from June 2001 Vol 42 30 40 ×103 km 50 60 256 224 192 160 128 96 64 32 0 8: 2-D kinematic modelling of bipolar flux emergence and the formation of a statistically steady magnetic network. Circles mark the positions of supergranules and boxes and crosses represent small flux elements with oppositely directed magnetic fields. Flux emerges near the centres of supergranules and is swept to their boundaries where it merges with the network. In the network, flux tubes with oppositely directed fields collide and reconnect, to reach a steady state. (Simon et al. 2001.) 3.13 Presidential Address 9: Dynamical modelling of three-dimensional nonlinear magnetoconvection in a wide box, with aspect ratio λ = 8. Narrow plumes in a fully compressible and strongly stratified layer undergo spatially modulated oscillations in the strong field regime (R = 105, Q = 1600). The left-hand panel shows |B| 2 at the top and (reflected) bottom of the layer, with the colour table running from violet to red as the field strength increases. Magnetic flux is swept aside by rising plumes at the upper boundary and drawn into them at the base. The right-hand panel shows the vertical temperature gradient, proportional to the intensity at the upper boundary. Temperature fluctuations are shown on the sidewalls and the arrows represent the tangential components of the velocity on the bounding surfaces. This pattern corresponds to behaviour in the dark nuclei of sunspot umbrae. (See Weiss et al. 2001.) observations (Simon et al. 2001). Despite its success, the limitations of this kinematic approach are, however, obvious. The real magnetic field is a solenoidal vector and three-dimensional effects cannot be ignored. Moreover, passive flux concentration produces fields that are strong enough to affect the motion so that the Lorentz force (j × B) can no longer be neglected. Thus it becomes necessary to carry out fully dynamical, three-dimensional computations – and they are inevitably much more complicated. Modelling 3-D compressible magnetoconvection There are two approaches to modelling astrophysical convection. One is to include as many properties of the real atmosphere as possible – chemical composition, ionization, radiative transfer and so forth – with the aim of comparing the simulation directly with observable features. This has been achieved successfully for photospheric granulation in the absence of magnetic fields (Stein and Nordlund 1998). The alternative method, which I shall pursue, is to investigate complex nonlinear behaviour in an idealized configuration. This approach yields qualitative results that can be used to interpret behaviour that is seen on the Sun but cannot supply a quantitative comparison with observations. The induction equation has to be augmented by adding an equation of motion, 3.14 Du ρ = –∇p + ρg + j × B + Fvisc (3) Dt where Fvisc is the viscous force and other symbols have their usual meanings, together with the continuity equation ∂ρ = –∇ . (ρu) (4) ∂t the perfect gas equation of state, p = RTρ, and an equation for the entropy S: ∂S (5) ρT = –∇ . (k∇T) + Qdiss ∂t where radiative transfer is treated in the diffusive limit and the last term represents the dissipative sources. Moreover, the induction equation has to be extended to include the effects of ohmic dissipation, with a magnetic diffusivity η, so that ∂B = curl u × B + η∇2B (6) ∂t The magnetic Reynolds number Rm = UL/η, where L and U are typical measures of length and velocity respectively, measures the ratio of advection to diffusion. In a star, Rm >> 1: Rm ≈ 106 in the photosphere and Rm ≈ 109 in the deep convective zone, but the currently attainable limit in numerical experiments lies in the range Rm ≈ 100–1000. We have systematically explored the effect of imposing an initially uniform vertical magnetic magnetic field B0 on convection in a strongly stratified layer of fluid contained in a box with square cross-section subject to standard illustrative boundary conditions (Weiss et al. 1996; Rucklidge et al. 2000). To obtain meaningful results it is essential that the aspect ratio λ of the box (its normalized width) should be sufficiently large: the runs presented here have λ = 4 or 8 and consequently require massive computing power. Fortunately, with support from PPARC, we have had access to the Hitachi SR-2201 at the University of Cambridge High Performance Computing Centre. The runs to be described here have required a resolution of up to 256 × 256 × 100 mesh points. The governing equations are solved within the domain, with idealized boundary conditions at the top and bottom of the box and periodic lateral boundary conditions. The thermal forcing is measured by a Rayleigh number R which is proportional to the mean superadiabatic temperature gradient. For the runs illustrated here we fix R = 105, a value high enough to generate turbulent convection in the absence of any magnetic field, and vary the strength of the imposed field B0 , which is measured by a Chandrasekhar number Q ∝ |B0|2 (Weiss et al. 2001). When Q is very large, convection is completely suppressed, but with our choice of parameters convection sets in at Q ≈ 4200. In the magnetically dominated regime, with 4000 > Q > 2000, we obtain steadily (but feebly) convecting solutions with very narrow plumes disposed in a hexagonal pattern. As Q is further decreased the motion gradually grows more vigorous, June 2001 Vol 42 Presidential Address 10: Flux separation in nonlinear magnetoconvection, for exactly the same parameters as the solution in figure 9 but started with different initial conditions. The cluster of broad and vigorous plumes is able locally to expel nearly all the magnetic flux, so that convection is unimpeded. In the rest of the region, where the average field strength is increased, only small-scale convection can occur. The whole pattern, which evolves chaotically, is related to the appearance of bright umbral dots in sunspots (cf. Weiss et al. 2001). 11: Vigorous convection with flux separation (Q = 1000). The box is split into two regions, one with vigorous convection and the other with strong fields and weaker, small-scale convective plumes. The pattern keeps changing but the separation persists indefinitely (cf. Weiss et al. 2001). until the plumes eventually become timedependent, undergoing irregular spatially modulated oscillations, in which adjacent plumes wax and wane aperiodically in strength. This pattern is illustrated in figure 9. The right-hand panel shows a snapshot of temperature fluctuations at the upper boundJune 2001 Vol 42 ary, with an array of slender plumes that vary in strength, as can be seen from the temperature fluctuations on the sidewalls. The lefthand panel shows the field strength, measured by |B|2, at the top and bottom of the layer. As a plume impinges upon the upper boundary there is a radial flow that carries field lines outwards, just as corks were transported in the kinematic calculation. Thus magnetic flux is compressed into a network that encloses the rising plumes, and the field strength is anticorrelated with the temperature. (Conversely, at the base of the layer, the field is concentrated into the centres of the plumes – but that is 3.15 Presidential Address 12: Formation of strong transient magnetic features in intergranular lanes for Q = 200 in a box with λ = 4. Magnetic flux moves rapidly along the intergranular lanes and the most intense magnetic fields appear at corners. This pattern is similar to that suggested by observations of intergranular magnetic fields, as shown in figure 7 (cf. Weiss et al. 2001). a consequence of introducing a rigid boundary, which is not present in a star.) The pattern in figure 9 corresponds to an attractor that is stable but is not unique. With different initial conditions but the same parameter values (Q = 1600) an entirely different pattern appears. Figure 10 illustrates this new phenomenon of flux separation. There is a cluster of broad and vigorous convective plumes from which most of the magnetic flux has been expelled, thereby strengthening magnetic fields in the rest of the region. Outside this cluster the field is therefore strong enough to limit motion into narrow plumes that are relatively ineffectual. The whole pattern is chaotic and the strong plumes migrate slowly through the region. This pattern, a form of phase separation with fronts that separate vigorous convection from strong fields, is characteristic of a strong field regime that exists in our model for 2000 ≥ Q ≥ 1000. For Q ≤ 1400 the narrow plumes become unstable and give way to larger scale convection. Figure 11 shows flux-separated behaviour for Q = 1000 (Tao et al. 1998). There is a constantly changing pattern, with several clusters of broad and vigorous plumes surrounded by magnetically dominated regions of small-scale weak convection. Note also that, although the active rising plumes are enclosed by cooler sinking fluid, the downward motion is rapidly focused into narrow plumes that plunge towards the lower boundary. This is a characteristic feature of convection in a strongly stratified layer 3.16 and has appeared in many turbulent simulations (cf. Spruit et al. 1990). When the imposed field strength is decreased yet further, the motion squeezes magnetic flux into lanes where the field is locally too strong to allow any small-scale motion to occur. Figure 12 shows a close-up (in a box with λ = 4) of a solution for Q = 200. Now there are strong magnetic fields in lanes enclosing clusters of active plumes. The magnetic flux migrates rapidly along these lanes, accumulating preferentially at the corners. Note, however, that there are no individual isolated flux tubes. Rather, magnetic flux at the surface behaves like a fluid that moves in spurts along the network of lanes. There are ephemeral concentrations, giving rise to fields that are locally intense. At such locations the magnetic pressure rises and the gas pressure and density consequently drop to values that correspond to almost complete evacuation. This results in numerical difficulties that set a current limit to this investigation. Turbulent dynamos What happens as the mean imposed field B0 is yet further reduced? In particular, what do we expect to find in the limit when there is no net imposed magnetic flux (B0 = 0)? It is known that even simple velocity fields possessing the property of Lagrangian chaos may act as fast dynamos in the limit as the magnetic Reynolds number Rm → ∞ (Childress and Gilbert 1995). In such flows the stretching of field lines, as neighbouring fluid elements move apart at an exponentially increasing rate, predominates over dissipation at the ever-decreasing scale of the intermittent fields that are produced. Figure 13 illustrates the elegant and complex structure that appears in such a model calculation (Brummell et al. 1995). So we might expect that highly turbulent convection is able to maintain a similarly disordered field. This process has indeed been demonstrated for convection in a Boussinesq (i.e. incompressible) fluid layer. (See the article by Cattaneo and Hughes in this issue, A&G 42 pages 3.18–3.22.) In this model calculation, in a box with λ = 10 and a very high Rayleigh number (R = 5×105), producing a magnetic Reynolds number Rm ≈ 1000, there is a gradual transition from magnetoconvection to dynamo action as B0 is progressively reduced (Cattaneo et al. 2001b). In the limiting case, when B0 = 0, a small seed field grows exponentially until the Lorentz force is able to saturate its growth. Thereafter the small-scale turbulent magnetic field, with zero net flux, is maintained indefinitely (Cattaneo 1999). Figure 7 of Cattaneo and Hughes (on page 3.21 of this issue) shows the structure of this intermittent field, which is strongest locally in isolated regions where the fluid flow converges; note that there is an up–down symmetry in this problem that was not present in the stratified layers of figures 9 to 12. These ambitious calculations (which required up to 1024 × 1024 × 96 gridpoints) confirm that sufJune 2001 Vol 42 Presidential Address 13: Intermittent field structures in fast dynamos. Results for a linear (kinematic) dynamo model with a largescale, chaotic velocity field and a magnetic Reynolds number Rm ≈ 105. The convoluted structure of the magnetic field, with rapidly alternating directions (indicated by black and white against a red background) illustrates the complex structure of the intermittent field. (From Brummell et al. 1995, courtesy of F Cattaneo.) ficiently vigorous convection is indeed able to generate a turbulent magnetic field that can be maintained forever, as opposed to amplifying pre-existing flux that must eventually decay. Relating theory to observations The next challenge is to relate these theoretical models to observational results. As we have seen, the action lies in small-scale structures, and in each case these fine structures have barely been resolved. Moreover, we have to extrapolate from numerical experiments with Rm ≤ 1000 to the solar atmosphere, where Rm ≈ 106. All the same, we can use results from model problems to interpret features that can be observed. We have already seen that the formation of the magnetic network that surrounds the supergranules can be modelled by a straightforward two-dimensional kinematic calculation. Given a total flux of 3×1022 Mx in the network and the observed rate of flux emergence (at 2×1022 Mx hr–1), the emerging flux is replaced after a residence time of only 12–24 hr. This flux appears over the entire surface of the Sun, at sunspot minimum as well as sunspot maximum. Moreover, the ephemeral regions are essentially oriented randomly, with only a weak bias towards the orientation of fields in active regions. That suggests that the emerging flux is not related to that in active regions, which must be formed near the tachocline at the base of the convection zone. Rather, it must owe its origin to small-scale dynamo action near the photosphere. Since the ephemeral regions emerge with fluxes of around 3×1018 Mx and are then split by granular convection into flux June 2001 Vol 42 elements with much smaller fluxes, of order 1017 Mx (Schrijver et al. 1997, 1998), this dynamo must operate at some level below that where the granules themselves are formed. That in turn suggests that the photospheric granulation may itself also be capable of acting as a turbulent dynamo, generating ultrasmall-scale fields. Is that perhaps the origin of the weak intranetwork fields whose existence has frequently been reported? Furthermore, if there are really three different scales of dynamo action, how do these scales interact, and do the larger scales feed the smaller scale activity? The persistence of plage regions, which are associated with distorted granulation and the presence of a significant average field, certainly suggests that their flux is anchored at some deeper level in the convection zone. There is a close correspondence between the observed behaviour of intergranular magnetic fields, as revealed by bright points in the continuum or in G-band images like those in figures 3 and 7, and the strong fields formed in the numerical experiment that was illustrated in figure 12. In each case there are flux concentrations that are locally intense but constantly changing. The bright points only indicate positions where some threshold is exceeded, but in the computations we can follow the continuous distribution of the field. Thus we can confirm that magnetic flux does indeed behave like a “magnetic fluid”. Finally, we come back to sunspots. In the dark central umbra there is a strong magnetic field that is only slightly inclined to the vertical. Nevertheless, the reduced flux of energy has to be carried by convection to just below the visible surface. In such a magnetically dominated regime we should expect to find small-scale time-dependent convection, with slender plumes like those in figure 9, that are concealed beneath a radiative blanket. Although there are indeed dark nuclei with no visible structure within the umbrae of sunspots, there are also bright features known as umbral dots. They seem to be examples of flux separation, which allows isolated bright plumes (like the clump in figure 10) to form and then to penetrate the radiative blanket. In conclusion, I should emphasize that this lecture has been restricted to complex and chaotic behaviour such as we see near the solar surface — the only place where these detailed local fields and velocities can be measured. As one moves on to study larger features it becomes necessary to introduce the effects of rotation, which brings some order out of all the chaos. Investigating deeper structures and large-scale systematic behaviour then leads one to apply dynamo theory to global fields in stars like the Sun. That, however, is a subject for my next address. ● Nigel Weiss FRAS, DAMTP, University of Cambridge, Cambridge CB3 9EW. Acknowledgments: The work I have described rests on collaborations with Sean Blanchflower, Derek Brownjohn, Fausto Cattaneo, David Hughes, Neal Hurlburt, Paul Matthews, Michael Proctor, Alastair Rucklidge, Louis Tao and Steven Tobias (all of whom have been in Cambridge at some stage); I have also benefited enormously from working with George Simon and Alan Title, both primarily observers. This research has relied on continuing support from PPARC. References Berger T et al.1995 ApJ 454 531. Berger T et al. 1998 ApJ 495 973. Berger T and Title A M 1996 ApJ 463 365. Brummell N et al. 1995 Science 269 1370. Cattaneo F 1999 ApJ 515 L39. Cattaneo F et al. 2001a ApJ submitted. Cattaneo F et al. 2001b in preparation. Childress S and Gilbert A D 1995 Stretch, Twist, Fold: the Fast Dynamo, Springer, Berlin. Hagenaar M 2001 in preparation. Hagenaar M et al. 1997 ApJ 481 988. Hathaway D et al. 2000 Solar Physics 193 299. Rucklidge A M et al. 2000 J.Fluid Mech. 419 283. Schrijver C J et al. 1997 ApJ 487 424. Schrijver C J et al. 1998 Nature 394 152. Schrijver C J and Zwaan C 2000 Solar and Stellar Magnetic Activity Cambridge University Press. Shine R A et al. 2000 Sol.Phys. 193 313. Simon G W et al. 1995 ApJ 442 886. Simon G W et al. 2001 ApJ submitted. Spruit H C et al. 1990 Ann. Rev. Astron. Astrophys. 28 26. Stein R and Nordlund Å 1998 ApJ 499 914. Stix M 1989 The Sun Springer, Berlin. Tao L et al. 1998 ApJ 496 L39. Thomas J H and Weiss N O 1992 in Sunspots: Theory and Observations, ed. J H Thomas and N O Weiss, Kluwer, Dordrecht. Weiss N O et al. 1996 MNRAS 283 1153. Weiss N O et al. 2001 in preparation. 3.17
© Copyright 2024 Paperzz