arX

iv:0

705.

1097

v1 [

astr

o-ph

] 8

May

200

7

The effect of the relative orientation between the coronal field and

new emerging flux: I Global Properties

K. Galsgaard1, V. Archontis4, F. Moreno-Insertis2,3 and A. W. Hood4

ABSTRACT

The emergence of magnetic flux from the convection zone into the corona is an importantprocess for the dynamical evolution of the coronal magnetic field. In this paper we extend ourprevious numerical investigations, by looking at the process of flux interaction as an initiallytwisted flux tube emerges into a plane parallel, coronal magnetic field. Significant differencesare found in the dynamical appearance and evolution of the emergence process depending onthe relative orientation between the rising flux system and any preexisting coronal field. Whenthe flux systems are nearly anti-parallel, the experiments show substantial reconnection anddemonstrate clear signatures of a high temperature plasma located in the high velocity outflowregions extending from the reconnection region. However, the cases that have a more parallelorientation of the flux systems show very limited reconnection and none of the associated features.Despite the very different amount of reconnection between the two flux systems, it is found thatthe emerging flux that is still connected to the original tube, reaches the same height as a functionof time. As a compensation for the loss of tube flux, a clear difference is found in the extent of theemerging loop in the direction perpendicular to the main axis of the initial flux tube. Increasingamounts of magnetic reconnection decrease the volume, which confines the remaining tube flux.

Subject headings: Sun: magnetic fields – Numerical experiments – Sun: active regions – Sun: corona

1. Introduction

Flux emergence is one of the manifestationsof the continuously changing solar magnetic field.In this process magnetic flux is transported upthrough the convection zone presumably by acombination of buoyancy and convection towardsthe photosphere. At the photosphere the phys-ical structure of the sun changes, from a con-vectively unstable to a convectively stable atmo-sphere, making the continued rise of magnetic flux,due to buoyancy, more difficult. Despite this, nu-merous observations in all wavelength ranges showthat the emergence process is a frequently occur-

1Niels Bohr Institute, Julie Maries vej 30, 2100 Copen-hagen Ø, Denmark

2Instituto de Astrofisica de Canarias (IAC), Via Lacteas/n, 38200 La Laguna (Tenerife), Spain

3Department of Astrophysics, Faculty of Physics, Uni-versidad de La Laguna, 38200 La Laguna (Tenerife), Spain

4School of Mathematics and Statistics, University of StAndrews, North Haugh, St Andrews, Fife KY16 9SS, UK

ring event in the solar atmosphere.

There are numerous observations of the pho-tosphere and lower transition region (Lites et al.1995, 1998; Strouse & Zwaan 1999; Kubo et al.2003; Pariat et al. 2004; Lites 2005). These show apicture whereby emerging magnetic fields changethe convective flows, allowing for initially horizon-tal magnetic fields to penetrate the photosphere.As the magnetic field expands into the transitionregion and lower corona, relatively cold plasma islifted by the magnetic field and eventually startsdraining back towards the photosphere, along themagnetic field lines. In isolated emergence eventstwo strong opposite polarities arise forming twomagnetically connected sunspots. In the regionbetween the two flux concentrations new flux con-tinues to emerge, with new positive - negativepairs arising at different positions. Lites et al.(1995) used observations covering temperaturesfrom photosphere to corona plasma to argue thata full coherent flux tube rises from the convection

1

zone. Strouse & Zwaan (1999) suggested that theflux emerges in ”parallel” sheets between the twomajor flux concentrations, and that the randomappearance of new small flux concentrations be-tween the two sunspots favor a picture wherebyundulating field lines emerge at different locationsalong the full length of the field line connectingthe two sunspots. Pariat et al. (2004) uses fieldextrapolation of the photospheric field to estab-lish a field line structure of the magnetic field andfind that undulating field lines exist in this model.Furthermore, the dips are associated with Eller-man bombs (Georgoulis et al. 2002), and they in-terpret these as being responsible for leaving densematerial in the part of the loop that dips belowthe photosphere, making the emergence for the re-maining upper part of the loop much easier.

The manifestation of flux emergence in the so-lar corona often results in a strong interaction be-tween the two initially disconnected flux systems.This interaction results in local brightenings ob-served in all wavelengths from white light throughEUV and into X-ray. Often these events are as-sociated with high-speed flows of the hot plasmaemerging from the (reconnection) region where thetwo flux systems interact. Longcope et al. (2005)investigated one such event. Using simple flux es-timates for the flaring structures, they showed howthe new flux regions changed its connectivity withtime and became connected to the existing coronalmagnetic field. This happened in such a way thatno dynamic activity took place for a initial periodof time, followed by a rapid change of connectivity,and finally a more quiet phase.

How does this interaction depend on the struc-ture of the two initially separate flux systems? Noclear analysis has been made on existing observa-tional data, and one can ask if this is actually pos-sible, with the present inability of directly trackingmagnetic field lines and their photospheric connec-tions. Further to this, each observed event is dif-ferent from previous ones in a number of ways, sohow can one quantify the most important reasonsfor these differences in evolution?

Such effects are much easier to investigate byundertaking a series of numerical experimentswhere one can control, in detail, the environmentinto which the flux emergence occurs. In previousexperiments (Fan 2001; Magara & Longcope 2003;Archontis et al. 2004, 2005; Manchester et al.

2004; Magara 2006), the rise of a twisted loop isinitiated inside the convection zone and followedin a self-consistent manner. In these investiga-tions, the structure of the emerging flux resemblesa situation where the outer layers of the flux tubeexpand into the corona. In a manner, this resem-bles the process of peeling off the outer layers of anonion and expanding these parts into the corona.Specifically Magara (2006), compared the devel-opment of the photospheric flux concentrationswith observations, finding the same characteris-tic evolution of the flux concentrations and theirassociated inversion polarity line as seen in obser-vations. Galsgaard et al. (2005); Archontis et al.(2005) and Isobe et al. (2005) presented the first3D MHD experiments in which a twisted flux tubeemerges from the convection zone and interactswith a pre-existing coronal magnetic field.

Archontis et al. (2005) analyzed in detail theinteraction between an emerging magnetic fluxsystem and a uniform horizontal coronal magneticfield, using for the rising tube the same initialconditions as in previous experiments (e.g. Fan(2001); Archontis et al. (2004)). Archontis et al.(2005) show how the sub-photospheric flux tubeemerges into the corona and pushes the magneticfield upward and outward. Given the initial al-most antiparallel, mutual orientation of the sys-tem at the time of first contact, a strong currentsheet is formed at the interface. The interactionof the two flux systems then follows a complicatedpattern that slowly changes in time. This inter-action depends also on the relative orientation be-tween the two systems, as has been pointed out byGalsgaard et al. (2005).

In a different approach, Fan & Gibson (2004)used a 2D magnetic arcade field embedded in aconstant temperature coronal region, into whichthey force a pre-defined curved loop structureusing an imposed boundary electric field. Thisshowed, for the setup with minimal reconnectionbetween the two flux systems, that the twistedemerging loop entered into the ”corona”, even-tually experienced a kink instability and strongcurrents were generated where reconnection takesplace due to non-ideal effects. The emergenceof a twisted loop into the corona is supportedby Lites et al. (1995). Therefore it is interest-ing to compare the structure of current concen-trations between Fan & Gibson (2004) and exper-

2

iments which included the full emergence process,knowing that differences in the coronal structureand the emergence process may provide very dif-ferent dynamical evolutions.

In this paper, we follow up on the work pre-sented in Galsgaard et al. (2005) and Archontis et al.(2005) and investigate, in more detail, the signif-icant deviations in the emergence process as theorientation between the two interacting magneticflux systems is changed. In Section 2 we brieflydescribe the numerical model. Sections 3-6 con-tain the various results of the experiments andSection 7 is a discussion of the implications ofthese results. Finally, highlights of the results aresummarized in Section 8.

2. Model setup and numerical approach

The parameters of the magnetic flux tube, thebackground stratification and the initial condi-tions follow the work presented in Galsgaard et al.(2005) and Archontis et al. (2005). As a referencefor the following discussion we start by summa-rizing the initial state of the experiments and thenumerical approach.

Our model consists of a highly stratified en-vironment and a horizontally twisted magnetictube. The backgroundmedium consists of an adia-batically stratified convection zone, an isothermallayer representing the photosphere, a region wherethe temperature steeply increases with height andrepresents the transition region and finally anisothermal layer with coronal temperatures. Thetube center is located almost 2 Mm below the baseof the photosphere. The longitudinal componentof the magnetic tube has a Gaussian profile with acentral field strength of 3.8 kG, while the twist isuniform around the axis of the tube. This particu-lar flux tube is stable towards the kink instability.This gives a plasma β = 12.8 at the axis of thetube. The rise of the tube is triggered by imple-menting a density deficit distribution that has amaximum at the middle of the axis of the tubeand with a Gaussian distribution along the tube.Initially, a horizontal magnetic field is included inthe atmosphere above the lower transition region.The orientation of the ambient field relative to themain axis of the tube is an important parameterin the experiments.

Figure 1 shows the gas pressure, temperature

Fig. 1.— Distribution of gas pressure (thicksolid), density (dashed), temperature (dash-dotted) and magnetic pressure (thin solid) alongthe central, vertical (y = 0,x = 0) line.

and density of the stratified environment as a func-tion of height. All the profiles are normalized ac-cording to the photospheric values: pph = 1.4 105

erg cm−3; ρph = 3 10−7 g cm−3; Tph = 5.6 103

K and Hph = 170 km. Other units used in thesimulations are: time, tph = 25 sec; velocity,

V ≡ (pph/ρph)1/2 = 6.8 km sec−1 and magnetic

field, Bph = 1.3 103 Gauss.

The distribution of the magnetic pressure inFig. 1 shows the magnetic flux tube and the am-bient field. The ambient field is given by

Bcor = Bc(z) [cos(φ), sin(φ), 0], (1)

where Bc(z) is described by an hyperbolic tangentprofile. The intensity of the coronal field is chosensuch that the local plasma β is close to 0.06.

The direction of the initial ambient field is givenby the polar angle φ, which is measured in a hor-izontal xy-plane from the positive x-axis. At thesame time the magnetic fieldlines at the top ofthe rising flux system are oriented in an approxi-mately antiparallel direction to the ambient field-lines when φ = 0 and they are almost parallelwhen φ = 180 (see Fig. 2). Thus, the relative hor-izontal angle between the two flux systems uponcontact is φ0 = 180 − φ deg. The polar angle, φ,of the coronal magnetic field is different in the fiveexperiments we have performed; it changes fromφ = 0 to φ = 180 (see Table (1)).

3

Z

Y

XZ

X

Y

Z

X

Y

Fig. 2.— Top view of an emerging twisted fieldline and the ambient coronal field (long horizontal arrows).The short arrow at the center of the box shows the direction of the emerging fieldline. The three panelsrepresent different relative orientations of the two flux systems: experiment A: φ = 0, experiment B: φ = 45and experiment C: φ = 90.

Experiment φ0 Line style OrientationA 180. Triple Dotted dashed AntiparallelB 135. Dot-Dashed SlantedC 90. Full PerpendicularD 45. Long-Dashed SlantedE 0. Dotted Parallel

Table 1: List of the numerical experiments performed in the simulations. The first column shows the referencenames of the experiments. The second column shows the relative horizontal angle of the two magneticsystems. The third column indicates the linestyle used in the different plots in this paper and the fourthcolumn the orientation of the coronal field relative to the upcoming field.

The evolution of the system is governed bythe three-dimensional, time-dependent and resis-tive MHD equations. These are solved using anumerical approach based on high order finite dif-ferencing on staggered grids. By using 6 neigh-boring data points, a 6th order accurate spatialderivatives and corresponding 5th order accurateinterpolations routines are used. The solutionis advanced in time using a 3rd order predictor-correction algorithm. Due to the high spatial or-der, special treatment of viscosity and resistivityare required to prevent numerical ringing in thevicinity of steep gradients in the physical quanti-ties. This is handled by a combined approach, thatis designed to remove numerical problems that canoccur in specific problems. These approaches arelocalized in space, implying that dissipation onlytake place over length scales of a few gridpoints.Using such an approach, makes it impossible to

assign a single characteristic Reynolds number forthe experiment (Nordlund & Galsgaard 1997).

The numerical resolution of the experiments is(148, 160, 218) in the (x, y, z) directions, with z be-ing the height. The size of the numerical domain is(−60, 60), (−70, 70) and (−22, 70), which is equiv-alent to a box of sides 20.4 Mm x 23.8 Mm x 15.6Mm. The resolution in the x and y directions is137.8 km/cell and 148.8 km/cell correspondingly.The grid in the vertical direction is stretched ina way that the highest resolution covers the re-gion from the top of convection zone to the bot-tom of the corona. Here the grid resolution is 47.7km/cell. The resolution has lower values close tothe top and bottom boundary of the numericaldomain.

4

3. Current sheet orientation

The location, orientation and strength of cur-rent sheets in 3D is vital for providing an environ-ment for fast magnetic energy release. This sec-tion is concerned with investigating the buildupof current sheets in the various experiments, withthe aim to study the relation between some basicmodel parameters and locations of reconnection.

As the emerging flux pushes its way into thecoronal magnetic field, stress builds up at the in-terface between the two flux systems. When thetwo flux systems are antiparallel (φ0 = 180), a cur-rent concentration is formed all over the emergingplasma hill. As the stress continues to build up thecurrent is concentrated into a narrow curved sheetthat reaches from the summit point of the plasmahill down its sides towards the photosphere al-most along the direction of the underlying emerg-ing flux tube in the y-direction. As the orientationof the coronal magnetic field changes in the dif-ferent experiments, the orientation and strengthof the current sheet changes too. The currentsheet is found to rotate around its vertical cen-tral axis as a monotonic function of the angleφ0 between the coronal magnetic field and theemerging magnetic field (see also Fig. 3), wherecosφ0 = (Bcor ·Bt)/|Bcor||Bt|, Bt represents thetube field at the summit point of the emergenceregion.

Fig. 3.— The orientation of the current sheet withrespect to the vertical yz midplane as a function ofthe angle between the two magnetic flux systems.

The orientation of the current sheet can befound using the following analysis. Assume we areonly interested in the orientation at the summit

point of the emergence, then the z component ofthe field can be ignored. In general, the field vec-tors can be expressed in terms of an orthogonal co-ordinate system, with one unit vector, e1, definedby the direction given by the sum of the magneticvectors, Bcor + Bt and the other unit vector, e2,orthogonal to this defining a right hand system.Thus,

Bcor = Bcor‖e1 +Bcor⊥e2, (2)

Bt = Bt‖e1 +Bt⊥e2 (3)

The components of Bcor and Bt along e1, namelyBcor‖ and Bt‖, represent the components of themagnetic field that cannot be annihilated and atthe same time determine the direction of the mainaxis of the current sheet. The components alonge2, Bcor⊥ and Bt⊥, provide two oppositely di-rected components that can annihilate in a recon-nection process. The integrated current across thesheet is simply given by |Bcor⊥| + |Bt⊥|. Visual-ization of the current sheet and illustration of itsorientation in the computational volume is shownin Section 6.

3.1. Magnetic Pressure Balance Across

the Current Sheet

In 2D the total pressure balance across a cur-rent sheet is simple, and it maintains a changebetween magnetic and gas pressure. In 3D such abalance may change with time as the relative ori-entation of the field that is advected into the sheetchanges. In this paper, the evolution of the totalpressure balance is used as an indirect indicator ofmagnetic reconnection.

As the tube rises the current sheet moves up-ward covering the upper part of the buoyant fluxsystem. Archontis et al. (2005) showed that themagnetic pressure distribution across the currentsheet changes with time in experiment A. Thischange in the temporal evolution of the magneticpressure was found to play an important role inthe reconnection process between the two flux sys-tems.

Figure 4 shows the pressure of the horizon-tal component of the magnetic field (Bhor) below(solid line, |Bt|) and above the current sheet (dot-dashed line, |Bcor|) as a function of time. Thedashed line represents the pressure from the Bx

component of the magnetic field below the current

5

Fig. 4.— Temporal evolution of the magnetic pressure just above and below the current sheet are shownfor experiments A, B, C and D from top-left to bottom-right panel. Full line represents the pressure belowthe sheet, the dot dashed line indicates the pressure above the sheet. Finally the dashed line indicate thepressure from the magnetic component perpendicular to the tube direction just below the sheet.

sheet.

At the beginning of the emergence process themagnetic pressure inside the rising plasma is muchlarger than that of the ambient coronal field. Att ≈ 60 the difference is more than one orderof magnitude. This pressure excess pushes theemerging tube upwards into the atmosphere - fordetails see the discussion of forces given in Sec-tion 5.2. After t ≈ 80 a transverse balance ofmagnetic pressure is achieved, independently ofthe different orientation of the ambient field, andthis balance remains until the end of the simula-tion.

Before t ≈ 95 the Bx and the Bhor of the ris-ing magnetic field in the tube, are approximatelyequal. For experiments A and B the two compo-nents separate around this time, indicating thatthe orientation of the emerging magnetic field just

below the interface changes with time. The reasonbeing that, as time proceeds, the uppermost fieldlines of the rising plasma reconnect with the ambi-ent field and this allows for different internal fluxlayers to come into contact with the overlying fluxsystem; the magnetic field vector in these internallayers points increasingly away from the transversedirection and this explains the decrease apparentin the dashed curve in experiments A and B. Onthe other hand, for experiment C and D the re-connection affects a much shallower region of therising tube, and, as a result, the x-component ofthe field as the interface does not decrease.

Finally, it is found that pressure at the end ofthe experiments, t = 120, has an almost linear de-pendence on the value of φ, providing the highestmagnetic pressure for the cases that do not showeffective reconnection in Fig. 4. The effects andimportance of reconnection is further discussed be-

6

low.

4. Magnetic Connectivity

In the previous section we showed that theemerging flux tube reconnects with the coronalmagnetic field, at least for the cases A - C. Asa result of this the magnetic pressure below thecurrent sheet changes with the relative horizontalangle φ0. Thus, it seems plausible that the emer-gence process will be strongly influenced by theorientation of the coronal field. For example, whenthe two fields are approximately parallel one mightexpect the reconnection process to be slowed sub-stantially and the emergence process possibly hin-dered. This section investigates the efficiency ofreconnection, from a global point of view, by

• measuring the height of the apex and axis ofthe emerging tube in time (Section 4.1),

• measuring the amount of horizontal and nor-mal flux that emerges into the corona (Sec-tion 4.2),

• studying the changes in field line connectiv-ity (Section 4.3), and finally by

• measuring the fraction of the tube flux thatreconnects in time (Section 4.4).

4.1. Height-time relation: apex and axis

To measure the dependence of the rising mo-tion of the tube on the orientation of the ambientfield, we find the height of the apex and the axisof the rising tube as a function of time for theexperiments listed in Table 1. More precisely, alarge number of field lines are traced from start-ing points along the central vertical line. Thenwe find those fieldlines that stay in the tube andthose that belong to the ambient magnetic field.The summit point is then the first point along thecentral line at which the connectivity changes.

Fig. 5 (left panel) shows the height-time re-lation of the apex and the center of the emerg-ing tube. The emergence starts with a slow risephase while the flux tube is below the photosphere(Murray et. al. (2006) investigate how this initialemergence phases depend on the tube parame-ters). This is followed by a rapid rise phase be-tween t = 55 and t = 80 during which the apex

Fig. 5.— Left: The height of the apex (uppercurves) and the center (lower curves) of the tube asa function of time. Right: The associated velocityof the apex of the rising system. For line styles seeTable 1.

of the tube rises through the transition region andinto the corona. After t = 80 the rise rate slowsdown and settles down to a lower rate that fluc-tuates with time and between the various experi-ments, but on average follows the same trend. Thechange in the height-time relation around t = 75corresponds to the time at which magnetic pres-sure balance across the current sheet is achieved,as seen in Section 3.1. At this time the width ofthe current sheet shrinks to the numerical resolu-tion limit.

The lower curves in the top panel of Figure 5show that the axis of the tube reaches the photo-sphere at t ≈ 45 and remains close to the lowerregion of the photospheric layer until the end ofthe simulation. A similar result is found in pre-vious experiments of flux emergence (Fan 2001;Magara & Longcope 2003; Archontis et al. 2004;Manchester et al. 2004).

Finally, the right panel of Fig. 5 shows the risevelocity of the apex of the tube. It is seen that

7

all experiments follow the same evolution untilt ≈ 75, after which a general decrease with super-imposed oscillations are seen in the rise velocity.

4.2. Emerging flux

Fig. 6.— The graphs show the fraction of fluxbelow z = 1.2 Mm as a function of time for thefive experiments mentioned in Table (1). Almost65% of the flux has emerged into the corona att=120 for experiment A.

Another global measurement of the amount ofthe emerging flux in time can be obtained by cal-culating the amount of new horizontal flux passingthrough the vertical midplane (x-z plane locatedat y = 0) and above a height of z = 1.2Mm. Thisparticular reference height is chosen to be abovethe initial flux tube and below the initial coronalmagnetic field. In this way, the initial coronal flux,independent of its orientation, does not contributeto the measurement.

Alternatively, we can measure the amount ofhorizontal flux that remains below the height ofz = 1.2Mm in time. This amount of flux is definedby:

Φ(t) =

∫ Lx

−Lx

∫ 1.2

−Lz

By(x, 0, z, t)dzdx. (4)

Fig. 6 shows Φ(t), normalized by its value at t = 0,as a function of time for the five experiments. Theprofiles for the five experiments are almost identi-cal until t = 90. This result shows that the amountof flux that emerges into the upper atmosphere isalmost independent of the relative horizontal an-gle, φ0. After t = 90, we find that the larger theinitial angle, φ0, is between the emerging flux and

the coronal magnetic field, the larger is the frac-tion of the magnetic flux that emerges into thecoronal regime at a given time. However, the dif-ference in emerged flux between experiment A andexperiment E is found to be small, close to 7%, atthe end of the experiment, at which time about60% of the initial flux has emerged into the outeratmosphere.

Observationally Kubo et al. (2003); Spadaro et al.(2004) and Zuccarello et al. (2005) have measuredthe development of the normal flux representedby the emerging region. They find a time depen-dent growth, that over the first few days is closeto linear. This is followed by a saturation of fluxand eventually a decrease. The grow rates of themagnetic flux in the three cases are naturally dif-ferent. One striking difference with the numericalexperiments is the time scale involved. In theobservations the timescale of the evolution of thesystem is measured in days, while the experimenthere only covers about 20 minutes. Therefore itwill not be possible to make a real comparison,but it is interesting to compare the structure ofthe comparable numerical measurement of theemerged magnetic flux.

If one assumes the emerging region is at the diskcenter, then the observed quantity is equivalentto integrating the positive(/negative) flux repre-sented by the Bz component in our experiments:

Φ(t) =

∫ Lx

−Lx

∫ Ly

−Ly

Bz(x, y, z, t)dydx. (5)

Fig. 7 shows the time evolution of the positivevertical flux through the transition region (z =1.2Mm). The plot shows two different phases inthe evolution: first a near linear increase of thevertical flux until about t = 75, followed by an-other phase with a continuously decreasing rate.The difference in the time evolution of the verti-cal flux between the different experiments is verysmall.

Despite significant differences in the timescalesbetween observations and these experiments, thebasic structure of the emerging flux appears verysimilar.

4.3. Field line connectivity

The magnetic energy released in reconnectionevents is bound to spread along the reconnected

8

field lines due to anisotropic heat conduction. Thebright structures found in EUV and X-ray obser-vations in association with flux emergence, andother energy release events, are therefore provid-ing vital information about the field line connec-tivity in the corona. Comparing the coronal struc-tures with models provides us with a possibility ofunderstanding the field line structure of dynamicalevents.

There are several ways to show, in a qualitativemanner, how the fieldline connectivity changes intime for the various experiments. We choose totrace field lines starting from one end of the sub-merged tube and see if they either connect to theother end of the tube or to the corona. A disk isselected at the tube end, centered on the initialtube axis, and the destination of a large numberof field lines is determined. Field lines going fromthe one end of the tube to the other are coloredgrey and field lines connecting to the corona arecolored black. This method indicates the globalconnectivity of the fieldlines between the two fluxsystems. Details of the method can be found inParnell et al. (2004).

Figure 8 consists of five columns. Each columncorresponds to a different experiment and showshow the field lines, which have been traced frominside the selected disk, change their connectiv-ity in time. The left column, for example, showsthe connectivity for experiment A. The four pan-els in the left column show that the connectivitychanges first at the outer layer of the disk andthen moves toward the center following a swirling

Fig. 7.— Time evolution of the normal flux of theemergence through the z = 1.2 Mm plane for thefive experiments. For line styles see Table (1).

motion. This is because the fieldlines which aretraced from the outer periphery of the disk, reachhigher levels in the atmosphere and reconnect firstwith the ambient field. At the end of the simula-tion only few fieldlines, which are found in a shortdistance around the center of the disk, have notchanged connectivity yet. All the other fieldlineshave already been reconnected with the ambientfield.

We also find that at each time the numberof fieldlines that do not change connectivity issmaller, as the relative horizontal angle betweenthe two flux systems increases from φ0 = 0 (rightcolumn) to φ0 = 180 (left column). As a result,there are very few fieldlines that have been recon-nected at the end of the simulation for experimentE, whereas most of the fieldlines inside the diskhave change their connectivity in experiment A.

It has to be noticed that the disk used for trac-ing the field lines is not changed in time, implyingthat the starting points do not represent exactlythe same footpoints in time. Investigating the driftvelocity of the flux pattern it is found to be far tosmall to account for the change in connectivity be-tween t = 60 and t = 80. A closer examination ofthe field line structure shows that field lines at thetop of the reconnection sheet continuesly changesconnectivity (as shown in Archontis et al. (2005)),being slowly pushed towards the flanks of the dif-fusion sheet. Here they re close and again becomepart of the flux tube. In other words, flux fromfootpoints at side of the emerging flux region isinvolved in reconnection processes more than onesalready in the early phase of the emerging process.

4.4. Flux connectivity

To further illustrate the results obtained in theprevious section, we calculate the amount of fluxthat remains in the tube and does not reconnect,normalized to the total flux within the disk, as afunction of time.

Φ(t) =

∫black areaBy(x, l, z, t)dxdz∫

diskBy(x, l, z, t)dxdz. (6)

Φ(t) is shown in the top panel of Fig. 9 for the fiveexperiments in Table (1).

Flux emergence through the photosphere startsat around t = 60 and reconnection between thetwo flux systems starts shortly after this for most

9

Fig. 8.— Connectivity plots for the five experiments. The five columns represent the A, B, C, D and E caserespectively. The rows show the connectivity at times close to 60, 80, 100 and 120. The disks show the areain the tube at the y−boundary from where the fieldlines are initially traced.

of the experiments. A measure of the reconnectionrate is given by the absolute value of the slope ofthe lines in the left frame of Fig. 9. This showsthat there is a short initial phase where the re-connection rate builds up, followed by a periodof time where the reconnection proceeds with dif-ferent, but almost constant, rates in all experi-ments. After t ≈ 100 the reconnection rate de-creases for the three fastest reconnecting experi-ments, to a lower level that is roughly maintaineduntil the end of the experiments. The reconnec-tion rates are simply given by the gradient of theconnected flux fraction. As already stated ear-lier, an unknown amount of flux reconnects morethan once, implying that global-double-separatorbifurcations may take place, Haynes et al. (2006),through which recycling of the flux may occur,Paranell et al. (2006). The estimates of the re-connection rate are, therefore, only providing aminimum value and cannot be used, to estimatethe reconnection speed in the reconnection pro-cesses. This implies that a quantitative compari-son with Longcope et al. (2005) results is not pos-

sible. Longcope et al. (2005) used TRACE andMDI observations to estimate the energy transferbetween a new emerging region and old coronalmagnetic field. Their results showed that recon-nection was not active for a long initial phase, afterwhich a large fraction of the emerging flux con-nected to the coronal field over a relative shorttimescale. In our simulations, we find a moresmooth increase in the flux interaction, with aclear leveling off towards the end of the experi-ment. Hidden in this may well be a significantrestructuring of the field that is not apparent dueto limitations in following the connectivity of in-dividual flux concentrations.

Despite this, the graphs still provide informa-tion about the general development of the exper-iments. The right frame of Fig. 9 indicates thatexperiment A has the highest rate of reconnectionand that the reconnection rate decreases as theangle between the two flux systems becomes lessfavorable and the magnitude of the reconnectingfield component decreases. This fact is unlikely

10

Fig. 9.— The left panel shows how the flux con-nectivity, as defined in Eq. (6), changes with timefor the different experiments. The right panel rep-resents the reconnection rate of the experimentsby simply estimating the gradient of the curves inthe left panel. The line style is given in Table 1.

to be changed since the multiple reconnection de-pends on field lines already having changed con-nectivity once. Thus, the amount of reconnectedflux in experiment A increases up to 65% by theend of the simulation while it remains close to zerowhen the two flux systems are parallel. Finally,we find that the amount of flux that remains inthe tube at the end of the experiments (t=120)decreases nearly as 1 − 0.65 sin(φ0/2). In otherwords, it scales with the orientation, and there-fore the strength, of the current sheet.

5. Dynamics of emergence

In the previous section we showed that theheight-time relation of the apex of the tube issimilar in all experiments. It is also found thatthe amount of flux that emerges through a certainheight as a function of time does not depend onthe orientation of the ambient field. However, the

Fig. 10.— Magnetic pressure (thin lines) and gaspressure (thick lines) distribution along height forthe experiments A and D along the central verticalline. Line style are given in Table (1)

amount of flux that changes connectivity betweenthe emerging flux and the coronal magnetic field,depends critically on the relative orientation of thetwo flux systems.

In the experiments where the rate of change ofconnectivity is high, the overlying coronal mag-netic field is constantly removed by the reconnec-

11

tion process. Thus, the volume above the risingtube is opened and the buoyant system can makeits way up into the upper atmosphere. On theother hand, in the experiments with very little re-connection, the coronal magnetic field is not easilyremoved but instead is pushed upwards and keepsup resistance to the rising motion of the tube (thefieldline topology at the top part of the emergingtube is illustrated in Section 6). Thus, one mayask why the rising motion of the tube is not influ-enced by the change of connectivity in the differentexperiments.

In fact, the height-time relation of the apex ofthe tube indicates that the process of flux emer-gence is predominantly governed by the dynam-ics of the rising magnetized plasma. Thus, in thefollowing sections, first we consider the gas andmagnetic pressure distribution along height, in-side the expanding rising volume and across thecurrent sheet, for two experiments with differentinitial relative angle (Section 5.1); then we studythe temporal evolution of forces that act on theupper part of the buoyant tube (Section 5.2).

5.1. Pressure distribution

It has been shown, (Fig.5 in Archontis et al.(2005)), that the three-dimensional current sheetwhich is formed between the two flux systems inexperiment A, is the location of a rapid changein the direction of the magnetic field. In the earlyphase of the evolution of the system the total mag-netic field vector goes through a tangential discon-tinuity across the current sheet with a clear mini-mum at the center of the sheet. As time proceedsthe direction of the field changes smoothly acrossthe interface of the two flux systems following arotational-like discontinuity.

Figure 10 shows the gas pressure and mag-netic pressure along the central vertical line forthe experiments A and D at t = 60, 70 and 100.The first panel of Fig. 10 (t = 60) shows that inboth experiments the magnetic pressure is higherthan the gas pressure inside the expanding vol-ume (5 < z < 10) by almost two orders of magni-tude. The plasma β in this region is therefore verylow. The magnetic pressure decreases across theinterface between the two flux systems and has aminimum value inside the current sheet. This ismost easily seen in the top left panel of Fig. 10 forthe triple dotted dashed line. This position corre-

sponds to the pronounced minimum of the mag-netic pressure occurring at the position of maxi-mum electric current, that is due to the tangentialdiscontinuity across the sheet. At the same timethe total pressure has a smooth change over thecurrent sheet. This implies that the plasma β in-creases in the current sheet and becomes largerthan unity. At this early stage of the experimentreconnection at the top of the rising tube has notstarted yet and thus the pressure distribution isalmost identical in the two experiments.

The top right panel of Fig. 10 shows the pres-sure distribution when reconnection occurs be-tween the tube and the ambient field. In experi-ment A, the magnetic pressure still goes througha sharp minimum, although the value of the min-imum is higher than at t=60. In experiment D,instead, then magnetic pressure has a smooth dis-tribution across the interface, with no minimum.In either case, the gas pressure supplements themagnetic pressure across the interface, so that thetotal pressure distribution has no extrema here.The reason for the different behavior of the mag-netic pressure profiles for the two experimentsis the presence of an important non-zero, non-reconnecting field component in the current sheetin experiment D which is nearly absent in experi-ment A.

Finally, the bottom panel shows the distribu-tion at a later time, at t = 100. Now, the direc-tion of the total magnetic field vector in experi-ment A changes smoothly across the current sheetfollowing a highly compressed rotational discon-tinuity. Thus, there is a finite non-reconnectingmagnetic component in the current sheet and themagnetic pressure there becomes higher comparedto the magnetic pressure at t = 70. The gas pres-sure in the top of the rising system has decreasedbecause the dense plasma which was carried up-wards has been reconnected. In experiment D,the gas pressure and the magnetic pressure do notchange dramatically compared to the same distri-bution at earlier times. Thus, the magnetic pres-sure inside the expanding volume in experimentD now is much higher than the magnetic pressurein experiment A. At all times, the magnetic pres-sure exceeds the gas pressure and dense materialis lifted up against gravity.

By looking at the pressure profiles with heightin the bottom panel of Fig. 10, one notice that

12

in experiments A the current sheet region experi-ences a pressure minimum. This is a consequenceof the Bernoulli effect associated with the lateralemission of reconnection outflows from the diffu-sion region: the gas pressure depletion in the re-gion helps accelerating the plasma from above andbelow into the the current sheet, thereby bringingnew magnetic flux in to the reconnection site.

From these plots it is also found that the to-tal pressure increases with time in experiment D.This is a pileup effect as the upper boundary isclosed and, thus, the growing excess pressure can-not propagate through the top boundary.

5.2. Forces

In this section, we focus our attention on theforces that act on the upper part of the expandingrising volume, just below the current sheet. In thefollowing, we consider experiment A.

Fig. 11.— Temporal evolution of the verticalforces acting below the current sheet along the(x = 0, y = 0) line. Shown are the mag-netic pressure gradient (dashed-rectangles), thetension force (dotted-triangles), the gravitationalforce (dotted-dashed), the Lorentz force (solid),the gas pressure gradient (dashed) and the totalforce (thick solid).

The vertical component of the forces is shown asa function of time in Fig. 11. The plot shows thatthe emergence of the buoyant tube is driven by themagnetic pressure force that exceeds all the otherforces at the early stage of the evolution. What isalso noticeable is that the magnetic pressure forceand the magnetic tension are equal in magnitudeand opposite in sign and that they are substan-

tially larger than either the pressure gradient orthe gravity force.

Fig. 12.— The variation of the magnetic pressureforce (thin lines) and tension force (thick lines)with height for experiment A (triple dotted dashedlines) and D (dashed lines) at t = 100.

The total force is clearly positive acceleratingthe emerging field against gravity until t ≈ 72.This is followed by a short period of deceleration.After t = 95, all the forces become very small andare essentially in balance. This result is consistentwith the motion of the plasma, which rises withan almost constant velocity after t = 95 (see rightpanel in Fig. 5). Notice that the temporal evolu-tion of the total force corresponds well with themotion of the apex of the tube shown in Fig. 5.Also, the time (t = 95) at which the forces takeon very small values corresponds to the time atwhich pressure balance is achieved across the cur-rent sheet.

An analysis of the forces for the experimentsB-D show that the temporal evolution of the totalacceleration is similar to the experiment A. Thisresult explains why the apex of the tube reachesalmost the same height at the same time in allexperiments. To further illustrate the point madeabove we examine the total force for the experi-ments A and D for t = 100. The z−component ofthe total force is

Fz = Flz + Fg + Fp, (7)

where Fg = −ρg is the gravitational force, Fp =

−∂P∂z is the gas pressure force and Flz is the verti-

cal component of the Lorentz force. The latter is

13

written as

Flz =1

4π(B · ∇)Bz −

1

8π

∂B2

∂z, (8)

where the first term describes the magnetic tensionand gives a force when the fieldlines are curved,while the second term is the magnetic pressureforce that acts from regions of high magnetic pres-sure to regions of low magnetic pressure.

Fig. 13.— Total force as a function of height forexperiment A (tripe dotted dashed line) and D(dashed) at t = 100.

Fig. 12 shows the distribution of the two termsin Eq. (8) along height for the experiments A andD for t = 100. On the one hand, the verticalcomponent of the magnetic pressure force belowthe current sheet (z < 40) is larger for experimentD. On the other hand, the tension force, which isa downward force, is also larger for experiment Dand, thus, the Lorentz force has comparable sizein both experiments.

Finally, Fig. 13 shows the distribution of thetotal force in Eq. (7) along height for the experi-ments A and D for t = 100. The total force at thetop part of the rising flux system is very small andalmost identical for A and D. Thus, it seems thatthe total acceleration that acts on the expandingrising volume below the current sheet does not de-pend on the structure of the overlying field and asa result the crest of the tube reaches almost thesame height at the same time for experiments withdifferent orientation of the coronal field.

6. Topology of the emerging region and

magnetic reconnection

The topology of the interacting magnetic fieldis important for understanding the structure of theemerging high velocity jets and associated hotterplasma distribution. In this section, we show howdifferent is the structure of the magnetic field thatappears in the corona when the relative orienta-tion of the interacting magnetic fields changes be-tween the experiments.

The three-dimensional geometry of the cur-rent sheet and the jets, emanating from the rimsof the current sheet, for experiment A and B,have been studied by Archontis et al. (2005) andGalsgaard et al. (2005). Here, we illustrate theprojection on horizontal xy planes of the three-dimensional structure of the sheet for experi-ments A-E. Fig. 14 shows five panels containingcolourmaps of the total magnitude of the current(|J |) on a horizontal cut at a height of 1.7 Mmabove the base of the corona at t = 100. Thearrows in the panels correspond to the projectionof the velocity field.

The bright patches in Fig. 14, show the loca-tion of the highest values of |J |, and correspondto the intersection of the horizontal cut with thearch-like current sheet. Effective reconnection oc-curs and high-velocity outflows are ejected side-ways from these sites. The velocities of the jetsreach values close to 200 Kmsec−1 in experimentA and B. In experiments A-D, the direction of thejets is aligned with the direction of the ambientmagnetic field. In experiment E there are no jetsbecause reconnection does not occur actively be-tween the two flux systems. The arrows in thiscase illustrate the drain of the plasma from theuppermost layers of the emerging plasma ball asit rises and pushes the ambient field upward.

The weaker current structure, forming an en-closure between the two locations of strong cur-rent, outlines the border region between the mag-netic flux totally connected to the emerging fluxtube (inside) and either the reconnected flux orthe original coronal flux (outside). From this it isclear that the horizontal volume of the emergingflux tube decreases significantly as reconnectionbecomes more favorably. It is only the flux con-tained inside this volume that eventually can endup outlining the structure of the emerging flux re-

14

Fig. 14.— Colourmaps of the total current at t = 100. Superimposed is the velocity field (arrows). Thepanels correspond to the experiments A-E.

gion, while the reconnected field lines connect thetwo flux concentrations to neighboring flux regions(here only the corona).

To further illustrate how the three-dimensionalreconnection works at the top of the emergingflux system we study the topology of the fieldlinesacross the current sheet at t = 100. Figures 15and 16 consist of ten panels that show the field-line topology for the five experiments. Three setsof fieldlines have been traced from different start-ing positions along height and across the interfaceof the two flux systems close to the center of theemerging region. The blue fieldlines are tracedfrom just below the current sheet and belong en-tirely to the rising tube. The red fieldlines aretraced from just above the sheet and are ambientfieldlines. Finally, the yellow fieldlines are tracedfrom inside the diffusion region and connect thetube with the coronal field. The current concen-tration at the interface is visualized with a trans-parent isosurface.

The general picture of the reconnection shows aclear difference from the traditional two-dimensionalconfiguration. This is because the magnetic fieldvector across the sheet resembles a rotational dis-continuity. On the one hand, the uppermost risingfieldlines have an orientation which is not perfectlyaligned with the x−axis, as has been explained inArchontis et al. (2005). On the other hand, theorientation of the ambient fieldlines changes fromexperiment A to experiment E so that the rel-ative horizontal angle between the two systemsincreases. The product of the reconnection be-tween these two sets of fieldlines is another set offieldlines, the yellow lines, which are ejected side-ways from the current sheet and establish linksbetween the solar interior and the outer atmo-sphere. Panels B2, C2 and D2 show that theyellow fieldlines at the top of the current sheethave an intermediate orientation between the blueand the red fieldlines and that they do not stayin a two-dimensional plane but they experience

full 3D-reconnection. The only experiment wherethe initial relative orientation is not favorable forreconnection is experiment E. Panels E1 and E2show that there is no reconnection (and thus, noyellow fieldlines in the panels). Instead, the coro-nal field is pushed upwards and a bended hill-likeshape interface is formed between the two fields.At the late stages of the evolution of the systemthe ambient field slides down along the sides ofthe hill of the emerging flux and some reconnec-tion occurs at low heights.

Through figures 15 and 16 we also get a con-firmation of the dependence of the volume of theemerged region on the orientation of the coronalfield that are shown in Fig. 14. More precisely,the blue fieldlines illustrate the geometry of theoutermost layers of the tube and show that theexpansion, in the transverse direction to the axisof the tube, is larger when reconnection is less ef-ficient. Indeed, in the case of experiment A theouter fieldlines that suffer a large expansion re-connect first and as the time goes on more internallayers, which expand less and are also less twisted,come into contact and eventually reconnect withthe ambient field. In contrast to this, experimentE shows a large expansion of the loops of the upperfieldlines as these have not been reconnected dueto the small angle φ0. This picture confirms theindications regarding the location of the currentstructure seen in Fig. 14.

Finally, if we focus on the shape of the emergingfieldlines we find that the blue lines in experimentA represent fieldlines which were initially locatedcloser to the main axis of the tube and their ori-entation was not far away from the y−axis. Whenthese fieldlines emerge (see Panel A1 and A2) keepan almost flat shape in the middle of their crest.On the other hand, the outermost fieldlines in ex-periments with less reconnection (see for examplethe panels E1 and E2) represent fieldlines whichwere initially located at the outskirts of the tubeand they had larger curvature. As these fieldlines

15

rise and expand, they keep their convex shapeand, thus, the magnetic tension force at the topof the emerging tube becomes larger compared tothe experiments with more efficient reconnection.In fact, this has been also shown in Fig. 12, wherethe magnetic tension is plotted against height forthe experiments A and D at t = 100.

7. Discussion

The present experiments were terminated ataround t=120 due the use of periodic boundaryconditions in the horizontal direction. For timesafter t=120 the effect of these conditions becomevery apparent on the experiments as the reconnec-tion jets have propagated across the system, andthe subsequent evolution is not showing a freelyexpanding magnetic flux concentration. In theSun, this scenario is possibly more realistic, thanhaving open boundary conditions. On the otherhand, it may cause numerical problems and in-fluences the dynamical evolution of the system.Thus the timescale of each numerical experimentdepends also on the boundary effects from the pe-riodic conditions.

Some observations seem to favor a situationwhere a flux rope is emerging into the corona ascoherent structure (Lites et al. 1995), while others(Strouse & Zwaan 1999; Pariat et al. 2004) indi-cate a pattern where undulating field lines makedifferent dynamical effects to release the dense ma-terial in the lower parts along them. In our sim-ulations, including our previous 3D numerical ex-periments, the initial flux tube becomes unstableto further expansion into the corona, by buildingup a dominating magnetic pressure force, due tothe buoyancy of the tube (Archontis et al. 2004).When the force becomes strong enough it ”blows”the layers close to the transition region up and intothe corona, where they rapidly push the overlay-ing material away and created a dense, cold mag-netically dominated plasma dome. These experi-ments show that almost 65% of the normal flux ofthe initial flux tube has emerged into the corona.We also find that the axis of the initial tube hasnot fully emerged yet above the photosphere. Itis wortwhile mentioning that in one of the ex-periments we find the formation of a horizontalcurrent sheet, first reported by Manchester et al.(2004), which drives internal reconnection of the

flux belonging to the emerging tube, allowing thelower parts of the emerging flux tube to discon-nect from the emerging flux system. This providesa mechanism to decouple the dense photosphericplasma from the field lines that expand into thecorona (as indicated by Strouse & Zwaan (1999)and Pariat et al. (2004)) and provides a possibilityfor forming a structure that looks like an emergedtwisted flux tube without emerging the entire mag-netic flux system.

As it is seen from the Sections above, the rel-ative orientation between the emerging magneticflux and the coronal magnetic field is of great im-portance when it comes to the dynamical evolutionof the flux interaction. The emergence process,eventually, produces high velocity plasma jets withtemperatures in excess of average coronal valuesonly in the cases of efficient reconnection. In lessfavorable situations the plasma will not be heatedmuch and the spectacular display often seen inTRACE movies will not take place. Thus, theemergence of an easily observed flux region intothe hostile coronal environment depends on a rel-ative narrow span of angles between the two sys-tems. On the Sun, this regime may be increasedcompared to the simple model presented here, dueto the much larger structural complexity of the so-lar environment.

The experiments discussed here relay stronglyto the evolution of driven magnetic reconnection.From analytical investigations of 2D reconnection(see Priest & Forbes (2000) and references therein), it is found that steady state reconnection de-pends critically on the structure of the magneticfield, the velocity flow and the value of the mag-netic resistivity. This makes it natural to expectsimilar dependences to be carry over to steadystate 3D reconnection. In relation to this, it isobvious that 3D numerical experiments are notable to resolve the Reynolds numbers present inthe coronal plasma. Why should it then be ex-pected that the results from the above mentionedexperiments represent an evolution that may takeplace in the solar corona? Using hyperdiffusion thesmallest length scale (the thickness) of the cur-rent sheet is alway only resolved by a few gridpoints, and it is only at these length scales thatdiffusion of the magnetic field becomes impor-tant. Increasing the numerical resolution, impliesa decrease of the thickness of the current sheet

16

and through this a slight delay in the time forthe initiation of reconnection. In the driven re-connection scenario presented here, the largescaleof the magnetic field and velocity flow does notchange significantly when the numerical resolutionis increased and, thus, the magnetic flux advectedinto the current sheet will remain approximatelythe same. This indicates that magnetic recon-nection provides the same new classes of connec-tivity. What may change is the local structureof the process, where increased numerical reso-lution can allow for reaching local turbulence inthe current sheet as it may become tearing un-stable before starting reconnecting. Such a pro-cess naturally makes it possible to reach a morecomplicated field line connectivity locally, but italso implies that the reconnection process becomesindependent of the magnetic resistivity and re-sponds directly to the amount of flux advected intothe current sheet (Galsgaard & Nordlund (1996);Hendrix et al. (1996); Eyink & Aluie (2006)). Fi-nally, the choise of resolution for the experimentsis important. Experiments with low resolutionyield low values for the energy release obtained inthe simulations and large values for the timescaleof the evolution of the system. High resolution,especially for the volume occupied by the initialtube, may cause problems with the entropy distri-bution and affect the buoyancy of the rising tube.

If the assumption of an emerging magnetic loopis correct, and the processes described in this pa-per and in our previous work are representative ofthe emergence of flux from the solar interior to theouter atmosphere of the Sun, then there are somesimple features that this model may predict.

• The process of flux emergence is usually ob-served using vector magnetograms, whichshow the appearance of bipolar structuresat the photosphere. In our experiments, dueto the highly twisted magnetic flux tube inthe subphotospheric layer, the orientation ofthe polarity patterns are, in the early phases,perpendicular to the main axis of the emerg-ing flux tube (North-South orientation). Astime progresses the two opposite polarityspots moving apart in the direction along themain tube axis (East-West orientation).

• The picture of an emerging bipolar regionwith its strong flux region located along the

axis of the main tube is only true for thephotospheric layer during the short time ofthese experiments. Slightly above the pho-tosphere the topology changes and the ini-tial large scale bipolar region structure go-ing across the magnetic loop is maintainedover the remaining time of the various exper-iments. Variations between the different ex-periments are seen, but the systematic bipo-lar patterns going across the main axis of theinitial flux tube is maintained for all exper-iments even for a height that is only 3400Km above the photosphere.

• Magnetic reconnection between the emerg-ing and coronal magnetic fields depends onthe relative orientation between the two fluxsystems. Significant plasma heating andhigh velocity jets will appear when efficientreconnection occurs between the two fluxsystems in a coronal environment. For thecases where the relative orientation of thetwo flux systems is close to parallel, a weakerreconnection may take place closer to thephotosphere.

• The emergence rate seems to depend on thestructure of the magnetic field below thephotosphere, which in our experiments hashigher field strength than the coronal mag-netic field. The excess magnetic pressuretherefore expels the flux upwards and, thus,opens the volume for a new bipolar structureto appear.

The intense reconnection in the near anti-parallel cases makes it possible for most of theinitial flux of the tube to reconnect with the coro-nal magnetic field. If this process continues forlong enough, then the initial connectivity of theflux tube will be totally disrupted. It is thereforepossible that by emerging a bipolar region intothe hostile coronal environment, that the initialconnectivity between the two opposite flux con-centrations will be, at the least partly, disrupted.How severe this change in connectivity becomesdepends on several factors, such as the structureof the coronal flux, the amount of the emergingflux and the duration of the reconnection process.A first estimate of the changes in the connectiv-ity may be obtained simply by using potentialmodels. On the other hand, it seems that these

17

models is not always the best method to followfor the study of the time evolution of magneticstructures in actively evolving regions, which iswhy we have not used this analysis here.

Further up in the atmosphere Longcope et al.(2005) have analyzed an interesting emergence ob-servation recorded by TRACE. They adopt theminimum current model to estimate how muchflux has reconnected. Using various assumptionsthey find it is possible to account accurately for thechanges in connectivity. They also find that thereconnection between the emerging region and theexisting coronal flux happens only after some time.As the interaction starts, it reaches a high activitylevel that last for a finite time before rapidly lev-eling off again. As most of the involved flux seemsto be accounted for, then it indicates that the sys-tem has to build up a significant amount of stressbefore they start interacting. A different possibil-ity is that the lower lying magnetic field structuresare not favorable for the reconnection process andthat the onset of the reconnection process doesnot start until some of the aligned flux has beenpushed away and the relative orientation betweenthe two flux systems has changed. It is also a pos-sibility that the initial weaker reconnection pro-cess takes place in regions where the plasma den-sity is still so high that the released dissipationcan not heat the plasma to coronal temperatures.Finally there is the possibility that the initial coro-nal field in the emergence region is so weak that itdoes not provide for any significant heating as thenew flux pushes its way up.

8. Summary

The emergence of new magnetic flux into anexisting coronal magnetic field can evolve in verydifferent ways. The coronal reaction on the emer-gence depends strongly on the relative orientationof the two flux systems. Efficient reconnectiontakes place only in the case where the two flux sys-tems have almost antiparallel orientations. Takingplace in the corona, this provides a number of veryclear signatures, such as, high velocity jets andintense heating. The active reconnection processclearly affects the volume occupied by the mag-netic flux totally connected with the underlyingmagnetic flux tube.

The reconnection between the two flux systems

can slow down mainly because of two reasons; Ei-ther because all flux in the emerging flux tube hasreconnected with the coronal field, and by this to-tally disrupted the structure of the emerging fluxrope. Or because the emerging process stops andthe two flux systems find a mutual balance witheach other where no, immediate, reconnection istaking place. Thus, the length and final structureof the new coronal magnetic field strongly dependson the total amount of emerging flux, its ability toreconnect and on the amount of the coronal mag-netic flux.

For situations where the two flux systems arenearly aligned, it will be possible for the emergingfield to penetrate far into the coronal magneticfield before it will create any significant coronalsignatures revealing its present. These situationswill only be easily spotted if simultaneous obser-vations are made in photospheric and transitionregion lines.

In our simulations, we have used a simple en-ergy equation that does not take into account op-tical thin radiation or anisotropic heat conductionin the corona. These effects are important ingre-dient to include in the experiments before moredetailed comparisons with observations can be ob-tained. These are effects that will be discussed infuture publication.

In a subsequent paper we are going to discuss indetail the local implications of the ongoing mag-netic reconnection process and its implication onthe dynamical evolution of the different experi-ments.

We are grateful for computational time on theUKMHD linux cluster in St. Andrews (Scotland,UK), funded by SRIF and PPARC, and on thelinux cluster at the IAC (Tenerife, Spain) partiallyfunded by the Ministry of Science and Technol-ogy. This work has also benefited from finan-cial support through the Platon European Re-search Training Network HPRN-CT-2000-00153 ofthe European Commission and the CICYT projectno. AYA2001-1649 of the Spanish Ministry of Sci-ence and Technology. K.G. was supported bythe Carlsberg Foundation in the form of a fellow-ship. FMI thankfully acknowledges the computerresources, technical expertise and assistance pro-vided by the Barcelona Supercomputing Center -Centro Nacional de Supercomputacion (Spain) We

18

thank the anonymous referee for helpful commentsthat improved this paper.

REFERENCES

Archontis, V., Moreno-Insertis, F., Galsgaard, K.,Hood, A. & O’Shea, E. 2004, Astron Astrophys,426, 1047

Archontis, V., Moreno-Insertis, F., Galsgaard, K.,& Hood, A. 2005, ApJ, 635, 1299

Archontis, V., Moreno-Insertis, F., Galsgaard, K.,& Hood, A. 2006, ApJ, 645, 161

Eyink, G. L. & Aluie H. 2006, Physica D, 223, 82

Fan, Y. 2001, ApJ, 554, L111

Fan, Y. & Gibson, S. E. 2004, ApJ, 609, 1123

Galsgaard, K., Moreno-Insertis, F., Archontis, V.,& Hood, A. 2005, ApJ, 618, 153

Galsgaard, K. & Nordlund, A. 1996, JGR, 101,13445

Georgoulis, M. K., Rust, D. M., Bernasconi, P. N.& Schmieder B., 2002, ApJ, 577, 506

Haynes, A., Parnell, C. E., Galsgaard, K. & Prist,E. R. 2007 Proc. Roy. Soc. Lond. A Math. Phys.Eng. Sci., 463, 1097

Hendrix, D. L., Van Hoven, G., Mikici Z. &Schnack, D. D. 1996, ApJ, 470, 1192

Isobe, H., Miyagoshi, T., Shibata, K., &Yokoyama, T. 2005, Nature, 434, 478

Kubo, M., Shimizu, T. & Lites, B. W., 2003, ApJ,595, 465

Leake, J. E. & Arber, T. 2006, A&A, 450, 805

Lites, B. W., 2005, ApJ, 622, 1275

Lites, B. W., Low, B. C., Martınez Pillet, V., Sea-graves, P., Skumanich, A., Frank, Z. A., Shine,R. A. & Tsuneta, S., 1995, ApJ, 446, 877

Lites, B. W., Skumanich, A. & Martınez Pillet,V., 1998, A&A, 333, 1053

Longcope, D. W., McKenzie, D. E. , Cirtain, J. &Scott, J. 2005, ApJ, 630, 596

Magara, T. 2006, ApJ, 653, 1499

Magara, T. & Longcope, D. W. 2003, ApJ, 586,630

Manchester, W., Gombosi, T., DeZeeuw, D., &Fan, Y. 2004, ApJ, 610, 588

Murray, M.J., Hood, A., Moreno-Insertis, F.,Galsgaard, K., & Archontis V. 2006, A&A, sub-mitted

Nordlund, A. & Galsgaard, K. 1997,http://www.astro.ku.dk/∼kg

Pariat, E., Aulanier, G., Schmieder, B., Geor-goulis, M.K., Rust, D.M. & Bernasconi, P.N.2004, ApJ, 614, 1099

Parnell, C., & Galsgaard, K. 2004, A&A, 428, 596

Parnell, C. E., Haynes, A. & Galsgaard, K. 2007,In Preperation

Priest, E. R. & Forbes, T. 2000, Magnetic Recon-nection, Cambridge University Press

Spadaro, D., Billotta, S., Contarino, L., Romano,P. & Zuccarello, F. 2004, A&A, 425, 309

Strouse, L. H. & Zwaan C., 1999, ApJ, 527, 435

Zuccarello, F., Battiato, V., Contarino, L., Ro-mano, P., Spadaro, D. & Vlahos, L. 2005, A&A,442, 661

This 2-column preprint was prepared with the AAS LATEXmacros v5.2.

19

A1

Z

XY

A2

X

YZ

B1

X

Z

Y

B2

X

YZ

C1

Y

Z

X

C2

X

YZ

Fig. 15.— 3D visualization of the fieldline topology across the current sheet at t = 100 for the experimentsA (panels A1, A2), B (panels B1, B2) and C (panels C1, C2). The left column is a side view and the rightcolumn is a top view of the same snapshot. The current sheet is visualized as transparent isosurface. Thearrows show the direction of the magnetic field vector.

20

D1

X

Z

Y

D2

X

YZ

E1

X

Z

YX

YZ

E2

Fig. 16.— Same as in Fig.15 but for the experiments D and E.

21