0.03, where MA is an Alfvén Mach number of the inflow.The Astrophysical Journal, 546:L69-L72, 2001 January 1
© 2001. The American Astronomical Society. All rights reserved. Printed in U.S.A.
Clear Evidence of Reconnection Inflow of a Solar Flare
T. Yokoyama
Nobeyama Radio Observatory, Minamimaki, Minamisaku, Nagano, 384-1305, Japan; yokoyama@solar.mtk.nao.ac.jp
K. Akita
Osaka Gakuin University, Suita, Osaka, 564-8511, Japan
T. Morimoto and K. Inoue
Department of Astronomy, Kyoto University, Kyoto, 606-8502, Japan
and
J. Newmark
Emergent Information Technologies East, NASA Goddard Space Flight Center, Greenbelt, MD 20771
Received 2000 April 13; accepted 2000 October 27; published 2001 December 29
ABSTRACT
We found an important piece of evidence for magnetic reconnection inflow in a flare on 1999 March 18. The flare occurred on the northeast limb, displaying a nice cusp-shaped soft X-ray loop and a plasmoid ejection typical for the long-duration events. The EUV observation of the same flare shows us a bubble-like void ejection. The core of this EUV void corresponds to the soft X-ray plasmoid. Moreover, as this void is ejected, magnetic reconnection occurs at the disconnecting point. A clear ongoing pattern toward the magnetic X-point is seen. The velocity of this apparent motion is about 5 km s-1, which is an upper limit on reconnection inflow speed. Based on this observation, we derive the reconnection rate as MA = 0.0010.03, where MA is an Alfvén Mach number of the inflow.
1. INTRODUCTION
Magnetic reconnection (Petschek 1964; Sweet 1958; Parker 1963)
the reorganization caused by local diffusion of antiparallel magnetic field lines and the consequent release of magnetic energyhas been thought to be the cause of solar flares (e.g., Shibata 1996). Many indirect pieces of evidence for this process have been found by recent spacecraft observations. There was, however, almost no direct evidence, such as inflow or outflow (reconnection jet) that carries the field lines toward or from the magnetic neutral point where the local dissipation occurs (except for McKenzie & Hudson 1999). We report here the first discovery of reconnection inflow during a flare on 1999 March 18.
Solar flares are now thought to be caused by magnetic reconnection (Fig. 1; e.g., Shibata 1996; Yokoyama & Shibata 1998). In this model, the antiparallel field lines dissipate in a certain local point in the corona. The tension force of the reconnected field lines then accelerates the plasma out of the dissipation point. In response to this outflow, the ambient plasma is drawn in. The inflowing plasma carries the ambient magnetic field lines into the dissipating point. These field lines continue the reconnection cycle. In this manner, the magnetic energy stored near the neutral point is released to become the thermal and bulk-flow energy of plasma.
 | Fig. 1 Schematic illustration of the reconnection model of a solar flare. Thick solid lines show the magnetic field. Arrows indicate the plasma bulk flow. This is for the simplest bipole case and approximates the structure observed in this study. |
The supporting evidence for this model is the observation of a cusp-shaped soft X-ray flare loop (Tsuneta et al. 1992). The tip of the cusp is thought to be the remnant of the kink of the reconnected field lines. This cusp-shaped flare loop increases its height and the distance between the footpoints, which might be the consequence of the piling up of the shrunk magnetic field lines (see also Forbes & Acton 1996; Hiei, Hundhausen, & Sime 1993). Flare observation by Masuda et al. (1994) demonstrates a hard X-ray source above the soft X-ray loop. This source suggests that some high-energy process, such as acceleration of electrons associated with reconnection, is occurring above the soft X-ray loop. Plasma blob ejections associated with flares were observed (Shibata et al. 1995; Ohyama & Shibata 1997). This ejection is the result of the outflow from the reconnecting point. Note that this plasma blob ejection is not the reconnection jet itself, although it is strongly related to the jet.
The above evidence supports the reconnection model. But it is indirect in the sense that we are not seeing the energy release process itself. So the search for the inflow or reconnection jet is still ongoing. Recently, McKenzie & Hudson (1999) found downward plasma motion above an arcade formed in a long-duration flare and coronal mass ejection. They suggest that this motion is a consequence of the retraction of the reconnected field lines, i.e., the outflow from the magnetic reconnection point. The study described in this Letter is the first successful result of the searching effort of the reconnection inflow.
2. FLARE 1999-3-18: DISCOVERY OF RECONNECTION INFLOW
This flare occurred on the northeast solar limb. According to the Yohkoh/soft X-ray telescope (SXT; Tsuneta et al. 1991) observation, it showed a nice cusp-shaped loop and a plasma blob (plasmoid) ejection (Fig. 2). The EUV observation of the same flare by the Solar and Heliospheric Observatory (SOHO)/EUV Imaging Telescope (EIT; Delaboudinière et al. 1995) shows us a bubble-like void ejection (Fig. 3). The core of this EUV void corresponds to the soft X-ray plasmoid and is bright in X-ray but dark in EUV because of its temperature, 
4 MK. As this void moves away from the limb, the lower part of the void becomes thinner and thinner. When the void finally detaches from the limb, an X-shaped structure appears at the detaching point. We believe a magnetic reconnection process occurs at this detaching point because of the following reasons: First, it is natural to consider that the void is a bubble of plasma surrounded by magnetic field lines. As the void detaches, the surrounding lines are also pulled up, and at the detaching point the antiparallel field lines meet. Second, as the void continues rising, the pattern of threads (we assume that the threads are field lines) merge toward the X-point. Figure 4 shows the evolution of the slice cut across the X-point. A clear merging pattern toward the X-line can be seen. We believe this is the inflow of the magnetic reconnection. Third, as a consequence of this process, reconnected field lines should shrink down below the X-point. We observe an evolving cusp-shaped loop on the limb in soft X-ray and EUV, which is the result of the piling up of the shrunk magnetic field lines (Tsuneta et al. 1992; Hiei et al. 1993; Forbes & Acton 1996). Through these observations, we conclude that we have finally found the last link demonstrating the reality of an ongoing magnetic reconnection process.
 | Fig. 2 Time evolution of soft X-ray images by Yohkoh/SXT and EUV (Fe 195 Å) images by SOHO/EIT. The field-of-view size of each panel is |
 | Fig. 3 Two-dimensional EUV (Fe 195 Å) images by SOHO/EIT. In each image, the data shown is the difference from that of 2:58:52 UT. For the right panel, Yohkoh/SXT soft X-ray intensity is overlaid as contour lines. The field-of-view size of each panel is |
 | Fig. 4 Evolution of the one-dimensional plot of EIT data nearly across the X-point. Left panel: A snapshot of an EUV (Fe 195 Å) image by SOHO/EIT. The field-of-view size of each panel is 670 |
The measurement of the inflow velocity is done by tracing the movement of the threadlike patterns in the EIT movie. In order to obtain an exact velocity, the measurement should be done in a frame that is moving with the X-point. It is, however, difficult to keep this frame because the position of the X-point in each EIT image is not clear. We, therefore, fix the frame for measurement, but the error added by this procedure is discussed in detail. We measure the speed of the movement of a pattern in a slice cut indicated in the left panel of Figure 4. The speed of the incoming pattern is about Vpattern = 4.7 km s-1 over the period 4:006:00 UT (Fig. 4, top right panel). This measured speed Vpattern is an upper limit of the real inflow speed because it is a mixture of the real inflow velocity Vinflow and an error velocity due to the movement of the X-point. Using simple geometrical consideration, one can show that Vpattern = Vinflow + Vxp tan 
, where Vxp is a velocity of the X-point in the fixed frame and is the angle between the measured threadlike pattern in the images and the perpendicular direction of the slice cut. It is still difficult to measure the values of Vxp and from the images. But from the measurement of the outer edge of the ejecting plasmoid, it is possible to obtain the upper limit values to these variables. Thus, we have tan < tan plasmoid 0.1 and Vxp < Vplasmoid 37 km s-1, where plasmoid is the angle of the outer edge of the plasmoid and Vplasmoid is the velocity of the plasmoid. Then, the derived lower limit of the inflow velocity is Vinflow > 1.0 km s-1.
3. ESTIMATION OF RECONNECTION RATE OF INFLOW
An important quantity for various models is the magnetic reconnection rate, MA. This is defined as the ratio of the inflow speed to the Alfvén velocity (the speed of the plasma when all the magnetic energy is converted to the bulk kinetic energy). The derivation of the Alfvén velocity is estimated from the soft X-ray observations. The emission measure and temperature is derived from the filter ratio method (Hara et al. 1992) as EM = 10301031 cm-5 and T = 2.74.2 MK, respectively. If we assume the depth (L) along the line of sight is 1.5 × 105 km (that is approximately the same size as the loop length), the density is n = (0.82.6) × 1010 cm-3[L/(1.5 × 105 km)]-1/2. Thus, the total thermal energy is Eth = 3nkBTL3 = (0.31.5) × 1031 ergs, where kB is the Boltzmann constant. Using the lifetime of this flare, 
3 × 104 s, we obtain the output rate of the thermal energy as Eth/ = (15) × 1026 ergs s-1. Note that we adopt the temperature and the density as the averaged value over the period 4:375:16 UT and over the intense spatial part, namely, the cusp-shaped X-ray loops. This might be acceptable for our present purpose, i.e., obtaining the rough value based on the order-of-magnitude estimation. Balancing the energy output is the released magnetic energy input rate, expressed as 2B2/(4
)VinflowL2 with unknown magnetic field strength B. This is twice (due to merging from two sides of the X-point) the product of the magnetic energy per unit volume and the inflow speed. Equating the increasing rate of thermal energy and the released magnetic energy rate, we obtain the magnetic field strength of B = 2.412 G. This is clearly an underestimate of the real magnetic field strength because we do not take into account radiation losses, the conduction of heat to the chromosphere, the kinetic energy of the outflow, etc. The second method to derive the magnetic field strength is to assume that the magnetic field outside the current sheet in the inflow region is equal to the gas pressure in the downstream region. Thus, the obtained value is B (16nkBT)1/2 = 1240 G[L/(1.5 × 105 km)]-1/4. Using the values derived by the second method, the Alfvén speed becomes CA = 160970 km s-1. Since the inflow speed is Vinflow = 1.04.7 km s-1, the reconnection rate is Vinflow/CA = 0.0010.03. This observed result gives us further information about the magnetic reconnection process itself. The value of the reconnection rate MA = 0.0010.03 is roughly consistent with Petschek's (1964) reconnection model, not with Sweet-Parker's model (Sweet 1958; Parker 1963), because the latter predicts a value much smaller than this.
We thank K. Shibata, H. S. Hudson, D. McKenzie, M. Shimojo, and the anonymous referee for helpful comments. The Yohkoh satellite is a project of Japan, the US, and the UK, launched and operated by the Institute of Space and Astronautical Science. SOHO is a project operated by ESA and NASA.
REFERENCES
