Do Long-Lived Features Really Exist in the Solar Photosphere? II. Contrast of Time-Averaged Granulation Images

The decrease in the rms contrast of time-averaged images with the averaging time is compared between four data sets: (1) a series of solar granulation images recorded at La Palma in 1993, (2) a series of artificial granulation images obtained in numerical simulations by Rieutord et al. (Nuovo Cimento 25, 523, 2002), (3) a similar series computed by Steffen and his colleagues (see Wedemeyer et al. in Astron. Astrophys. 44, 1121, 2004), (4) a random field with some parameters typical of the granulation, constructed by Rast (Astron. Astrophys. 392, L13, 2002). In addition, (5) a sequence of images was obtained from real granulation images by using a temporal and spatial shuffling procedure, and the contrast of the average of n images from this sequence as a function of n is analysed. The series (1) of real granulation images exhibits a considerably slower contrast decrease than do both the series (3) of simulated granulation images and the series (4) of random fields. Starting from some relatively short averaging times t, the behaviour of the contrast in series (3) and (4) resembles the t ^−1/2 statistical law, whereas the shuffled series (5) obeys the n ^−1/2 law from n=2 on. Series (2) demonstrates a peculiarly slow decline of contrast, which could be attributed to particular properties of the boundary conditions used in the simulations. Comparisons between the analysed contrast-variation laws indicate quite definitely that the brightness field of solar granulation contains a long-lived component, which could be associated with locally persistent dark intergranular holes and/or with the presence of quasi-regular structures. The suggestion that the random field (4) successfully reproduces the contrast-variation law for the real granulation (Rast in Astron. Astrophys. 392, L13, 2002) can be dismissed.     

Photon Spectroscopy with Imaging X-Ray Instruments

Individual X-ray photons in the keV energy range produce hundreds of photoelectrons in a single pixel of a CCD array detector. The number of photoelectrons produced is a linear function of the photon energy, allowing the measurement of spectral information with an imaging detector system. Most solar X-ray telescopes, such as Yohkoh/SXT and Hinode/XRT, use CCD detectors in an integrating mode and are designed to make temperature estimates from multiband filter photometry. We show how such instruments can be used in a new way to perform a limited type of this photon spectroscopy. By measuring the variance in intensity of a series of repeated images through a single filter of an X-ray source, the mean energy per detected photon can be determined. This energy is related to the underlying coronal spectrum, and hence it can be used to deduce the mean plasma temperature. We apply this technique to data from the Yohkoh Soft X-Ray Telescope and compare the temperatures obtained with this technique with the temperatures derived using the standard filter ratio method for a postflare loop system. Given the large dynamic range of the soft X-ray flux observed from the Sun, we describe the requirements for a future instrument that would be better suited to performing photon spectroscopy. B.J. Labonte deceased 24 October 2005.     

A New Look at Mode Conversion in a Stratified Isothermal Atmosphere

Recent numerical investigations of wave propagation near coronal magnetic null points (McLaughlin and Hood: Astron. Astrophys. 459, 641, 2006) have indicated how a fast MHD wave partially converts into a slow MHD wave as the disturbance passes from a low-β plasma to a high-β plasma. This is a complex process and a clear understanding of the conversion mechanism requires the detailed investigation of a simpler model. An investigation of mode conversion in a stratified, isothermal atmosphere with a uniform, vertical magnetic field is carried out, both numerically and analytically. In contrast to previous investigations of upward-propagating waves (Zhugzhda and Dzhalilov: Astron. Astrophys. 112, 16, 1982a; Cally: Astrophys. J. 548, 473, 2001), this paper studies the downward propagation of waves from a low-β to high-β environment. A simple expression for the amplitude of the transmitted wave is compared with the numerical solution.     

Numerical Simulation of a Solar Active Region. I: Bastille Day Flare

We present three-dimensional unsteady modeling and numerical simulations of a coronal active region, carried out within the compressible single-fluid MHD approximation. We focus on AR 9077 on 14 July 2000, and the triggering of the X5.7 GOES X-ray class “Bastille Day” flare. We simulate only the lower corona, although we include a virtual photosphere and chromosphere below. The boundary conditions at the base of this layer are set using temperature maps from line intensities and line-of-sight magnetograms (SOHO/MDI). From the latter, we generate vector magnetograms using the force-free approximation; these vector magnetograms are then used to produce the boundary condition on the velocity field using a minimum energy principle (Longcope, Astrophys. J. 612, 1181, 2004). The reconnection process is modeled through a dynamical hyper-resistivity which is activated when the current exceeds a critical value (Klimas et al., J. Geophys. Res. 109, 2218, 2004). Comparing the time series of X-ray fluxes recorded by GOES with modeled time series of various mean physical variables such as current density, Poynting energy flux, or radiative loss inside the active region, we can demonstrate that the model properly captures the evolution of an active region over a day and, in particular, is able to explain the initiation of the flare at the observed time.     

Transient Coronal Sigmoids and Rotating Erupting Flux Ropes

To determine the relationship between transient coronal (soft X-ray or EUV) sigmoids and erupting flux ropes, we analyse four events in which a transient sigmoid could be associated with a filament whose apex rotates upon eruption and two further events in which the two phenomena were spatially but not temporally coincident. We find the helicity sign of the erupting field and the direction of filament rotation to be consistent with the conversion of twist into writhe under the ideal MHD constraint of helicity conservation, thus supporting our assumption of flux rope topology for the rising filament. For positive (negative) helicity the filament apex rotates clockwise (counterclockwise), consistent with the flux rope taking on a reverse (forward) S shape, which is opposite to that observed for the sigmoid. This result is incompatible with two models for sigmoid formation: one identifying sigmoids with upward arching kink-unstable flux ropes and one identifying sigmoids with a current layer between two oppositely sheared arcades. We find instead that the observations agree well with the model by Titov and Démoulin (Astron. Astrophys. 351, 707, 1999), which identifies transient sigmoids with steepened current layers below rising flux ropes.     

Lorentz Force: A Possible Driving Force for Sunspot Rotation

Zhao and Kosovichev (Astrophys. J. 591, 446, 2003) found two opposite sub-photospheric vortical flows in the depth range of 0 – 12 Mm around a fast rotating sunspot. So far there is no theoretical model explaining such flow motions. In this paper, we try to explain this phenomenon from the point of view of magnetic flux tubes interacting with large-scale vortical motions of plasma. In the deeper zone under the photosphere, the magnetic force may be less than the nonmagnetic force of plasma. The vortical flow located there twists the flux tube and magnetic free energy is built up in the tube. In the shallower zone under the photosphere, the magnetic force may be greater than the nonmagnetic force. Thus, part of the stored magnetic free energy is released to drive the plasma to rotate in two opposite directions, e.g., in the depth ranges of 0 – 3(5) and 9 – 12 Mm. In addition, we also define a vector of nonpotential magnetic stress τ, which can be related to flare occurrence. It is calculated for the active region NOAA 10930 on 11 December 2006. We find that: i) the integral of its line-of-sight (LOS) stress successively increases around the magnetic neutral line (MNL) prior to and during the flare and decreases to a minimum after the flare; ii) the integral of its transverse stress exceeds the integral of its LOS component by one order of magnitude over the whole field of view; iii) the transverse stress first points toward the MNL, then along it, and finally it points away from it. We need other data to verify whether or not the magnetic energy is transported in the horizontal direction to the neutral line, and then partly changes into the energy in LOS direction before and during the flare.     

A Study of Surges: II. On the Relationship between Chromospheric Surges and Coronal Mass Ejections

Liu et al. (Astrophys. J. 628, 1056, 2005a) described one surge – coronal mass ejection (CME) event showing a close relationship between solar chromospheric surge ejection and CME that had not been noted before. In this work, large Hα surges (>72 Mm, or 100 arcsec) are studied. Eight of these were associated with CMEs. According to their distinct morphological features, Hα surges can be classified into three types: jetlike, diffuse, and closed loop. It was found that all of the jetlike surges were associated with jetlike CMEs (with angular widths ≤30 degrees); the diffuse surges were all associated with wide-angle CMEs (e.g., halo); the closed-loop surges were not associated with CMEs. The exclusive relation between Hα surges and CMEs indicates difference in magnetic field configurations. The jetlike surges and related narrow CMEs propagate along coronal fields that are originally open. The unusual transverse mass motions in the diffuse surges are suggested to be due to magnetic reconnections in the corona that produce wide-angle CMEs. For the closed-loop surges, their paths are just outlining stable closed loops close to the solar surface. Thus no CMEs are associated with them.     

Short-Term Periodicities in Sunspot Activity and Flare Index Data during Solar Cycle 23

The short-term periodicities in sunspot numbers, sunspot areas, and flare index data are investigated in detail using the Date Compensated Discrete Fourier Transform (DCDFT) for the full disk of the Sun separately over the rising, the maximum, and the declining portions of solar cycle 23 (1996 – 2006). While sunspot numbers and areas show several significant periodicities in a wide range between 23.1 and 36.4 days, the flare index data do not exhibit any significant periodicity. The earlier conclusion of Pap, Tobiska, and Bouwer (1990, Solar Phys. 129, 165) and Kane (2003, J. Atmos. Solar-Terr. Phys. 65, 1169), that the 27-day periodicity is more pronounced in the declining portion of a solar cycle than in the rising and maximum ones, seems to be true for sunspot numbers and sunspot area data analyzed here during solar cycle 23.     

Effect of Gravity on the Fast Modes of a Horizontal Coronal Slab

Theoretical studies of the normal modes of a coronal slab often neglect gravity, as in Edwin and Roberts (Solar Phys. 71, 239, 1982). Here we study analytically the effect of gravity acting on a horizontal slab as a first step away from a homogeneous medium. Because of the inclusion of gravity, the symmetry of a homogeneous slab is lost, so the normal modes cannot be classified into kink and sausage modes. The presence of gravity also modifies the oscillatory frequencies of the slab, as well as the lower cutoff frequency, resulting in the possible transition between surface and body modes. For general coronal parameters, the dimensionless gravity term turns out to be small, so these effects are also small. A.J. Díaz’s current affiliation: Universitat de les Illes Balears, Palma, E-07122, Spain.     

Direct Propagation of Photospheric Acoustic <Emphasis Type="Italic">p</Emphasis> Modes into Nonmagnetic Solar Atmosphere

Solar p modes are one of the dominant types of coherent signals in Doppler velocity in the solar photosphere, with periods showing a power peak at five minutes. The propagation (or leakage) of these p-mode signals into the higher solar atmosphere is one of the key drivers of oscillatory motions in the higher solar chromosphere and corona. This paper examines numerically the direct propagation of acoustic waves driven harmonically at the photosphere, into the nonmagnetic solar atmosphere. Erdélyi et al. (Astron. Astrophys. 467, 1299, 2007) investigated the acoustic response to a single point-source driver. In the follow-up work here we generalise this previous study to more structured, coherent, photospheric drivers mimicking solar global oscillations. When our atmosphere is driven with a pair of point drivers separated in space, reflection at the transition region causes cavity oscillations in the lower chromosphere, and amplification and cavity resonance of waves at the transition region generate strong surface oscillations. When driven with a widely horizontally coherent velocity signal, cavity modes are caused in the chromosphere, surface waves occur at the transition region, and fine structures are generated extending from a dynamic transition region into the lower corona, even in the absence of a magnetic field.     

Prediction of Space Weather Using an Asymmetric Cone Model for Halo CMEs

Halo coronal mass ejections (HCMEs) are responsible of the most severe geomagnetic storms. A prediction of their geoeffectiveness and travel time to Earth’s vicinity is crucial to forecast space weather. Unfortunately, coronagraphic observations are subjected to projection effects and do not provide true characteristics of CMEs. Recently, Michalek (Solar Phys. 237, 101, 2006) developed an asymmetric cone model to obtain the space speed, width, and source location of HCMEs. We applied this technique to obtain the parameters of all front-sided HCMEs observed by the SOHO/LASCO experiment during a period from the beginning of 2001 until the end of 2002 (solar cycle 23). These parameters were applied for space weather forecasting. Our study finds that the space speeds are strongly correlated with the travel times of HCMEs to Earth’s vicinity and with the magnitudes related to geomagnetic disturbances.     

Simulating Solar Cycles in Northern and Southern Hemispheres by Assimilating Magnetic Data into a Calibrated Flux-Transport Dynamo

We use the flux-transport dynamo prediction scheme introduced by Dikpati, de Toma, and Gilman (Geophys. Res. Lett. 33, L05102, 2006) to make separate simulations and predictions of sunspot cycle peaks for northern and southern hemispheres. Despite the division of the data, the skill level achieved is only slightly lower than that achieved for the sum of both hemispheres. The model shows skill at simulating and predicting the difference in peaks between North and South, provided that difference is more than a few percent. The simulation and prediction skill is achieved without adjustment to any parameters of the model that were used when peaks for the sum of North and South sunspot areas was simulated. The results are also very insensitive to the averaging length applied to the input data, provided the simulations and predictions are for peaks defined by averaging the observations over at least 13 rotations. However, in its present form, the model is not capable of skillfully simulating or predicting short-time-scale features of individual solar cycles.     

Evidence that Synchrotron Emission from Nonthermal Electrons Produces the Increasing Submillimeter Spectral Component in Solar Flares

We investigate the origin of the increasing spectra observed at submillimeter wavelengths detected in the flare on 2 November 2003 starting at 17:17 UT. This flare, classified as an X8.3 and 2B event, was simultaneously detected by RHESSI and the Solar Submillimeter Telescope (SST) at 212 and 405 GHz. Comparison of the time profiles at various wavelengths shows that the submillimeter emission resembles that of the high-energy X rays observed by RHESSI whereas the microwaves observed by the Owens Valley Solar Array (OVSA) resemble that of ∼50 keV X rays. Moreover, the centroid position of the submillimeter radiation is seen to originate within the same flaring loops of the ultraviolet and X-ray sources. Nevertheless, the submillimeter spectra are distinct from the usual microwave spectra, appearing to be a distinct spectral component with peak frequency in the THz range. Three possibilities to explain this increasing radio spectra are discussed: (1) gyrosynchrotron radiation from accelerated electrons, (2) bremsstrahlung from thermal electrons, and (3) gyrosynchrotron emission from the positrons produced by pion or radioactive decay after nuclear interactions. The latter possibility is ruled out on the grounds that to explain the submillimeter observations requires 3000 to 2×10⁵ more positrons than what is inferred from X-ray and γ-ray observations. It is possible to model the emission as thermal; however, such sources would produce too much flux in the ultraviolet and soft X-ray wavelengths. Nevertheless we are able to explain both spectral components at microwave and submillimeter wavelengths by gyrosynchrotron emission from the same population of accelerated electrons that emit hard X rays and γ rays. We find that the same 5×10³⁵ electrons inferred from RHESSI observations are responsible for the compact submillimeter source (0.5 arcsec in radius) in a region of 4500 G low in the atmosphere, and for the traditional microwave spectral component by a more extended source (50 arcsec) in a 480 G magnetic field located higher up in the loops. The extreme values in magnetic field and source size required to account for the submillimeter emission can be relaxed if anisotropy and transport of the electrons are taken into account.     

Spike-like Bursts as Fine Structure of Zebras

We studied the characteristics of the zebra-associated spike-like bursts that were recorded with high time resolution at 1420 MHz in four intervals (from 12:45 to 12:48 UT) during 5 August 2003. Our detailed analysis is based on the selection of more than 500 such spike-like bursts and it is, at least to our knowledge, the first study devoted to such short-lived bursts. Their characteristics are different from those pertinent to “normal” spike bursts, as presented in the paper by Güdel and Benz (Astron. Astrophys. 231, 202, 1990); in particular, their duration (about 7.4 ms at half power) is shorter, so they should be members of the SSS (super short structures) family (Magdalenić et al., Astrophys. J. 642, L77, 2006). The bursts were generally strongly R-polarized; however, during the decaying part of interval I a low R-polarized and L-polarized bursts were also present. This change of polarization shows a trend that resembles the peculiar form of the zebra lines in the spectral dominion (“V” like). A global statistical analysis on the bursts observed in the two polarimetric channels shows that the highest cross-correlation coefficient (about 0.5) was pertinent to interval I. The zebras and the bursts can be interpreted by the same double plasma resonance process as proposed by Bárta and Karlicky (Astron. Astrophys. 379, 1045, 2001) and Karlicky et al. (Astron. Astrophys. 375, 638, 2001); in particular, the spikes are generated by the interruption of this process by assumed turbulence (density or magnetic field variations). This process should be present in the region close to the reconnection site (e.g., in the plasma reconnection outflows) where the density and the magnetic field vary strongly.     

Inferring Small-Scale Flatfields from Solar Rotation

We present a method to infer small-scale flatfields for imaging solar instruments using only regular-observation intensity images with a fixed field of view. The method is related to the flatfielding method developed by Kuhn, Lin, and Loranz (Publ. Astron. Soc. Pac. 103, 1097 – 1108, 1991), but does not require image offsets. Instead, it takes advantage of the fact that the solar image is changing in the CCD reference frame due to solar rotation. We apply the method to data sets of MDI filtergrams and compare the results to flat fields derived with other methods. Finally, we discuss the planned implementation of this method in the data processing for Helioseismic and Magnetic Imager on the Solar Dynamics Observatory.     

A New Forecasting Index for Solar Wind Velocity Based on EIT 284 Å Observations

Various solar wind forecasting methods have been developed during the past decade, such as the Wang?–?Sheeley model and the Hakamada?–?Akasofu?–?Fry Version 2 (HAFv2) model. Also, considerable correlation has been found between the solar wind speed v and the coronal hole (CH) area A _M on the visible side of the Sun, showing quantitative improvement of forecasting accuracy in low CME activity periods (e.g., Vr?nak, Temmer, and Veronig, Solar Phys. 240, 315, 2007a). Properties of lower layers of the solar atmosphere are good indications of the subsequent interplanetary and geomagnetic activities. We analyze the SOHO/EIT 284 Å images and construct a new forecasting factor (Pch) from the brightness of the solar EUV emission, and a good correlation is found between the Pch factor and the 3-day-lag solar wind velocity (v) probed by the ACE spacecraft. The main difference between the Pch and A _M factor is that Pch does not depend on the CH-boundary estimate and can reflect both the area and brightness of CH. A simple method of forecasting the solar wind speed near Earth in low CME activity periods is presented. Between Pch and v from 21 November until 26 December 2003, the linear correlation coefficient is R=0.89. For comparison we also analyze the data in the same period (DOY 25?–?125, 2005) as Vr?nak, Temmer, and Veronig (Solar Phys. 240, 315, 2007a), who used the CH areas A _M for predicting the solar wind parameters. In this period the correlation coefficient between Pch and v is R=0.70, whereas for A _M and v the correlation coefficient is R=0.62. The average relative difference between the calculated and the observed values is $\overline{|\delta|}\approx 12.15\%Various solar wind forecasting methods have been developed during the past decade, such as the Wang – Sheeley model and the Hakamada – Akasofu – Fry Version 2 (HAFv2) model. Also, considerable correlation has been found between the solar wind speed v and the coronal hole (CH) area A _M on the visible side of the Sun, showing quantitative improvement of forecasting accuracy in low CME activity periods (e.g., Vršnak, Temmer, and Veronig, Solar Phys. 240, 315, 2007a). Properties of lower layers of the solar atmosphere are good indications of the subsequent interplanetary and geomagnetic activities. We analyze the SOHO/EIT 284 ? images and construct a new forecasting factor (Pch) from the brightness of the solar EUV emission, and a good correlation is found between the Pch factor and the 3-day-lag solar wind velocity (v) probed by the ACE spacecraft. The main difference between the Pch and A _M factor is that Pch does not depend on the CH-boundary estimate and can reflect both the area and brightness of CH. A simple method of forecasting the solar wind speed near Earth in low CME activity periods is presented. Between Pch and v from 21 November until 26 December 2003, the linear correlation coefficient is R=0.89. For comparison we also analyze the data in the same period (DOY 25 – 125, 2005) as Vršnak, Temmer, and Veronig (Solar Phys. 240, 315, 2007a), who used the CH areas A _M for predicting the solar wind parameters. In this period the correlation coefficient between Pch and v is R=0.70, whereas for A _M and v the correlation coefficient is R=0.62. The average relative difference between the calculated and the observed values is . Furthermore, for the ten peaks during the analysis period, Pch and v show a correlation coefficient of R=0.78, and the average relative difference between the calculated and the observed peak values is . Moreover, the Pch factor can eliminate personal bias in the forecasting process, which existed in the method using CH area as input parameter, because CH area depends on the CH-boundary estimate but Pch does not. Until now the CH-boundary could not be easily determined since no quantitative criteria can be used to precisely locate CHs from observations, which led to differences in forecasting accuracy.     

The Solar Optical Telescope for the Hinode Mission: An Overview

The Solar Optical Telescope (SOT) aboard the Hinode satellite (formerly called Solar-B) consists of the Optical Telescope Assembly (OTA) and the Focal Plane Package (FPP). The OTA is a 50-cm diffraction-limited Gregorian telescope, and the FPP includes the narrowband filtergraph (NFI) and the broadband filtergraph (BFI), plus the Stokes Spectro-Polarimeter (SP). The SOT provides unprecedented high-resolution photometric and vector magnetic images of the photosphere and chromosphere with a very stable point spread function and is equipped with an image-stabilization system with performance better than 0.01 arcsec rms. Together with the other two instruments on Hinode (the X-Ray Telescope (XRT) and the EUV Imaging Spectrometer (EIS)), the SOT is poised to address many fundamental questions about solar magnetohydrodynamics. This paper provides an overview; the details of the instrument are presented in a series of companion papers. M. Otsubo is a former NAOJ staff scientist.     

Statistical Assessment of Photospheric Magnetic Features in Imminent Solar Flare Predictions

In this study we use the ordinal logistic regression method to establish a prediction model, which estimates the probability for each solar active region to produce X-, M-, or C-class flares during the next 1-day time period. The three predictive parameters are (1) the total unsigned magnetic flux T _flux, which is a measure of an active region’s size, (2) the length of the strong-gradient neutral line L _gnl, which describes the global nonpotentiality of an active region, and (3) the total magnetic dissipation E _diss, which is another proxy of an active region’s nonpotentiality. These parameters are all derived from SOHO MDI magnetograms. The ordinal response variable is the different level of solar flare magnitude. By analyzing 174 active regions, L _gnl is proven to be the most powerful predictor, if only one predictor is chosen. Compared with the current prediction methods used by the Solar Monitor at the Solar Data Analysis Center (SDAC) and NOAA’s Space Weather Prediction Center (SWPC), the ordinal logistic model using L _gnl, T _flux, and E _diss as predictors demonstrated its automatic functionality, simplicity, and fairly high prediction accuracy. To our knowledge, this is the first time the ordinal logistic regression model has been used in solar physics to predict solar flares.     

Relationship between Powerful Flares and Dynamic Evolution of the Magnetic Field at the Solar Surface

Powerful flares are closely related to the evolution of the complex magnetic field configuration at the solar surface. The strength of the magnetic field and speed of its evolution are two vital parameters in the study of the change of magnetic field in the solar atmosphere. We propose a dynamic and quantitative depiction of the changes in complexity of the active region: E=u×B, where u is the velocity of the footpoint motion of the magnetic field lines and B is the magnetic field. E represents the dynamic evolution of the velocity field and the magnetic field, shows the sweeping motions of magnetic footpoints, exhibits the buildup process of current, and relates to the changes in nonpotentiality of the active region in the photosphere. It is actually the induced electric field in the photosphere. It can be deduced observationally from velocities computed by the local correlation tracking (LCT) technique and vector magnetic fields derived from vector magnetograms. The relationship between E and ten X-class flares of four active regions (NOAA 10720, 10486, 9077, and 8100) has been studied. It is found that (1) the initial brightenings of flare kernels are roughly located near the inversion lines where the intensities of E are very high, (2) the daily averages of the mean densities of E and its normal component (E _n) decrease after flares for most cases we studied, whereas those of the tangential component of E (E _t) show no obvious regularities before and after flares, and (3) the daily averages of the mean densities of E _t are always higher than those of E _n, which cannot be naturally deduced by the daily averages of the mean densities of B _n and B _t.     

Microwave Type III-Like Bursts as Possible Signatures of Magnetic Reconnection

Magnetic reconnection is commonly accepted to play a key role in flare energy release, but only poor information about the main characteristics of this process is available so far. An intrinsic feature of reconnection is plasma density enhancement in current sheets. A unique method to detect this effect is provided by analysis of drifting bursts, whose emission frequency is close to the local Langmuir frequency or its harmonics. With this purpose, we analyze a series of several tens of drifting microwave bursts during the 30 March 2001 flare. The burst drift rates range from −10 to 20 GHz s⁻¹. Using one-dimensional scans recorded with the SSRT interferometer at two different frequencies near 5.7 GHz, we have measured relative positions of burst sources and their velocities along a flare loop revealed from soft X-ray and extreme-ultraviolet images. It is argued that the contribution of the increasing density effect into the observed frequency drift rates is about 6 GHz s⁻¹, which is shown to be consistent with theoretical models of magnetic reconnection with reasonable boundary conditions.