temporal spectral shift and polarization of a band-splitting ... - IOPscience

25 downloads 47 Views 2MB Size Report
3 Department of Space Science and CSPAR, University of Alabama in Huntsville, Huntsville, AL 35899, USA. Received 2014 September 2; accepted 2014 ...
The Astrophysical Journal Letters, 793:L39 (5pp), 2014 October 1  C 2014.

doi:10.1088/2041-8205/793/2/L39

The American Astronomical Society. All rights reserved. Printed in the U.S.A.

TEMPORAL SPECTRAL SHIFT AND POLARIZATION OF A BAND-SPLITTING SOLAR TYPE II RADIO BURST Guohui Du1 , Yao Chen1 , Maoshui Lv1 , Xiangliang Kong1 , Shiwei Feng1 , Fan Guo2 , and Gang Li3 1

Shandong Provincial Key Laboratory of Optical Astronomy and Solar-Terrestrial Environment, and Institute of Space Sciences, Shandong University, Weihai 264209, China; [email protected] 2 Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM 87545, USA 3 Department of Space Science and CSPAR, University of Alabama in Huntsville, Huntsville, AL 35899, USA Received 2014 September 2; accepted 2014 September 8; published 2014 September 18

ABSTRACT In many type II solar radio bursts, the fundamental and/or the harmonic branches of the bursts can split into two almost parallel bands with similar spectral shapes and frequency drifts. However, the mechanisms accounting for this intriguing phenomenon remain elusive. In this study, we report a special band-splitting type II event in which spectral features appear systematically earlier on the upper band (with higher frequencies) than on the lower band (with lower frequencies) by several seconds. Furthermore, the emissions carried by the splitting band are moderately polarized with the left-hand polarized signals stronger than the right-hand ones. The polarization degree varies in a range of −0.3 to −0.6. These novel observational findings provide important constraints on the underlying physical mechanisms of band-splitting of type II radio bursts. Key words: shock waves – Sun: coronal mass ejections (CMEs) – Sun: radio radiation Online-only material: color figures

plasma densities and, therefore, the emission frequencies are different (Smerd et al. 1974, 1975). The similarity between the two bands in most events and the fact that the frequency ratio between the two bands do not vary too much from one event to another implies that the coronal structure proposal is unlikely. The other proposal, referred to as the upstream–downstream (UD) scenario, assumes that the lower band with a smaller frequency is from the shock upstream and the upper band is from the shock downstream or the shock ramp. This scenario has been employed as a working assumption by many authors in their efforts of diagnosing the shock and coronal parameters, such as the shock compression ratio, the coronal Alfv´en speed, and the magnetic field strength (e.g., Smerd et al. 1974; Vrˇsnak et al. 2002, 2004; Cho et al. 2007; Liu et al. 2009; Ma et al. 2011; Vasanth et al. 2014). However, the UD scenario does not receive strong support from theoretical and observational studies (see Cairns 2011, for a recent review). In most, if not all, theoretical models of solar type II radio bursts, radio emissions are produced by energetic electrons and enhanced Langmuir oscillations in the shock upstream (Cairns & Melrose 1985; Cairns 1988; Knock & Cairns 2005). The mechanism for electron acceleration and production of bump-on-tail distribution required to excite Langmuir waves is thought to be shock reflection to the upstream region (Wu 1984; see also Guo & Giacalone 2010). In the downstream region, the electron distribution is rather isotropic and generation of enhanced Langmuir oscillations (compared to the thermal level) is not observed downstream of interplanetary shocks or the bow shock (Filbert & Kellogg 1979; Treumann et al. 1986; Cairns 1986; Bale et al. 1999; Pulupa & Bale 2008). Furthermore, with this scenario, the splitting width should strongly depend on the shock compression ratio, which may be considerably different from event to event. This is inconsistent with the observed nearly constant relative splitting width for metric type IIs (e.g., Vrˇsnak et al. 2002). In addition, the compression ratio deduced with this scenario is relatively small, in a range of 1.2–1.7. It is unlikely for such weak shocks to

1. INTRODUCTION It is generally believed that solar type II radio bursts are excited by plasma emissions of energetic electrons accelerated at shocks in the corona (metric-type IIs) and interplanetary (IP) space (e.g., Wild et al. 1963; Ginzberg & Zhelezniakov 1958; Wild & Smerd 1972; Nelson & Melrose 1985). According to the plasma emission mechanism, type II frequencies and, therefore, the overall spectral shape are largely determined by plasma densities along the shock path. Type IIs frequently display fundamental and harmonic branches. Both branches may further split into two almost parallel upper (higher frequency) and lower (lower frequency) bands with correlated variations of spectral features like the intensity and shape changes, frequency drift, and other intermittent spectral structures (cf. Nelson & Melrose 1985). The relative splitting width (defined as the ratio of their frequency difference to the frequency of the lower band) is found to vary within a relatively fixed range of values (∼0.1−0.3) for metric type II events, and to increase to a slightly larger value (0.4−0.6) for hecto-decametric to kilometric type IIs (also called IP type IIs), according to a series of studies by Vrˇsnak et al. (2001, 2002, 2004). Such spectral substructures contain important information on type II radio emission and the shock-acceleration process. However, the exact physical mechanism accounting for this fascinating phenomenon remains unknown. Proposed scenarios for band splitting type II bursts can be divided into two classes. In the first class, the splitting is caused by geometrical effects (McLean 1967) and in the second class, the splitting is inherently associated with the emission process (cf. Treumann & Labelle 1992). For the geometric explanation, it has been suggested that the splitting originates either at (1) two separate sources along the shock fronts, possibly due to interactions with different coronal structures (McLean 1967; Wild & Smerd 1972; Nelson & Robinson 1975), or (2) at different locations ahead (upstream) of or behind (downstream from) the shock front where the 1

The Astrophysical Journal Letters, 793:L39 (5pp), 2014 October 1

Du et al.

Figure 1. (a) EIT 171 Å image taken at 13:00 UT, superposed by the contours of the NRH image shown in panel (c). (b) LASCO C2 difference image with the blue arrow pointing to the eruption front. (c) NRH image of the brightness temperature (TB ) at 150.9 MHz and 14:44:20 UT with contours given by the 80% (black), 90% (purple), and 95% (green) levels of the corresponding intensity maximum. (A color version of this figure is available in the online journal.)

including (1) two narrow and slowly drifting spectral bands, (2) similar intensity and shape characteristics along the two bands, and (3) a frequency ratio of the upper and lower bands that is considerably less than 2 to avoid confusion with the fundamental and harmonic branches (see, e.g., Vrˇsnak et al. 2001). In addition, we require that both split bands last for several minutes, possessing prominent intensity or shape changes such as curved, blob- or patch-like or intermittent features. This is necessary for a successful identification of band-split events with temporal shift of spectral features. Among the many BS events, we identified one specific event with convincing spectral shift features. Such a shift can be defined when correlated spectral features such as intensity variations and shape changes appear systematically at an earlier or later time in one band than the other. The following factors severely limit the number of candidate events for the purpose of our study. In many cases, (1) no appropriate prominent spectral features exist. For instance, in events where the split bands are rather smooth or straight, it is not possible to identify any spectral shift even if such a shift does exist. (2) The interference from other radio bursts, artificial signals, or the ionospheric effects are too strong to allow the identification of clean spectral features. (3) The band-splits are too wide or the temporal shift is too small. In these cases, we are unable to judge whether a temporal shift of spectral features exists or not. Because of these limitations, only one event with a clear temporal shift of spectral features is identified. The event occurred on 2005 May 31. We examine it in detail in the next section.

strongly accelerate electrons according to the well-known shock drift acceleration mechanism (Wu 1984). For the other type of band-splitting (BS) scenario, i.e., the band splits are caused by an intrinsic emission mechanism; a Langmuir caviton emission process has been proposed (Treumann & Labelle 1992). The caviton represents density depletion formed by the nonlinear evolution of Langmuir waves that are induced by energetic electron beams. The BS can be explained as simultaneous emissions from outside (upper band) and inside (lower band) the Langmuir caviton. Another possibility, as suggested by Cairns (1994) based on similar splitting features produced by the Earth’s bow shock, is that the frequency difference represents the halves of the electron cyclotron frequencies (see also Cairns 2011). Neither possibility has received any direct observational or theoretical verification. Given the above controversial scenarios, it is important to explore the type II radio data searching for further observational constraints on the underlying physics. Here we start by looking for a possible consequence to the splitting spectra under the UD scenario. It is known that large density inhomogeneities or structures along the shock path can affect the type-II spectral shape according to the plasma emission mechanism (Ginzberg & Zhelezniakov (1958); Nelson & Melrose (1985), see Kong et al. (2012), and Feng et al. (2012, 2013) for some of the latest observational studies). A density structure, when swept by a shock, will first present in the upstream region and then in the downstream region. Thus, according to the UD scenario, the corresponding spectral feature will first appear in the splitting lower band and then in the upper band. If the distance between the BS sources is large enough, this may lead to observable temporal shift of correlated spectral features on the two split bands. This provides a test for the UD scenario. Here we examine the UD scenario with one band-splitting event where clear temporal shifts are present.

3. THE 2005 MAY 31 EVENT This type II event was associated with a coronal mass ejection (CME) and a solar flare from NOAA AR10771 at N24W47. The pre-eruption image of the corona obtained by EIT/SOHO at 171 Å is shown in Figure 1(a). During the eruption, the EIT only took images at 13:00 and 19:00 UT, yielding no useful data for the erupting process. The LASCO/C2 took images at 14:32, 15:32, and 18:32 UT. The difference image of the LASCO/C2 data obtained at 14:32 and 15:32 UT is shown in Figure 1(b) with the eruption front clearly observed and denoted by the blue arrow. The GOES X-ray profiles (http://cdaw.gsfc.nasa.gov/images/goes/xrs/2005/05/31/) show a C2.4 flare with start, peak, and end times at 14:30, 14:42,

2. EVENT IDENTIFICATION To select candidate events for our study, we examined the online database of metric to decametric type IIs recorded by the Green Bank Solar Radio Burst Spectrometer (GBSRBS: http://www.astro.umd.edu/∼white/gb/), the ARTEMIS IV Multichannel solar radiospectrograph (Caroubalos et al. 2001), and the Nancay Decemetric Array (NDA: Lecacheux, A. 2000). BS events are identified according to their general characteristics 2

The Astrophysical Journal Letters, 793:L39 (5pp), 2014 October 1

Du et al.

Figure 2. Dynamic spectra recorded by Artemis for 170–70 MHz and GBSRBS for 70–20 MHz. The bottom orange solid line is the fitting curve using the 1 × Saito density model and a shock speed of 633 km s−1 . The line on the fundamental upper band is given by 1.26 times this fitting curve, and the upper two lines are given by 1.92 times the lower two lines. (A color version of this figure is available in the online journal.)

the C2.4 flare and the CME shown in Figure 1. Due to the lack of high-cadence imaging data of the eruption, we are not able to tell whether the type II was driven by the CME or by the flare. Besides the band-splits, the most obvious feature of the type II spectra is that both bands present very similar and clear morphology features (such as a sudden change in drift rate and bumped spectra) and intermittent intensity variations. This makes it possible to identify the time shift between the split bands. Visual inspection of the spectra clearly shows that various features on the upper band appear earlier than those on the lower band. To better show the correlation between the spectral features on the split bands, in Figure 3 we superpose the contour of the lower band onto the spectra. The contours are given by intensity levels of 12%, 40%, and 60% of the corresponding intensity maximum and are shown in yellow, red, and black, respectively. We then shift these contours in time (x-axis) and log frequency (corresponding to a multiplication of frequency) until most of the features of the upper band can be covered with the shifted contours. The shift in time is found to be ∼7 s and the multiplication in frequency is 1.26. The shifted contours are shown in Figure 3 on the upper band. This exercise shows clearly that the same spectral features appeared ∼7 s earlier in the upper band than in the lower band. This is the most important finding of this paper. A two-dimensional (2D) correlation analysis confirms the above result, as shown in Figure 4. To obtain this figure, we calculate the correlation coefficients between two images, one containing only the upper band spectrum and the other containing the “shifted” lower band spectrum. The “shifted” lower band image is obtained using the same procedure as

and 15:20 UT, respectively. In Figure 1(c), we present the Nancay Radioheliograph (NRH) image obtained at 14:44:20 UT and 150.9 MHz. The intensity of the radio burst is indicated by the color; red denotes the strongest with a brightness temperature (Tb ) close to 6 × 108 K. The spatial resolution of the NRH data at this frequency is 5.5–3.2 arcmin and the projected location of the maximum intensity is at N19W34, about 5◦ southward and 13◦ westward of AR10711. The properties of this NRH burst can be seen from the dynamic spectrum of this type II burst (see Figure 2), which is given by the composite of data from Artemis IV (170 MHz–70 MHz) and GBSRBS (70 MHz–20 MHz). The temporal resolution of both data sets is 1 s. We see that the type II exhibits very clear fundamental (F) and harmonic (H) branches. The F branch showed nice band splits. In comparison, the upper band of the H branch is only marginally discernible. The F branch of the radio burst starts from 14:47 UT at 70 (58) MHz and ends at ∼14:54 UT and 25 (20) MHz for the upper (lower) band. The frequency drift rate of the upper (lower) band of the F branch is about 0.11 MHz s−1 (0.09 MHz s−1 ). The average frequency ratio of the two bands is ∼1.26. The bottom orange solid line is the fitting curve using the 1 × Saito density model and a shock speed of 633 km s−1 . The line on the fundamental upper band is given by 1.26 times this fitting curve and the upper two lines are given by 1.92 times the lower two lines. These lines are drawn to better identify the splitting counterparts on both the F and H branches. The plus sign indicates the frequency (150.9 MHz) and time (14:44:20 UT) of the NRH image shown in Figure 1(c). It corresponds to the lower band of the H branch of this type II burst. This is consistent with the type II being associated with 3

The Astrophysical Journal Letters, 793:L39 (5pp), 2014 October 1

Du et al.

Figure 3. Dynamic spectrum recorded by GBSRBS. The contour of the lower band is given by intensity levels of 12% (yellow), 40% (red), and 60% (black) of the maximum intensity. The corresponding contours on the upper band are the shifted contours of these lower band contours (see the text for details). (A color version of this figure is available in the online journal.)

1992; P. Zarka 2014, private communication). In Figures 5(a) and (b), we show the right-handed and left-handed polarized signals of this burst from 14:40 to 15:02 UT with a temporal resolution of 1 s. The polarization degree is measured by the ratio of the difference between the two linear polarization intensities and their sum (i.e., the total linear intensity). In our convention, positive means right-handed polarized and negative means lefthanded polarized. We select three subsequent spectral segments, labeled as A (14:46–14:48 UT), B (14:48–14:51 UT), and C (14:51– 14:53 UT) in Figure 5, to calculate their average splitting widths and polarization levels. It is found that the average relative splitting width varies in a relatively small range of 0.25–0.27, with little variation during the event, and the average polarization degree varies from −32% to −58%, with an average of the three segments being −50% for the upper band and −64% for the lower band. 4. CONCLUSIONS AND DISCUSSION In this study we report a unique solar radio type II burst with a band-splitting feature that occurred on 2005 May 31. The split bands present prominent morphological structures. This allows us to determine whether a temporal shift of these spectral features exists. We find that the spectral features carried by the upper band appear systematically earlier by several (∼7) seconds than those on the lower band, and the band-splitting signals are moderately polarized with left-handed polarized emission that is stronger than the right-handed polarized emission and the polarization degree is found to vary between −0.3 and −0.6. To our knowledge, no previous works concerning time shifts in band splitting exist. The event reported here provides important observational constraints on the underlying physics. Indeed, the obtained shift puts up a critical challenge to the popular UD scenario of the band splitting, which predicts that certain spectral features caused by the shock-sweeping plasma structures should always appear first in the lower band (emitted from the upstream region) and followed in the upper band (emitted from the downstream region), contrary to our finding.

Figure 4. Correlation coefficients as a function of the time shift (Δt) and the frequency multiplication factor (α) deduced by the correlation analysis. (A color version of this figure is available in the online journal.)

described above; namely it is shifted by a certain time (Δt) and its frequencies are multiplied by a factor of α. The correlation coefficient between the shifted lower band image and the upper band image is then deduced and shown in Figure 4 as a 2D function of Δt and α. The maximum of the correlation coefficients is 0.77, occurring at Δt = −7 s and α = 1.26. These two values are the same as those obtained with the contour method. We further note that the α at the maximum correlation equals the average splitting width deduced above. The type II spectrum of this event was also recorded by NDA. NDA consists of 144 conical helix antennas, divided into two groups of 72 antennas, each group detecting one sense of circular polarization (left-handed or right-handed) signals. The accuracy of the polarization level measured by NRH is ∼10% (Dulk et al. 4

The Astrophysical Journal Letters, 793:L39 (5pp), 2014 October 1

Du et al.

Figure 5. Left-handed and right-handed polarized signals recorded by NDA. The spectra are separated into three segments (labeled as A (14:46–14:48 UT), B (14:48–14:51 UT), and C (14:51–14:53 UT)) to calculate the average of parameters. (A color version of this figure is available in the online journal.)

This may suggest that the UD explanation is incorrect. Of course, more events with high spectral and temporal resolution data that contain prominent spectral features should be further examined before a more definite conclusion can be reached.

Filbert, P. C., & Kellogg, P. J. 1979, JGR, 84, 1369 Ginzburg, V. L., & Zhelezniakov, V. V. 1958, SvA, 2, 653 Guo, F., & Giacalone, J. 2010, ApJ, 715, 406 Knock, S. A., & Cairns, I. H. 2005, JGR, 110, A01101 Kong, X. L., Chen, Y., Li, G., et al. 2012, ApJ, 750, 158 Lecacheux, A. 2000, in Radio Astronomy at Long Wavelengths, ed. R. G. Stone, K. W. Weiler, M. L. Goldstein, & J.-L. Bougeret (Washington, DC: AGU), 321 Liu, Y., Luhmann, J. G., Bale, S. D., & Lin, R. P. 2009, ApJL, 691, L151 Ma, S., Raymond, J. C., Golub, L., et al. 2011, ApJ, 738, 160 McLean, D. J. 1967, PASAu, 1, 47 Nelson, G. S., & Melrose, D. B. 1985, in Solar Radiophysics, ed. D. J. McLean & N. R. Labrum (Cambridge: Cambridge Univ. Press), 333 Nelson, G. J., & Robinson, R. D. 1975, PASA, 2, 370 Pulupa, M., & Bale, S. D. 2008, ApJ, 676, 1330 Smerd, S. F., Sheridan, K. V., & Stewart, R. T. 1974, in IAU Symp. 57, Coronal Disturbances, ed. G. A. Newkirk (Dordrecht: Reidel), 389 Smerd, S. F., Sheridan, K. V., & Stewart, R. T. 1975, ApL, 16, 23 Treumann, R. A., Bauer, O. H., Labelle, J., et al. 1986, AdSpR, 6, 93 Treumann, R. A., & Labelle, J. 1992, ApJL, 399, L167 Vasanth, V., Umapathy, S., Vrˇsnak, B., et al. 2014, SoPh, 289, 251 Vrˇsnak, B., Aurass, H., Magdaleni´c, J., & Gopalswamy, N. 2001, A&A, 377, 321 Vrˇsnak, B., Aurass, H., Magdaleni´c, J., & Mann, G. 2002, A&A, 396, 673 Vrˇsnak, B., Magdaleni´c, J., & Zlobec, P. 2004, A&A, 413, 753 Wild, J. P., & Smerd, S. F. 1972, ARA&A, 10, 159 Wild, J. P., Smerd, S. F., & Weiss, A. A. 1963, ARA&A, 1, 291 Wu, C. S. 1984, JGR, 89, 8857

We are grateful to the SOHO, NRH, NDA, GBSRBS, and the Artemis IV teams for making their data available to us. This work was supported by grants NSBRSF 2012CB825601, NNSFC 41274175, 41331068, and U1431103. G.L.’s work at the University of Alabama Huntsville is supported by NSF grants ATM-0847719 and AGS1135432. REFERENCES Bale, S. D., Reiner, M. J., Bougeret, J.-L., et al. 1999, GeoRL, 26, 1573 Cairns, I. H., & Melrose, D. B. 1985, JGR, 90, 6637 Cairns, I. H. 1986, JGR, 91, 2975 Cairns, I. H. 1988, JGR, 93, 858 Cairns, I. H. 1994, JGR, 99, 23505 Cairns, I. H. 2011, in The Sun, the Solar Wind, and the Heliosphere, ed. M. P. Miralles & J. S. Almeida (Berlin: Springer), 267 Caroubalos, C., Maroulis, D., Patavalis, N., et al. 2001, ExA, 11, 23 Cho, K. S., Lee, J., Moon, Y. J., et al. 2007, A&A, 461, 1121 Dulk, G. A., Lecacheux, A., & leblanc, Y. 1992, A&A, 253, 292 Feng, S. W., Chen, Y., Kong, X. L., et al. 2012, ApJ, 753, 21 Feng, S. W., Chen, Y., Kong, X. L., et al. 2013, ApJ, 767, 29

5