Waiting times of Quasihomologous Coronal Mass Ejections from Super Active Regions【推荐论文】 .doc

精品论文Waiting times of Quasi-homologous Coronal Mass Ejections from Super Active RegionsWANG Yuming , LIU Lijuan , SHEN Chenglong , LIU Rui , WANG Shui(CAS Key Laboratory of Geospace Environment, Department of Geophysics and PlanetarySciences, University of Science and Technolgy of China, Hefei, Anhui 230026)Abstract: Why and how may some active regions (ARs) frequently produce coronal mass ejections(CMEs)? It is one of the key questions to deepen our understanding of the mechanisms and processes10of energy accumulation and sudden release in ARs and to improve our capability of space weather prediction. Although some case studies have been made, the question is still far from fully answered.This issue is now being tried to address statistically through an investigation of waiting times ofquasi-homologous CMEs from super ARs in solar cycle 23. It is found that the waiting times of quasi-homologous CMEs have a two-component distribution with a separation at about 18 hours, the15peak waiting time of the first component is at about 7 hours, and the likelihood of occurrences of two or more CMEs faster than 1200 km /s within 18 hours is about 20%. Furthermore, the correlationanalysis among CME waiting times, CME speeds and CME occurrence rates reveals that these quantities are independent to each other, suggesting that the perturbation by preceding CMEs ratherthan free energy input be the direct cause of quasi-homologous CMEs. The peak waiting time of 720hours probably characterize the time scale of the growth of instabilities triggered by preceding CMEs.This study uncovers more clues for us to understand quasi-homologous CMEs as well as CME-richARs.Key words: Space Physics; Sun; coronal mass ejections; instabilities250IntroductionMagnetic free energy is thought to be the energy source of coronal mass ejections (CMEs). Active regions (ARs) carry a huge amount of free energy and therefore are the most probable place where CMEs come out. Lots of efforts have been devoted to the triggering mechanisms of CMEs. Flux emergence, shear motion and mass loss all could be the initial cause of an isolated30CME 1234. No matter which one takes effect, the determinative factor of the CMEs launch isthe force balance between the inner core field and the outer overlying arcades 567. Free energy stored in the source region will be consumed when a CME launches 8.The picture of isolated CMEs is somewhat clear. However, it is still a question how CMEs could lift successively in a limited region within a relatively short interval. Usually the energy35accumulation is a gradual process in time scale of hours to days 910, while a CME is a suddenprocess releasing accumulated energy in minutes. Why and how could some ARs frequently produce CMEs? Does the occurrences of successive CMEs from the same AR mean that the source AR accumulate free energy quickly? The waiting time distribution of quasi-homologous CMEs contains clues.40Homologous CMEs were defined by Zhang and Wang(2002)11after the definition ofhomologous flares 12. Strictly speaking, homologous CMEs must originate from the same region, have similar morphology, and be associated with homologous flares and EUV dimmings. Here, we use the term quasi-homologous to refer to successive CMEs originating from the same ARs within a short interval, but may have different morphology and associates.Foundations: CAS Key Research Program(KZZD-EW-01), 100-Talent Program, NSFC(41131065,40904046,40874075,41121003), 973 key project(2001CB811403), Fundamental research funds for the central universities.Brief author introduction:Yuming Wang(1976-), Male, Professor, Associate Dean,School of Earth&SpaceScience,University of Science&Technology of China. Deputy Director, Key Laboratory of Geospace Environment, Chinese Academy of Sciences.Main research: Geospace Environment & Solar Physics. E-mail:ymwang- 10 -45A previous study on 15 CME-rich ARs during the ascending phase of the last solar cycle from 1998 to 1999 have suggested that quasi-homologous CMEs occurred at a pace of about 8 hours, and there was at most one fast CME within 15 hours13. These result are important for space weather prediction, and did imply that the accumulation rate of free energy in an AR may not support such frequently occurrences of quasi-homologous CMEs, and the triggering50mechanisms of the first and the following CMEs are probably different. Three scenarios were proposed to interpret the averagely 8-hour waiting time of quasi-homologous CMEs.Before deepening our understanding of such a phenomenon, we need to check if a similar waiting time distribution of quasi-homologous CMEs could be obtained for the whole solar cycle. In this paper, we extend the period of interest to the whole solar cycle 23 from 1996 to 2006.55Instead of searching all ARs and the associated CMEs, which are too many to be identified manually, we investigate super ARs that were reported in literatures. Super ARs are those with larger area, stronger magnetic field and more complex pattern, and thought to be the representative of CME producers. In the following section, we present the selected data and the method. In Sec.3, an analysis of waiting times of quasi-homologous CMEs from these super ARs during the last60solar cycle is performed. Finally, conclusions and discussion is given in the last section.1Data Preparationa)Super ARs and Associated CMEsSuper ARs were studied by many researchers 14151617. It was first defined by Bai 1418as a region producing four and more major flares. In most studies, super ARs were selected based on65several parameters, such as the largest area of sunspot group, the soft X-ray flare index, the 10.7 cm radio peak flux, the short-term total solar irradiance decrease, the peak energetic proton flux, the geomagnetic Ap index, etc. No matter which one or more criteria are used, most selected super ARs are CME-productive (that could be seen at the last paragraph of this sub-section).In our study, we focus on super ARs during solar cycle 23. Instead of identifying super ARs70by ourselves, we simple use existent lists of super ARs in literatures. To our knowledge, there are three lists regarding to super ARs in solar cycle 23. The first one is given by Tian et al. 15, who found 16 super ARs from 1997 to 2001 base on their selection criteria. The second one is given by Romano and Zuccarello)16, which contains 26 super ARs from 2000 to 2006. The last one is in paper by Chen et al. 17, in which 12 super ARs were identified during the last solar cycle. Since75Chen et al. 17 used stricter criteria, the last list is actually a subset of the other two. Totally, wehave 37 super ARs from 1996 to 2006.To identify the CMEs originating from these super ARs, we examine imaging data from Large Angle and Spectrometric Coronagraph (LASCO)19 and Extreme Ultraviolet Imaging Telescope (EIT)20 onbard Solar and Heliospheric Observatory (SOHO). The identification80processis the same as that applied by Wang et al. 21and Chen et al. 13. The CMEs listed in theCDAW LASCO CME catalog (refer to http:/cdaw.gsfc.nasa.gov/CMElist/)22 are our candidates. Through a careful identification, it is found that a total of 285 CMEs are associated with these super ARs. Figure 1 shows the distribution of the CME productivity of super ARs, in which the number of super ARs almost linearly decreases with increasing CME number though there is a85sharp decrease below the CME productivity of 3.It should be mentioned that there are 7 super ARs with too many large data gaps in LASCO and/or EIT observations, and therefore their CME productivity cannot be obtained. Except them, there were 28 super ARs producing 3 or more CMEs (called CME-rich ARs), among which 149095100105110115super ARs generated at least 10 CMEs. The other 2super ARs produced only one or two CMEs though sporadic data gaps existed. This fact suggests that not all of super ARs are CME productive. But it is definite that super ARs are more likely to be CME productive. Chen et al.(2011b)13 identified 108 ARs during 19971998 and found that only 14% of these ARsproduced 3 or more CMEs. This percentage is much lower than that for super ARs, which is about93% (28/30). In this study we focus on the 28 CME-rich ARs, which produced 281 CMEs in total. A list of all the CMEs associated with these CME-rich super ARs can be retrieved from Fig. 1. Distribution of CME productivities of super ARs.b)Waiting TimesAs long as there is no large data gap, we tentatively believe that all CMEs originating from a super AR of interest are recognized based on combined observations from SOHO LASCO and EIT. The waiting time of each CME is obtained according to the times of first appearance of the CME and its preceding one from the same super AR in the field of view of LASCO/C2. However, data gaps exist, and some CMEs may missed. If there was a large data gap between two CMEs from the same super ARs, the waiting time of the second CME cannot be obtained. Here, all data gaps less than 3 hours are ignored, because it is almost impossible for a CME to stealthily escape the field of view of LASCO in 3 hours.Before analyze the waiting times of these CMEs from the super ARs, it has to be noted that there are probably about 32% of frontside CMEs missed by SOHO 2123. Of course, these missed CMEs might be generally weak and faint. The statistical study by Chen et al. 13 have suggested that the properties of ARs have effects on the CME productivity, but do little with the kinetic properties of CMEs. Thus, it is possible that some CMEs originating from the super ARs are missed in our study, though such CMEs might be very weak and erupt in a gradual manner. So far. it is hard to evaluate how significant an influence this error will cause, and one maybear it in mindthat the following analysis is performed with a bias of normal to strong CMEs.2Results120125130135140145150a)Waiting Time DistributionThe average value of the waiting times is about 17.8 hours. The waiting time distribution is shown in Figure 2. Similar to that shown in Figure 10 of the paper by Chen et al. 13, the distribution consists of two components. One component locates less than 18 hours and looks like a gaussian distribution, and the other beyond 18 hours. For the first component distribution, the peak waiting time is about 7 hours. In Chen et al.13, the separation of the two components of the distribution is near 15 hours, and the first component distribution peaked near 8 hours, which are both slightly different than those obtained here. These slight differences might be caused by the solar cycle variation.An interesting result in Chen et al. 13 is that any AR cannot produce two or more CMEsfaster than 800 km s-1 within 15 hours. In other words, the time intervals between fast CMEs are longer than 15 hours. If this result obtained during the last solar minimum also holds for the whole solar cycle, we could expect that any AR cannot produce two or more CMEs faster than a certain speed threshold within 18 hours. However, such a speed threshold cannot be found. The blue linein Figure 2 shows the waiting time distribution for CMEs faster than 1200 km s-1 . Note that all theslower CMEs are ignored when we calculate waiting times for CMEs faster a certain speed threshold. Some fast CMEs did occur in the same ARs within 18 hours. For example, there were four CMEs from the super AR 10720 on 2005 January 15 at 06:30 UT, 23:06 UT, on January 17 at09:30 UT and 09:54 UT, respectively, which were all faster than 2000 km s 1.The first two CMEswere separated by about 16.6 hours, and the other two by about only 24 minutes. These fast CMEs caused ground-level enhancement (GLE) event 24.Although a similar result cannot be obtained, we find that the likelihood for an AR producing two or more fast CMEs within 18 hours is much smaller than normal. For all CMEs, 68% of the waiting times are shorter than 18 hours, while for CMEs faster than 1200 km s 1 , the fraction decreases to only about 18%. The dependence of the likelihood on the CME speed threshold isgiven in Figure 3. Generally, the likelihood monotonically decreases as the speed threshold increases. When the threshold reaches to about 1200 km s 1 , the likelihood stops decreasing and stays between 15%25%, suggesting a limit likelihood of approximate 1/5.The waiting time distribution for all CMEs from 1999 February to 2001 December was investigated by Moon et al.25, which is significantly different from the distribution for quasi-homologous CMEs obtained here (see Figure 1 in their paper). This difference reveals that the occurrence of CMEs follows a poisson process 2627, but that of quasi-homologous CMEs does not. In a statistical view, we may conclude that there are tight physical connections between quasi-homologous CMEs, but for CMEs from different source regions, the connection is quite loose.Fig. 2. Waiting time distributions for all quasi-homologous CMEs (black line) and for quasi-homologous CMEs faster than 1200 km s1(blue line).155160Fig. 3. Dependence of likelihood of quasi-homologous CMEs occurring within 18 hours onCME speed.b)Role of Free Energy Input in Causing Quasi-Homologous CMEsSufficient free energy is a necessary condition for an AR to produce CMEs. However, it is questionable if free energy input is a direct cause of quasi-homologous CMEs. This issue is considered from two aspects. First, we investigate the correlation between the CME speeds and waiting times. If free energy input is a direct cause, it is expected that there is some regulation165170175between CMEs speeds and their waiting times, as a long waiting time may lead to more free energy in an AR. Figure 4a shows the dependence of CME speed on the waiting time. Overall, no clear correlation could be found between them, except that there is seemingly an upper limit in CME speed depending on the CME waiting time. However, although the distribution is statistically true, it does not imply that an AR is difficult to produce a fast CME if it had waited too long. It is a result simply