1: \documentclass[12pt,preprint]{aastex}
2:
3: %\usepackage[dvips]{graphicx}
4: %\usepackage{verbatim}
5: %\usepackage{multirow}
6: %\usepackage{lscape}
7: %\usepackage[square]{natbib}
8: %\usepackage{subfigure}
9:
10: \def\gsim{\;\lower4pt\hbox{${\buildrel\displaystyle >\over\sim}$}\;}
11: \def\lsim{\;\lower4pt\hbox{${\buildrel\displaystyle <\over\sim}$}\;}
12: \def\grls{\;\lower4pt\hbox{${\buildrel\displaystyle >\over <}$}\;}
13:
14: \newcommand{\unsure}[1]{{\color[rgb]{1,0.3,0}\it #1}}
15: \newcommand{\modpart}[1]{{\bf #1}}
16: %\setlength{\parskip}{5pt plus 0.2pt minus 0.1pt} %\setlength{\parindent}{16pt}
17:
18: \shortauthors{WANG \& ZHANG} \shorttitle{Fast-CME ARs}
19:
20:
21: \begin{document}
22:
23:
24: \title{A Statistical Study on Solar Active Regions Producing Extremely Fast
25: Coronal Mass Ejections}
26:
27: \author{Yuming Wang\altaffilmark{1, 2}, and Jie Zhang\altaffilmark{1}}
28: \altaffiltext{1}{Department of Computational and Data Sciences,
29: College of Science, George Mason University, 4400 University Dr.,
30: MSN 6A2, Fairfax, VA 22030, USA, Email: ywangf@gmu.edu,
31: jzhang7@gmu.edu}
32:
33: \altaffiltext{2}{School of Earth \& Space Sci., Univ. of Sci. \& Tech. of
34: China, Hefei, Anhui 230026, China, Email: ymwang@ustc.edu.cn}
35:
36: \begin{abstract}
37: We present a statistical result on the properties of solar source regions
38: that have produced 57 fastest front-side coronal mass ejections (CMEs)
39: (speed $\geq$1500 km s$^{-1}$) occurred from 1996 June to 2007 January. The
40: properties of these fast-CME-producing regions, 35 in total, are compared
41: with those of all 1143 active regions (ARs) in the period studied. An
42: automated method, based on SOHO/MDI magnetic synoptic charts, is used to
43: select and characterize the ARs. For each AR, a set of parameters
44: are derived including the areas (positive, negative and total, denoted by
45: $A_P$, $A_N$ and $A_T$ respectively), the magnetic fluxes (positive,
46: negative and total, $F_P$, $F_N$ and $F_T$ respectively), the average magnetic
47: field strength ($B_{avg}$), quasi-elongation ($e$) characterizing the overall
48: shape of the AR, the number and length of polarity inversion lines (PILs,
49: or neutral lines, $N_{PIL}$ and $L_{PIL}$ respectively), and the average and
50: maximum of magnetic gradient on the PILs ($GOP_{avg}$ and $GOP_{max}$
51: respectively). Our statistical analysis shows a general trend between the
52: scales of an AR and the likelihood of producing a fast CME, i.e., the larger
53: the geometric size ($A_T$), the larger the magnetic flux ($F_T$), the stronger
54: the magnetic field ($B_{avg}$), and/or the more complex the magnetic configuration
55: ($N_{PIL}$ and $L_{PIL}$), then the higher the possibility of producing a fast
56: CME. When all ARs are sorted into three equally-numbered groups with low, middle
57: and high values of these parameters, we find that, for all these AR parameters,
58: more than 60\% of extremely fast CMEs are from the high-value group. The two
59: PIL parameters are the best indicators of producing fast CMEs, with more than
60: 80\% from the high value group.
61: \end{abstract}
62: \keywords{Sun: coronal mass ejections -- Sun: active regions}
63:
64: \section{Introduction}
65:
66: Coronal mass ejections (CMEs) are the most spectacular eruptive phenomenon in
67: the solar corona, which ejects a large amount of mass and magnetic flux into
68: the interplanetary space and may cause adverse disturbances in the geo-space.
69: From 1996 to 2007, more than 10 thousand CMEs have been observed by the LASCO
70: (Large Angle and Spectrometric Coronagraph, \citet{Brueckner_etal_1995})
71: instrument on board SOHO (Solar and Heliospheric Observatory) spacecraft.
72: Based on the CDAW CME catalog\footnote{http://cdaw.gsfc.nasa.gov/CME\_list/},
73: the CME speeds measured could be as low as $\sim21$ km s$^{-1}$ and as high as
74: $\sim3387$ km s$^{-1}$, and the average speed is about 450 km s$^{-1}$. Among
75: them, fast CMEs are of particular interest. Fast CMEs originate from solar
76: active regions with a large amount of free magnetic energy that gives rise to
77: the kinetic energy of a CME. Further, fast CMEs are usually much more
78: geoeffective than slow CMEs, e.g, producing large solar energetic particle events
79: and causing major geomagnetic storms. In this paper, we focus on the extremely
80: fast CMEs originating from the front-side solar disk, i.e.,
81: faster than 1500 km s$^{-1}$ or about 3
82: times faster than the average speed. These fast CMEs occupy a very small
83: fraction ($\sim0.5\%$) of all CMEs.
84:
85: The speed of a CME may be closely tied to the magnetic properties of its surface
86: source region. While a CME could originate from a quiet Sun region with filament
87: (or filament channel), it is believed that most CMEs, like most flares, originate
88: from active regions (ARs), where strong magnetic field occupies a relatively
89: large area. In a statistical study of 32 CME events, \citet{Subramanian_Dere_2001}
90: found that 84\% CMEs were associated with ARs. \citet{Yashiro_etal_2005} suggested
91: that almost all CMEs associated with sizable flares (flare larger than C3.0)
92: occurred in ARs, and the percentage could reach up to 99\% (private communication).
93: According to the flare catalogs from 2004 to 2007 during which the surface locations
94: of almost all flares are listed (compiled by NOAA/SWPC, Space Weather Prediction
95: Center), 79\% of all recorded flares originate from ARs, and the percentage rate is
96: higher for larger flares. It is 89\% for C-class and above flares, and 98\% for M
97: and X-class flares. Thus, the properties of ARs are the major concern in studying
98: the production of CMEs and flares.
99:
100: The relation between flare productivity and AR properties has been studied
101: extensively by many researchers \citep[e.g.,][]{Sammis_etal_2000, Leka_Barnes_2003b,
102: Leka_Barnes_2007, Maeshiro_etal_2005, Jing_etal_2006, Ternullo_etal_2006,
103: Schrijver_2007, Georgoulis_Rust_2007}. In general, larger flares tend to occur in
104: ARs with more complicated morphology, larger magnetic flux, larger magnetic energy,
105: larger helicity, and stronger and longer neutral lines.
106:
107: Nevertheless, CMEs are different phenomena from flares even though they are
108: related in many aspects \citep[e.g.,][]{Harrison_1995, Harrison_2003,
109: Zhang_etal_2001a}. The relations between CME productivity/properties and ARs
110: properties have not been well pursued until recently. Examining 6 active regions
111: with vector magnetogram observations, \citet{Falconer_etal_2006} showed a good
112: correlation between AR magnetic properties (including the total magnetic flux,
113: three other parameters related with main neutral lines, and two more parameters
114: related with magnetic field twist) and CME productivity. \citet{Guo_etal_2007}
115: investigated 55 flare-CME productive active regions with four magnetic parameters
116: (tilt angle, total magnetic flux, length of main neutral lines and effective distance),
117: and found that fast CMEs tend to occur in ARs with large magnetic flux and long
118: length of main PILs.
119:
120: In this paper, we focus on the top 0.5\% fastest front-side CMEs and characterize their
121: source ARs. Further, we use an automated method to select and characterize
122: all ARs observed by MDI from 1996 June to 2007 January, in order to create a
123: matrix for comparison. The events and ARs studied cover almost the whole
124: solar cycle 23. The data gathering and processing methods are introduced
125: in the following section. The statistical results are presented in
126: section \ref{sec_results}. Section \ref{sec_summary} summarizes the work.
127:
128:
129: \section{Data and Methods}
130:
131: \subsection{CME Selection and Source Region Identification}
132:
133: Based on the CDAW CME catalog, the numbers of CMEs with speed over
134: 1000, 1500, 2000, and 2500 km s$^{-1}$ from 1996 June to 2007 January
135: are 496, 122, 37 and 9, respectively. For the sake of avoiding
136: the time-consuming manual examination of all fast CMEs, we choose the CME category of
137: speed $\geq1500$ km s$^{-1}$, which has a limited population of 122 events and
138: is more than 3 times faster than the CME average speed. Through visual
139: examination of SOHO/EIT (EUV Imaging Telescope, \citet{Delaboudiniere_etal_1995})
140: and LASCO movies, 57 CMEs are found to originate from the front-side of solar
141: disk that can be used to study the source regions. Table \ref{tb_cmes_ars}
142: lists these front-side fast CMEs, including their speeds and source region
143: locations. These CMEs (except the 55th and 56th CMEs which are determined
144: according to the flare information in absence of EIT data) have definite
145: eruptive signatures, including large scale dimming, compact brightening,
146: and/or post-eruption loop arcades as seen in the SOHO/EIT corona images.
147: Figure \ref{fg_cme_loc} shows the distribution of the heliographic coordinates
148: of the surface source region of these CMEs. Apparently, the sample is uniform in
149: longitude, along which they are almost equally distributed; there is no bias
150: toward the limbs, from which a CME may be faster due to smaller projection
151: effect on speed measurement.
152:
153: \clearpage
154: \begin{figure}[tb]
155: \centering
156: \includegraphics[width=1.\hsize]{f1.eps}
157: \caption{Distribution of heliographic coordinates of the surface source
158: regions of the extremely fast CMEs. ARs are indicated in diamond-plus symbols
159: and queit-Sun regions are indicated in diamond symbols.}\label{fg_cme_loc}
160: \end{figure}
161: \clearpage
162:
163: \subsection{Determining the Properties of Active Regions}\label{sec_AR_method}
164:
165: A catalog of solar ARs is compiled on a daily basis by NOAA/SWPC
166: \footnote{http://www.sec.noaa.gov/ftpmenu/forecasts/SRS.html}. It
167: is based on the active region report in near-real time from
168: approximately half a dozen ground-based observatories around the
169: globe, including observations of white light, $H_\alpha$ and
170: magnetogram. The compiled Solar Region Summary (SRS), for each
171: identified AR (or a sunspot group) on the front disk, lists the
172: assigned NOAA AR number, its heliographic coordinate (rotated to
173: 24:00 UTC), Carrington longitude, total corrected area of the
174: sunspot group, Modified Zurich classification
175: \citep{McIntosh_1990}, longitudinal extent of the group in
176: heliographic degrees, total number of visible sunspots in the
177: group, and magnetic classification of the group. For the 11 year
178: period from 1996 to 2006 inclusive, there are in total 2999 active
179: regions listed by NOAA. Apparently, the NOAA AR catalog does not
180: provide the necessary quantitative information, such as magnetic
181: field strength and total magnetic flux, for the purpose of
182: studying the source regions of fast CMEs. In this paper, we have
183: generated one version of our own AR catalog from the long term SOHO/MDI
184: (Michelson Doppler Imager) observations from 1996 to 2006.
185:
186: To simplify the process of identifying and quantifying ARs, we
187: have used the Carrington rotation (CR) synoptic charts of MDI
188: magnetogram instead of individual magnetograms. When an AR
189: trans-crosses the front disk of the Sun in about 2 weeks, its
190: appearance may vary due to the intrinsic evolution and as well the
191: projection effect in the time-lapse magnetograms. On the other
192: hand, a synoptic chart is assembled from the stripe of data in the
193: central meridian from individual magnetograms observed over the
194: course of one Carrington rotation. By using synoptic charts the
195: projection effect is largely reduced. However, the possible
196: evolution of ARs is simply ignored. Since the launch of SOHO
197: spacecraft in Dec. 1995, high resolution synoptic charts of
198: photospheric magnetic fields from MDI are routinely made by the
199: MDI team at Stanford University
200: \footnote{http://soi.stanford.edu/magnetic/index6.html}. There are
201: 143 synoptic charts from 1996 May to 2007 January, covering almost
202: the whole solar cycle 23.
203:
204: We have developed an automated method to uniformly identify the ARs in MDI
205: synoptic charts. Attempts to detect ARs automatically have been made before,
206: for example, based on simple thresholdings \citep[e.g.,][]{Worden_etal_1998,
207: Brandt_Steinegger_1998, Preminger_etal_2001}. \citet{Turmon_etal_2002}
208: utilized pattern recognition techniques to distinguish different features on
209: the solar surface. \citet{Benkhalil_etal_2006} used a more sophisticated
210: approach by applying the region-growing segmentation technique in addition
211: to thresholding. In this study, we use the region-growing method to automate
212: the AR identification and characterization processes as described in the
213: following.
214:
215: (1) Find the kernel or seed pixels in the synoptic chart. We
216: choose the threshold of 15 times of the standard deviation,
217: $\sigma$, of the quite Sun region (typically $|B_{los}|\leq50$ G).
218: (2) Applying the morphological closing procedure. Isolated small
219: features, including anomaly noisy pixels, are effectively removed
220: by applying a morphology erosion operation followed by a
221: morphology dilation operation with a size of 15 pixels on the
222: $3600\times1080$ pixel map. (3) Applying the region-growing procedure.
223: Starting from the kernels defined above, all connected pixels with
224: $|B_{los}|\geq 5\sigma$ are included in the grown region. (4)
225: Applying the morphological opening procedure. Regions that are very close in space are
226: merged, by first applying the dilation operation followed by the
227: erosion operation with a size of 60 pixels.
228:
229: \clearpage
230: \begin{figure*}[tb]
231: \centering
232: \includegraphics[width=1.\hsize]{f2.eps}
233: \caption{An example showing the original MDI synoptic chart (top) and
234: the map showing only ARs extracted with our automated method (bottom).}
235: \label{fg_ars_example}
236: \end{figure*}
237: \clearpage
238:
239: Figure \ref{fg_ars_example} illustrates the result of the morphological analysis
240: processes described above. The top panel shows the MDI synoptic chart of
241: CR 1948 corresponding to the period from 00:54 UT on April 4, 1999 to 07:11
242: UT May 1, 1999, and the bottom panel highlights all the extracted active
243: regions. Ten individual ARs are indicated by the lines with different colors.
244: The synoptic chart actually processed is an enlarged chart which is patched on
245: the right-most with a left-most portion of the previous synoptic chart
246: separated with the vertical black line; this treatment is to prevent those
247: ARs crossing the Carrington rotation boundary from being cropped. The details
248: of the imaging processing methods, including how sensitive the results depend
249: on various intensity thresholds and area boxes chosen, will be pursued in another paper.
250: Changing thresholds on AR selection will not have significant impact on the
251: results of this paper, since this study mainly concerns the relative properties
252: of ARs.
253:
254: Once an AR is identified and its boundary is determined, it is
255: straightforward to extract many useful parameters of the AR. We
256: use one AR in CR 1948 (Fig. \ref{fg_ar_example}) to illustrate the
257: characterization. The white and black patches, enclosed by the
258: blue and yellow lines, correspond to the regions of positive and
259: negative polarities in that active region. We calculated the area size
260: of the positive polarities ($A_P$), the size of the negative
261: polarities ($A_N$) and the total size ($A_T = A_P + A_N$). We also
262: calculated the total positive magnetic flux ($F_P$), total
263: negative magnetic flux ($F_N$) and the total unsigned flux
264: ($F_T$). We derived the average magnetic field strength ($B_{avg}
265: = F_T/A_T$). We also defined a quasi-elongation parameter of ARs,
266: ($e=1-\frac{D_s}{D_{max}}$, where $D_s$ is the radius of an assumed
267: circle of total size $A_T$ and $D_{max}$ is the largest distance
268: between two points in an AR). For the AR shown in Figure
269: \ref{fg_ar_example}, we found that the $A_P$, $A_N$, $A_T$, $F_P$,
270: $F_N$, $F_T$, $B_{ave}$ and $e$ are $1.9\times10^4$ Mm$^2$,
271: $1.2\times10^4$ Mm$^2$, $3.1\times10^4$ Mm$^2$,
272: $505.7\times10^{12}$ wb, $280.7\times10^{12}$ wb,
273: $786.4\times10^{12}$ wb, 72.8 G, and 0.32, respectively. These
274: parameters quantify the intensity and morphology of the AR, which
275: may be useful in relating ARs with eruptive phenomena.
276:
277: \clearpage
278: \begin{figure}[tb]
279: \centering
280: \includegraphics[width=1.\hsize]{f3.eps}
281: \caption{An AR of Carrington rotation of 1948 showing the positive (white
282: patches with blue boundaries), negative (black with yellow boundaries)
283: polarities and inversion lines (green curves) between them.}
284: \label{fg_ar_example}
285: \end{figure}
286: \clearpage
287:
288: Further, we extracted the polarity inversion lines (PIL, or
289: neutral lines) of the host AR. It is generally believed that the
290: properties of PILs are related with the stored free energy of the
291: AR studied. For each AR, we determined the number of PIL
292: ($N_{PIL}$) and the length of these PILs ($L_{PIL}$). The PILs of
293: AR are indicated by the green curves in Figure \ref{fg_ar_example}.
294: A PIL is a gathering of linearly linked pixels, which are between pixels that have opposite
295: polarities. In this AR there are 6 separate PILs with a total
296: length ($L_{PIL}$) of 185.2 Mm. We also determined the average
297: magnetic gradient across these PILs ($GOP_{avg}$) and the maximum
298: gradient ($GOP_{max}$); the numbers are 152.7 and 496.4 G
299: Mm$^{-1}$ respectively.
300:
301: In summary, a total of 12 magnetic parameters are derived.
302: $A_T$ indicates the size of an AR; $F_T$ and $B_{avg}$ suggest
303: how strong an AR is; $e$ indicate show much the overall shape of
304: an AR deviates from a perfect circle;
305: $N_{PIL}$ and $L_{PIL}$ indicate the morphological complexity of an AR; and
306: $GOP_{avg}$ and $GOP_{max}$ is a measure of the free energy stored between
307: two opposite polarities. These parameters are expected to have certain
308: relations with the properties of eruptive events resulted. For instance,
309: $F_T$ is ranked number one in the effectiveness of predicting flare occurrence
310: from a comprehensive discriminant data analysis \citet{Leka_Barnes_2007}. A
311: similar positive correlation between $F_T$ and large flare productivity was
312: also obtained by \citet{Schrijver_2007}. Besides, ARs with larger
313: area and/or more complicated configuration tend to produce larger flares
314: \citep[e.g.,][]{Sammis_etal_2000, Ternullo_etal_2006}.
315:
316:
317: \subsection{Comparison with NOAA Active Regions}
318:
319: \clearpage
320: \begin{figure}[tb]
321: \centering
322: \includegraphics[width=1.\hsize]{f4.eps}
323: \caption{Comparison of our identified ARs with NOAA ARs from CR1911 to
324: CR2052 in terms of the count and area. Upper panel shows the AR counts and
325: the lower one shows the total AR area. The solid lines present our results
326: and the dotted lines NOAA results. The horizontal solid and dashed lines
327: indicate the average ratio of our results to NOAA results and the standard
328: deviations. The heavy solid line in the upper panel denotes the number of
329: ARs hosting extremely fast CMEs.}
330: \label{fg_comp_noaa}
331: \end{figure}
332: \clearpage
333:
334: We have compared our MDI AR catalog with the NOAA AR catalog (Fig.
335: \ref{fg_comp_noaa}). The upper panel displays the AR counts of
336: each CR from our catalog (solid line) and NOAA catalog(dotted line). The
337: variation of the counts along the solar cycle seems the same for the two
338: lines: small near the solar minimum and reaching the peak near the
339: solar maximum. However, the numbers of our ARs are smaller than that of
340: NOAA ARs. The ratio of the two counts is 0.42 averaged over the entire
341: period studied with a standard deviation of 0.20. The smaller
342: number of ARs identified by our method is attributed
343: to several selection effects.(1) The usage of magnetogram data.
344: Regions that are sufficiently close to each
345: other in the distribution of magnetic field may be grouped into a
346: single large AR. The selection of NOAA ARs is largely based on the visual appearance of
347: sunspots, which are usually compact in size and discrete in distribution. On
348: the other hand, the surface magnetic field, which we use to define the
349: size, may extend to a much larger size than that of the sunspots. (2)
350: ARs with weak magnetic fields are not selected because of the adoption of a high kernel
351: threshold. (3) Some short lived ARs (emerged from the western
352: hemisphere and lasting less than one Carrington rotation) are not
353: caught in synoptic charts. A full comparison study between our catalog and
354: the NOAA catalog will be made in a separate paper.
355:
356: The lower panel of Figure \ref{fg_comp_noaa} presents the total AR
357: areas during each CR for both MDI ARs we identified and the NOAA
358: ARs. In contrast to the AR counts, the areas from our method
359: are much larger than those from NOAA. The ratio suggests that the
360: total AR area in our method is on average 27.2 times that by NOAA ARs.
361: The AR area estimated by NOAA concerns only sunspots, which
362: usually correspond to the regions with extremely strong magnetic fields. On
363: the other hand, we include all regions with $|B|\geq5\sigma$
364: connecting to the AR kernels.
365:
366:
367: \section{Results}\label{sec_results}
368:
369: Using our method, we have identified and characterized 1143 MDI ARs
370: for Carrington rotations 1911 to 2051 covering the period from 1996
371: June 28 to 2007 January 8. It seems that the 5 parameters that
372: characterize the overall scales of an AR ($A_T$, $F_T$, $B_{avg}$,
373: $N_{PIL}$ and $L_{PIL}$) are all fairly related with the production of fast CMEs.
374: The histogram distributions of AR numbers of these five parameters are
375: shown in the five panels in Figure \ref{fg_ars_histo} (upper panel),
376: respectively. All parameters except $N_{PIL}$ are plotted in
377: log-scale and divided into nine bins. One can find that all the
378: distributions, except for $L_{PIL}$, are unimodal with a peak appearing
379: at a certain value in between. The
380: histogram (a) and (b), which are for $A_T$ and $F_T$, indicate the
381: distributions of the geometric size and magnetic flux sizes of ARs,
382: respectively. Most ARs have a size of $\sim10^{3.8}$ Mm$^2$ with the
383: total magnetic flux peaked at $\sim10^{14}$ wb. This kind of
384: distribution with a central peak is much different from that of the
385: distribution of the size of sunspot groups studied before, which was
386: exponential \citep[e.g.,][]{Schrijver_1988, Howard_1996} or following
387: a power law \citep[e.g.,][]{Harvey_Zwaan_1993}. A detailed analysis on
388: the causes of these differences is beyond the scope of this paper and
389: will be pursued separately. In this paper, we focus on the ARs hosting
390: extremely fast CMEs.
391:
392: Table \ref{tb_cmes_ars} lists all the fast CMEs studied, the heliographic
393: coordinates of the surface source regions and the NOAA AR numbers of the
394: host ARs. 55 of the 57 events studied were originated from NOAA ARs. The
395: rest two CMEs, numbered 2 and 20 in the table, were not originated in any
396: NOAA ARs, nor in our MDI ARs. Careful inspection showed that these two
397: fast CMEs were associated with the eruption of giant filaments. Four more
398: CMEs (numbered 6, 23, 51 and 52) were not originating from our MDI ARs.
399: The surface source regions of these CMEs are too small and/or weak in the
400: MDI synoptic charts that they do not qualify to be an MDI AR using our
401: method. A summary is given in Table \ref{tb_ar_cme_number}. In total,
402: 89.5\% (51 out of 57) fast CMEs
403: were originated from our MDI ARs (the locations of these CMEs have been
404: indicated in diamond symbol with plus sign in Figure \ref{fg_cme_loc}). The
405: percentage of AR origin is as high as 96\% when NOAA ARs are used.
406:
407: These fast CMEs were originated from 35 individual MDI ARs. Compared
408: with a total number of 1143 MDI ARs during the period studied, the
409: percentage of ARs hosting these extremely fast CMEs is considerably small.
410: It is only about 3.1\%. The
411: distribution with solar cycle of these fast-CME-hosting ARs has been
412: plotted in the thick solid line in Figure \ref{fg_comp_noaa}. Apparently,
413: the majority of these ARs occurred during the solar maximum.
414:
415: \clearpage
416: \begin{figure*}[tb]
417: \centering
418: \includegraphics[width=1.\hsize]{f5.eps}
419: \caption{Histogram distributions of AR numbers on various AR parameters:
420: (a) total area, (b) total magnetic flux, (c) average magnetic field, (d)
421: number of PILs, and (e) total length of PILs. The first three and the last
422: parameters are plotted in log-scale along the x-axis. In each panel, the
423: upper sub-panel shows the histogram for all ARs in nine bins, while the lower
424: sub-panel shows the ratios of the ARs hosting extremely fast CMEs to all ARs
425: in three equally-numbered general AR groups (see text for details). The
426: numbers in parentheses give the number of ARs in the three groups. The
427: symbols of triangles are for the extremely fast CMEs and
428: the squares are for the fast CMEs originated between longitude of $\pm30^\circ$.}
429: \label{fg_ars_histo}
430: \end{figure*}
431:
432: \clearpage
433: \thispagestyle{empty}
434: \setlength{\voffset}{-15mm}
435: \begin{table*}[t]%\vskip -100pt
436: \linespread{1.5} \caption{Front-side fast CMEs\tablenotemark{*} in solar cyle 23 and their surface source regions}
437: \label{tb_cmes_ars}
438: \tabcolsep 2pt
439: \footnotesize %\centering
440: \begin{tabular}{r|ccccc||r|ccccc}
441: \hline
442: No.& Date Time\tablenotemark{a}& $V_{CME}$\tablenotemark{b}& Location& AR$_{NOAA}$ &AR\tablenotemark{c} & No.& Date Time\tablenotemark{a}& $V_{CME}$\tablenotemark{b}& Location& AR$_{NOAA}$ & AR\tablenotemark{c}\\
443: \hline
444: 1& 1997/11/06 12:10:41& 1556& S17W62& 8100& 1 & 30& 2003/06/17 23:18:14& 1813& S07E58& 10386& 24 \\
445: 2& 1998/04/20 10:07:11& 1863& S42W63& - & N & 31& 2003/10/26 17:54:05& 1537& N02W37& 10484& 25 \\
446: 3& 1998/12/18 18:09:47& 1749& N26E35& 8415& 2 & 32& 2003/10/28 11:30:05& 2459& S16E06& 10486& 26 \\
447: 4& 1999/05/03 06:06:05& 1584& N20E45& 8525& 3 & 33& 2003/10/29 20:54:05& 2029& S17W04& 10486& 26 \\
448: 5& 1999/06/04 07:26:54& 2230& N19W75& 8552& 4 & 34& 2003/11/02 17:30:05& 2598& S16W57& 10486& 26 \\
449: 6& 2000/01/06 07:31:24& 1813& N22W34& 8816& N & 35& 2003/11/04 19:54:05& 2657& S19W81& 10486& 26 \\
450: 7& 2000/05/15 08:50:05& 1549& N22E85& 9002& 5 & 36& 2003/11/18 08:50:05& 1660& N02E18& 10501& 27 \\
451: 8& 2000/06/25 07:54:13& 1617& N15W54& 9046& 6 & 37& 2004/01/07 04:06:07& 1581& N01E75& 10537& 28 \\
452: 9& 2000/07/14 10:54:07& 1674& N17W01& 9077& 7 & 38& 2004/01/07 10:30:29& 1822& N06E73& 10537& 28 \\
453: 10& 2000/09/12 11:54:05& 1550& S19W06& 9163& 8 & 39& 2004/01/08 05:06:05& 1713& N00E65& 10537& 28 \\
454: 11& 2000/11/08 23:06:05& 1738& N18W71& 9213& 9 & 40& 2004/04/11 04:30:06& 1645& S15W45& 10588& 29 \\
455: 12& 2000/11/25 01:31:58& 2519& N07E49& 9240& 10 & 41& 2004/11/07 16:54:05& 1759& N09W16& 10696& 30 \\
456: 13& 2001/01/20 21:30:08& 1507& S09E37& 9313& 11 & 42& 2004/11/09 17:26:06& 2000& N09W49& 10696& 30 \\
457: 14& 2001/04/02 22:06:07& 2505& N16W67& 9393& 12 & 43& 2004/11/10 02:26:05& 3387& N10W49& 10696& 30 \\
458: 15& 2001/04/10 05:30:00& 2411& S24W07& 9415& 13 & 44& 2005/01/15 06:30:05& 2049& N14E03& 10720& 31 \\
459: 16& 2001/06/11 04:54:05& 1647& S13E81& 9501& 14 & 45& 2005/01/15 23:06:50& 2861& N14W00& 10720& 31 \\
460: 17& 2001/07/19 10:30:22& 1668& S05W60& 9537& 15 & 46& 2005/01/17 09:30:05& 2094& N14W23& 10720& 31 \\
461: 18& 2001/09/24 10:30:59& 2402& S17E25& 9632& 16 & 47& 2005/01/17 09:54:05& 2547& N13W23& 10720& 31 \\
462: 19& 2002/04/21 01:27:20& 2393& S15W81& 9906& 17 & 48& 2005/01/19 08:29:39& 2020& N16W45& 10720& 31 \\
463: 20& 2002/05/22 03:50:05& 1557& S14W56& - & N & 49& 2005/05/13 17:12:05& 1689& N11E10& 10759& 32 \\
464: 21& 2002/07/23 00:42:05& 2285& S12E69& 10039& 18 & 50& 2005/07/30 06:50:28& 1968& N10E52& 10792& 33 \\
465: 22& 2002/08/16 12:30:05& 1585& S11E19& 10069& 19 & 51& 2005/08/22 17:30:05& 2378& S12W60& 10798& N \\
466: 23& 2002/09/05 16:54:06& 1748& N10E28& 10102& N & 52& 2005/08/23 14:54:05& 1929& S12W77& 10798& N \\
467: 24& 2002/11/09 13:31:45& 1838& S10W27& 10180& 20 & 53& 2005/09/09 19:48:05& 2257& S12E67\tablenotemark{d}& 10808\tablenotemark{d}& 34\\
468: 25& 2002/11/10 03:30:11& 1670& S11W37& 10180& 20 & 54& 2005/09/10 21:52:07& 1893& S13E47\tablenotemark{d}& 10808\tablenotemark{d}& 34\\
469: 26& 2003/03/18 12:30:05& 1601& S14W46& 10314& 21 & 55& 2005/09/11 13:00:53& 1922& & 10808\tablenotemark{e}& 34 \\
470: 27& 2003/03/23 12:06:05& 1505& S13E71& 10318& 22 & 56& 2005/09/13 20:00:05& 1866& & 10808\tablenotemark{e}& 34 \\
471: 28& 2003/06/02 00:30:07& 1656& S06W85& 10365& 23 & 57& 2006/12/13 02:54:04& 1774& S07W22& 10930& 35 \\
472: 29& 2003/06/15 23:54:05& 2053& S14E82& 10386& 24 & \\
473: \hline
474: \end{tabular}
475: \tablenotetext{a}{First appearance of CMEs in the field of view of LASCO/C2.}
476: \tablenotetext{b}{Linear speed of CMEs in the combined fields of view of LASCO/C2 and C3, in units of km s$^{-1}$.}
477: \tablenotetext{c}{Whether or not a CME originated from a MDI AR we identified. The sequenced numbers indicate a AR, and `NĄŻ indicates negative association with AR}
478: \tablenotetext{d}{Information is based on the NOAA/SEC flare list.}
479: \tablenotetext{*}{Date, time and $V_{CME}$ are adopted from the CDAW LASCO CME Catalog.}
480: \end{table*}
481:
482: \clearpage
483: \setlength{\voffset}{0mm}
484:
485:
486: \begin{table}[t]%\vskip -100pt
487: \linespread{1.5} \caption{Statistical Numbers of extremely fast CMEs and the host active regions}
488: \label{tb_ar_cme_number}
489: %\tabcolsep 2pt
490: \footnotesize %\centering
491: \begin{tabular}{ll}
492: \hline
493: Number of all ARs& 1143\\
494: \multicolumn{2}{c}{For all the front-side fast CMEs}\\
495: Number of front-side CMEs& 57 \\
496: Number of front-side CMEs originating from ARs& 51 (89.5\%)\\
497: Number of ARs hosting these CMEs& 35 ($3.1\%$)\\
498: \multicolumn{2}{c}{For the fast CMEs originating from E30$^\circ$ to W30$^\circ$}\\
499: Number of front-side CMEs& 17\\
500: Number of front-side CMEs originating from ARs& 16\\
501: Number of ARs hosting these CMEs & 12\\
502: \hline
503: \end{tabular}
504: \end{table}
505: \clearpage
506:
507: The overall distributions of these fast-CME-hosting ARs are shown along with
508: that of all ARs in Figure \ref{fg_ars_histo}. While the distributions of all
509: ARs (upper panel) are shown in 9 bins thanks to the large number, the
510: distributions of fast-CME-hosting ARs are displayed in only three groups with
511: low, middle and high values of the derived parameters, because of the limited
512: number of ARs. The minimum-to-maximum values of the parameters in each group
513: are such chosen that there are equal number of general ARs in each of three
514: groups. The values used for such equal divisions are listed in the 2nd and 3rd
515: columns in Table \ref{tb_AR_percentage}. Note that for $N_{PIL}$ the numbers can not be
516: exactly equal, but are most close to each other, because $N_{PIL}$ is highly discrete.
517: Then we determine, in each of the three groups, the number of ARs producing
518: fast CMEs, which indicate the effectiveness of the parameters used in discriminating
519: fast-CME-hosting ARs from general population. The triangles shows the ratios of ARs
520: hosting fast CMEs to all ARs, i.e., the probability of an AR producing fast CMEs.
521: The horizontal bars indicate the value range of the groups, and the vertical
522: bars indicate the uncertainty of the calculated percentage probability, defined
523: as $\sqrt{\frac{p(1-p)}{N}}$, where $p$ is the probability and $N$ is the total
524: sample number; this uncertainty is based on the assumption that the occurrence
525: of fast CMEs obeys a binomial distribution.
526:
527: It is obvious that the probability of producing fast CMEs increases as the
528: value of the AR parameters increases. This trend is true for all the
529: five parameters plotted in the Figure. The differences between low/middle and
530: high value groups all exceed the estimated uncertainty. In particular,
531: $N_{PIL}$ and $L_{PIL}$ manifest the largest increase from low/middle to high
532: value groups; the probability increases by a factor as large as 8. This
533: kind of trend indicates that the larger the AR parameters, i.e., the larger
534: the geometric size, the larger the magnetic flux, the stronger the magnetic field,
535: and the more complex the magnetic configuration, the higher the likelihood of it
536: producing fast CMEs.
537:
538: \clearpage
539: \begin{table*}[t]%\vskip -100pt
540: \linespread{1.5} \caption{Percentages of fast-CME-hosting ARs in three different value groups. }
541: \label{tb_AR_percentage}
542: \tabcolsep 5pt
543: \footnotesize %\centering
544: \begin{tabular}{c|cc|ccc|ccc}
545: \hline
546: Quantity& \multicolumn{2}{c|}{Separations}& $n^{low}$& $n^{mid}$& $n^{hig}$& $n^{low}_{cen}$& $n^{mid}_{cen}$& $n^{hig}_{cen}$\\
547: \hline
548: $A_{tot}$& 3.90\tablenotemark{a}& 9.05\tablenotemark{a}& 2 ( 5.7\%)& 8 (22.9\%)& 25 (71.4\%)& 1 ( 8.3\%)& 2 (16.7\%)& 9 (75.0\%) \\
549: $F_{tot}$& 102.1\tablenotemark{b}&263.1\tablenotemark{b}& 1 ( 2.9\%)& 8 (22.9\%)& 26 (74.3\%)& 1 ( 8.3\%)& 1 ( 8.3\%)& 10 (83.3\%) \\
550: $B_{avg}$& 238\tablenotemark{c}& 298\tablenotemark{c}& 3 ( 8.6\%)& 11 (31.4\%)& 21 (60.0\%)& 0 ( 0.0\%)& 2 (16.7\%)& 10 (83.3\%) \\
551: $N_{PIL}$& 1& 3& 5 (14.3\%)& 2 ( 5.7\%)& 28 (80.0\%)& 1 ( 8.3\%)& 1 ( 8.3\%)& 10 (83.3\%) \\
552: $L_{PIL}$& 13.3\tablenotemark{d}& 62.4\tablenotemark{d}& 2 ( 5.7\%)& 4 (11.4\%)& 29 (82.9\%)& 0 ( 0.0\%)& 1 ( 8.3\%)& 11 (91.7\%) \\
553: \hline
554: \end{tabular}
555: \tablenotetext{a}{in units of $10^3$ Mm$^2$.}
556: \tablenotetext{b}{in units of $10^12$ wb.}
557: \tablenotetext{c}{in units of Gauss.}
558: \tablenotetext{d}{in units of Mm.}
559: \tablenotetext{e}{in units of G Mm$^{-1}$.}
560: \end{table*}
561: \clearpage
562:
563: Table \ref{tb_AR_percentage} lists the occurrence numbers and
564: percentages of fast-CME-hosting ARs in the three groups discussed
565: earlier (the fourth through sixth column). For all the parameters studied, more than 60\%
566: fast-CME-hosting ARs lie in the high value group. The length of PIL
567: ($L_{PIL}$) has the largest percentages ($82.9\%$). These results indicate
568: that the PIL-related parameters are the best in determining whether an AR
569: would produce extremely fast CMEs.
570:
571: As mentioned earlier, the evolution of ARs has not been taken into account when
572: the AR parameters are derived, since the calculations were based on synoptic
573: charts in order to minimize the projection effect. An AR may have evolved to
574: some extent between the time of the CME occurrence and the time of
575: central-meridian crossing; the time difference may be up to 7 days for those
576: CMEs originating from the limbs. To find out how significant the temporal
577: evolution affects our statistical results, we made a similar statistical study
578: but considering only those CMEs that are close to the central meridian. The
579: squares in Figure \ref{fg_ars_histo} and the last three columns, named
580: $n_{cen}^{low}$, $n_{cen}^{mid}$ and $n_{cen}^{hig}$, in Table \ref{tb_AR_percentage}
581: show the results for these near-central meridian CMEs (longitude
582: between $\pm30^\circ$). It is found that there is no
583: significant difference from those of all front-side CMEs.
584:
585: Some studies showed that flux variations (e.g., emergence or cancellation)
586: are related with the CME initiation~\citep[e.g.,][]{Feynman_Martin_1995,
587: Lara_etal_2000, Green_etal_2003, Sterling_etal_2007}. We check the Hale
588: classes of all the corresponding NOAA ARs producing those fast CMEs studied.
589: During the interval between the CME launch time and the central-meridian
590: crossing of the corresponding source AR, it is found that 15 out of 35 ARs
591: changed from one Hale class to another. However, for these 15 ARs with
592: changes, nine are located very close to the limb when CMEs erupted,
593: indicating a maximal projection effect which may affect the AR classification.
594: Therefore, most of the ARs interested probably did not changed their Halo
595: classes. While it may be related with CME initiation, the small scale
596: flux variation may not alter the overall structure (and the stored energy)
597: of the coronal magnetic field where CMEs originate. Considering that this
598: study is a rather coarse quantitative statistical work, we believe that
599: the AR evolution effect, when fully addressed, would not change our results.
600:
601:
602: \section{Summary}\label{sec_summary}
603:
604: By applying the region-growing segmentation method, a total of 1143 ARs are
605: extracted from 141 MDI synoptic charts covering the period from 1996 Jun 28
606: to 2007 January 8. Twelve quantities ($A_P$, $A_N$, $A_T$, $F_P$, $F_N$, $F_T$,
607: $B_{avg}$, $e$, $N_{PIL}$, $L_{PIL}$, $GOP_{avg}$ and $GOP_{max}$) are derived
608: from the photospheric magnetic field distribution, concerning the size, strength, morphology,
609: complexity and free energy of ARs. These AR regions produced more than 10000 CMEs,
610: among which 122 CMEs were faster than or equal to 1500 km s$^{-1}$. We
611: studied 57 of these extremely fast CMEs that originated from front-side ARs.
612: Through a comparison of these fast-CME-hosting ARs with all other ARs, the
613: following conclusions are reached.
614:
615: (1) 89.5\%(51 out of 57) of fast CMEs are originated from our MDI ARs.
616: The number of ARs hosting these fast CMEs is 35, occupying 3.1\% of
617: all ARs. A majority of these ARs occurred during the solar maximum.
618:
619: (2) The distributions of the parameters, except $L_{PIL}$, are unimodal with a peak
620: at a certain middle value. A typical MDI AR has a geometric size of $\sim10^{3.8}$
621: Mm$^2$ and a magnetic flux of $\sim10^{14}$ wb.
622:
623: (3) The derived AR parameters, $A_T$, $F_T$, $B_{avg}$, $N_{PIL}$ and $L_{PIL}$,
624: all showed a positive correlation with the probability of an AR producing a
625: fast CME. When all ARs are equally divided into three groups with low, middle and high
626: value of the parameters studied, we find that the majority of fast-CME-hosting ARs
627: resides in the high-value groups, e.g., the PIL-related parameters ($N_{PIL}$ and
628: $L_{PIL}$) are particularly effective in indicating the occurrence of fast CMEs;
629: the percentage can be as high as 82.9\%. Compared with the low-to-middle-value AR
630: groups, the probability of producing a fast CME from a high-value AR can increase
631: by a factor as large as 8. These results suggest
632: that the larger, the stronger, and/or the more complex an AR, the more likely it
633: produce an extremely fast CME.
634:
635: The ARs hostingfast CMEs occupy a small fraction of all ARs
636: ($\sim3.1\%$). The fraction is about 8\% even for only the high-value AR groups.
637: It indicates that there are many ARs having large-value parameters, but never
638: producing an extremely fast CME. The current method is
639: based on photospheric magnetic field observations only. However, we believe that
640: coronal magnetic fields play a direct role in determining the dynamic or
641: kinematic properties of CMEs. In a statistical study of 99 halo
642: CMEs, \citet{Liu_2007} suggested that CMEs under unidirectional open field
643: structures are significantly faster than those under the heliospheric current
644: sheet. To develop a better capacity of predicting the occurrence of fast CMEs, one
645: need to study the AR parameters from both the photosphere and the corona as well.
646:
647:
648: \acknowledgments
649: We acknowledge the use of the solar data from the
650: MDI, LASCO, and EIT instruments on board SOHO spacecraft. The
651: SOHO/LASCO data used here are produced by a consortium of the
652: Naval Research Laboratory (USA), Max-Planck-Institut fuer
653: Aeronomie (Germany), Laboratoire d'Astronomie (France), and the
654: University of Birmingham (UK). SOHO is a project of international
655: cooperation between ESA and NASA. We also acknowledge the use of
656: CME catalog generated and maintained at the CDAW Data Center by
657: NASA and The Catholic University of America in cooperation with
658: the Naval Research Laboratory, and the flare list compiled by the
659: Space Environment Center of NOAA. This work is supported by
660: NSF SHINE grant ATM-0454612 and NASA grant NNG05GG19G. Y. Wang is
661: also supported by the grants from NSF of China (40525014) and
662: MSTC (2006CB806304), and J. Zhang is also supported by NASA
663: grants NNG04GN36G.
664:
665:
666: \bibliographystyle{kluwer}
667: \bibliography{../../ahareference}
668:
669:
670: \end{document}
671:
672: