Two-step development of geomagnetic storms

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 103, NO. A4, PAGES , APRIL 1, 1998 Two-step development of geomagnetic storms Y. Kamide, N. Yokoyama, W. Gonzalez,, 2 B.T. Tsurutani,, 3 I.A. Daglis, 4 A. Brekke, and S. Masuda Abstract. Using the Dst index, more than 1200 geomagnetic storms, from weak to intense, spanning over three solar cycles have been examined statistically. Interplanetary magnetic field (IMF) and solar wind data have also been used in the study. It is found that for more than 50% of intense magnetic storms, the main phase undergoes a two-step growth in the ring current. That is, before the ring current has decayed significantly to the prestorm level, anew major particle injection occurs, leadingto a further development of the ring current, and making Dst decrease a second time. Thus intense magnetic storms may often be the result of two closely spaced moderate storms. The corresponding signature in the interplanetary medium is the arrival of double-structured southward IMF at the magnetosphere. 1. Introduction causes. These studies followed essentially the same approach as Sugiurand Chapman's, where the variability in duration for In view of the increasingly wide recognition of the importance different storms was obscured in their averaging process. of"space weather" research in the scientifi community, studies In going through the recent literature, however, we find that of geomagnetic storms have recently been revived [e.g., D. intense magnetic storms often develop in two steps during the Knipp, Coordinated study of November 3-4, 1993, magnetic main phase [e.g., Tsurutani et al., 1988]. It is of great interest to storm, posted on world wide web site examine how often the ring current develops in such a two-step ssc.igpp.ucla.edu/gem/event_nov 93.html, 1996]. The main fashion during magnetic storms and to look for the corresponding objective of the "space weather" program is to understand the signatures in the solar wind and to discuss possible causes of magnetic storms in the Sun and the interplanetary magnetospheric processes. medium and to trace energy flow associated with storms from the Sun to the Earth's upper atmosphere. The present paper addresses the following major questions: What magnetospheric 2. Procedure parameter or parameters represent quantitatively the intensity of magnetic storms? How one can define the magnetic storm A total of 1252 geomagnetic storms were identified for the strength on the basis of available data? What parameters in the period from 1957 to 1991, covering nearly three solar cycles. solar wind best determine how intense the upcoming magnetic The entire data set was grouped into three classes: weak (Dstmin > storms will be? -50 nt), moderate (-50 > Dstmin >-100 nt), and intense (Dstmin Because of the close theoretical relationship between the total <-100 nt), according to the magnitude of the storms, which was energy of ring current particles and the geomagnetic Dst index defined by the peak Dst values. Visual inspection of Dst was first [Dessler and Parker, 1959; Sckopke, 1966; Siscoe, 1970], the employed to identify periods of magnetic storms. This inspection minimum Dst value at the main phase of magnetic storms has was necessary because we did not wish to miss gradual storms been used extensively in the literature as a measure of the storm that commence without a clear indication of an SSC [Akasofu, intensity [see Joselyn and Tsurutani, 1990]. In their extensive 1965]. There was no indication of any clear relationship between study using 364 magnetic storms, Sugiura and Chapman [1960] the intensity of magnetic storms and whether storms commenced divided magnetic storms into three categories based on peak Dst with or without SSCs. values: weak, moderate, and intense storms. In that "classic" We further classified each of the three classes of geomagnetic statistical study, they identified magnetic storms on the basis of storms into two types: and, according to how Dst the existence of storm sudden commencements (SSCs), thus reaches the peak through the main phase. Figure 1 shows excluding the so-called gradual storms. In more recent studies, schematically these two types of geomagnetic storms. Taylor et al. [1994], Loewe and Pr6lss [1997], and Yokoyama represents a "normal" magnetic storm that consists of a main and Kamide [1997] have conducted statistical studies of phase and a subsequent recovery phase. During the main phase, geomagnetic storms in which Dst variations were compared with the magnetic field on the Earth's surface is significantly auroral electrojet activity, as well as with their interplanetary depressed. This depression is caused by an enhancement of the trapped particle population in the magnetosphere and thus by a 1 Solar-Terrestrial Environment Laboratory, Nagoya University, proton ting current flowing westward. This sequence is at times Toyokawa, Japan. preceded by an initial phase during which Dst shows a positive 2Instituto Nacional de Pesquisas Espacias, S o Jos6 dos Campos, Sao change responding to a ram pressure increase the solar wind. Paulo, Brazil. On the other hand, magnetic storms are those which 3jet Propulsion Laboratory, California Institute of Technology, have a two-step growth in the ting current, that is, a two-step Pasadena. decrease in Dst. To differentiate properly from, 4Institute of Ionospheric and Space Research, National Observatory of several parameters are introduced, as represented in Figure 1. Athens, Palea Penteli, Greece. Most importantly, the following two conditions are required: 5 Auroral Observatory, University of Tromso, Tromso, Norway. 1. The first decrease in Dst should partly subside before the second decrease follows some time later. Parameter A represents Copyright 1998 by the American Geophysical Union. the magnitude of the first Dst decrease, while C quantifies Dst recovery. It should be noted that A > C > 0 nt. Furthermore, if Paper number 97JA C/A > 0.9, it is not classified as a storm, but simply a /98/97JA storm with a magnitude of A. 6917

2 6918 KAMIDE ET AL.: TWO-STEP DEVELOPMENT OF STORMS identifying these times, that is, the storm start, peak, and end, a superposed-epoch study was conducted in an attempto identify major characteristics common to different magnetic storms. In each of the three storm intensities, the average durations of the main and recovery phases were determined. The time scales and the Dst intensifies of each storm were then stretched/contracted according to the average values. Figure 3 shows the results. Figures 3a and 3b show the average Dst profile of moderate and intense magnetic storms, respectively. It is not shown for weak magnetic storms in Figure 3, but the essential statistical nature for weak storms is the same as that for moderate and intense storms, excepthat the time scale for weak storms is shorter: the average main phase durations for are 21, 15, and 11 hours for intense, moderate, and weak storms, respectively. As expected, the dominant features in the upper panels are nearly identical to what Loewe and PrOlss [1997] have shown statistically: see their Figure 3. More precisely, the average diagrams of Loewe and PrOlss are a mixture of our two diagrams (Types 1 and 2 in our Figure 3). It should be noted that there is no obvious difference between moderate and intense storms in terms of the overall difference between and 2 storms, except for their durations and the peak intensities. Note that because the distance between the two peaks in Dst in varies considerably from storm to storm, the recovery of the first intensification is not very clear in the Figure 1. Schematic representation of Dst for and superposed plots. geomagnetic storms. See text for parameters that differentiate Figures 4a and 4b show the corresponding variations in auroral and magnetic storms. electrojet activity in AL and in the B z component of the IMF, respectively. Both quantities consist of two peaks in. This effect is particularly pronounced in the AL plot, where the 2. The two peaks in Dst must be separated by more than 3 second peak is more intense than the first. For both and 2 hours, T + Y> 3 hours. This condition was employed in order to storms, peaks in AL and IMF B z occur well before (> 1 hour)the exclude cases where apparent decreases in the Dst magnitude corresponding peaks in Dst. Also note that the two peaks in Bz were caused by such substorm effects as the so-called current are almost equal, whereas the second AL peak seems to be more wedge, not by a true decrease in the storm time ring current. We note, however, that there are cases where the two-step decreases November 13, 1985 are closely spaced; these will be missed in the present study. It is easy to see, as an extreme case, that if T = 0 and C = 0, becomes, whose intensity is A + B. There are cases in where the second peak is less pronounced than the first: B is significantly smaller than C. These constitute, -1{3 however, a very small fraction (8.5%) of all storms. We admit -20 ø t that even with these quantitative criteria, there are a number of "uncertain" magnetic storms in our data set. The corresponding AE indices and the IMF/solar wind data were also examined whenever they were available. 3. Results Figure 2 shows two examples of storms. In Figure 2a the two peaks in Dst (labeled as I and//) are separated by 7 hours, while those in Figure 2b are separated by only 4 hours. This difference in separation time is also deafly identified in the corresponding B z component of the IMF, although both cases include other fluctuations as well. It is interesting to note that the Figure 2b case is a C - 0 magnetic storm. Table 1 summarizes our statistics. Two points of interest are noted: First, more than 50% of all magnetic storms are found to go through two steps in Dst during the main phase. Second, the percentage of occurrence increases statistically as the peak intensity in Dst increases. About 67% of intense storms have a two-step growth, whereas a relatively simple growth in Dst is found in less than 30% of the magnetic storms. For each of the 1252 magnetic storms we have defined the main and recovery phases. Time 0 is defined as the time when Dst crosses zero, and the end of a storm is defined as the time when Dst recovers to one-tenth the level of its peak value. After Time (hrs) After T=0 May 5, " N 0 II m _ Time (hrs) After T=0 Figure 2, Two typical examples of magnetic storms on (a) November 13, 1985, and (b) May 5, 1989, along with the corresponding B z component variations in the interplanetary magnetic field.

3 KAMIDE ET AL.: TWO-STEP DEVELOPMENT OF STORMS 6919 Table 1. Classification of Geomagnetic Storms Into Two Types Uncertain All Number of Cases Weak Medium Intense All Percentages Weak Medium Intense All intense than the first. In individual cases, AL often returns to a very quiet state close to zero between the two peaks. Since the variability of the "quiet" interval between the two Dst minima is quite high, the average "recess" value is finite (nearly-350 nt) in Figure 4a Solar Wind Conditions The present study indicates that having a single, large disturbance in the solar wind is neither necessary nor sufficiento generate an intense geomagnetic storm. Our future efforts should then be directed toward identifying the cause for a two-stage structure in the southward IMF, not one large southward turning. This structure has in fact been observed in some of the intense magnetic storms [see Gonzalez and Tsurutani, 1987; Tsurutani et al., 1988, 1992; Gonzalez et al., 1989; Kamide et al., 1997] (see section 4). The importance of both sheath (or draped) fields and driver gas fields, carrying southward IMFs, was pointed out by Tsurutani et al. [1988] for the generation of major geomagnetic storms, displaying two-stage development characteristics. Grande et al. [1996], following this suggestion, have recently shown that CRRES heavy ion charge states were distinctly different during the two particle injections (which led to the two main phases) of the March 1991 great storm. Their interpretation was that these represent ion populations from two different coronal regions, corresponding to sheath and driver gas plasmas. In connection with finding the double IMF B z structure responsible for storms, one important candidate is a 4. Discussion Moderate storms In this paper we have statistically studied more than 1200 geomagnetic storms. It has been determined that the increase in the ring current during the main phase of an intense geomagnetic storm often goes through two steps. This may be surprising because the study of geomagnetic storms has a long history, establishing their average features in which there is a smooth, single main phase followed by a slow and relatively smooth recovery phase. With this simple picture in mind, the minimum Dst value at the main phase has long been utilized as the magnitude of magnetic storms Intensity of Magnetic Storms When one observes an intense magnetic storm, it is natural to assume that some single major event occurred at the Sun and something intense propagated through the interplanetary medium to the Earth. The present study clearly demonstrates, however, thathis picture may be oversimplified. What actually happens in many cases is that before a Dst decrease has fully recovered to the prestorm level, a second decrease often follows. In fact, auroral electrojet activity at high latitudes is found to go through two steps as well. The IMF also has a structure of two southward field regions (seen in Figure 4b). This means that some of the "largest" geomagnetic storms consist of two or more superposed medium-size storms. Thus an "intense" magnetic storm in terms of the peak Dst value may result from the superposition effect, rather than a single, intense disturbance in the interplanetary field. This raises an important question regarding how one can - / Intense storms relied define on the intensity maximum of a Dst geomagnetic magnitude storm. observed One at has the customarily end of the stor main phase, buthis clearly may not be correct. The present rely on study the suggests minimum that Dst it is value not physically to define very storm meaningful intensity, to. Type I (b) o Type2 particularly for intense magnetic storms. It is interesting to o 0 o speculate as to why earlier studies did not reveal that intense magnetic storms often go through two steps during the main phase. It may well be that studies picked up only the peak value Figure 3. Results of a superposed-epoch analysis of Dst for (top) of Dst in identifying magnetic storms without paying special and (bottom) magnetic storm shown in solid lines: attention to how Dst reached the peak. Even though a double (a) moderate and (b) intense magnetic storms. Dotted lines above structure in the Dst development was found, it might have been and below the solid lines show the standard deviations. Two treated as two magnetic storms that occurred within a short vertical dotted lines in each diagram indicate the startime and the interval. end time of the main phase. -

4 ._ KAMIDE ET AL.: TWO-STEP DEVELOPMERqT OF STORMS :: AL index (a) ': :: ,', O 4O 6O Interplanetary Magnetic Field...,. would predictably result in fast ejecta events with south-north magneticloud fields having greater intensities (statistically). This is because there would be less separation between the first Bs event and the second (cloud) event. We are now in the process of testing this idea, It should be noted that the above scenario applies mainly for solar maximum intervals when CMEs are frequent. We expect that at solar minimum, high-speed streams from coronal holes interacting with slower streams can also produce fairy large B s structures, especially due to compression of large-amplitude Alfv6n waves in corotating interaction regions. However, in this latter case storms are expected to be more frequent due to the difficulty in obtaining additional large B s structures as in the CME case. It is also important to note that the time separation between the B s structures in the interplanetary CME case can vary from case to case, leading to a shorter larger spacing the corresponding Dst enhancements. Finally, when the interplanetary extension of a CME does not involve a shock, and one does not have a shock compressed B s structure, one could still have a dual Bs structure if the "draping" component [see Zwan and Wolf, 1976] ahead the ejecta are substantial Magnetospheric Processes, As for the cause of magnetic storms, there is at least one important candidate process in the magnetosphere we need to consider. There are two major particle sources for the ring current: the solar wind and the ionosphere. The ionospheric source has recently been found to become dominant in the inner magnetosphere during the main phase of the largest magnetic -2O 0 20 storms, particularly near solar maximum [Hamilton et al., 1988; Daglis, 1997]. Daglis and Axford [1996] have shown that the ionosphere responds to enhanced substerm activity with a fast feeding of the inner magnetotail, which occasionally results in a transient localized dominance of O + ions. It is quite possible that two distinct processes play the leading role in the two successivenhancements in the ring current. The first enhancement in the ring current, that is, the first Dst decrease, may be due to the magnetospheric convection driven by the southward IMF [e.g., Burton et al., 1975; McPherron, 1997], while the second ring current enhancement, that is, the second Dst decrease, may be due to the substorm-associated accumulation of Figure 4. Results of a superposed-epoch analysis of (a) the AL a new O + population [Daglis, 1997]. This second growth of the index and (b) the B z component of the interplanetary magnetic ring current must be driven by "highly fluctuating" electric fields field for (top) and (bottom) magnetic stormshown [e.g., Chen et al., 1994], resulting from substerm expansions. It in solid lines. Dotted lines above and below the solid lines show is well known that toward the end of the main phase of magnetic the standard deviations. Two vertical dotted lines in each diagram storms, the occurrence of intense substerms is very frequent and indicate the start time and the end time of the main phase. that the O + energy density in the inner magnetosphere is strongly correlated with these substerm activities [Daglis et al., 1994]. Thus the first development phase seems to prime the ring current, setting up a precondition for the second phase, which is shocked Bs (negative Bz) field followed by a magneticloud field dominated by injections of ionospheric ions. in the interplanetary extension of a coronal mass ejection (CME). When the solar ejecta propagates at a speed greater than the upstream slow solar wind such that the speed differential is 5. Summary greater than the magnetosonic speed, a fast forward shock develops. The greater the speed differential, the stronger (in Our statistical investigation of Dst variations for more than Math number)the shock: the shock will compress the upstream 1200 geomagnetic storms indicates the existence of a distinct magnetic fields and create a high-intensity field sheath region class of two-step main phase storms. Contrary to the tacit downstream from the shock. If the upstream field is originally assumption that an intense storm is caused by an intense southward, shock compression will lead to intense Bs in the solar/interplanetary disturbance, our result suggests that an sheath (there are also other mechanisms to create B s sheath intense storm can result from the superposition of two successive, fields). Following the sheath, the internal field of the ejecta itself, moderate storms, driven by two successive, southward IMF often called a magneticloud, can take on a helical structure with structures. An alternative interpretation is that intense storms can a cross- sectional rotation in the Z-X plane, showing a rotation result from a two-step development in the ring current, which is from south to north (or vice versa) in the IMF. The southern part initially the result of large-scale convection in the magnetosphere, of that field can become the second large B s structure, responsible and eventually substorm-associated injection of ionospheric O + for the second stage of a storm. The above scenario ions into the inner magnetosphere.

5 KAMIDE ET AL.: TWO-STEP DEVELOPMENT OF STORMS 6921 We suggesthat the classification of geomagnetic storms by their minimum Dst only is not sufficient in a comprehensive geomagnetic storm of Febmary 1986,,/. Geophys. Res., 93, 14343, study of storm dynamics. In any case, southward IMF is an Joselyn, J. A, and B. T. Ts, Geomagnetic sudden impulses and essential precondition of intense magnetic storms, although the storm sudden commencements (abstract), Eos Trans. AGU, 71, 1808, exact cause-and-effect relationship is not yet fully clear. The Kamide, Y., R. L, McPherron, W. D. Gonzalez, D.C. Hamilton, H. S. topic promises exciting future work within the framework of Hudson, J. A Joselyn, S. W. Kahler, L, R. Lyons, H. Lundstedt, and space weather efforts. E. Szuszczewicz, Magnetic storms: Current understanding and outstanding questions, in Magnetic Storms, Geophys. Monogr. Set., Acknowledgments. We would like to thank L. R. Lyons, S. vol. 98, edited by B. T. Tsurutani et al., pp. 1-19, AGU, Washington, Kokubun, and L Bargatze for their illuminating discussions throughout D.C., the present study. The work at the Solar-Terrestrial Environment Loewe, C. A, and O. W. Pr61ss, Classification and mean behavior of Laboratory was supported in part by the Ministry of Education, Science, magnetic storms, J. Geophys. Res., 102, 14209, Culture and Sports (Monbusho) under a Grant-in-Aid for Scientific McPherron, R. L., The role of substorms in the generation of magnetic Research, Category B. Portions of this work performed at the Jet storms, in Magnetic Storms, Geophys. Monogr. Set., vol. 98, edited Propulsion Laboratory, California Institute of Technology, Pasadena, by B. T. Tsurutani et al., pp , AGU, Washington, D.C., under contract with the National Aeronautics and Space Administration The Editor thanks S. W. H. Cowley and another referee for their Sckopke, N., A general relation between the energy of trapped particles assistance in evaluating this paper. and disturbance field near the Earth, d. Geophys. Res., 71, 3125, References Sistoe, G. L., The vidal theorem applied to magnetospheric dynamics, J. Geophys. Res., 75, 5340, Sugiura, M., and S. Chapman, The average morphology of geomagnetic Akasofia, S.-I., The development of geomagnetic storms without a preceding enhancement of the solar plasma pressure, Planet. Space Sci., 13,297, Burton, R. K., R. L. McPherron, and C. T. Russell, r-lan empirical relationship between interplanetary conditions and Dst, J. Geophys. Res., 80, 4204, Chen, M. W., M. Schulz, and L. R. Lyons, Simulations of phase space storms with sudden commencement, Abhandl. Akad. Wiss. GOttingen. Math. Phys. KL Sondehefi, 4, Taylor, J. R., M. Lester, and T. K. Yeoman, A superposed epoch analysis of geomagnetic storms, Ann. Geophys., 12, 612, Tsumtani, B. T., W. D. Gonzalez, F. Tang, S.-I. Akasofia, and E. J. Smith, Solar wind southward B z features responsible for major magnetic storms of , J. Geophys. Res., 93, 8519, distribution of stormtime proton ting current, J. Geophys. Res., 99, Tsumtani, B. T., W. D. Gonzalez, F. Tang, and Y. T. Lee, Great 5745, Daglis, I. A., The role ofmagnetosphere-ionosphere coupling in magnetic storm dynamics, in Magnetic Storms, Geophys. Monogr. Ser., vol. 98, edited by B. T. Tsumtani et al., pp , AGU, Washington, D. C., magnetic storms, Geophys. Res. Lett., 19, 73, Yokoyama, N., and Y. Kamide, Statistical nature of geomagnetic storms, J. Geophys. Res., 102, 14215, Zwan, B. J., and R. A. Wolf, Depletion of the solar wind plasma near a planetary boundary, J. Geophys. Res., 81, 1636, Daglis, I. A, and W. I. Axford, Fast ionospheric response to enhanced activity in geospace: Ion feeding of the inner magnetotail, d. Geophys. Res., 101, 5047, Daglis, I. A., S. Livi, E. T. Sarris, and B. Wilken, Energy density of ionospheric and solar wind origin ions in the near-earth magnetotail A Brekke, Auroral Observatory, University of Tmmso, Tmmso N- during substorms, J. Geophys. Res., 99, 5691, , Norway. ( asgeir@windows.phys.uit. no) Dessler, A J., and E. N. Parker, Hydromagnetic theory of geomagnetic I. A. Daglis, Institute of Ionospheric and Space Research, National storms, J. Geophys. Res., 64, 2239, Observatory of Athens, Palea Penteli, Greece. ( Gonzalez, W. D., and B. T. Tstmatani, Criteria of interplanetary daglis@creator. space.noa. gr) parameters causing intense magnetic storms (Dst < -100 nt), Planet. W. Gonzalez, Instituto National de Pesquisas Espacias, Postal 515, Space Sci., 35, 1101, S o Jos6 dos Campos, Sao Paulo, Brazil. ( Gonzalez, W. D., B.T. Tsurutani, A L. Clua de Gonzalez, F. Tang, E. J. gonzalez magnet. dge.inpe.br) Smith, and Y. Kamide, S. Masuda, and N. Yokoyama, Solar-Terrestrial S.-I. Altasofia, Solar wind-magnetosphere coupling during intense Environment Laboratory, Nagoya University, Toyokawa, Aichi 442, geomagneitc storms ( ), J. Geophys. Res., 94, 8835, Japan. ( kamide@stelab.nagoya-u. ac.jp; masudat stelab.nagoya u. ac.jp; nobu@stelab.nagoya-u. ac.jp) Grande, M., C. H. Perry, J. B. Blake, M. W. Chen, J. F. Fennell, and B. B. T. Tsmarta, Jet Propulsion Laboratory, California Institute of Wilken, Observations of iron, silicon, and other heavy ions in the Technology, Pasadena, CA ( btsurutani@jplspl.jpl.nasa. geostationary altitude region during late March 1991, J. Geophys. gov) Res., 101, 24707, Hamilton, D.C., G. Gloeckler, F. M. Ipavich, W. Studemam B. Wilken, and G. Kremser, Ring current development during the great (Received June 9, 1997; revised September 25, 1997; accepted November 13, 1997)