GEOPHYSICAL RESEARCH LETTERS, VOL.19, NO. 1, PAGES 73-76, JANUARY 3, 1992 GREAT MAGNETIC STORMS Bruce T. Tsurutani 1, Walter D. Gonzalez 2, Frances Tang 3, and Yen Te Lee 4 Abstract. The five largest magnetic storms thatoccurred between 1971 to 1986 are studied to determine their solar and interplanetary causes. All of the events arefoundto be associated with high speedsolar wind streams led by collisionless shocks. The high speed streams are clearly related to identifiable solar flares. It is foundthat: 1) it is the extreme values of the southward interplanetary magnetic fieldsratherthan solarwind speeds that are the primary causes of greatmagnetic storms, 2) shocked and draped sheath fields preceding thedriver gas (magnetic cloud) areat leastas effective in causing the onsetof great magnetic storms (3 of 5 events)as the strong fields within the driver gas itself, and 3) precursor southward fields ahead of the high speed streams allow theshock compression m•chanism (item 2) to beparticularly geoeffective. Introduction It has been demonstrated that there are a variety of interplanetary phenomena associatedwith southward magnetic fields which cause magnetic storms (Burlaga et al., 1987; Tsurutani et al., 1988; McComas et al., 1989, Gosling et aI., 1990; 1991). Some of these features/mechanisms are: the shock compression of quietupstream Bs, sheath field draping, waves and turbulence, compressed heliospheric current sheets, driver gases/magnetic clouds (coronal mass ejections), andcompound streams. Recently, Gosling et al. (1991), from a statistical analysis, have indicated that the subset of interplanetary causes are different for different levels of storm intensities. In this paper we wish to briefly address • the interplanetary and solarcauses of the very largest (gmat) magnetic storms. We have selected thefive largest magnetic storms that have occurred in the interval of 197I to 1986, using the Dsz index. Prior to 1971, there is little interplanetary data to use for an analysis of this type. The Dsz index is constructed from near-equatorial magnetic data and is a superior storm index in comparison to midlatitude indices such as Kp and Ap which are sensitive to substorms as well (Tsurutani andGonzalez, 1990;Joselyn and Tsurutani, 1990). The great storm onsets occurred on: July 13, 1982 (peak Dsz--- 325 nT), April 13, 1981 (Dsz = - 311 nT), February7, 1986 (DsT= - 312 nT), September 5, 1982 (DsT =- 289 nT) and December 19, 1980 (Dsz = - 249 nT). It should be noted thatfour of the storms occurred in the latter portion of (orslightly after) the1979- 1981solar maximum, andoneevent (February, 1986) near solar minimum. The biggest event of our study, July 13, !982, was the8th largest within the DsT tape interval of 1957 to 1986. The largest event wasthe July 15, 1959 storm (peak DsT = - 429 nT). Six of the seven most intense events since 1942 occurred during the 1957-1960 epoch when, unfortunately, interplanetary data were not available. We have chosen to study the very largest magnetic storms, not only to determine if there are different interplanetary andsolar causes (or if there areeven finer subsets of causes), but also to understand thephysics of 1Jet Propulsion Laboratory, Calif. Inst. of Technology 2Instituto Nacional de Pesquisas Espacias 3California Institute of Technology 4Jet Propulsion Laboratory, Calif. Inst. of Technology Copyright 1992 bythe American Geophysical Union. Paper number 91 GL02783 0094-8534 / 92/91GL-02783 $03.00 such events to beable to eventually predict their occurrence. Predictions will help preventmagnetospheric satellite damage, urban poweroutages and will give warnings of impending radiation hazards to humans flying at high altitudes (Alienet al., 1989). Approach We have used the Boulder World Data Center A Ds:r tape (courtesy of H. Kroehl)to identifythe five largest magnetic storms whichhaveoccurred from 1971 until 1986. High and midlatitude geomagnetic indices such asAL, A.E, and Kp werealsoobtained from theWDC A. The spacecraft plasma and field data (ISEE-3, IMP-8, and ISEE-1) were obtained from the Space Science Data Center, GSFC, courtesy of J. King. The severe (peak) Dsz values used in this study are higher thanthe thresholds of any other prior study. Burton et al. (1975) studiedstorms with Dsz between -40 nT and - 130 nT. Tsurutaniet al. (1988) examined stormswith - 220 nT < Ds•: _< - 100 nT, intensities almost totally complementary with theBurton et al. study andthe present one. Burlaga et al. (1987)studied "major" storms with Ap > 90, a threshold somewhat comparable to theTsurutani et al. study, but with overlapof only 4 out of 17 events. The Gosling et al. (1990;1991) studies used a criteria of Kp > 8- or Kp > 6- for at least three 3-hour periods within a 24-hour interval, the highest storm criteria other than this present study. Although the criteria of the above storm studieswere lowerthan thepresent study, several publications (Burlaga et al., !987; Gosling et al., 1990, 1991) have included information on the events. However, none of the studies address the specific physical causes of the intense interplanetary sheath Bswhich led to the onsets of the great magnetic storms.This will be onethe important focuses of this note. Results Figure I illustrates the September 5, (Day 248), 1982 event. From topto bottom, thepanels are: the solar wind velocity, theplasma density, electron temperature, magnetic field magnitude, theGSM Bz component, Dsz, AE, andthe epsilon parmeter. This figure indicates thatthe onset of the storm main phase (indicated by theabrupt decrease in Ds-r) is coincident with the passage of the shock. The shockis emphasized bythedashed vertical lineat-• 2100UT day 248 (September 5), 1982. The fast forward shockis identified bythe abrupt increase in velocity from --- 500km s -1to •- 800 kms -1 and in the magnetic field magnitude, from ~ 8 nTto~ 26 nT. Thepropagation time delay of thefield and plasma at ISEE-3 to the magnetosphere have not been taken into account in theFigure. This delay should be muchless than 1 hour for these highsolar wind speeds and is negligible for dataplots shown on these scales. The cause of the Dsx decrease and AE increase is thesouthward IMF component at andimmediately behind the shock. "Precursor" geomagnetic activity exist prior to thestorm onset. From the middle of day 247 until the storm onset, Dsr was --- 50 nT. Fromthebeginning of day 246 until midday 247, Dsz was -- - 30 nT. To demonstrate that this is not simply an artifact of determining the "zero level"of the Ds.. index,theAE electrojet indexis alsoshown. The latter indicates that substantial auroral activity was ongoing throughout day246 to day248, with AE typically having values of ~ 500nT. This is consistent withthe nonzero Dsz values. 73