The ELF (10–1500 Hz) electromagnetic emissions in the midnight sector of the outer magnetosphere have been studied using Ogo 5 search coil magnetometer data. Chorus was detected in conjunction with magnetospheric substorms throughout the region from L = 5 to L = 9 but only during postmidnight hours. No chorus was seen in the 3 hr preceding midnight even when substorms were in progress. The postmidnight chorus occurred only within ±15° of, and most frequently at, the geomagnetic equator. Time‐averaged intensities varied from 10−8 to 10−6 γ²/Hz, which is more than an order below the maximum values reported previously for dayside chorus. The chorus occurred in narrow frequency bands having a bandwidth of 50–200 Hz. Chorus frequencies varied from less than one‐fourth to as high as three‐fourths of the equatorial electron gyrofrequency as determined by the on‐board magnetometer. All frequencies in this range were detected except for a narrow band near one‐half the electron gyrofrequency where the chorus appeared to be strongly attenuated. Chorus was often observed as two distinct bands above and below one‐half the gyrofrequency. The two most common types of chorus were narrow band chorus without structure and falling tones. Rising tones and hooks were also observed. Chorus pulsations were observed often with quasi‐periods of 5–15 s. A correlation was sought, but none was found, between the micropulsations in the ambient magnetic field and the chorus pulsations. Many features of postmidnight chorus can be explained by a cyclotron resonant interaction between the waves and the substorm electrons. The distribution of the chorus as a function of local time and L is strikingly similar to the distribution of enhanced, trapped, and precipitated substorm electrons with energies >40 keV. The postmidnight occurrence of chorus is attributable to the eastward curvature and gradient drift of the injected electrons. Cyclotron resonant interactions should be strongest near the equator, as was observed. The confinement to within 15° of the equator is attributed to Landau damping by low‐energy (1–10 keV) auroral electrons. The attenuation band at one‐half the electron gyrofrequency may result from Landau damping by electrons that have energy corresponding to cyclotron resonance but are traveling in the same direction as the waves. A close correspondence is expected between the occurrence of postmidnight and dayside chorus. The maximum in dayside chorus intensity at approximately 1000 LT, which is correlated with the dayside maximum of energetic electron precipitation, may represent further precipitation of the substorm electrons injected near midnight.

Extremely low frequency (10-1500 Hz) magnetospheric chorus has been analyzed to investigate a possible dependence on substorms. Care was taken to separate spatial effects from temporal effects by analyzing an entire year of data acquired by the Ogo 5 search coil magnetometer. A major finding of the study of spatial dependences is that chorus occurs principally in two magnetic latitude regions. Equatorial chorus is detected near the equator, and high-latitude chorus is found at magnetic latitudes above 15 ø. When chorus in these two regions is analyzed separately, substorm dependences become apparent. Comparisons with A E indicate that equatorial chorus occurs primarily during substorms. High-latitude chorus is not strongly dependent on AE and often occurs during intervals of prolonged quiet with •E < 100 ,y for the previous 12 hours or more. The dependence of equatorial chorus on local time, magnetic latitude, and L is consistent with generation by a cyclotron resonance between the whistler mode chorus and 10-to 100-keV trapped substorm electrons. Equatorial chorus has an abrupt onset in the postmidnight sector and a second enhancement from dawn to noon, a pattern which is similar to that of energetic electron precipitation. The occurrence frequency of equatorial chorus peaks at the equator, I magnetic latitudel < 5 ø, a region where cyclotron resonance is most efficient. The L value of maximum chorus occurrence increases from 5-8 postmidnight to 7-11 postdawn, a dependence which is consistent with generation by electrons which have undergone drift shell splitting. Delay times between substorms and the onset of equatorial chorus are consistent with a gradient drift of •25-keV electrons. Equatorial postmidnight chorus and postdawn chorus have similar occurrence rates and wave intensities. The maximum chorus occurrence rates are 54% postmidnight and 56% postdawn. Time-averaged equatorial chorus intensities >_ l0 -8 ,V•' are detected up to 17% of the time for 6 _< L _< 7 postmidnight and up to 14% of the time for 7 _< L _< l0 from dawn to noon. Such wave intensities are sufficient to cause near-strong pitch angle diffusion of electrons for L > 6 and strong diffusion for L > 8. Instantaneous diffusion rates may be considerably higher owing to the discrete burstlike nature of the chorus. The spatial and temporal dependences of high-latitude chorus are considerably different from those of equatorial chorus. Highlatitude chorus occurs in local day and evening and at large L. The emission is detected primarily on the dayside, at 0800 _< LT _< 1600, and often within 1-2 RE of the magnetopause. The occurrence of highlatitude chorus during quiet intervals is consistent with local generation within 'minimum B pockets.' Jackerott, 1963; Egeland et al., 1965], at ionospheric altitudes [Taylor and Gurnett, 1968; Barrington et al., 1971; Thorne et al., 1977], and in the outer magnetosphere, the region where chorus is believed to be generated [Russell et al., 1969; Dunckel and Helliwell, 1969; Tsurutani and Smith, 1974; Burtis, 1...

The origins of the interplanetary southward B z which cause the 10 major (Dst • --100 nT) magnetic storms detected during the 500 days of study (August 16, 1978, to December 28, 1979 of the Gonzalez and Tsurutani (1987) work are examined in detail. A full complement of ISEE 3 plasma and field data, an ll-station AE index and the near-equatorial Dst index, are used in this analysis. It is found that the origins of the interplanetary southward B.. events are quite varied. If it is defined that the B z event which leads to Dst • --100 nT is "the cause" of the storm, then one of the storm intensifications is caused by shock compression of preexisting southward interplanetary magnetic fields, four (or five) are related to driver gas magnetic fields, one (or two) are caused by shocked kinky heliospheric current sheets, two (or three) by turbulence or waves behind interplanetary shocks, and one possibly by draped fields associated with a noncompressive density enhancement event (without a shock or a high-speed stream). In simplistic terms, four (or five) storms are caused by driver gas fields, four by shocked (sheath) fields, and one possibly by high-intensity draped fields. In actuality, many of the interplanetary southward B z and corresponding magnetic storm (Dst) structures are more complex than stated above. At least four of the interplanetary events have both major sheath and driver gas southward B z events. In two storms, sheath southward Bz features led to Dst reaching levels of --90 nT prior to driver gas southward B z features; the following driver gas fields then caused Dst to exceed our storm criteria of _< -100 nT. In two other cases, sheath B.. features led to magnetic storm onsets (Dst • --100 nT); the following driver gas southward B z features cause further storm intensifications. The above magnetic storms therefore displayed two-stage development characteristics. The results of this study indicate the equal importance of both sheath fields or draped fields and driver gas fields for the generation of major geomagnetic storms. Because of the importance of the sheath fields the intensity and duration of geomagnetic storms cannot be predicted by solar observations of active regions alone. Tang et al. (1988) will address this topic in detail. Paper number 7A9404. 0148-0227/88/007 A-9404 $05.00 when oriented in the vertical direction, have large southward (then northward) field components (or vice versa). The criteria of a "cloud" are a radial dimension of .-•0.25 AU at 1 AU, high, > 10 nT, field magnitudes, and magnetic field directional changes by a rotation in a plane. Klein and Burlaga note that this field geometry is consistent with a magnetic loop or bubble [see Burlaga and Behannon, 1982] but cannot be uniquely determined because of the limitation of singlespacecraft measurements. In a sense the above classification is unfortunately too broad for the purpose of this present study.As examples, kinky heliospheric current sheets [Smith, 1981;Akasofu, 1981;Tsurutani et al., 1984 that are present in the co...

A relatively steady band of ELF hiss has been detected by the Ogo 5 search coil magnetometer on almost every passage through the plasmasphere; except for an anomalous region on the dayside at high geomagnetic latitudes, the emissions terminate abruptly at the plasmapause, and we therefore refer to them as 'plasmaspheric hiss.' A preliminary statistical study of the properties of the observed whistler mode turbulence has yielded the following characteristics' the waves are band limited with a sharp lower-frequency cutoff and a more diffuse upper-frequency cutoff; power spectra show a well-defined maximum near a few hundred hertz, the peak intensities generally ranging between 10 -7 and 10 -5 v"Hz; the wave energy is spread over a bandwidth of a few hundred hertz, and cerresponding wide band amplitudes are 5-50 my; the waves are highly turbulent in nature and show little tendency of definite polarization. The above properties remain essentially constant throughout the plasmasphere. Observed properties of the hiss are con. sistent with generation at all local times in • restricted L range just within the plasmapause. Waves subsequently propagate on complex paths to fill the plasmasphere. The most probable generation mechanism is cyclotro n resonant instability wit h lo.w-energy electrons that continually diffuse inward from the outer radiation zone. At lower L, hiss resonates with higherenergy electrons, and thus the electrons are scattered in pitch angle and hence lost to the atmosphere throughout the 'slot' between the inner and outer radiation belts. Extremely low frequency whistler mode wavesare an important constituent of the magnetosphere, since they can resonate with radiation belt electrons and thus induce pitch angle scattering and a concomitant precipitational loss to the atmosphere. To date, the most complete in ritu measurements of such waves have been made on the Ogo satellites [.Dunckel and Helliwell, 1969; Russell et al., 1969]. Valuable information has also been obtained from loweraltitude polar-orbiting satellites [Taylor and Gurnett, 1968; Gurnett et al., 1969; Barrington, 1971; Muzzio and Angerami, 1972], but an extrapolation of such observations deep into the magnetosphere is complicated by the susceptibility of the waves to internal reflection [Ki-

E. J. SMITH ,let Propulsion Laboratory, California Institute of Technology, PasadenaNormally the •> 80-eV electrons which carry the solar wind electron heat flux are collimated along the interplanetary magnetic field (IMF) in the direction pointing outward away from the sun. Occasionally, however, collimated fluxes of •> 80-eV electrons are observed traveling both parallel and antiparallel to the IMF. Here we present the results of a survey of such bidirectional electron heat flux events as observed with the plasma and magnetic field experiments aboard ISEE 3 at times when the spacecraft was not magnetically connected to the earth's bow shock. The onset of a bidirectional electron heat flux at ISEE 3 usually signals spacecraft entry into a distinct solar wind plasma and field entity, most often characterized by anomalously low proton and electron temperatures, a strong, smoothly varying magnetic field, a low plasma beta, and a high total pressure. Significant field rotations often occur at the beginning and/or end of bidirectional heat flux events, and, at times, the large field rotations characteristic of "magnetic clouds" are present. Approximately half of all bidirectional heat flux events are associated with and follow interplanetary shocks, while the other events have no obvious shock associations. When shock associated, the delay from shock passage typically is ~ 13 hours, corresponding to a radial separation of ,-•0.16 AU. Independent of any shock association, bidirectional heat flux events typically are ~0.13 AU thick in the radial direction, although considerable variability is evident from one event to another. Near solar activity maximum, bidirectional heat flux events occurred at a rate of •3 per month, and the solar wind electron heat flux was bidirectional ~ 5% of the time. Bidirectional heat flux events often contain strong out-of-the-ecliptic field components and thus can be effective in producing geomagnetic disturbances. This is particularly true for shock-associated events where the intrinsically strong fields in the leading portions of the events are amplified by compression in transit from the sun and where strong out-of-the-ecliptic field components resulting from compression and draping of the ambient field are often present within the shocked plasma immediately ahead. Consistent with previous work we interpret the bidirectional heat flux as evidence for a closed field topology in interplanetary space. Further, we suggest that these events are one of the more prominent signatures of coronal mass ejection events in the solar wind at 1 AU.On occasion it is observed that solar wind electrons with energies •<80eV are collimated both parallel and antiparallel to the IMF. At such times the phase space densities of •> 80-eV electrons traveling in opposite directions along the field are roughly comparable, though not necessarily equal.