Foster John : Storm Time Plasma Transport at Middle and High Latitudes

INTRODUCTION

The large-scale convection electric field which couples the Earth's ionosphere to magnetospheric processes has been observed by ground-based and satellite-borne techniques with sufficient precision and regularity that quantitative models of its global extent and characteristics have been generated for a variety of geophysical circumstances. Satellite-based data sets have resulted in the empirical models of Heppner [1977], Heelis et al. [1982], and Heppner and Maynard [1987], and radar observations of ionospheric plasma drifts led to the models of Evans et al. [1980], Foster [1983], Foster et al. [1986], and Holt et al. [1987], while inversion techniques have determined the electric fields from the magnetic perturbation due to electric field-driven ionospheric currents [Kamide et al., 1981; Friis-Christensen et al., 1985]. These global electric field models indicate a dominant two-cell pattern of plasma convection whose equatorward portion transports ionospheric plasma sunward toward the noon meridian from both the dawn and dusk auroral and subauroral regions. The effects of plasma transport, and the linkage to the overlying magnetosphere, dominate ionospheric characteristics at these latitudes. With increasing activity level, and particularly during geomagnetic storms, the equatorward extent of such effects expands to lower latitudes [e.g., Foster et al., 1986]. At high latitudes in the noon sector, this sunward convecting plasma becomes a source for plasma entering the polar cap, across the regions of the cusp and cleft [Foster, 1989]. Plasma patches arising from this source populate polar latitudes where they are responsible for enhanced airglow and scintillations [Buchau et al., 1985].

The Dusk Effect at Middle Latitude

During the early stages of many magnetic storms the total electron content at subauroral latitudes (TEC), integrated vertically through the ionospheric F region, first increases (the storm positive phase) and often exhibits a pronounced maximum near sunset (the dusk effect) [Mendillo et al., 1970; Evans, 1970]. Evans [1970] related this increase to an uplifting of the F layer to altitudes where recombination proceeds more slowly. The theoretical work of Jones and Rishbeth [1971] suggested that an enhanced equatorward neutral wind, driven by atmospheric heating at auroral latitudes, would drive ionization up the magnetic field line, resulting in decreased recombination and a total density enhancement. Papagiannis et al. [1971] suggested that the dusk effect could be produced as an increased convection electric field at mid-latitudes during the storm acted to oppose plasma corotation, causing plasma to Ûpile upÙ and compressing the magnetic field in the dusk sector, resulting in a dumping of plasma from the plasmasphere. This idea was investigated and rejected by Rishbeth and Hanson [1974] on the grounds that the geomagnetic field is virtually incompressible. They suggested that the major effect was advection of relatively dense plasma into the volume monitored by the experiments. Evans [1973], using incoherent scatter radar, observed just such a westward (sunward) plasma drift (Ê200 m sÚ1) associated with the dusk TEC enhancement and concluded that the density increase was caused largely by horizontal transport of ionization from the evening into the late afternoon sector. It is significant to note that these early papers which defined and discussed the storm time dusk effect utilized data obtained at longitudes near the east coast of the North American continent.

Subsequent investigations included a large number of storms [e.g., Buonsanto et al., 1979] and confirmed the persistence of such features. Anderson [1976] modeled the mid-latitude ionospheric storm and showed that a combination of a storm-driven equatorward neutral wind, which acts to drive plasma up magnetic field lines to a region of reduced recombination, and westward convection, driven by a storm-enhanced northward electric field, can produce both the strong positive phase (enhanced TEC) near dusk and the sharp transition to the negative phase of the storm (depleted density) which follows it. In that model the northward electric field acted to slow corotation, but plasma transport from other longitudes was not included. During one event, Prolss et al. [1991] noted that the dusk effect density increase was preceded by a nearly 100-km increase in the height of the F layer peak and that the onset of the perturbation progressed from high to low latitudes at a speed of 500 m sÚ1. Those authors attributed the dusk effect ionization enhancement to an equatorward propagating traveling atmospheric-ionospheric disturbance (TAID [e.g., Bowman, 1965]) initiated at high latitudes during the storm.

Plasma Entry into the Polar Cap

During disturbed conditions, rapid sunward convection from the postnoon mid-latitude ionosphere carries high-density solar-produced F region plasma through the dayside cleft and into the polar cap. Radar observations from a number of sites have indicated the repeatability of this feature of the dayside ionosphere both in average studies and individual cases. Within the poleward convecting feature, plasma densities can be enhanced throughout the topside by a factor of 5 or more, and the total electron content increased by a factor of 2 to 4 [Foster, 1989]. This plasma is observed to spread out along convection trajectories within the polar cap where it constitutes a source for the observed polar cap F region density enhancements and their effects. The plasma tongue carried through the cleft from lower latitudes serves as a tracer of polar cap convection away from the cusp and cleft.

Foster [1984] examined the average diurnal pattern of convection and F region density at auroral latitudes observed from Chatanika, Alaska. In inertial coordinates, convection stagnates in the postnoon region of strong solar production before turning sunward toward the high-latitude noontime cleft. A tongue of enhanced density was seen to follow the convection contours toward the cleft and the polar cap, and it was conjectured that the high-density plasma which enters polar latitudes at noon is convected rapidly antisunward across the polar cap where it contributes to the enhancement of F region density seen above 70°Å near midnight. In a study of the nighttime F region density enhancements which were consistently seen at different radar sites during solar maximum, de la Beaujardiere et al. [1985] concluded that this plasma was of solar-produced origin and had been convected across the polar cap from a sunlit source region on the dayside.

A more detailed study of an individual noontime event observed from Chatanika was presented by Foster and Doupnik [1984], who observed high-density F region plasma convecting poleward through the cleft from a source at lower latitudes in the afternoon sector. They found that the prenoon convection cell, bringing plasma sunward from the predawn darkness, was associated with low F region densities, while the postnoon convection cell was clearly marked by the higher densities of the lower-latitude afternoon sector F region. While the radar probed poleward in the vicinity of the noontime convection convergence and the ionization tongue, high temporal resolution observations revealed discrete patches of topside plasma moving with the observed convection velocity into the polar cap. The study of Foster and Doupnik [1984] suggested that convection through the cleft can result in spatially discrete patches of enhanced F region density and that the motion of these patches can be used to delineate dayside convection trajectories. Such patches had been identified within the polar cap by Weber et al. [1984]. Weber et al. [1986] tracked such ionization patches flowing in the antisunward direction from the center of the polar cap to the edge of the nightside auroral oval.

In 1983, the former Chatanika radar began operations in the near vicinity of the noontime cleft at Sondrestrom, Greenland. From this much closer vantage point the features of the cusp and cleft convection patterns and their ionospheric signatures could be examined in detail. Foster et al. [1985] used radar azimuth scans to determine the noontime convection pattern over a span of 2 hours of local time and 10° of latitude with 20-min time resolution. As had been seen at Chatanika, a region of solar-enhanced ionization was observed to be carried through the cleft and into the polar cap by the postnoon convection cell. Detailed radar observations of E region density enhancements and F region electron temperatures and direct satellite observations of precipitating particles revealed that precipitation effects were confined to the near vicinity of the convection reversal and verified that the high-density plasma being carried into the polar cap was not locally produced by particle precipitation. A similar conclusion was reached by Buchau et al. [1985], who examined the F region ionization patches at polar cap latitudes.

Foster [1989] used Millstone Hill radar observations during a major storm on January 31, 1982, to examine the dayside plasma entry region from a position equatorward of the cleft and clearly observed a sunward convecting ionization feature extending along the equatorward edge of the postnoon trough and into the polar cap. Radar elevation scans intersected this feature, providing density/altitude profiles through the plasma tongue and mapping its local time/latitude behavior. Within the plasma tongue, TEC was enhanced by over a factor of 3 to a level of 75 TEC units (75Y10Ú16 mÚ2), and peak densities of 2.5Y1012 mÚ3 at 400 km altitude and convection into the polar cap at speeds in excess of 1500 m sÚ1 were observed.

STORM TIME EXPERIMENTS

The Millstone Hill radar facility is located at mid-latitude near L=3 (55°Å) on the east coast of the North American continent. The longitude of the facility is significant (288.5°E) because the north geomagnetic pole is offset from the geographic pole along a magnetic meridian near the site with the result that auroral magnetic latitudes reach their lowest extent in geographic latitude in the sector observed by the radar. Use of a fully steerable 46-m antenna permits the Millstone Hill UHF incoherent scatter radar to monitor the ionospheric F region over a 25° to 30° span of latitude centered on the site. This field of view encompasses important boundaries and regions of the magnetosphere-ionosphere system from the polar cap and cleft, across auroral latitudes, to the trough, the plasmapause, and the mid- and low-latitude ionosphere.

A standard storm time experimental mode has been developed for use at Millstone Hill consisting of full south to north elevation scans to monitor ionospheric features over as wide a range of latitude as possible, detailed local observations to investigate storm time effects specific to mid-latitudes, and a low elevation angle azimuth scan from north to west to south to measure the vector convection velocity over a wide two-dimensional area. Thermospheric and ionospheric parameters observed across the span of latitudes accessible from Millstone Hill have been presented for the duration of multiday storm time experiments by Buonsanto et al. [1990, 1992], and details of the convection electric field observations by Yeh et al. [1991]. These experiments provide full spatial coverage in both latitude and altitude with 30- to 45-min time resolution and are well suited to study the large-scale features, variability, and activity dependence which characterize ionospheric storms.

In the following sections we present observations which locate and identify the ionospheric feature which constitutes the storm time ionospheric dusk effect in both latitude and local time and show its clear relationship to sunward EYB convection. Simultaneous two-dimensional mapping of the convection pattern and ionospheric density shows the continuity of this convecting plasma feature between its mid- or low-latitude source region and the cleft and polar cap at high latitude. It is shown that the latitude of occurrence of this feature has a clear and repeatable dependence on the storm time disturbance level, as measured by Kp, that can be explained by the equatorward expansion of the region of sunward convection electric field during storms. The offset of the geomagnetic and geographic poles combines with the shape of the convection pattern in magnetic coordinates and the Earth's rotation to create the storm time dusk effect at mid-latitudes and a low-latitude source for the enhanced plasma found in the polar cap and at nightside auroral latitudes during disturbed magnetic conditions.

OBSERVATIONS

Millstone Hill storm mode radar experiments were performed in February 1986 and March 1990 in response to real-time alerts of intense solar activity. The 24-hour Kp index, dKp, was 60 for the interval centered on 0 UT on February 8-9, 1986, and was 43 for the similar period on March 20-21, 1990. Quiet time data were obtained on March 17-18, 1990. A description of these storm periods, and the radar coverage for each, is provided by Yeh et al. [1991] and Buonsanto et al. [1992]. Severe storm conditions at local noon (17 UT) were encountered on January 31, 1982, during regularly scheduled radar operations which looked to the north of the site only. dKp for that UT day was 34, and details of the Millstone Hill observations of the cleft and cusp and of the magnetic conditions encountered are provided by Foster et al. [1989]. Table 1 lists the observing periods used in this study.

Scan Maps

Antenna elevation scans in the magnetic meridian were made at Ê30-min intervals throughout each of the storm experiments. These provide altitude/latitude maps of ionospheric density and temperature between 100 km and 1000 km altitude and over a span of latitude which varies with altitude. At 500 km altitude, geodetic latitudes between 25° and 60° are covered. In Figure 1 we present event maps which detail the temporal evolution of the F region electron density at 500 km altitude along the meridian through Millstone Hill for four 24-hour periods centered on 0 UT (19 MLT along the Millstone meridian). This presentation reveals the combined signatures of the regular diurnal variation of the topside F region density and storm-dependent effects which vary with latitude and the intensity of the disturbance. The data for March 17-18, 1990, depict typical quiet day conditions (average Kp = 1Ú) with sunrise near 14 UT, sunset near 02 UT, and fairly uniform daytime densities (5Y1011 mÚ3) across the latitude range displayed. Strong magnetic storms occurred on each of the other days, resulting in a considerable perturbation of the topside F region density. The greatly depleted density within the ionospheric trough dominates each pattern at latitudes above 45° after sunset on each disturbed day. For disturbed conditions, the trough is seen at lower latitudes and at earlier local times, well into the sunlit F region. Whalen [1989] has examined the daytime trough and showed that it results from the advection of low-density nightside plasma sunward into the postnoon ionosphere. Of particular interest in our study is the appearance of a latitudinally narrow region of enhanced density equatorward of the trough in the postnoon sector which descends in latitude as the evening progresses. The occurrence of this feature at any latitude is seen as the Ûdusk effectÙ enhancement in total electron content discussed in previous studies, and the latitude-time position of this feature, as seen in Figure 1, will be shown to map the equatorward extent of storm-enhanced sunward convection which transports plasma from lower latitudes toward the region of the cusp at local noon.

In each of the disturbed-day density maps, the SED (storm-enhanced density) is first observed near the highest latitude sampled (60° geodetic, 72° magnetic) near local noon (17 UT) where it constitutes a significant enhancement over the background or quiet day density level. On the most disturbed day (February 8, 1986; average Kp = 8+), the SED signature shown in Figure 1b is distinctly ÛC-shapedÙ with the SED first seen near 55° at 16 UT and later (around 17 UT) simultaneously at both higher and lower-latitudes (see also Figure 2b and discussion, below). On this day, the lower-latitude region of the SED is produced by the sunward flow of plasma from its low-latitude source while the second region of SED at higher latitudes delineates the flow of this plasma away from the noontime cusp in the polar cap. (Radar azimuth scans to the west of the Millstone meridian (not shown) confirm this interpretation of the plasma flow pattern.) On each day there are sudden equatorward excursions of the region of SED which accompany intensifications in storm time activity (e.g., 2030 UT on February 8, 1986 and 2230 UT on March 20, 1990). Mid-latitude densities (below 35°) are elevated on each of the storm days with respect to the quiet day. This region of increased density constitutes the ionospheric storm positive phase enhancement, initially described by Matsushita [1959], and is a feature separate from the advecting SED discussed in this paper.

Elevation Scans

Storm-enhanced density often appears as a distinct region (blob) of plasma near the poleward edge of the mid-latitude F region. Latitude-altitude elevation scan maps spanning the transition between the mid-latitude F region and the SED are presented in Figure 2. At the left of the figure, scans near the same universal time (local noon) and for the same activity conditions (Kp =6) during magnetic storms in 1990 and 1982 (Figures 2a and 2c) observed the SED situated at 54° and separated from the mid-latitude F region by a trough of 5° width.

At the right of the figure, two scans during the extreme disturbance on February 8, 1986, are presented. In Figure 2b, at local noon, two distinct cuts through the SED, at both polar cap (>53°) and auroral latitudes (44°-48°), are seen poleward of the mid-latitude region of positive phase enhancement (<35°). Radar azimuth scans reveal that these two portions of the SED are interconnected to the west of the meridian of the elevation scan and that the more equatorward region of SED is convecting northwestward (sunward) while the high-latitude region is convecting antisunward at polar cap latitudes. Three successive elevation scans observed these dual regions of SED, and their gradual separation in latitude (cf. Figure 1b), as the observing meridian rotated across the noontime region of plasma entry into the polar cap. Some 4 hours later at 2030 UT (Figure 2d) the low-latitude SED appeared as an isolated blob of ionization near 40° geodetic latitude, when intersected by the N-S elevation scan, which then moved rapidly equatorward out of the radar field of view, as discussed below, during a further enhancement of the storm activity.

SED CHARACTERISTICS

In this section we show that the postnoon SED is a latitudinally distinct region of (rapidly) sunward convecting F region plasma whose latitude of occurrence decreases with both increasing Kp index and local time. Electron temperature is low within the region of SED, indicating that local production by particle precipitation is not responsible for the enhanced topside density observed. Vertical profiles through the SED reveal both an elevated F region peak altitude and much enhanced topside density with respect to the ambient plasma seen in the elevation scan data of Figure 2.

Kp Dependence

Synoptic storm time observations from Millstone Hill have been made throughout the last solar cycle spanning the 1980s and 1990s. As seen in Figure 1, SED is regularly seen throughout the afternoon sector at Millstone's longitude. In the discussion section, below, we will show that the Millstone Hill longitude in the northern hemisphere is a preferred location for the occurrence of strong SED and for the observation of the dusk effect. In order to study occurrence characteristics we have identified the SED center latitude at each UT for 12 experiments which provided appropriate latitude and temporal resolution. In all cases the latitude of the SED decreased with increasing local time and with increasing level of storm activity as measured by the Kp index. Postnoon SED can be identified at all disturbance levels down to Kp = 2. From the Millstone Hill observations, we have determined the average latitude of occurrence for SED at each universal time as a function of Kp, and this relationship is plotted in Figure 3. As seen in Figure 1, SED cannot be clearly identified in the Millstone Hill data after about 24 UT. For disturbed conditions (Kp · 4) the Kp dependence curves are nearly parallel through the afternoon sector with the latitude of SED occurrence at any local time decreasing by 6° for an increase of 2 in the Kp index. The curve for Kp = 8 is largely made up of data from the February 8, 1986, storm and reflects the severe intensification which took place after 20 UT.

The latitude at which the SED is observed at each level of Kp decreases with increasing local time at a rate of 2.5° hÚ1. The equatorward progression of the SED through the afternoon sector is very similar to that of the low-latitude extent of the convection electric field observed along the Millstone Hill meridian. Foster et al. [1986] determined average electric field characteristics in this sector, and the equatorward extent of electric field magnitude > 10 mV mÚ1 for Kp = 6 conditions taken from that work is indicated in Figure 3. Both the location and the equatorward progression in latitude with time of the SED closely match those of the equatorward limit of the sunward convection, suggesting the interrelationship of those phenomena.

Relationship to Sunward Convection

Near dusk, the SED occurs at latitudes immediately equatorward of the main ionospheric trough in a region where strong sunward convection (poleward directed electric field) overlaps the higher-density solar-produced mid-latitude plasma. Nearer noon, it is seen as a detached region following the convection trajectory into the polar cap.

Radar azimuth scans to the west of the Millstone Hill meridian provide a direct observation of the EYB plasma convection associated with the SED. Evans [1973] noted a Ê200 m sÚ1 westward velocity associated with the dusk effect enhanced ionization. Our experiments measure both F region density and velocity over a span of latitudes and allow us to associate the SED with strong sunward convection in all cases. Figure 4 presents convection equipotential contours and plasma density observed in the postnoon sector during the intense disturbance on January 31, 1982. An elevation scan through the ionosphere poleward of Millstone Hill at this time was presented in Figure 2c, which showed the extended altitude profile of the distinct SED region at 54° latitude. Dark shading in Figure 4 indicates the region of high-density F region plasma which is embedded in a region of 500 m sÚ1 sunward convection. Further poleward, within the depleted region of the trough, the convection speed reached 1300 m sÚ1. It is clear that the F region plasma seen at 56°N, 275°E has been advected to that longitude from the postnoon ionosphere at 310°E or beyond.

During the February 8, 1986, experiment (Figure 1b), the westward directed antenna position sampled the ionosphere near 42°, which was equatorward of the SED until Ê2100 UT when the SED descended to this latitude. The lower portion of Figure 5 records the topside density values observed during radar azimuth scans across this sector as a function of universal time and clearly shows the onset of the SED as a 5 Y increase after 2030 UT. At the top of the figure, a scatterplot of log density versus line-of-sight velocity for all the data samples shown in the lower figure indicates that the enhanced density which heralded the arrival of the SED (·105 cmÚ3) was associated with sunward velocities in excess of 800 m sÚ1. Such a relationship holds for all the Millstone Hill observations of SED events. Typical velocities associated with the enhanced plasma transport in the SED are 500 m sÚ1 to 1000 m sÚ1 and result in a sunward flux of Ê1014 mÚ2 sÚ1.

Profiles

Vertical profiles of plasma density have been constructed for the latitudes of the principal features seen in the noontime elevation scan at 1645 UT on February 8, 1986 (presented in Figure 2b) and these are presented in Figure 6. The positive phase enhancement at mid-latitude (Figure 6a at 30°) resulted in a peak density near 1012 mÚ3 at 400 km altitude. A daytime trough, where the peak density fell to Ê1011 mÚ3, separated this region from the sunward convecting SED at 46° (Figure 6b) where densities near 1012 mÚ3 were seen at altitudes between 400 km and 700 km. Low topside density characterized the region near 50° (Figure 6c) while the more poleward region of SED at 57° (Figure 6d) again exhibited an enhanced topside. Similar density profiles were seen by Foster and Doupnik [1984] and Foster [1989], who reported that a greatly enhanced topside density characterized plasma entering the polar cap through the noontime cleft.

Observations with the zenith-directed antenna at Millstone Hill detailed the vertical profile associated with the occurrence of SED at mid-latitudes (42° geodetic, 55° magnetic) as the feature moved equatorward over the site during the February 1986 event. Figure 7 presents contours of electron density and temperature versus altitude and time which show that the SED was seen continuously over the site during a 2-hour period from 18 UT to 20 UT. The altitude of the F region peak density decreased from >450 km as the SED first appeared over the site to Ê300 km at its poleward edge. Electron temperature was low within the SED plasma, indicating that local production of enhanced ionization by particle precipitation did not contribute significantly to the elevated density within the SED.

MID-LATITUDE PHENOMENA

The characteristics of the mid-latitude ionosphere are generally determined by the diurnal, seasonal, and solar cycle variations in solar production and heating and by the coupling to the neutral atmosphere. During large magnetic disturbances, this regular, predictable variability can be dramatically overturned as the forces which drive the high-latitude regime expand equatorward.

The large-scale enhancement of ionospheric density at mid-latitudes on the first day of a magnetic storm is termed the ionospheric storm positive phase. A further enhancement of the mid-latitude density localized near 18 LT has been termed the dusk effect and has usually been described as a part of the positive phase. Observations from Millstone Hill separate these two storm time effects and reveal the SED dusk effect as a distinct feature related to plasma transport along the poleward extent of the positive phase enhancement. The two-dimensional picture afforded by the radar elevation scan data, as shown in Figure 1, identifies the SED as spanning a wide extent of local time and latitude. Figure 8 presents topside F region density near 500 km altitude observed along the Millstone Hill meridian during the February 8, 1986, storm. Originally observed near 60° latitude at 16 UT, the SED rapidly descended to near 45° and remained there until its sudden equatorward excursion after 20 UT. During the initial hours of the disturbance, a distinct region of increased density was observed at latitudes equatorward of 38°, constituting the usual positive phase enhancement. As individual ground stations rotated under the enhanced density regions identified in the radar scan data, they recorded the separate signatures of the positive phase and dusk effect. The time and latitude of observation of dusk effect enhancements in TEC from the sites discussed in conjunction with Figure 9 are indicated by X in Figure 8. Observations taken looking parallel to the magnetic field direction (at 40.5° geodetic latitude) at 1700 UT reveal a downward plasma velocity component of 200 m sÚ1 throughout the region of the positive phase enhancement while azimuth scans indicate a poleward and eastward EYB convection at 500 m sÚ1. At 2030 UT, the radar observed 140 m sÚ1 upward plasma velocity parallel to the magnetic field in the region of SED and magnetic westward EYB convection at 1800 m sÚ1.

In all cases observed from Millstone Hill, the SED is associated with enhanced sunward (westward and poleward) convection while the positive phase enhancement is not. An analysis of the event scan maps shown in Figure 1 serves to further differentiate these phenomena. During the event of March 20-21 (Figure 1d) densities equatorward of the radar are enhanced by a factor of 2 over the quiet day background beginning around 1500 UT and are accompanied by upward field-aligned velocities of 40 m sÚ1. Near 00 UT a further enhancement to a density in excess of 1012 mÚ3 occurred in association with a poleward velocity component of 300 m sÚ1 at latitudes around 32° and no change in the velocity parallel to the magnetic field. At the same time, the azimuth scan observed the onset of a SED with density of 1012 mÚ3 at 38° latitude to the west of the site. If we assume that convection equipotentials connect the two regions of SED, as seen in Figure 4, and combine the Doppler velocity components from the two observing directions, we determine a vector velocity of 1200 m sÚ1 directed at a bearing 60° west of magnetic north, identical to the bearing between the two sites at which the SED was observed. The transit time for plasma carried poleward between these sites would be 20 min, and the flux 10 15 mÚ2 sÚ1.

Regular measurements of total electron content (TEC) are made from a series of ground stations which record the Faraday rotation of transionospheric signals from satellite beacons in geosynchronous orbit. Figure 9 indicates the ground sites and ionospheric penetration points at 350 km altitude for four such stations which monitor mid-latitude TEC along the North American east coast. TEC recordings from these sites are shown in Figure 10 for the February 8, 1986, event (the monthly-averaged variation at each site indicates the diurnal background). During that event, the mid-latitude stations at Goose Bay (L=3.9) and Hamilton (L=2.8) observed increased ionization at significantly different times at different latitudes, and the lower-latitude stations at Kennedy (L=1.7) and Ramey (L=1.3) observed both a large early enhancement and a smaller increase at later times. Since this chain of TEC observations is closely aligned with the Millstone Hill scan meridian, an X has been placed on Figure 8 at the latitude and universal time corresponding to the TEC observations of the dusk effect enhancements seen in Figure 10. It is clear that the dusk effect in TEC is associated with the convecting plasma of the SED and that the more equatorward region of increased density seen early in the event in the radar data corresponds to the positive phase enhancement of TEC seen at Kennedy and Ramey. Although the radar observations of the spatial extent of the SED during this event extend only to the latitude of Kennedy, the Ramey TEC observations indicate that this dramatic high-latitude effect penetrated equatorward to at least L=1.3 (28° invariant latitude) during this event, giving further evidence for a low-latitude source for plasma advected poleward to very high latitudes during such intense disturbances.

HIGH-LATITUDE PHENOMENA

The density pattern of Figure 8 suggests that there is a continuous stream of plasma advected from low latitudes in the evening sector to the cusp and polar cap at noon. Those data, and the data of Figure 1, however, are not snapshots of the ionosphere, but show latitudinal variations seen at one longitude during the course of the events. Single azimuth scans from Millstone Hill, spanning nearly 6 hours of local time and 20° of latitude during a 20-min observing period, reveal that the ÛinstantaneousÙ pattern of ionospheric densities can be strikingly similar to the time-averaged pattern seen in Figure 8. Figure 11 presents an azimuth scan on January 31, 1982, which shows enhanced-density topside plasma streaming along the observed convection trajectories through the cleft and well into the polar cap. This event was discussed in detail by Foster [1989] and Foster et al. [1989], who used precipitating particle data observed during a simultaneous satellite overflight of the radar field of view to identify the cleft/low-latitude boundary layer signature at 67°Å, immediately to the west of the convecting SED feature. The longitude at which the SED plasma entered polar cap latitudes during that event was observed to change rapidly, following changes in the configuration of the convection pattern. This observation gives dramatic proof that enhanced storm time convection carries low-latitude solar-produced plasma into the noon sector polar cap. The positive phase density enhancement is a true mid-latitude effect with widespread latitudinal extent, but the dusk effect is, in fact, a high-latitude phenomenon driven by the convection electric field and is symptomatic of rapid transport of ionospheric plasma from low to high latitudes during the early phase of a magnetospheric storm or substorm.

The SED plasma convected through the cleft/cusp near noon constitutes a major source for F region plasma features at polar cap latitudes. Foster and Doupnik [1984] reported Chatanika radar observations of enhanced topside density streaming along the convection direction through the cusp at noon. High-temporal-resolution density/range observations along a radar beam direction parallel to the convection flow, taken from that earlier work, are presented in Figure 12. Discrete patches of enhanced density are seen, and their change in range with time corresponds to the 500 m sÚ1 poleward component of the convection velocity deduced from the radar Doppler observations at that time. Multiple ionization patches separated by Ê200 km in range can be seen at a given time (e.g., observation at 2028 UT), and Lockwood [1992] has interpreted this observation as giving evidence that the poleward convection through the cusp is pulsed with a 7- to 8-min period, consistent with the expectations of transient magnetopause reconnection. Such a mechanism could help account for the discrete nature of the F region patches observed at polar cap latitudes [e.g., Weber et al., 1984].

DISCUSSION

Millstone Hill radar observations have revealed the region of storm-enhanced density as a spatially continuous feature spanning a large portion of the postnoon/midnight sector at North American longitudes. The SED is closely associated with sunward convection, and the latitude of the SED decreases with increasing local time and with increasing disturbance level, (cf. Figure 3) in a way similar to that of the equatorward edge of the convection pattern [e.g., Foster et al., 1986]. The region of SED is produced as the storm-enhanced convection electric field picks up solar-produced ionospheric plasma at middle and low latitudes and carries it toward the noontime cusp. The occurrence and characteristics of SED can be understood in terms of this process. The large-scale convection electric field mainly results from magnetospheric processes and is best organized in the magnetic coordinates appropriate to such phenomena. Solar production of ionospheric density depends on solar zenith angle, which is a function of geographic latitude and local time. The 11° offset of the magnetic pole from the geographic pole creates a diurnal wobble of the convection pattern in geographic coordinates such that the convection electric field dips to its furthest equatorward extent at each local time near the Millstone Hill meridian. As the Earth rotates under the convection pattern, the region of sunward convection extends to an increasingly equatorward geographic latitude as time progresses from 12 UT to 24 UT. During the subsequent 12-hour period, the reverse is true, and the extent of the convection pattern retreats poleward in geographic coordinates. As the convection pattern moves to lower geographic latitudes in the postnoon sector, it continuously encounters fresh, corotating, solar-produced F region plasma which is picked up in a snowplow effect and carried noonward and poleward. Although the effects of SED are most clearly seen during magnetic storms, when the convection electric field expands to relatively low latitudes, our observations identify this feature at Kp levels as low as 2 (e.g., Figure 3), which implies that the convection of solar-produced plasma into the polar cap occurs during nonstorm conditions, as well.

The regularity and repeatability of the latitude at which the SED is observed from Millstone Hill as a function of universal time (cf. Figure 3) indicates that SED is associated with a general characteristic of the ionosphere-magnetosphere system, in this case the convection pattern, and is not the result of a propagating ionospheric disturbance (TAID) initiated by an impulsive, and randomly timed, high-latitude event as proposed by Prolss et al. [1991]. We have shown that for constant activity conditions, the latitude at which SED occurs varies in a manner very similar to that of the equatorward extent of the convection electric field [e.g., Foster et al., 1986]. Accordingly, we expect the latitude/local time characteristics of SED occurrence to differ at different longitudes and to be related to the extent of the penetration of the convection pattern into the sunlit ionosphere at any universal time. To illustrate these points, we have taken the magnetic latitude/local time shape of the Kp=6 SED observations of Figure 3 (local time = UT Ú 5 hours on the Millstone meridian and magnetic latitude = geographic latitude + 12°) and in Figure 13 show how this feature, held fixed in magnetic coordinates, would appear in geographic coordinates at different universal times. After 12 UT, the SED convection boundary begins to descend into the lower-latitude ionosphere with the point of most equatorward penetration beginning near 64° geographic latitude near local noon at 12 UT and reaching 43° geographic latitude at 21 LT by 24 UT. After 24 UT the SED convection boundary retreats rapidly poleward and encounters at its most equatorward extent flux tubes which have been previously exposed to strong convection and are thus already depleted of ionization. This is consistent with the observation that the SED effects observed on the Millstone Hill longitude end abruptly near 00 UT.

The characteristics of storm-enhanced density are very similar to those of the daytime F layer trough reported by Whalen [1989]. That study found that low-density ionospheric plasma from later local times was carried toward noon along the evening sector convection cell, resulting in a low-density trough which extended into the noon sector from beyond the dusk terminator. Whalen observed the daytime trough to be a continuous two-dimensional feature in latitude and local time which shifted equatorward with increasing Kp index. Data from a network of ionosondes spanning high latitudes in the northern hemisphere were used, and a pronounced longitude dependence was noted for the daytime trough's characteristics. Whalen concluded that this was due to the fact that the convection pattern resides in solar geomagnetic coordinates so that the appearance of the trough in the daytime F layer, as observed in solar terrestrial coordinates, is subject to the systematic longitudinal variation of the two coordinate systems with respect to each other.

The ionospheric trough is seen immediately poleward of the storm-enhanced density feature at each local time in the observations presented in this paper. Both the SED and the daytime trough reported by Whalen [1989] result from enhanced EYB sunward plasma transport in the evening sector convection cell during disturbed conditions combined with offset of the geomagnetic and geographic poles, which bring strong storm time electric fields to low geographic latitudes in the North American sector.

SUMMARY

Storm time SED is a pronounced feature of the Millstone Hill ionospheric observations, and its characteristics as seen in the radar observations explain the ionospheric storm dusk effect and provide a source for the enhanced density features which have been found to populate the polar cap. It is separate from the usual positive phase effect and is associated with the transport of solar-produced plasma from lower latitudes toward noon. The SED is observed equatorward of the evening sector trough, in the region of sunward plasma convection. The integrated column density (TEC) is enhanced by a factor of 2 to 4 and is spatially extended along the sunward convection trajectory. SED is associated with velocities of 800 m sÚ1, giving a Ê2-hour transit time from its source at low latitudes to the polar cap at noon and a plasma flux of 1014 mÚ2 sÚ1.

The characteristics and mechanism of SED explain the storm time dusk effect reported from mid-latitude TEC observations. Two-dimensional radar mapping shows the continuity of the region of SED from mid-latitudes through the cusp. The plasma carried sunward is fragmented on entering the polar cap and constitutes a source for F region patches within the polar cap. Its latitude of occurrence expands with increasing Kp, and it can be identified at L=1.3 during large storms. As a result of the offset between the geomagnetic and geographic poles, the afternoon sector region of strong sunward convection is shifted to increasingly lower geographic latitude throughout the interval between 12 UT and 24 UT. A snowplow effect occurs in which the convection cell continually encounters fresh corotating ionospheric plasma along its equatorward edge, producing a latitudinally narrow region of storm-enhanced plasma density (SED) and increased total electron content (TEC) which is advected toward higher latitudes in the noon sector. The east coast of the United States is a preferred longitude for the occurrence of SED due to the offset of the poles.

Acknowledgments. Millstone Hill radar operations and analysis are supported by National Science Foundation Co-operative Agreement ATM-91-02445 with the Massachusetts Institute of Technology. The assistance of the Atmospheric Sciences staff at the MIT Haystack Observatory is gratefully acknowledged. Special thanks are due to M. J. Buonsanto for many fruitful discussions. Total electron content (TEC) data from the North American east coast stations were provided by J. A. Klobuchar.

The Editor thanks J. P. Heppner and J. A. Whalen for their assistance in evaluating this paper.

REFERENCES

Anderson, D. N., Modeling the midlatitude F-region ionospheric storm using east-west drift and a meridional wind, Planet. Space Sci., 24, 69-77, 1976.

Bowman, G. G., Traveling disturbances associated with ionospheric storms, J. Atmos. Terr. Phys., 27, 1247-1261, 1965.

Buchau, J., E. J. Weber, D. N. Anderson, H. C. Carlson, Jr., J. G. Moore, B. W. Reinisch, and R. C. Livingston, Ionospheric structures in the polar cap: Their origin and relation to 250-MHz scintillation, Radio Sci., 20, 325-328, 1985.

Buonsanto, M. J., M. Mendillo, and J. A. Klobuchar, The ionosphere at L=4: Average behavior and response to geomagnetic storms, Ann. Geophys., 35, 15-26, 1979.

Buonsanto, M. J., J. C. Foster, A. D. Gallasso, D. P. Sipler, and J. M. Holt, Neutral winds and thermosphere/ionosphere coupling and energetics during the geomagnetic disturbance of March 6-10, 1989, J. Geophys. Res., 95, 21033-21050, 1990.

Buonsanto, M. J., J. C. Foster, and D. P. Sipler, Observations from Millstone Hill during the geomagnetic disturbances of March and April, 1990, J. Geophys. Res., 97, 1225-1243, 1992.

de la Beaujardiere, O., V. B. Wickwar, G. Caudal, J. M. Holt, J. D. Craven, L. A. Frank, L. H. Brace, D. S. Evans, J. D. Winningham, and R. A. Heelis, Universal time dependence of nighttime F region densities at high latitudes, J. Geophys. Res., 90, 4319-4332, 1985.

Evans, J. V., The June 1965 magnetic storm: Millstone Hill observations, J. Atmos. Terr. Phys., 32, 1629-1640, 1970.

Evans, J. V., The causes of storm-time increases of the F-layer at mid-latitudes, J. Atmos. Terr. Phys., 35, 593-616, 1973.

Evans, J. V., J. M. Holt, W. L. Oliver, and R. H. Wand, Millstone Hill incoherent scatter observations of auroral convection over 60° u Å u 75°, 2, Initial results, J. Geophys. Res., 85, 41-54, 1980.

Foster, J. C., An empirical electric field model derived from Chatanika radar data, J. Geophys. Res., 88, 981-987, 1983.

Foster, J. C., Ionospheric signatures of magnetospheric convection, J. Geophys. Res., 89, 855-865, 1984.

Foster, J. C., Plasma transport through the dayside cleft: A source of ionization patches in the polar cap, in Electromagnetic Coupling in the Polar Clefts and Caps, edited by P. Sandholt and A. Egeland, pp. 343-354, Kluwer Academic, Boston, Mass., 1989.

Foster, J. C., and J. R. Doupnik, Plasma convection in the vicinity of the dayside cleft, J. Geophys. Res., 89, 9107-9113, 1984.

Foster, J. C., J. M. Holt, J. D. Kelly, and V. B. Wickwar, High resolution observations of electric fields and F-region plasma parameters in the cleft ionosphere, in The Polar Cusp, edited by J. A. Holtet and A. Egeland pp. 349-364, D. Reidel, Boston, Mass., 1985.

Foster, J. C., J. M. Holt, R. E. Musgrove, and D. S. Evans, Ionospheric convection associated with discrete levels of particle precipitation, Geophys. Res. Lett., 13, 656-659, 1986.

Foster, J. C., H.-C. Yeh, J. M. Holt, and D. S. Evans, Two-dimensional mapping of dayside convection, in Electromagnetic Coupling in the Polar Clefts and Caps, edited by P. Sandholt and A. Egeland, pp. 115-125, Kluwer Academic, Boston, Mass., 1989.

Friis-Christensen, E., Y. Kamide, A. D. Richmond, and S. Matsushita, Interplanetary magnetic field control of high-latitude electric fields and currents determined from Greenland magnetometer data, J. Geophys. Res., 90, 1325-1338, 1985.

Heelis, R. A., J. K. Lowell, and R. W. Spiro, A model of the high-latitude ionospheric convection pattern, J. Geophys. Res., 87, 6339-6345, 1982.

Heppner, J. P., Empirical models of high-latitude electric fields, J. Geophys. Res., 82, 1115-1125, 1977.

Heppner, J. P., and N. C. Maynard, Empirical high-latitude electric field models, J. Geophys. Res., 92, 4467-4489, 1987.

Holt, J. M., R. H. Wand, J. V. Evans, and W. L. Oliver, Empirical models for the plasma convection at high latitudes from Millstone Hill observations, J. Geophys. Res., 92, 203-212, 1987.

Jones, K. L., and H. Rishbeth, The origin of storm increases of midlatitude F-layer electron concentration, J. Atmos. Terr. Phys., 33, 391-401, 1971.

Kamide, Y., A. D. Richmond, and S. Matsushita, Estimation of ionospheric electric fields, ionospheric currents, and field-aligned currents from ground magnetic records, J. Geophys. Res., 86, 801-813, 1981.

Lockwood, M., Incoherent scatter radar measurements of the cusp, in Proceedings of ESA Cluster Workshop, European Space Agency, Nevilly, France, 1992.

Matsushita, S., A study of the morphology of ionospheric storms, J. Geophys. Res., 64, 305-321, 1959.

Mendillo, M., M. D. Papagiannis, and J. A. Klobuchar, Ionospheric storms at midlatitudes, Radio Sci., 5, 895-898, 1970.

Papagiannis, M. D., M. Mendillo, and J. A. Klobuchar, Simultaneous storm-time increases of the ionospheric total electron content and the geomagnetic field in the dusk sector, Planet. Space Sci., 19, 503-511, 1971.

Prolss, G. W., L. H. Brace, H. G. Mayr, G. R. Carignan, T. L. Killeen, and J. A. Klobuchar, Ionospheric storm effects at subauroral latitudes: A case study, J. Geophys. Res., 96, 1275-1288, 1991.

Rishbeth, H., and W. B. Hanson, A comment on plasma "pile up" in the F region, J. Atmos. Terr. Phys., 36, 703-706, 1974.

Weber, E. J., J. Buchau, J. G. Moore, J. R. Sharber, R. C. Livingston, J. D. Winningham, and B. W. Reinisch, F layer ionization patches in the polar cap, J. Geophys. Res., 89, 1683-1694, 1984.

Weber, E. J., J. A. Klobuchar, J. Buchau, H. C. Carlson, Jr., R. C. Livingston, O. de la Beaujardiere, M. McCready, J. G. Moore, and G. J. Bishop, Polar cap F layer patches: Structure and dynamics, J. Geophys. Res., 91, 12121-12129, 1986.

Whalen, J. A., The daytime F layer trough and its relation to ionospheric-magnetospheric convection, J. Geophys. Res., 94, 17169-17184, 1989.

Yeh, H.-C., J. C. Foster, F. J. Rich, and W. Swider, Storm-time electric field penetration observed at mid-latitude, J. Geophys. Res., 96, 5707-5721, 1991.

Fig. 1. Event maps prepared from radar elevation scan data detail the temporal evolution of the F region electron density at 500 km altitude along the meridian through Millstone Hill for four 24-hour periods centered on 0 UT (19 MLT along the Millstone meridian). A region of storm-enhanced density (SED), which descends in latitude as the evening progresses, is seen equatorward of the trough on the disturbed days. The data for March 17-18, 1990, depict typical quiet day conditions.

Fig. 2. Storm-enhanced density often appears as a distinct region (blob) of plasma near the poleward edge of the mid-latitude F region. Latitude-altitude elevation scan maps spanning the transition between the mid-latitude F region and the SED are shown. (Ground clutter distorts the isodensity contours below 150 km altitude near 42°.)

Fig. 3. Average latitude of occurrence of SED on the Millstone Hill meridian is shown for several levels of the Kp index. Local time = UT Ú 5 hours. The equatorward extent of the average convection electric field for Kp · 6 is indicated by squares.

Fig. 4. Convection equipotential contours (2-kV spacing) and plasma density observed in the postnoon sector during the intense disturbance on January 31, 1982. Dark shading indicates the region of high-density F region plasma which is seen to be embedded in a region of 500 m sÚ1 sunward convection. (See Figure 2c for an elevation scan through this feature.)

Fig. 5. (Top) Scatterplot of log density versus line-of-sight velocity for all the data samples shown below indicates that the enhanced density which heralded the arrival of the SED (·105 cmÚ3) was associated with velocities in excess of 800 m sÚ1. (Bottom) The increase in topside density near 2030 UT in the zenith over Millstone Hill signaled the arrival of the SED over the site.

Fig. 6. Vertical profiles of plasma density for the latitudes of the principal features seen in the noontime elevation scan at 1645 UT on February 8, 1986 (presented in Figure 2c).

Fig. 7. Contours of electron density and temperature versus altitude and time. The SED was seen continuously over the site during a 2-hour period from 18 UT to 20 UT. Electron temperature was low within the SED plasma.

Fig. 8. Topside F region density near 500 km altitude observed along the Millstone Hill meridian during the February 8, 1986, storm. Originally observed near 60° latitude at 16 UT, the SED rapidly descended to near 45° and remained there until its sudden equatorward excursion after 20 UT.

Fig. 9. Regular measurements of total electron content (TEC) are made from a series of ground stations which record the transionospheric phase variations of signals from satellite beacons in geosynchronous orbit. Ground sites and ionospheric penetration points at 350 km altitude are shown for four stations operated by the Air Force Geophysics Laboratory which monitor mid-latitude TEC along the North American east coast.

Fig. 10. TEC recordings from the sites shown in Figure 9 for the February 8, 1986, event (the monthly-averaged variation at each site indicates the diurnal background). The occurrence of SED at each site is indicated by X, and P denotes the large positive phase enhancement at the lower-latitude stations.

Fig. 11. Azimuth scan on January 31, 1982, which shows enhanced-density topside plasma streaming along the observed convection trajectories through the cleft and well into the polar cap.

Fig. 12. High-temporal-resolution density/range observations taken along a radar beam direction parallel to the convection flow into the polar cap across the noontime cleft, taken from Foster and Doupnik [1984]. Discrete patches of enhanced density are seen, and their change in range with time corresponds to the 500 m sÚ1 poleward component of the convection velocity at that time.

Fig. 13. The magnetic latitude/local time shape of the Kp=6 SED observations of Figure 3 (local time = UT Ú 5 hours on the Millstone meridian and magnetic latitude = geographic latitude + 12°), held fixed in magnetic coordinates, is shown as it would appear in geographic coordinates at different universal times.