The dynamic cusp aurora on 30 November 1997: response to southward turning of the IMF

We document the detailed dynamics of the dayside aurora in the ≈1200–1600 MLT sector in response to a sharp southward turning of the interplanetary magnetic field (IMF) under negative IMF B_y conditions. Features not documented in previous work are elucidated by using two meridan scanning photometers (separated by 2 h) and an all-sky auroral imager in Ny Ålesund, Svalbard (75.5^MLAT) in combination with magnetograms from stations on Svalbard, covering the latitude range 71^–75^MLAT. The initial auroral response may be divided into three phases consisting of: (1) intensification of both the red (630.0 nm) and green (557.7 nm) line emissions in the cusp aurora near 1200 MLT and ≈100 km equatorward shift of its equatorward boundary, at ≈75^MLAT, (2) eastward and poleward expansions of the cusp aurora, reaching the 1430 MLT meridian after 5–6 min, and (3) east-west expansion of the higher-latitude aurora (at ≈77^–78^MLAT) in the postnoon sector. The associated magnetic disturbance is characterized by an initial positive deflection of the X-component at stations located 100–400 km south of the aurora, corresponding to enhanced Sunward return flow associated with the merging convection cell in the post-noon sector. The sequence of partly overlapping poleward moving auroral forms (PMAFs) during the first 15 min, accompanied by corresponding pulsations in the convection current, was followed by a strong westward contraction of the cusp aurora when the ground magnetograms indicated a temporary return to the pre-onset level. These observations are discussed in relation to the Cowley-Lockwood model of ionospheric response to pulsed magnetopause reconnection.     

The relationship between VHF radar auroral backscatter amplitude and Doppler velocity: a statistical study

A statistical investigation of the relationship between VHF radar auroral backscatter intensity and Doppler velocity has been undertaken with data collected from 8 years operation of the Wick site of the Sweden And Britain Radar-auroral Experiment (SABRE). The results indicate three different regimes within the statistical data set; firstly, for Doppler velocities <200 m s⁻¹, the backscatter intensity (measured in decibels) remains relatively constant. Secondly, a linear relationship is observed between the backscatter intensity (in decibels) and Doppler velocity for velocities between 200 m s⁻¹ and 700 m s⁻¹. At velocities greater than 700 m s⁻¹ the backscatter intensity saturates at a maximum value as the Doppler velocity increases. There are three possible geophysical mechanisms for the saturation in the backscatter intensity at high phase speeds: a saturation in the irregularity turbulence level, a maximisation of the scattering volume, and a modification of the local ambient electron density. There is also a difference in the dependence of the backscatter intensity on Doppler velocity for the flow towards and away from the radar. The results for flow towards the radar exhibit a consistent relationship between backscatter intensity and measured velocities throughout the solar cycle. For flow away from the radar, however, the relationship between backscatter intensity and Doppler velocity varies during the solar cycle. The geometry of the SABRE system ensures that flow towards the radar is predominantly associated with the eastward electrojet, and flow away is associated with the westward electrojet. The difference in the backscatter intensity variation as a function of Doppler velocity is attributed to asymmetries between the eastward and westward electrojets and the geophysical parameters controlling the backscatter amplitude.     

Polar cap convection patterns inferred from EISCAT observations

From data of the European incoherent scatter radar EISCAT, and mainly from its tristatic capabilities, statistical models of steady convection in the auroral ionosphere were achieved for various levels of magnetic activity. We propose here to consistently extend these models to the polar cap, by avoiding the use of a predefined convection pattern. Basically, we solve the second-order differential equation governing the polar cap convection potential with the boundary conditions provided by these models. The results display the classical twin-vortex convection pattern, with the cell centres around 17 MLT for the evening cell and largely shifted towards midnight (3–3.5 MLT) for the morning cell, both slightly moving equatorward with activity. For moderate magnetic activities, the convection now appears approximately oriented along the meridian from 10:00 MLT to 22:00 MLT, while in more active situations, it enters the polar cap at prenoon times following the antisunward direction, and then turns to exit around 21:00 MLT. Finally, from these polar cap patterns combined with the auroral statistical models, we build analytical models of the auroral and polar convection expected in steady magnetic conditions.     

Observations of the response time of high-latitude ionospheric convection to variations in the interplanetary magnetic field using EISCAT and IMP-8 data

We have combined ∼300 h of tristatic measurements of the field-perpendicular F region ionospheric flow measured overhead at Tromsø by the EISCAT UHF radar, with simultaneous IMP-8 measurements of the solar wind and interplanetary magnetic field (IMF) upstream of the Earth’s magnetosphere, in order to examine the response time of the ionospheric flow to changes in the north-south component of the IMF (B_z). In calculating the flow response delay, the time taken by field changes observed by the spacecraft to first effect the ionosphere has been carefully estimated and subtracted from the response time. Two analysis methods have been employed. In the first, the flow data were divided into 2 h-intervals of magnetic local time (MLT) and cross-correlated with the “half-wave rectifier” function V²B_s, where V is the solar wind speed, and B_s is equal to IMF B_z if the latter is negative, and is zero otherwise. Response delays, determined from the time lag of the peak value of the cross-correlation coefficient, were computed versus MLT for both the east-west and north-south components of flow. The combined data set suggests minimum delays at ∼1400 MLT, with increased response times on the nightside. For the 12-h sector centred on 1400 MLT, the weighted average response delay was found to be 1.3 ± 0.8 min, while for the 12-h sector centred on 0200 MLT the weighted average delay was found to increase to 8.8 ± 1.7 min. In the second method we first inspected the IMF data for sharp and enduring (at least ∼5 min) changes in polarity of the north-south component, and then examined concurrent EISCAT flow data to determine the onset time of the corresponding enhancement or decay of the flow. For the case in which the flow response was timed from whichever of the flow components responded first, minimum response delays were again found at ∼1400 MLT, with average delays of 4.8 ± 0.5 min for the 12-h sector centred on 1400 MLT, increasing to 9.2 ± 0.8 min on the nightside. The response delay is thus found to be reasonably small at all local times, but typically ∼6 min longer on the nightside compared with the dayside. In order to make an estimate of the ionospheric information propagation speed implied by these results, we have fitted a simple theoretical curve to the delay data which assumes that information concerning the excitation and decay of flow propagates with constant speed away from some point on the equatorward edge of the dayside open-closed field line boundary, taken to lie at 77° magnetic latitude. For the combined cross-correlation results the best-fit epicentre of information propagation was found to be at 1400 MLT, with an information propagation phase speed of 9.0 km s⁻¹. For the combined event analysis, the best-fit epicentre was also found to be located at 1400 MLT, with a phase speed of 6.8 km s⁻¹.     

The ionospheric signature of transient dayside reconnection and the associated pulsed convection return flow

Three SuperDARN coherent HF radars are employed to investigate the excitation of convection in the dayside high-latitude ionosphere in response to transient reconnection occurring in the cusp region. This study demonstrates the existence of transient antisunward-propagating backscatter features at the expected location of the ionospheric footprint of the cusp region, which have a repetition rate near 10 min. These are interpreted as the ionospheric signature of flux transfer events. Moreover, transient sunward-propagating regions of backscatter are observed in the convection return flow regions of both the pre- and post-noon sectors. These patches are observed to propagate towards the noon sector from at least as far around the auroral zone as 07 MLT in the pre-noon sector and 17 MLT in the post-noon sector, travelling with a veloCity of approximately 1.5 to 2 km s⁻¹. These return flow patches have a repetition rate similar to that of the transient features observed at local noon. While providing supporting evidence for the impulsive nature of convection flow, the observation of sunward-propagating features in the return flow region is not consistent with current conceptual models of the excitation of convection.     

EISCAT measurements of the solar wind

EISCAT observations of interplanetary scintillation have been used to measure the velocity of the solar wind at distances between 15 and 130R (solar radii) from the Sun. The results show that the solar wind consists of two distinct components, a fast stream with a velocity of 800 km s^–1 and a slow stream at 400 kms^–1. The fast stream appears to reach its final velocity much closer to the Sun than expected. The results presented here suggest that this is also true for the slow solar wind. Away from interaction regions the flow vector of the solar wind is purely radial to the Sun. Observations have been made of fast wind/slow wind interactions which show enhanced levels of scintillation in compression regions.     

Multi-instrument observations of the electric and magnetic field structure of omega bands

High time resolution data from the CUTLASS Finland radar during the interval 01:30–03:30 UT on 11 May, 1998, are employed to characterise the ionospheric electric field due to a series of omega bands extending 5° in latitude at a resolution of 45 km in the meridional direction and 50 km in the azimuthal direction. E-region observations from the STARE Norway VHF radar operating at a resolution of 15 km over a comparable region are also incorporated. These data are combined with ground magnetometer observations from several stations. This allows the study of the ionospheric equivalent current signatures and height integrated ionospheric conductances associated with omega bands as they propagate through the field-of-view of the CUTLASS and STARE radars. The high-time resolution and multi-point nature of the observations leads to a refinement of the previous models of omega band structure. The omega bands observed during this interval have scale sizes 500 km and an eastward propagation velocity 0.75 km s^–1. They occur in the morning sector (05 MLT), simultaneously with the onset/intensification of a substorm to the west during the recovery phase of a previous substorm in the Scandinavian sector. A possible mechanism for omega band formation and their relationship to the substorm phase is discussed.     

2-D Crustal Velocity Structure Along Hirapur-Mandla Profile in Central India: An Update

The 2-D crustal velocity model along the Hirapur-Mandla DSS profile across the Narmada-Son lineament in central India (Murty et al., 1998) has been updated based on the analysis of some short and discontinuous seismic wide-angle reflection phases. Three layers, with seismic velocities of 6.5–6.7, 6.35–6.40 and 6.8 km s^–1, and upper boundaries located approximately at 8, 17 and 22 km depth respectively, have been identified between the basement (velocity 5.9 km s^–1) and the uppermost mantle (velocity 7.8 km s^–1). The layer with 6.5–6.7 km s^–1 velocity is thin and is less than 2-km deep between the Narmada north (at Katangi) and south (at Jabalpur) faults. The upper crust shows a horst feature between these faults, which indicates that the Narmada zone acts as a ridge between two pockets of mafic intrusion in the upper crust. The Moho boundary, at 40–44 km depth and the intra-crustal layers exhibit an upwarp suggesting that the Narmada faults have deep origins, involving deep-seated tectonics. A smaller intrusive thickness between the Narmada faults, as compared to those beyond these faults, suggests that the intrusive activities on the two sides are independent. This further suggests that the two Narmada faults may have been active at different geological times. The seismic model is constrained by 2-D gravity modeling. The gravity highs on either side of the Narmada zone are due to the effect of the high velocity/high density mafic intrusion at upper crustal level.     

CUTLASS/IMAGE observations of high-latitude convection features during substorms

The CUTLASS Finland HF radar has been operational since February 1995. The radar frequently observes backscatter during the midnight sector from a latitude range 70–75° geographic, latitudes often associated with the polar cap. These intervals of backscatter occur during intervals of substorm activity, predominantly in periods of relatively quiet magnetospheric activity, with Kp during the interval under study being 2-and K_P for the day being only 8-. During August 1995 the radar ran in a high time resolution mode, allowing measurements of line-of-sight convection velocities along a single beam with a temporal resolution of 14s, and measurement of a full spatial scan of line-of-sight convection velocities every four minutes. Data from such scans reveal the radar to be measuring return flow convection during the interval of substorm activity. For three intervals during the period under study, a reduction in the spatial extent of radar backscatter occurred. This is a consequence of D region HF absorption and its limited extent in the present study is probably a consequence of the high latitude of the substorm activity, with the electrojet centre lying between 67° and 71° geomagnetic latitude. The high time resolution beam of the radar additionally demonstrates that the convection is highly time dependent. Pulses of equatorward flow exceeding 600 m s^–1 are observed with a duration of 5 min and a repetition period of 8 min. Their spatial extent in the CUTLASS field of view was 400–500 km in longitude, and 300–400 km in latitude. Each pulse of enhanced equatorward flow was preceded by an interval of suppressed flow and enhanced ionospheric Hall conductance. The transient features are interpreted as being due to ionospheric current vortices associated with field aligned current pairs. The relationship between these observations and substorm phenomena in the magnetotail is discussed.     

A statistical study of diurnal, seasonal and solar cycle variations of F-region and topside auroral upflows observed by EISCAT between 1984 and 1996

A statistical analysis of F-region and topside auroral ion upflow events is presented. The study is based on observations from EISCAT Common Programmes (CP) 1 and 2 made between 1984 and 1996, and Common Programme 7 observations taken between 1990 and 1995. The occurrence frequency of ion upflow events (IUEs) is examined over the altitude range 200 to 500 km, using field-aligned observations from CP-1 and CP-2. The study is extended in altitude with vertical measurements from CP-7. Ion upflow events were identified by consideration of both velocity and flux, with threshold values of 100 m s^–1 and 10¹³ m^–2 s^–1, respectively. The frequency of occurrence of IUEs is seen to increase with increasing altitude. Further analysis of the field-aligned observations reveals that the number and nature of ion upflow events vary diurnally and with season and solar activity. In particular, the diurnal distribution of upflows is strongly dependent on solar cycle. Furthermore, events identified by the velocity selection criterion dominate at solar minimum, whilst events identified by the upward field-aligned flux criterion dominated at solar maximum. The study also provides a quantitative estimate of the proportion of upflows that are associated with enhanced plasma temperature. Between 50 and 60% of upflows are simultaneous with enhanced ion temperature, and approximately 80% of events are associated with either increased F-region ion or electron temperatures.     

Intercalibration of HRDI and WINDII wind measurements

The High Resolution Doppler Imager (HRDI) and the Wind Imaging Interferometer (WINDII) instruments, which are both on the Upper Atmosphere Research Satellite, measure winds by sensing the Doppler shift in atmospheric emission features. Because the two observation sets are frequently nearly coincident in space and time, each provides a very effective validation test of the other. Discrepancies due to geophysical differences should be much smaller than for comparisons with other techniques (radars, rockets, etc.), and the very large sizes of the coincident data sets provide excellent statistics for the study. Issues that have been examined include relative systematic offsets and the wind magnitudes obtained with the two systems. A significant zero wind position difference of 6 m s^–1 is identified for the zonal component, and it appears that this arises from an absolute perturbation in WINDII winds of -4 m s^–1 and in HRDI of +2 m s^–1. Altitude offsets appear to be relatively small, and do not exceed 1 km. In addition, no evidence is found for the existence of a systematic wind speed bias between HRDI and WINDII. However, considerable day-to-day variability is found in the quality of the agreement, and RMS differences are surprisingly large, typically in the range of 20-30 m s^–1.     

Stationary magnetospheric convection on November 24, 1981. 2.Small-scale structures in the dayside cusp/cleft

A case study of the dayside cusp/cleft region during an interval of stationary magnetospheric convection (SMC) on November, 24, 1981 is presented, based on detailed measurements made by the AUREOL-3 satellite. Layered small-scale field-aligned current sheets, or loops, superimposed to a narrow V-shaped ion dispersion structure, were observed just equatorward from the region of the “cusp proper”. The equatorward sheet was accompanied by a very intense and short (less than 1 s) ion intensity spike at 100 eV. No major differences were noted of the characteristics of the LLBL, or “boundary cusp”, and plasma mantle precipitation during this SMC period from those typical of the cusp/cleft region for similar IMF conditions. Simultaneous NOAA-6 and NOAA-7 measurements described in Despirak et al. were used to estimate the average extent of the “cusp proper” (defined by dispersed precipitating ions with the energy flux exceeding 10⁻³ erg cm⁻² s⁻¹) during the SMC period, as ≈0.73∼ ILAT width, 2.6–3.4 h in MLT, and thus the recently merged magnetic flux, 0.54–0.70 × 10⁷ Wb. This, together with the average drift velocity across the cusp at the convection throat, ≈0.5 km s⁻¹, allowed to evaluate the cusp merging contribution to the total cross-polar cap potential difference, ≈33.8–43.8 kV. It amounts to a quite significant part of the total cross-polar cap potential difference evaluated from other data. A “shutter” scenario is suggested for the ion beam injection/penetration through the stagnant plasma region in the outer cusp to explain the pulsating nature of the particle injections in the low- and medium-altitude cusp region.     

What can we tell about global auroral-electrojet activity from a single meridional magnetometer chain?

The AE indices are generally used for monitoring the level of magnetic activity in the auroral oval region. In some cases, however, the oval is either so expanded or contracted that the latitudinal coverage of the AE magnetometer chain is not adequate. Then, a longitudinal chain in the key region would give more information of the real situation, but, of course, only during some limited UT-period. In order to find out the UT coverage of a single meridional chain, we have compared the global AL and AU indices with corresponding local indices determined using data from the meridional part of the EISCAT Magnetometer Cross during the years 1985–1987. A statistical study shows that the local indices are close (within relative error of 0.2) to the global AU and AL during periods 1500–2000 UT ( 1730–2230 MLT) and 2130–0130 UT (000–0400 MLT), respectively. In the middle of these optimal MLT-sectors the EISCAT Cross sees more than 70% of the cases when the global AE chain records activity. Then, also the correlation between the local and global indices is generally good (>0.7). Thus we conclude that five to six evenly located meridional chains are needed for covering all the UT-periods. On the other hand, already the combination of IMAGE, CANOPUS, and the Greenland chains catches 50% of the substorms. Case-studies show that usually during 2130 – 1100 UT the AL achieved from these chains reproduces the real AL with good timing, although it does not follow all transient variations.     

CUTLASS HF radar observations of high-latitude azimuthally propagating vortical currents in the nightside ionosphere during magnetospheric substorms

High-time resolution CUTLASS observations and ground-based magnetometers have been employed to study the occurrence of vortical flow structures propagating through the high-latitude ionosphere during magnetospheric substorms. Fast-moving flow vortices (800 m s^–1) associated with Hall currents flowing around upward directed field-aligned currents are frequently observed propagating at high speed (1 km s^–1) azimuthally away from the region of the ionosphere associated with the location of the substorm expansion phase onset. Furthermore, a statistical analysis drawn from over 1000 h of high-time resolution, nightside radar data has enabled the characterisation of the bulk properties of these vortical flow systems. Their occurrence with respect to substorm phase has been investigated and a possible generation mechanism has been suggested.     

Relation of auroral substorm onset to local AL index and dispersionless particle injections

In this study, we investigate the relation of auroral substorm onset to the sharp decrease in the local AL index (IL) during substorms. With a database of over 4200 onsets determined from auroral images, we have statistically examined the timing between the auroral substorm onset and the sharp decrease in the IL index, as determined with data from the IMAGE magnetometer network. From the database of onsets, 54 substorms were determined to be within 6° of the central meridian of the IMAGE ground array. Our superposed epoch median curve shows that the IL index begins to sharply decrease 3 min before the auroral onset, which is twice the 2 min resolution of the auroral imager. However, the mean difference determined by measuring the time between the start of the IL decrease and the auroral substorm onset is about 1.1±0.6 min. An analysis of the superposed epoch median curves of the SOPA particle data for the LANL spacecraft closest to the auroral onset meridian indicates that both the electron and proton injections begin about 3 min before the auroral onset. However, the mean time of the difference between the minimum of the particle dispersionless injection and the auroral onset is simultaneous within the uncertainty of the auroral onset and the error of the mean for the injection. The location of the electron injection relative to the IMAGE ground array seems to be 01–04 MLT, while the proton injection appears to be in the 22–01 MLT sector. These statistical results support the idea that the field aligned and ionospheric currents of the substorm current wedge begin to flow before the auroral onset.     

Scattering attenuation and intrinsic absorption using uniform and depth dependent model – Application to full seismogram envelope recorded in Northern Chile

Two seismic wave attenuation factors, scatteringattenuation Q _s ^-1 and intrinsicabsorption Q _i ^-1 are measured using theMultiple Lapse Time Window (MLTW) analysis method forthree different frequency bands, 1–2, 2–4, and 4–8 Hz.Data from 54 temporally deployed seismic stationslocated in northern Chile are used. This methodcompares time integrated seismic wave energies withsynthetic coda wave envelopes for a multiple isotropicscattering model. In the present analysis, the waveenergy is assumed to decay with distance in proportionto1/GSF·exp(- (Q _s ^-1+Q _i ^-1)· r/v), where r, and v are the propagationdistance, angular frequency and S wave velocity,respectively, and GSF is the geometricalspreading factor. When spatial uniformity of Q _s ^-1, Q _i ^-1 and v isassumed, i.e. GSF = 4r ², theestimates of the reciprocal of the extinction length,L _e ^-1 (= (Q _s ^-1+Q _i ^-1)·/v), are 0.017,0.012 and 0.010 km^-1, and those of the seismicalbedo, B ₀ (= Q _s ^-1/ (Q _s ^-1+Q _i ^-1)), are 0.48, 0.40and 0.34 for 1–2, 2–4 and 4–8 Hz, respectively, whichindicates that scattering attenuation is comparable toor smaller than intrinsic absorption. When we assumea depth dependent velocity structure, we also findthat scattering attenuation is comparable to orsmaller than intrinsic absorption. However, since thequantitative estimates of scattering attenuationdepend on the assumed velocity structure (strength ofvelocity discontinuity and/or Moho depth), it isimportant to consider differences in velocitystructure models when comparing attenuation estimates.     

Two-dimensional kinematic and rheological modeling of the 1912 pyroclastic flow,Katmai, Alaska

Pyroclastic flow emplacement is strongly influenced by eruption column height. A surface along which kinetic energy is zero theoretically connects the loci of eruption column collapse with all coeval ignimbrite termini. This surface is reconstructed as a two-dimensional energy line for the 1912 Katmai pyroclastic flow in the Valley of Ten Thousand Smokes from mapped flow termini and the runup of the ignimbrite onto obstructions and through passes. Extrapolation of the energy line to the vicinity of the source vent at Novarupta suggests the eruption column which generated the ignimbrite eruption was approximately 425 m high. The 1912 pyroclastic flow travelled about 25 km downvalley. Empirical velocity data calculated from runup elevations and surveyed centrifugal superelevations indicate initial velocities near Novarupta were greater than 79–88 m s^–1. The flow progressively decelerated and was travelling only 2–8 m s^–1 when it crossed a moraine 16 km downvalley. The constant slope of the energy line away from Novarupta suggests the flow was systematically slowed by internal and basal friction. Using a simple physical model to calculate flow velocities and a constant kinetic friction coefficient (Heim coefficient) of 0.04 derived from the reconstructed energy line, the flow is estimated to have decelerated at an average rate of –0.16 m s^–2 and to have taken approximately 9.5 minutes to travel 25 km down the Valley of Ten Thousand Smokes. The shear strength of the flowing ignimbrite at the moraine was approximately 0.5 kPa, and its Bingham viscosity when it crossed the moraine was 3.5 × 10³ P. If the flow was Newtonian, its viscosity was 4.2 × 10³ P. Reynolds and Froude numbers at the moraine were only 41–62 and 0.84–1.04, respectively, indicating laminar, subcritical flow.     

Differences between Mean Sea Levels for the Pacific,Atlantic and Indian Oceans From Topex/Poseidon Altimetry

Geopotential values W of the mean equipotential surfaces representing the mean ocean topography were computed on the basis of four years (1993 - 1996) TOPEX/POSEIDON altimeter data: W = 62 636 854.10m ² s ^–2 for the Pacific (P), W = 62 636 858.20m ² s ^–2 for the Atlantic (A), W = 62 636 856.28m ²s^–2 for the Indian (I) Oceans. The corresponding mean separations between the ocean levels were obtained as follows: A – P = – 42 cm, I– P = – 22 cm, I – A = 20 cm, the rms errors came out at about 0.3 cm. No sea surface topography model was used in the solution.     

Dayside convection and auroral morphology during an interval of northward interplanetary magnetic field

We investigate the dayside auroral dynamics and ionospheric convection during an interval when the interplanetary magnetic field (IMF) had predominantly a positive B_z component (northward IMF) but varying B_y. Polar UVI observations of the Northern Hemisphere auroral emission indicate the existence of a region of luminosity near local noon at latitudes poleward of the dayside auroral oval, which we interpret as the ionospheric footprint of a high-latitude reconnection site. The large field-of-view afforded by the satellite-borne imager allows an unprecedented determination of the dynamics of this region, which has not previously been possible with ground-based observations. The location of the emission in latitude and magnetic local time varies in response to changes in the orientation of the IMF; the cusp MLT and the IMF B_y component are especially well correlated, the emission being located in the pre- or post-noon sectors for B_y < 0 nT or B_y > 0 nT, respectively. Simultaneous ground-based observations of the ionospheric plasma drift are provided by the CUTLASS Finland HF coherent radar. For an interval of IMF B_y 0 nT, these convection flow measurements suggest the presence of a clockwise-rotating lobe cell contained within the pre-noon dayside polar cap, with a flow reversal closely co-located with the high-latitude luminosity region. This pattern is largely consistent with recent theoretical predictions of the convection flow during northward IMF. We believe that this represents the first direct measurement of the convection flow at the imaged location of the footprint of the high-latitude reconnection site.