Observations of polar patches generated by solar-wind-driven
Pc5 field line resonances and pulsed magnetic reconnection 
at the dayside magnetopause

P. Prikryl1, I. F. Grant2, J. W. MacDougall2, C. W. S. Ziesolleck3, D.
P. Steele4,5, G. J. Sofko5, and R. A. Greenwald6

1Communications Research Centre, Ottawa, ON
2Canadian Network for Space Research, Department of Physics, University of
Western Ontario, London, ON
3Herzberg Institute of Astrophysics, National Research Council of Canada, Ottawa,
ON
4Department of Physics and Astronomy, University of Calgary, Calgary, AB
5Institute of Space and Atmospheric Studies, Department of Physics, University of
Saskatchewan, Saskatoon, SK
6The Johns Hopkins University Applied Physics Laboratory, Laurel, MD

Abstract. A long series of polar patches was observed by ionosondes and
an all-sky camera during a disturbed period (Kp=7- and IMF Bz<0). The
ionosondes measured electron densities of up to 9x1011 m-3 in the patch
center, an increase above the trough density by a factor of ~4.5. Bands
of F-region irregularities generated at the equatorward edge of the patches
were tracked by HF radars. The elongated backscatter bands were swept
northward and eastward across the polar cap in a fan-like formation as the
morning convection cell expanded due to the IMF By>0. Near the North
magnetic pole, a polar camera observed the 630-nm emission patches of
a distinctly bandlike shape drifting northeastward to eastward. The 630-nm
emission patches were associated with the density patches and backscatter
bands. The patches originated in, or near, the cusp where they were
formed by convection bursts (flow channel events, FCEs) structuring the
solar EUV and auroral/cusp ionization by segmenting it into elongated
patches. It is suggested that FCEs were associated with the dayside
magnetic reconnection modulated into pulses by field line resonances
(FLRs) on the magnetic shells adjacent to the magnetopause. Multi-peak
power spectra of the IMF By-component closely matched the power spectra
of the ground magnetic Z-component observed equatorward of the cusp.
The measured time delays between the IMF By and the ground
disturbances progressively increased with latitude across the open/closed
field line boundary. These correlations in the time and frequency domains
indicated that the FLRs were directly driven by the IMF By oscillations.
The FLR ionospheric signatures that were identified in the E-region VHF
and HF backscatter just equatorward of the ionospheric cusp evolved into
FCEs at the footprint of newly reconnected field lines. The FLRs/FCEs
were associated with poleward progressing DPY currents (Hall currents
modulated by the IMF By) and riometer absorption enhancements. The
absorption indicated that particle precipitation contributed to formation of
the patches. Relative importance of possible sources of ionization forming
patches is discussed. It is concluded that both particle precipitation
(including the FLR related precipitation) and redistribution of existing
ionization by processes associated with FCEs are important in the
production of patches. 

1. Introduction
The regions of electron density enhancements known as polar patches
drifting antisunward in the polar cap ionosphere can be observed by
various techniques, principally radio or optical ones. Initially, polar
patches were identified as enhancements in the 630-nm airglow emission
due to dissociative recombination of O while the enhanced F-region
ionization in the polar patches was tracked by sophisticated ionosondes
[Buchau et al., 1983; Weber et al., 1984]. Scintillations measurements
[Buchau et al., 1985] showed that the patches are highly structured and
implied the presence of small-scale irregularities with considerable
irregularity amplitude [Basu et al., 1994]. These techniques, plus total
electron content and incoherent scatter radar measurements were combined
to study the structure and dynamics of polar patches [Weber et al., 1986].
A review of past research on polar cap patches was presented by Tsunoda
[1988]. A more recent advance in the study of patches was brought about
by imaging riometers and HF radars that turned out to be rather successful
in detecting and tracking the patch related absorption and irregularities
across the polar cap [Rosenberg et al., 1993]. The HF radars, together
with other techniques, were used to discover some new features of polar
patches, such as their close relationship to the cusp and the hemispherical
conjugacy [Rodger et al., 1994ab].
In spite of these efforts to understand the phenomenon, the question of
polar patch formation remains unresolved. Several mechanisms have been
proposed and, while each of them can explain some aspects of polar
patches, none of them alone seems to provide a complete picture.
However, one should keep in mind that several mechanisms may in fact
be needed to explain this rather complex phenomenon. Anderson et al.
[1988] proposed a mechanism that assumes a large solar induced
enhancement in the electron density close to the edge of the convection
pattern. When the cross-polar-cap potential rises abruptly, the polar cap
expands bringing in sunlit plasma from low geomagnetic latitudes which
thereafter convects into the polar cap. The polar cap boundary then
retracts, "breaking off" a polar patch. Lockwood and Carlson [1992] using
a new concept of flow excitation developed by Cowley and Lockwood
[1992] proposed that time-dependent magnetic reconnection and
convection produce polar cap patches from pre-existing enhancements of
the electron density and associated density gradient produced by solar
photoionization near the terminator. They invoked transient bursts of
magnetic reconnection, so called flux transfer events (FTEs), in their
mechanism of patch production. It should be noted that patches were
tentatively associated with FTEs earlier by Walker et al. [1986] who used
an HF radar to study pulsations near the cusp. Rodger et al. [1994a]
suggested that enhanced ionization of the convecting plasma by particle
precipitation in the cusp and subsequent disruption of the convection by
flow channel events (FCEs) [Pinnock et al., 1993] can form polar patches.
In models of polar cap patches that rely on the solar induced ionization
the problem becomes one of understanding how the patches are produced
from a tongue of ionization (TOI) drawn into the polar cap by convection.
Sojka et al. [1993] modeled polar cap patches and suggested that they can
be formed without a complex plasma source and can occur naturally as the
magnetospheric convection varies in time. The TOIs that are extended into
the polar cap are often invoked as sources of patches. Valladares et al.
[1994] pointed out that the fast plasma jets are collocated with regions of
low F region density and enhanced plasma temperature. They concluded
that the latter resulted in an enhanced recombination rate thus eroding a
substantial volume of the ionization tongue and segmenting it into patches.

All models that assume solar induced ionization to form TOIs and
patches predict seasonal variation in the occurrence of patches because the
location of the solar terminator with respect to the convection pattern
varies significantly with season. However, Buchau and Reinisch [1991]
observed no significant difference in the occurrence of patches in summer
and winter. On the other hand, Basu et al. [1995] observed the seasonal
and UT dependence of the occurrence of 250-MHz scintillations in the
central polar cap. 
There is strong evidence suggesting that patches originate in the dayside
auroral [Buchau et al., 1985] or even subauroral ionosphere [Tsunoda,
1988] although these authors emphasized the (EUV) solar-produced
plasma as a source of patches. Foster and Doupnik, [1984] have inferred
that significantly enhanced density blobs convect into the polar cap at the
dayside throat region. Sojka and Schunk, [1986] modeled significant
density enhancements produced by soft and hard precipitation in the
dayside auroral oval. Local enhancement of F-region densities (near the
cusp before a patch was formed) was attributed to soft elctrons
[Valladares et al., 1994]. 
In the present paper it is suggested that the electron precipitation
associated with field line resonances (FLRs) equatorward of the cusp can
significantly contribute to patch formation. Also, further evidence in
support of the results presented in a compagnion paper [Prikryl et al.,
submitted to Journal of Geophysical Research, 1996] (later referred to as
paper 1). Namely, it is shown that when excited on the magnetic shells
adjacent to the magnetopause FLRs modulate the dayside magnetic
reconnection into pulses. However, in the present paper (paper 2) we
concentrate on the question of polar patch formation and suggest that
precipitation and convection bursts associated with the FLR modulating
the reconnection structure the dayside polar ionosphere into bandlike
patches. 

2. Instruments and Techniques
The Kapuskasing-Saskatoon pair of HF radars (Figure 1) is a part of
an extended network of HF radars called SuperDARN (extended Dual
Auroral Radar Network) [Greenwald et al., 1995]. The FoV of this pair
of radars extends from 65o to 85o north magnetic latitude and covers ~3
hours in magnetic local time. The radars employ linear phased arrays of
16 log-periodic antennas and the operational frequency is between 8 and
20 MHz. Each radar forms a beam which is narrow in azimuth (2.5-6o)
but broad in elevation (up to ~40o at 8 MHz). The beams were stepped
through 16 adjacent azimuthal positions every 96 s. In near-real time, the
radars measure the backscatter power, line-of-sight velocity and Doppler
spectral width by fitting the auto-correlation functions (ACFs) for 70
range bins starting at the slant range of 180 km (normally with the range
resolution of 45 km).
The Canadian Advanced Digital Ionosonde (CADI) installed at Rabbit
Lake, Resolute Bay and Eureka observed patches drifting overhead. Each
CADI [MacDougall et al., 1995] uses a receiving antenna array consisting
of four dipoles arranged along the sides of a square 60 m on a side, with
each antenna attached to a dedicated receiver.  A fixed transmitter
frequency mode (usually 3 frequencies) was interleaved with ionosonde
frequency-sweep mode generating an ionogram every 15 min. The drift
was measured using the fixed transmitter frequency mode [Grant et al.,
1995].
A polar camera operated in Eureka is a pair of CCD-based all-sky
imagers [Steele and Cogger, 1995]. The 630.0-nm images to be discussed
here are 60-s time exposures at intervals of two to three minutes. 
The 48.5-MHz Bistatic Auroral Radar System (BARS) was part of the
CANOPUS ground-based network [Rostoker et al., 1995] before the
radars were shut down in summer 1994. The BARS normal mode of
operation was described by McNamara et al. [1983], and André et al.
[1988]. Figure 1 shows the Nipawin and Red Lake radar field-of-view
(FoV) projected onto a map that also shows the FoVs of a pair of
SuperDARN radars and the CANOPUS sites (referred to later in this
report). After the summer of 1992 BARS was operated in a mixed mode.
The main operation mode (mode 4, which we call the normal mode)
employed an alternating single-pulse/double-pulse technique to measure
the intensity and mean Doppler velocity in 20-km range gates out to a
maximum of 1300 km. In every 30-second time interval, the normal mode
integrated 450 single pulses for the intensity and 450 double pulses for the
mean Doppler determination. The normal mode was interleaved at 5-
minute intervals with mode 6 (the spectral mode) which was a burst of 34
pulses at a pulse repetition frequency (PRF) of 500 Hz. In the spectral
mode the echo ranges were aliased by the 300-km long interval which was
established by the 500-Hz PRF. Fourteen of the 15 range gates in each of
the 8 beams were usable within the 300-km interval and were sampled
simultaneously. Normally, the high rate burst analysis used the last 32
pulses as a time series over an interval of 64 milliseconds. Fourier
transformations of the 112 simultaneous time series yield Doppler spectra
for all range gates with a spectral resolution of approximately 16 Hz (50
m/s). In this mode, spectral aliasing can occur for Doppler shifts greater
than 250 Hz (773 m/s). Range and spectral ambiguities can usually be
resolved with the aid of the intensity and mean Doppler data from the
neighboring 30-second intervals of integration where there are no range
ambiguities.
Another CANOPUS instrument from which data are used in this paper
is the Magnetometer and Riometer Array (MARIA). Each MARIA site is
equipped with a three-component fluxgate magnetometer of the ring-core
type and a 30-MHz zenith riometer with an antenna composed of two half-
wave dipoles separated by half a wavelength and equipped with a
reflector. The magnetometer data are acquired at a rate of one sample per
5 s. In the case study presented here (December 2, 1993, 1800-2200 UT)
quiet day baselines derived from data on December 9 and 13 are used for
magnetometer and riometer data, respectively. Mainly the data from the
North-South array (Pinawa (PINA), Island Lake (ISLL), Gillam (GILL),
Back (BACK), Fort Churchill (FCHU), Eskimo Point (ESKI) and Rankin
Inlet (RANK)) are used in this paper although some data from East-West
array are mentioned (Fort Smith (FSMI) and Dawson (DAWS)).
Geographic, geomagnetic coordinates of the sites and associated L values
have been listed e.g. by Rostoker et al., [1995]. 

3. Dayside Event of December 2, 1993: A Case Study
On December 2, 1993 (1800-2200 UT), the IMF conditions (section
3.1) were favorable for dayside magnetic merging and polar patch
formation. The cusp shifted from ~71o down to at least ~68o of magnetic
latitude (sections 3.2, 3.3 and 3.8). During this disturbed period (Kp=7-),
BARS observed a series of poleward progressing Pc5 backscatter bands
entering the cusp from the auroral oval. The bands were associated with
discrete riometer absorption enhancements and Pc5 magnetic pulsations
(section 3.4) as a result of field line resonances (FLRs) (section 3.5) near
the polar cap boundary (magnetopause) and correlated with the solar-wind-
driven poleward progressing ionospheric disturbances (DPY currents)
(section 3.3) observed in the cusp and poleward of it [Stauning et al.,
1995]. Also, the IMF By is found to have modulated the the DPY (Hall)
currents equatorward of the cusp (section 3.4). The ground-based data
coverage from regions equatorward of the cusp (which the latter authors
did not discuss in detail) to the central polar cap during this event support
evidence presented in paper 1 that magnetosphere-ionosphere coupling
processes on the open and closed field lines are closely interrelated
(sections 3.6 and 4.2). The F-region irregularities associated with polar
patches that formed in, or near, the cusp were observed with SuperDARN
(section 3.7). The patches 100-200 km wide were detached from the cusp
by FCEs (section 3.8) and observed with CADIs and an all sky imager as
they drifted poleward in the polar cap (section 3.9). Combining
observations using various techniques we conclude that some of the
patches extended over at least 2500 km in length (section 3.10). 
  
3.1. The IMF Conditions
The IMP 8 spacecraft measured the interplanetary magnetic field
(Figure 2) and was located at (-6.5, -31.0, 6.4 RE) in GSM coordinates at
2000 UT. The IMF Bz component was negative from 1800 to 2050 and
from 2315 to at least 2400 UT (except for a brief reversal at 1850 UT).
There were no IMF data available between 2050 and 2315 and several
hours before 1800 UT. During the time of the event, the By component
was positive except for a few brief negative excursions. These are
favorable conditions for dayside magnetic merging and polar patch
formation [Rodger et al., 1994a]. 
The IMF fluctuated on the scale from a few to several minutes. Fast
Fourier transform (FFT) power spectra of the IMF (particularly the By
component) resolved a number of discrete frequencies. The multi peak
power spectra are similar to those discussed in paper 1. 

3.2. The Cusp Location
No DMSP satellites passed over the geographical area covered by
SuperDARN radars during the time of the event. Later, the poleward edge
of the auroral oval in the northern hemisphere could be estimated from
F11 satellite particle data to be located near Fort Churchill at 2207 UT.
In the southern hemisphere, the cusp was identified from the particle data
[Rich, 1995, priv. comm.] during two passes of the F10 DMSP satellite.
It was located at -69o of magnetic latitude at 1851 and just before 2030
UT. These observations of the particle cusp in the southern hemisphere
are in agreement with the radar cusp signatures discussed below.
Baker et al., [1995] showed that the Doppler spectra of the ionospheric
backscatter from within the cusp obtained by HF radars are more complex
than the spectra typically observed and that the spectral widths can be
used to identify the cusp. Plate 1 (top) shows spectral widths for the
Saskatoon radar beam #7. The large widths (> 250 m/s) before 1950 UT
indicate that the cusp equatorward boundary was located near 69o. Just
equatorward of the radar cusp is a region identified by lower spectral
widths which are typical of the low latitude boundary layer (LLBL) [Baker
et al., 1995]. Before 1900 UT (not shown), the boundary between large
and moderate spectral widths shifted from ~71o magnetic latitude at
~1730 UT, when the Saskatoon radar showed a weak cusp signature,
down to ~69o at ~1820 UT, when the Kapuskasing radar measured
spectral widths of >200 m/s at magnetic latitudes above 69o. Plate 1 show
the radar cusp observed with the Saskatoon radar at ~69o from 1900 UT.
The last clear radar cusp signature was identified with the Kapuskasing
radar near 68o at 2000 UT. At this time, the Saskatoon radar showed large
spectral widths at latitudes >70o indicating somewhat fuzzy poleward
boundary of the cusp [Baker et al., 1995]. It should be noted that
"pockets" of enhanced spectral widths can also be observed in the LLBL
footprint [Pinnock et al., 1995] with an implication that large spectral
widths are not exclusively associated with the cusp. A band of large
spectral widths observed by the Kapuskasing radar in the E-region (at
~500 km) at 2000 UT may or may not be related to the cusp. The widths
of the E-region radar spectra are not necessarily an indicator of the cusp
because they reflect a different plasma regime and are due to Farley-
Buneman/gradient drift instabilities [Farley, 1963; Buneman, 1963; Farley
and Fejer, 1975] which can result in a variety of spectral shapes including
narrow peaks at Doppler velocities ranging from small to very high
depending on the horizontal density gradient scales and orientations [St.-
Maurice et al., 1994, Prikryl et al., 1995]. Thus the spatial and temporal
averaging of E-region backscatter may result in broad spectra even in
auroral oval. Nevertheless, it was observed that the region of very narrow
widths (Plate 1) kept shifting equatorward. This and the fact that FCEs
were observed at progressively lower latitudes (Figure 15) clearly indicate
that the cusp also shifted further equatorward after 2000 UT.
Furthermore, the polar cap expanded as far as 70o magnetic latitude at
2045 UT as indicated by significantly reduced spectral widths associated
with patches (Plate 1). The equatorward shift of the cusp also agreed with
observations of DPY currents associated with the FCEs discussed below.
Also, it should be noted that the cusp can span up to 6 hours of local time
[Crooker et al., 1991; Newell and Meng, 1994]. At 2100 UT, most of the
radar data are from magnetic local times before 1400 MLT and the
northern hemisphere cusp is expected to be shifted into the afternoon
sector for the IMF By>0. This supports the argument that the cusp/cleft
was present and observed by the radars at near ranges until at least 2200
UT (~1500 MLT). Similarly to the event described in paper 1, the IMF
conditions during this event were favorable for magnetic reconnection at
the dayside magnetopause (at least until 2050 UT). In both cases, the
polar cap clearly expanded as more and more closed magnetic flux was
eroded by flux transfer events. 

3.3. Poleward Edge of the ionospheric cleft currents (DPY currents)
Figure 3 shows magnetograms (X and Z components) from the
CANOPUS meridional chain. Also shown is the IMF By-component which
is discussed below (section 3.6). During the first hour shown (1700-1800
UT), poleward progressing currents were observed poleward of FCHU.
At ~1730 UT, as we mentioned above, the Saskatoon radar showed a
weak signature of a radar cusp with a equatorward boundary at ~71o
magnetic latitude. After 1800 UT, the poleward progressing disturbances
were correlated with the IMF By-component and thus were due to eastward
DPY (Hall) currents [Friis-Christensen and Wilhjelm, 1975; Clauer et al.,
1995, Stauning et al., 1995] poleward, and near, the cusp that now shifted
further equatorward. The thick arrows in Figure 3a show corresponding
phases of the long-period variations of the IMF By- and ground X-
components. A major equatorward shift of the poleward boundary of the
DPY current system is noted at 1755 UT (from RANK to ESKI) followed
by a further equatorward expansion (~1840 UT) to FCHU. This
boundary briefly recovered back to ESKI at 1920 UT but then suddenly
expanded equatorward again at 1935 UT through FCHU and BACK to
GILL and later shifted even further south towards ISLL. This was
confirmed by obtaining the latitude profiles of magnetic perturbation
(Figure 4) that can be used to estimate the three-dimensional current
system [Hughes and Rostoker, 1977]. The position of the poleward edge
of the eastward DPY current system as indicated by the location of the
negative extremum of the Z component approximating the position of the
polar cap boundary is shown in Figure 5b. Mapped along the leftmost
beam #1 of the Nipawin radar it approximately matched the poleward edge
of the BARS backscatter (Figure 5a). This boundary and the boundary
between large and small spectral widths observed by the SuperDARN
radars approximately bracketted the cusp but possibly included parts of the
LLBL ionospheric footprint (see section 3.2).

3.4. Pc5 Pulsations
The Pc5 pulsations that were observed with radars, magnetometers and
riometers clearly spanned the regions of closed and open field lines. We
shall concentrate on two periods (1800-1830 UT and 1940-2030 UT;
refered to as event 1 and event 2, respectively) when the pulsations
appeared to be most coherent and the BARS data were the least
contaminated with sidelobe leakage due to very intense backscatter. 
Associated with the sudden expansion of the polar cap at ~1755 UT
the CANOPUS magnetometers (ISLL, GILL, BACK, and FCHU)
observed a strong burst of a linearly polarized oscillation in the north-
south direction with frequency of ~5 mHz (Figure 3). This 5-mHz
oscillation that was also observed with riometers (Figure 6), was the
strongest around the magnetic shell L~7 (GILL/BACK) and had nearly
zero azimuthal wave number. Virtually no phase shift either in latitude or
longitude could be identified before 1800 UT during this initial burst of
the pulsation (centered at ~1755 UT in Figure 6) that maximized further
westward at FSMI (not shown) where the amplitude of the oscillation in
absorption was ~2 dB. A weak backscatter was observed with the Red
Lake radar during this earlier time. 
After 1800 UT, the BARS radars observed poleward drifting
backscatter bands a-h between GILL and FCHU (Figures 5a and 7). The
Nipawin radar power was correlated with the IMF By component: the IMF
By leading the Nipawin radar power by ~5 min (as discussed later with
Figure 10). The IMF By component is superposed in Figure 5a and shifted
by 5 min which is the average time lag (BARS vs. IMF By) that was
inferred from the cross-correlation analysis (discussed below). The
pulsations continued after 1830 UT but were more complex (composed of
several oscillation frequencies superposed) while the IMF By remained
strongly positive (By~+10 nT) and Bz briefly turned positive at ~1850
UT. The riometer absorption fluctuated similarly to the BARS backscatter
power but the fluctuations were more irregular. This can be explained by
a broad riometer antenna intercepting a band of latitudes with a latitude
dependent temporal and spatial ionospheric structure due to field line
resonances at various frequencies (discussed below). As a result it was not
possible to determine a one-to-one correspondence between the BARS
backscatter bands and riometer absorption enhancements with certainty.
Figure 6 shows a series of discrete absorption events (up to a few tenths
of a decibel) superposed on a strong background absorption (>1 dB). 
Associated with the second major equatorward shift of the eastward
current system at 1935 UT (Figures 3a and 5b) when the IMF By
component sharply turned negative and back to positive again was another
coherent train of Pc5 pulsations (event 2). BARS observed a series of
poleward progressing backscatter bands (labeled A-O in Figure 5a). Figure
8 shows an example of BARS maps of mean backscatter power and mean
line-of-sight velocities for the Red Lake radar. Note that some backscatter
bands occurred in pairs with a narrow gap between the two bands of
opposite Doppler shifts which indicated the electric field to be directed
into an arc that was likely filling the gap (e.g. band E'E). However, most
of the bands which reached the zenith at GILL at the same time when the
riometer absorption subsided ("equatorward" edge of the absorption band;
see Figure 6) were associated with only one polarity of Doppler shift thus
suggesting that the backscatter was located near the equatorward edge of
an arc. (There were no riometer data from ISLL, a gap in the riometer
data from BACK, and no optical data available at auroral latitudes during
this daytime event.) BARS spectral mode (Figure 8; bottom panels)
showed broad and complex spectra as these "arcs" entered the cusp/cleft
region. The spectra are averaged for two adjacent beams and also for all
ranges. The latter average is shown in the line plot at the bottom. An
aliased (in range and frequency) type-4 peak (-900 m/s) is superposed on
a broad component. The range aliasing can be easily resolved because the
spectra are clearly associated with the strongest backscatter band D (to get
the true range 600 km should be added to the aliased range shown in
Figure 8). Also, the frequency aliasing can be resolved because the mean
Doppler velocity was highly negative during the standard mode of BARS
operation 30 s earlier (Figure 8; top panels) clearly indicating that the
peak near +600 m/s must be aliased as shown (Figure 8; bottom left).
Such spectra are explained by strong electric fields, electron heating,
and/or steep horizontal gradients in the electron density associated with
auroral arcs [Prikryl et al., 1995].
The backscatter bands A-H were closely associated with riometer
absorption (backscatter trailing the absorption enhancements; see Figure
6). Unlike the complex variations  in absorption that included several
frequency components at earlier times the low frequency oscillations
dominated after 1930 UT (see below) as the cusp/cleft shifted further
equatorward. The absorption was most likely due to ionization in the arcs
produced by FLRs. Previously, a large BARS database during the times
when there were optical data available was examined and it was found that
backscatter bands such as those shown in Figure 8 are commonly
associated with arcs [Prikryl and Cogger, 1992]. 
For SuperDARN, at ranges less than 800 km, the limited refraction
usually limits observations to E-region irregularities. Similarly to BARS,
SuperDARN observed the poleward drifting Pc5 bands although with
lower spatial and temporal resolution. There was considerably less
SuperDARN E-region backscatter before 2000 UT, when the Kapuskasing
radar (Figure 11a) observed only a trace of a couple of Pc5 bands. After
2000 UT, a direct comparison between BARS and SuperDARN was
possible for the Nipawin radar beam #1 (Figure 5a) and the Saskatoon
SuperDARN beam #12 (Figure 11b). These RTI plots showed a good
correspondence between the E-region backscatter bands observed with
both radars.
Figures 9a and 9b show 16- and 32-point FFT power spectra of the
IMP-8 IMF, ground magnetic perturbations and VHF radar backscatter
power for event 1 and 2, respectively. The low resolution spectra for
event 1 show two main bands at low (<2 mHz) and high (4-6 mHz)
frequencies with an increase in spectral power near 3 mHz. The IMF
spectra are correlated with those for the ground instruments where the low
frequencies dominated at high latitudes (ESKI) while the higher frequency
pulsations were more pronounced at lower (subcusp) latitudes (GILL). The
FFT spectra of the riometer data (not shown) were similar. The 32-point
FFT power spectra for event 2 (Figure 9b) show a low frequency
component near ~1 dominating all of the spectra. The IMF spectral
power appears to be non-uniformly spread across higher frequencies with
a clear peak at 2.5 mHz and a less coherent peak near 2 mHz. However,
the spectral resolution is insufficient to further resolve these and other
possible spectral peaks. At subcusp latitudes (ISLL) the FFT power is
clearly distributed into several discrete and relatively coherent peaks
showing up in two or three components, particularly those at frequencies
>2 mHz. Near the cusp (GILL) low frequencies (1 and 2 mHz)
dominated the spectra. The same frequencies also modulated the intensity
of the Nipawin radar backscatter bands near GILL (Figure 9b; bottom
panel). The similarity between the spectra for the IMF and ground data is
quite clear, particularly for event 1. This was confirmed by using the
maximum entropy method (MEM) to compute spectra (not shown). The
MEM spectra clearly resolved some of the dominant (e.g., 1- and 2-mHz)
frequency components including the broad peak for the Red Lake radar
data that is not resolved in the FFT spectra. However, this method
resulted in what appeared to be unrealistically narrow peaks while
sometimes producing unexpected shifts but, most importantly, it neglected
other frequency oscillations with smaller amplitudes that were clearly
resolved by the FFT method. 
Above, we concentrated on the events 1 and 2 for which radar data
were useful for comparison with other data sets and we will refer to these
events again when discussing SuperDARN and CADI data. While there
were notable differences between events 1 and 2, it should be emphasized
that these two "events" were not isolated pulsation events. The IMF
oscillated and ground pulsations were observed throughout the period
shown in Figure 5a. The low resolution (16-point) FFT power spectra
were already discussed above (Figure 9a). Concentrating on the IMF data
first, when the FFT window was doubled in length more peaks were
resolved. However, apart from the resolution, the 16-, 32- and 64-point
spectra were all very similar in shape. Also, four consecutive 16-point
spectra (not shown) were similar to the one shown in Figure 9a. Figure
9c show the 64-point FFT power spectra for a 130-min long IMF time
series. The lowest resolution spectra formed a fitting envelope to the 64-
point FFT spectra (an average of four consecutive 16-point power spectra
is shown in Figure 9c). The peak near 0.5 mHz in the 64-point FFT is
missing in the low resolution 16-point FFT spectra because it was filtered
out by a detrending procedure (the 2nd order polynomial is subtracted
from time series before FFT). Similarly, the averaged 16-point spectra for
other two components (not shown) enveloped the corresponding high
resolution spectra in Figure 9c. Thus we conclude that the time series had
approximately the same spectral content (multimode oscillations
superposed) throughout the 130-min period. For example, the broad
looking By FFT spectrum (Figure 9a) is simply a result of a number of
unresolved discrete peaks at ~1, 1.5, 2, 3, and 4 mHz that are resolved
in the 64-point spectrum (ignoring some weaker peaks at this point and
excluding the 0.5-mHz peak). Similar results were obtained for other
events discussed in paper 1 where it was concluded that these multi-mode
IMF oscillations are vestiges of dominant solar oscillation modes in solar
wind. This possibility is also suggested here and Figure 9c shows the best
matching eigenfrequencies for spherical harmonic order l~40 for a typical
solar model [Christensen-Dalsgaard et al., 1985]. These multi-mode IMF
oscillations caused the pulsations in the magnetosphere and excited field
line resonances equatorward of the cusp. The best correlation was between
the FFT power spectra between for the ISLL Z- and IMF By-component
(cross-correlation coefficient of 0.80). Note that virtually all of the major
and many minor peaks in these two spectra coincide. It should be noted
that the ground Z component is a good indicator of FLRs (poleward
progressing Hall current intensifications associated with FLRs). The
correlation between the FFT power spectra for the IMF and ground
magnetic data was also fairly good for other observing sites (see Figure
9ab). Poleward of the cusp (RANK and ESKI), low frequency components
clearly dominated the spectra but some spectral power at frequencies >
2 mHz was also correlated with the IMF By FFT power.

3.5. Evidence for FLRs
In paper 1, we have shown that the latitude dependent FLR structure
impressed on the ionospheric flow was extended from the closed field line
region in the LLBL footprint to the cusp which is threaded by field lines
that are being opened (reconnected with IMF). A series of intense FCEs
showed characteristics very similar to what is expected of FLRs.
However, the FCE characteristics (the enhanced flows, peak width of the
resonance, and the phase) deviated from a clear-cut FLR signature (see
paper 1) because the FCEs result from FLRs on the magnetic field lines
that were about to be opened while the resonance was acting on them,
i.e., the resonance was strongly disturbed (interrupted) by reconnection.
Furthermore, it is suggested (see paper 1) that the reconnection is
modulated by the Alfvén wave associated with the resonance. 
The response of the magnetosphere and ionosphere to multi-mode
oscillations in the solar wind IMF observed on Dec 2, 1993 showed
similar evidence for FLRs. The latitude dependent frequency response
which is a typical characteristic of FLRs was already pointed out in
section 3.4. FLR signatures were identified in the ground magnetometer
and ionospheric signals. It should be noted that this was not possible for
all data subsets. For example, there was not enough SuperDARN E-region
backscatter during the first pulsation event. BARS Doppler data was not
always reliable because of a sidelobe contamination problem (Figure 7a).
Also, note that the F-region SuperDARN backscatter which is usually very
useful to look for FLRs [see e.g. Ruohoniemi et al., 1991] could not be
used during this very disturbed event because most of the F-region
backscatter was in the polar cap on the open field lines. However, the E-
region backscatter data (both SuperDARN and BARS) showed evidence
for FLRs. 
Figure 12 shows the spectral analysis results for the Saskatoon
SuperDARN radar (beam #11) LoS velocity. Three distinct peaks at about
1, 2 and 2.5 mHz can be seen in the spectral power density of the LoS
velocity (Figure 12, top panel). The position of these peaks is clearly
latitude dependant, i.e. the highest frequency peak appeared at the lowest
latitude and vice versa. The latitude profile of the spectral power at 2
mHz (Figure 12, bottom panel) shows a peak with halfpower width of
about 1o centered near Gillam. The phase decreased by 180o across the
latitudinal width of the power peak. These signal characteristics are fully
consistent with discrete low-m field line resonances [e.g. Ruohoniemi et
al., 1991; Samson et al., 1991, 1992ab; Ziesolleck and McDiarmid, 1994,
1995; Fenrich et al., 1995]. 
The phase increase between 66.3o and 66.7o is probably due to a 2*
phase wrapping. Away from the power peak, the phase continues to
change so that the total phase change between 65o and 69o latitude appears
to be considerably larger than 180o. However, the phase estimates at
latitudes with low spectral power density must be considered less reliable.
We note, though, that field line resonances with a latitudinal phase shift
larger than the theoretically predicted 180o [e.g., Chen and Hasegawa,
1974; Southwood, 1974] have been observed previously at auroral
latitudes (see for instance Ziesolleck and McDiarmid, 1994; their Figures
7 and 15). 
It should be noted that the peak at 2.5 mHz was also identified at ISLL
(Figure 9b) and RABB (not shown) and this frequency was also found in
the IMF signal (Figures 9b and 9c). In general, the observed magnetic
FLR signatures were not as clear cut as those for the radar data. This
could be because a magnetometer averages over much larger volume than
a radar and also because the pulsation region was rather close to the
magnetopause where the FLR were disturbed by reconnection. At lower
latitudes, the magnetometer array is sparsely populated and FLR
signatures could not be studied in detail. 
The results discussed in this section and section 3.4 are consistent with
the theory of FLRs. Because the IMF FFT power was strongly correlated
with the FFT power of the ground magnetic components on the closed
field lines (the correlation between the time series of the IMF By
component and the ground and ionospheric data is further discussed
below), we conclude that the resonances were directly driven from the
solar wind particularly by the IMF By oscillations.

3.6. The Correlation Between the IMP-8 IMF By and the Ground and
Ionospheric Data
It is well known that the IMF By component controls the dayside cleft
current system [Banks et al., 1984] and plays an important role in the
dayside magnetic reconnection [Crooker, 1979; Cowley et al., 1991;
Taguchi et al., 1993; Newell et al., 1995]. A number of studies have
clearly demonstrated the solar-wind-driven modulation of the dayside
ionospheric DPY currents and convection on the open field lines [Clauer
et al., 1984; Clauer and Banks, 1986; Greenwald et al., 1990; Clauer et
al., 1995; Stauning et al., 1995]. The results presented here and in paper
1 show that the IMF By has a strong influence on the ionosphere at the
footprint of the closed field lines near the magnetopause where the FLRs
can be directly driven by oscillations in the solar wind IMF.
Cross-correlation functions (CCFs) were computed for the time series
of the IMF By and the Nipawin radar backscatter power (beam #1) for
each range gate with significant backscatter using eleven 30-min windows
with the centers of the adjacent windows separated by 15 min (overlapping
by 15 min). Apart from two short intervals around 1900 and 2000 UT, the
backscatter power associated with the poleward progressing bands was
well correlated with the IMF By oscillations (Figure 10). Note that the
correlation was poor for the strong backscatter signals between 1830 and
1900 UT at ranges from 400 to 700 km which were associated with bands
moving equatorward rather than poleward. Because this backscatter feature
is quite distinct from the poleward progressing bands we choose to ignore
it in the discussion. The time lags (CCF maxima) with significant
correlation (coefficients often significantly >0.5) that are plotted in
Figure 10 increased with range/latitude from a few minutes (at the low-
latitude edge of the backscatter where the E-region bands started to
appear) up to several minutes near the polar cap boundary or cusp (Figure
5b). These radar results are in good agreement with those for the
magnetometer data. Time delays obtained from CCFs for IMF By-/ground
X-component data sets that were highly correlated are also shown in
Figure 10 (see also Figure 3a). The largest values of the cross-correlation
coefficient are printed for each window for BARS (first row) and MARIA
(second row). The cross-correlation analysis was performed similarly to
the radar data except that the IMF By component was smoothed [Stauning
et al., 1995] for BACK, FCHU, ESKI and RANK data analysis.
However, such smoothing was not applied for GILL and ISLL where the
higher frequency pulsations were strong and correlated with the IMF By
component. This abundance of many frequency components in the ISLL
time series which is shown in Figure 9d (discussed in section 3.4) may
explain why the time series (IMF By and ISLL X) were not always well
correlated when compared with the CCF results for stations in high
latitudes where the low frequencies dominated the geomagnetic
disturbances.
These results are similar to the observations of poleward progressing
By-related ionospheric disturbances [Stauning et al., 1995]. The latter
authors closely examined the time delays and found a good agreement
with the values they calculated. IMP 8 was located somewhat unfavorably
during the event presented in this paper so the theoretical estimation of the
time delays would be difficult if possible at all. However, the observed
delays are significantly smaller (by ~10 minutes for cusp locations) than
those obtained by Stauning et al. [1995]. For BARS data, the minimum
time lag of the ionospheric disturbances was 1-2 min near the equatorward
edge of the backscatter region. It should be noted that the E-region
irregularities are generated only if electric field exceeds some threshold
value (~15 mV/m). So in fact, at lower latitudes the BARS/IMP-8 delays
would be <1 min or even negative if there was backscatter observed
there. At 1830 UT, the ISLL magnetometer delay of -1 min with respect
to unaveraged IMP-8 data suggests such possibility of negative delays
considering the position of IMP 8 downstream of the Earth.
These cross-correlation results (the smallest delays found at lower
latitude equatorward of the cusp) combined with the strong correlation in
the frequency domain (Figure 9d) between the IMF and the ground
magnetic pulsations at ISLL strongly indicate that FLRs were directly
driven by the IMF By oscillations. While Stauning et al., [1995] mainly
discussed the dayside convection disturbances observed poleward of the
cusp, our results show that these disturbances spanned both the regions
poleward and equatorward of the cusp. Further evidence that coupling
processes on the open and closed field lines are closely interrelated is
discussed in paper 1.

3.7. The F-region Backscatter Bands Observed by SuperDARN
The F-region density irregularities that caused the HF radar backscatter
are expected to be generated along the equatorward border of the patches
by the gradient-drift instability [Tsunoda, 1988]. The E-region
backscatter/precipitation bands faded away near the cusp latitudes but the
remanent F-region ionization and the steep density gradients (indicated by
BARS spectra discussed in section 3.4) provided favorable conditions for
production of such irregularities in the trailing edge of patches. Poleward
progressing E-region backscatter bands trailed the riometer absorption
enhancements (Figure 6). The E-region backscatter bands were associated
with the equatorward edge of the riometer absorption and showed Doppler
spectra indicative of sharp horizontal density gradients (in auroral arcs).
The F-region backscatter bands observed by SuperDARN were mapped
along and/or connected with the E-region backscatter bands (see, e.g.,
Figures 13 and 15). In other words, the E- and F-region irregularities
were associated with (generated by) the same density ridges (gradients) at
the equatorward edge of patches at times and/or locations when the
patches were formed near the cusp. These density gradients were
convected into the polar cap where they continued to generate the F-region
irregularities associated with patches. Unfortunately, there were no
ionosonde data available from Cambridge Bay and the backscatter bands
did not extend as far as Resolute Bay to further provide stronger evidence
about the location of the backscatter with respect to patches.  However,
some of the backscatter bands were associated with equatorward edge of
patches observed by the riometer and CADI in RABB after 2000 UT. 
Before 1840 UT, the Kapuskasing radar observed 3 major bands that
were detached from the cusp every 11-13 minutes (Figure 11a) and there
was an indication of intermediate weaker bands. The flow bursts (FCEs)
near the cusp (Figures 15a and 7c) were similarly modulated by two
frequency components of FLRs (see section 3.8). This was a result of the
FLR signature that was imposed on the ionospheric structure at auroral
latitudes (Figure 7). Also, when patches were observed by CADI in
Resolute Bay ~20 min later (Figure 18; discussed below) some of this
complex structure was found impressed on the patches.
After 2000 UT, many F-region backscatter bands were observed
(mainly with the Saskatoon radar; Figure 11b) and there was a clear one-
to-one correspondence between the E- and F-region bands, which was
confirmed by comparing the BARS and SuperDARN radar maps. The
patches and associated backscatter bands retained their bandlike form
while they drifted in the polar cap. Apparent drift velocities of up to 400
m/s for the E-region bands along the BARS Nipawin radar beam #1 were
estimated from the slope of the bands in the RTI plots. Because of the tilt
of the bands this value is an overestimate of the actual progression speed
perpendicular to the bands. In the polar cap, the polar patches/bands
moved with convection (~600 m/s as inferred from the translational
motion of a series of bands observed with the Saskatoon radar). Just
before about 2020 UT the backscatter bands were only slightly tilted with
respect to L shells and drifted predominantly poleward. After the IMF By-
component turned strongly positive the polar cap convection changed from
poleward to north-eastward. The far (west) ends of the bands drifted
faster, the bands rotated clockwise, become nearly sun-aligned and
eventually drifted eastward. Figure 13 shows two radar maps of
backscatter power where a series of bands G-K is identified at two
different times as the bands drifted eastward. At these times, the band J
was still observed with BARS at lower latitudes. The gap between the E-
and F- region HF backscatter widened because the cusp/cleft (where the
enhanced ionization would help the refraction that is needed to intercept
field-aligned irregularities) was shifted further equatorward (Plate 1 and
Figure 5b).
Convection maps were obtained by merging the LoS velocities from a
pair of SuperDARN radars after 2000 UT (Figure 14). The direction and
magnitude of the convection agreed with the bulk motion of backscatter
bands (Figure 13) derived by tracking individual bands. The 200-s (two-
sweep) average radially smoothed velocity maps (Figure 14) show the
initially poleward flow turning eastward. The change of the convection
from poleward to eastward at these high latitudes is an expected result of
the IMF By-component changing from small values about zero to larger
(positive) values after 2020 UT.
SuperDARN observations of backscatter associated with patches
spanned the gap between the regions of the patch formation in, or near,
the cusp and that of the central polar cap where the patches were observed
by ionosondes and a polar camera. At close ranges (E-region and near F-
region backscatter) the radar (BARS and SuperDARN) observations
provided evidence that the structure imposed on the ionosphere by FLRs
was propagated into the polar cap. The large spatial coverage of the radars
allowed mapping of the patch related irregularities and the spatial extent
of patches/bands could be determined. The convection velocities agreed
with the motion of the bands (Figure 13), optical patches (Figure 17), and
the drift velocities measured by CADIs (briefly discussed in section 3.9).

3.8. Convection Flow Bursts (Flow Channel Events)
Similarly to the event described in paper 1 FCEs were observed in the
cusp and associated with FLRs. Unfortunately, the radar backscatter was
often fragmented (most likely for reasons discussed in paper 1) and only
some of the FCEs were identified (Figures 5a and 15). Rodger et al.
[1994a] proposed that patches can be detached from the cusp by FCEs.
The results presented here confirm that but also provide evidence that the
plasma is "pre-structured" on the closed field lines before it enters into the
polar cap as a patch.
Before 1810 UT, there was not enough HF backscatter and the
Kapuskasing radar observed only very weak and inconspicuous flow
enhancements that showed a band structure but could not be associated
with the BARS backscatter bands at this time. Between 1812 and 1815
UT, the first FCE was observed by the northwestward pointed
Kapuskasing radar beams showing LoS velocities up to 1 km/s that
spanned E-region (BARS) bands c and d. At ~1820 UT, when the IMF
By became strongly positive, westward flows became even more enhanced
(LoS speeds exceeding 1 km/s) but were often patchy in appearance and
occurred in a band of magnetic latitudes between 68 and 72o of magnetic
latitude. Nevertheless, a few radar scans identified individual FCEs which
were associated with the E-region bands in the BARS backscatter. Figure
15a shows an FCE near 71o of magnetic latitude associated with an E-
region backscatter band f that had already faded away in the BARS FoV
by this time. A weaker FCE that has started to form at 69o magnetic
latitude was associated with the next E-region backscatter band g which
moved poleward to replace the previous band that has faded away. The
LoS flow velocity for the stronger FCE exceeded 1.2 km/s. Considering
the tilt of the flow channel with respect to the radar beams the actual flow
along the band likely exceeded 2 km/s in this case.
Moderately enhanced flow bursts were also observed with the BARS
Nipawin radar (Figure 7c) near the cusp. It should be noted that the E-
region radar measurements often strongly underestimate the actual plasma
flows because of the saturation of the Farley-Buneman instability near the
ion-acoustic speed [Nielsen and Schlegel, 1983]. Just after 1800 UT, the
first flow burst appeared to be associated with a weak backscatter band a
(observed with the Red Lake radar) and riometer absorption enhancements
progressing poleward (Figure 7). The second flow enhancement spanned
bands c and d (also observed with the Kapuskasing radar). The strong
flow bursts associated with bands f and h sandwiched a weaker flow burst
(g). Similarly to the latter band, the flow bursts associated with bands b
and e were weak or missing. The BARS flow bursts were closely
associated with riometer absorption enhancements in ESKI as well as the
enhanced SuperDARN flows observed immediately poleward or at the
poleward edge of the BARS backscatter. While the "narrow" absorption
events at BACK (Figure 7b) suggested precipitation associated with the
Pc5 bands (auroral arcs), the absorption events at ESKI (Figure 7c) were
more smeared in time than those at lower latitudes. They could have been
caused by the heated E- and/or F-region plasma similarly to observations
of patches near the cusp that were reported by Rosenberg et al., [1993].
The positive bays in the IMF By component (shown unshifted in Figure
7c) lead the Nipawin radar power and enhanced LoS velocity by ~5 min
at ~70o (900 km). Most of the By minima were correlated with sharp
decreases in the flow (and backscatter power) near 70o magnetic latitude
about 5 min later.
After 1940 UT, there were several flow bursts (FCEs) observed with
SuperDARN radars (mainly the Kapuskasing radar). At ~1940 and 1950
UT, two patchy flow bursts (~1 km/s) were observed in the western part
of the Saskatoon radar associated with the large amplitude fluctuation of
By and possibly with bands A and B (Figure 5a). The next FCE that started
at ~2003 UT showed an evidence of westward motion of the FCE which
is consistent with the expected westward and poleward motion of the FTE
signature [e.g., Smith et al., 1992]. After a deep negative bay the By
component turned positive again and a high flow burst (>1 km/s) was
associated with the band C. Figure 15b shows the last of the three
consecutive radar scans taken as the flow burst propagated westward with
a bulk speed that was comparable with the plasma flow speed. Note that
the gap between the E-region bands C and C' which was associated with
the riometer absorption enhamcement (auroral precipitation) over GILL
and FCHU was located at the poleward edge of the FCE. The next FCE
that was associated with the band D, is shown in Figure 15c. Poleward of
this FCE one can still recognize a remnant of the previous flow burst that
was associated with the patch C. Only less intense flow bursts (~500 m/s)
associated with the bands E and F at 2017 and 2022 UT, respectively,
were observed at somewhat lower latitudes (between 67 and 68o of
magnetic latitude). However, at this time, the F-region backscatter in the
western part of the Kapuskasing radar FoV become rather fragmented or
completely absent. Two poorly defined bursts of high flow (>1000 m/s)
were observed at 2042 UT and 2047 UT which might have been
associated with the bands that followed but their identity could not be
determined with certainty. Later, between 2113 and 2117 UT a flow burst
associated with the patch K (Figure 13b) was observed by the westernmost
Saskatoon radar beams.
Similarly as for the earlier event, most of the FCEs identified after
1940 UT can be approximately associated with maxima in IMF By (see
Figure 5a while noting that the delays between IMF By and the cusp
ionosphere were measured to be ~7-8 min rather than 5 min as shown in
Figure 10). While there was some uncertainty of the time delay that makes
this correlation appear less convincing there was a very good
correspondence between the FCEs and the maxima of the ground magnetic
X-component near the cusp footprint (Figure 3). The latter were due to
Hall (DPY) current enhancements associated with FCEs observed
overhead. The modulation of DPY currents by the IMF By-driven FLRs
equatorward of the cusp is further discussed in section 4.2.
Two dominant frequency components/bands (<2 mHz and 4-6 mHz)
were identified in the pulsation data between 1800 and 1830 UT (event 1)
and these frequencies modulated the flow bursts. In particular, note that
the separations between the strongest flow bursts (Figure 7c) were 10-11
min (between c and f), 7-8 min (e.g., between f and h) and separations of
3-4 minutes are found for less intense bursts. These separations are the
periodicities approximately corresponding to frequencies ~1.5, 2 and 4-5
mHz that are resolved in 64-point FFT power spectra of the IMF (Figure
9c). This result supports our previous conclusion that these particular
frequency modes were present but unresolved in the 16-point FFT
spectrum (Figure 9a; top panel). This presence of many competing FLR
modes contributed to some structural complexity of polar patches during
event 1. After 1940 UT (event 2), when the ratio of the two dominant
frequency components was approximately 1:2, the FCEs (as long as they
were observed) were associated with individual E-region bands and there
was a one-to-one correspondence between the patches and E-region bands.


3.9. CADI and Optical Observations of Patches
Figure 16 shows fixed frequency CADI  records (virtual height vs
time) obtained in Rabbit Lake (5 MHz), Resolute Bay (4 MHz) and
Eureka (4 MHz). We will concentrate on the period after 2000 UT
(By>0) and only briefly discuss the earlier period when By was more
variable turning negative at times. There was strong absorption before
2000 UT at Rabbit Lake which was located near the equatorward
boundary of the radar cusp (section 3.2) at that time. Although more
noisy, the riometer data from RABB was similar to that from GILL (only
0.5o south of RABB) showing relatively strong absorption (~1 dB) before
2000 UT. With the absorption being inversely proportional to the square
of the frequency this means that at 3 MHz the absorption would have been
about 21 dB. In agreement with the inferred absorption, there were
virtually no ionosonde echoes before 2000 UT seen by the RABB CADI
which had a rather limited dynamic range. At about 2000 UT, when the
equatorward boundary of the radar cusp moved near and possibly south
of RABB, the absorption decreased and the CADI data on the highest of
the fixed frequencies showed a few patches soon after they were formed
near the cusp. However, these CADI observations were still somewhat
fragmented because of the absorption. Half an hour later the patches were
observed drifting northeastward over Resolute Bay and a few more
minutes later (almost simultaneously) over Eureka. Between 1800 and
1900 UT the CADI signature in Eureka was weak but the 630-nm
emission showed irregular patches (not shown) associated with CADI
echoes. After the By IMF become strongly positive no patches were
observed by CADIs until ~1930 UT (suggesting that a change in the
polar cap convection prevented these patches to reach Resolute Bay and
Eureka. However, a couple of well-separated patches that were most
likely associated with the two sharp dips in the IMF By component at 1906
and 1916 UT drifted over Resolute Bay at 1942 and 1951 UT. About 5
minutes after the patches passed Resolute Bay weak optical signatures
were observed in Eureka in 630 nm (not shown) but the Eureka CADI
echoes were too weak at this time.
After 2000 UT, many bandlike patches that were observed with
SuperDARN (Figure 11b and Plate 1) were also observed with CADIs
(Figure 16) and the all-sky camera (Figure 17). For example, between
2100 and 2140 UT, CADIs in Resolute Bay and Eureka observed a series
of strong patches (G-K) that were seen with SuperDARN at lower latitudes
as bands moving eastward (Figure 13). The transit times of the optical
bands (Figure 17) across the zenith in Eureka (indicated by arrows in the
top panel of Figure 16) were correlated with patches observed with CADI
at Eureka. CADI measured horizontal plasma drift velocities for the
patches ranging between 500 and 1000 m/s northeastward to eastward.
The average velocity agreed with the mean translational motion observed
by SuperDARN (~600 m/s). Note that the latter speed was estimated for
the band centers as observed by SuperDARN. The poleward ends of these
bands moved faster as they rotated turning eastward with convection.
Figure 18 shows the ionograms, taken at 15-minute intervals by the
Resolute Bay CADI, over the period 2100-2145 UT. Approximate values
of fxF2 are shown for patches and/or troughs separating the patches. The
value of foF2 can be obtained by subtracting 0.8 MHz (half of the electron
gyrofrequency). At 2100 UT, oblique echoes from the patch G
approaching the zenith and a trough between patch F and G that were
recorded. At 2115 UT, the center of the patch H was overhead (foF2~8.5
MHz) indicating the maximum electron density of about 9x1011 m-3. This
is an increase by a factor of about 4.5 from the trough density estimated
in the previous ionogram. The ionogram at 2130 UT showed spread
echoes which were due to oblique echoes from edges of the patch J
(receding) and the patch K (approaching). Note that a couple of minutes
later the poleward end of the bands J and K swept over Eureka and were
also observed optically (Figure 17).
As mentioned above, the RABB CADI observed patches soon after they
formed near the cusp (before 2030 UT). At 2015 UT, a full ionogram (not
shown) was taken when the patch E was overhead. It showed almost the
same foF2 value (electron density) as that of the patch H over Resolute
Bay (Figure 18) suggesting that the patches were fully formed in the cusp.
Also, there is a possibility that the electron precipitation associated with
FLRs contributed to patches. However, as further discussed in section 4.4,
the electron precipitation alone is not sufficient to build up enough of the
patch ionization relative to the background. In fact, looking at quiet days
(November 30 and December 4) the maximum ionization density at RABB
between 2000 and 2100 UT would correspond to fxF2 of 8-9 MHz which
is not too different from the 9.4-MHz maximum shown for patch E
overhead at Rabbit Lake at 2015 UT. Therefore, a combination of a small
enhancement due to auroral precipitation on top of the solar EUV
ionization and redistribution of the plasma into enhancements and
depletions by means of ionospheric currents could explain the patches.
Further evidence for restructuring is that the ionization between patches
was much lower than quiet day values thus indicating that it had been
depleted (see section 4.4). For example, at 2100 UT the Resolute Bay
CADI ionogram showed an F-region peak density corresponding to about
6 MHz at the edge of a patch and 4 MHz in the overhead trough.
The peak electron densities of nearly 1012 m-3 (factor ~5 above the
background/trough density) are similar to previous measurements [e.g.,
Weber et al., 1984 and Rodger et al., 1994a]. However, the patches,
particularly those observed after 2000 UT, were distinctly bandlike in
appearance. 

3.10. The Spatial Extent of Bands/Patches
At 1750 UT, the riometer absorption enhancements extended from
BACK to FSMI and just before 1800 UT even farther to DAWS (>3
hours of local time). In the BARS FoV the absorption enhancements were
associated with irregularity bands. The observed F-region patches were
more irregular than those observed after 2000 UT and there was no clear
one-to-one correspondence between E-region bands and F-region patches
(see section 3.7). Also, the optical patches in Eureka were structured and
distorted (likely a result of the large amplitude IMF By-component
oscillation affecting the polar cap convection).
Figures 13 and 15 show several examples when the E- and F-region
radar (BARS and SuperDARN) backscatter bands were joined into one
structure extended in azimuth and altitude. The CADI observations in
Resolute Bay and Eureka indicated that, as the bands turned eastward,
they were extended poleward even farther than observed with the
SuperDARN radars. Furthermore, the F-region patches were imaged (630-
nm emission) from Eureka (Figure 17). 
Combining all these observations one can conclude that some of the
bands/patches were extended over ~3 hours of local time and/or 2500 km
in observed length. The widths of the backscatter bands were mostly
between 100 and 200 km except for the strongest band H which, at one
time, may have been over 300 km wide in the F region (Figure 13a). The
widths of the optical and density patches were similar and the separations
between neighboring patches were between 200 to 300 km. 

4. Discussion
4.1. Dayside Poleward Moving Auroral Forms
The dayside poleward moving auroral forms (PMAFs) are frequently
observed near the auroral oval around magnetic local noon (see e.g.
Sandholt et al., [1990] and Fasel [1995] and references therein). The arcs
appear near the equatorial edge of the auroral oval with a repetition rate
ranging from a few to several minutes, drift poleward spanning about two
degrees of latitude and then usually brighten and/or disappear. Vorobjev
et al. [1975] were first to suggest that these events are the manifestations
of the transfer of magnetic field lines from the dayside to the nightside,
which are now called FTEs. A statistical study by Fasel [1995] found that
the PMAFs have the mean lifetime of 5 min and the mean time between
successive PMAFs is 6 minutes. Such PMAF recurring rates, similar to
Pc5 periodicities, were also found by Lockwood et al. [1989] and Sandholt
et al. [1990] who called the events dayside auroral break up events or
simply cusp/cleft auroral activity [Sandholt et al., 1994]. McHarg and
Olson [1992] correlated such events with the ULF wave activity and
Leontyev et al. [1992] showed magnetometer data for one of their
"PMAF" event but did not discuss the associated Pc5 pulsation that can
clearly be identified in their Figure 5. Lockwood et al. [1989] found that
the mean repetition period of the ionospheric flow and auroral burst events
is about 8.5 min when the IMF is continuously and strongly southward.
They suggested that transient momentum exchange between the
magnetosheath and the ionosphere occurs quasi-periodically with a mean
period similar to that of FTE occurrence rate at the magnetopause.
While there were no optical data during the day time event presented
here, the similarities of the radar and riometer signatures (near cleft/cusp
locations, recurrence rates, lifetimes, and drift velocities) with those of
PMAFs are rather suggestive of poleward drifting auroral forms. The
BARS radar signatures alone (backscatter structure and the Doppler
spectra) combined with riometer observations provide a strong evidence
for the presence of auroral arcs. Even a more subtle detail of the motions
observed in the radar backscatter agrees with the results by Sandholt et
al., [1990] and Fasel [1995]. These authors described two phases of
PMAF evolution: In the phase I, the PMAF tends to elongate in the east-
west dimension and rapidly moves westward (progressing poleward
[Fasel, 1995]). During phase II, the PMAF just moves poleward and the
optical intensities fades. The BARS radar backscatter exhibited very
similar signatures: shortly after the bands appeared some of them showed
a westward moving intensification along the band. Later the bands just
drifted poleward and faded away. Another feature of PMAFs that was
reported in literature [Fasel et al., 1992] is a multiple brightening of a
PMAF. The Nipawin radar backscatter power (Figures 5a and 7b)
indicated that some bands (e.g., f, g, and h) showed a couple of
intensifications as they progressed poleward. A multiple flaring of
poleward drifting arcs and possibly even splitting of such arcs due to
higher FLR frequencies that were mentioned (section 3.5) but not further
discussed could explain the complex riometer signature (Figure 7). The
similarities between the radar observations of FLRs/FCEs and the 557.7-
nm emissions observed by Fasel et al. [1992] are further discussed in
paper 1. 
 
4.2. The Modulation of DPY currents on the closed field lines
Stauning et al., [1994 and 1995] reported observations of poleward
progressing DPY currents (IMF By-related dayside ionospheric
disturbances) in the polar cap region poleward of the cusp where the IMF
disturbance is directly coupled to the polar ionosphere by connecting
(newly merged) field lines. Their observations are further discussed in
paper 1 where it is suggested that the coupling of the IMF oscillations to
the closed field lines is also very efficient and the DPY currents are
modulated by FLR frequencies (which are similar or the same as those
found in the IMF). The event described in the present paper was similar
to those observed by Stauning et al., [1995]. The ground magnetometers
observed the IMF By-modulated DPY currents near, and poleward of, the
cusp (Figure 3). At these latitudes, the correlation between IMF By and
the ground X-component was good (Figure 10) and rather obvious from
a direct comparison of time series. In contrast, at ISLL the magnetometer
measuring the perturbations due to ionospheric (Hall) currents across the
closed field lines showed more complex fluctuations. The correlation
between the time series of IMF By and the ground X-component was not
always very clear (Figures 3 and 10). However, there was moderate to
very strong correlation in the frequency domain between the IMF By and
the ground X- and Z-components (Figures 9c and 8d). We conclude that
the IMF By-driven FLRs on magnetic shells equatorward of the cusp
modulated the DPY (Hall) currents on the closed field lines. This is
further discussed in paper 1 where it is suggested that there is no
fundamental difference between the IMF By-related disturbances poleward
and equatorward of the cusp except that the solar wind IMF oscillations
are directly transmitted to the ionosphere along the connecting "open"
field lines in the polar cap while the same oscillations drive FLRs on the
closed field lines that are associated with oscillating FACs feeding the
Hall (DPY) currents. These ionospheric disturbances intensify in the cusp
where the DPY current modulations coincide with FCEs (Figure 3a) on
the field lines that are being merged with IMF in FTEs (modulated by
FLRs on the outermost magnetic shells as discussed below and paper 1).
Because the low frequency IMF oscillations have the largest amplitudes
the DPY currents maximize on the freshly reconnected field lines
poleward of the cusp [Stauning et al., 1995]. 

4.3. A Link Between Pc5 FLRs and Dayside Magnetic Reconnection
The phenomena mentioned above (PMAFs, DPY currents, FCEs, and
polar patches) have been linked to the process of magnetic merging
(reconnection) at the dayside magnetopause which often appears to be
pulsed. In paper 1, it is shown that the magnetic reconnection is
modulated into pulses by FLRs. Similarly, the present results suggest such
reverse feedback from the resonating magnetic shells near the
magnetopause to the reconnection region at the magnetopause (see
schematic diagram in paper 1). The cross-correlation analysis of IMP-8
By component vs. the ground/ionospheric data in the time and frequency
domains suggest that the FLRs were directly driven from the solar wind.
As discussed in paper 1, the solar wind disturbance would propagate to
the resonant L shells via the fast compressional mode and then couple to
the shear Alfvén mode, thus driving the resonance. While the
magnetospheric magnetic field oscillations (due to FLRs) were
predominantly in the azimuthal direction they contained a significant radial
component as well. The amplitudes of the corresponding magnetic
oscillations on the ground that were discussed in section 3.4 exceeded 100
nT in the north-south direction (X component) but reached up to 50 nT in
the east-west direction (Y component). The electric field component of a
resonant Alfvén wave on the outermost closed magnetic flux tubes (see
paper 1) would modulate the magnitude of the electric field along the
reconnection line which is believed to be the most important measure of
the reconnection rate.
The above, and also a less likely possibility of the IMF oscillation
directly driving the pulsed reconnection (FTEs) causing compressional
mode wave that would excite the FLRs are discussed in paper 1.
However, the latter interpretation is not consistent with the correlation
results presented here. These results showed satellite-ground delays
(Figure 10) suggesting that FLRs were excited first and modulated the
reconnection process (FTEs) as suggested above. Also, the IMF FFT
power spectra were strongly correlated with the ground magnetic
perturbations at subcusp latitudes (ISLL). The FLRs that were excited on
the outermost magnetic shells and modulated magnetic reconnection also
contributed to polar patches formed by FCEs in the cusp.
 
4.4. Patch formation
The polar patches have been found breaking off from the cusp region
shortly after quasiperiodically occurring FCEs [Rodger et al., 1994a] and
drifted poleward in trains (series of patches separated by a few hundred
kilometers) [Foster, 1993; Rosenberg et al., 1993]. Lockwood et al.
[1993] observed the ionospheric signatures of pulsed FTEs and suggested
the role of low-energy precipitation in the polar patch formation.
Valladares et al. [1994] emphasized the role of increased recombination
rate [Banks et al., 1974] associated with enhanced ion temperature in a
fast plasma jet that results in a deep electron density depletion breaking
off a patch. These observations were supported by numerical simulations
[Valladares et al., 1996] showing that patches can be produced by plasma
jets associated with poleward progressing twin vortices superposed on a
standard background convection pattern. These results showed that the
plasma at the base of the tongue of ionization (TOI) can be structured into
elongated patches by drawing in low density plasma from earlier local
times and increasing the O+ loss rate thus producing a density depletion
adjacent to the patch. 
Because the observations presented in this paper appear to be consistent
in many aspects with all the above results we have to reconcile and/or
qualify some apparent differences between the above mentioned
"mechanisms" of patch formation. We showed that polar patches were
associated with Pc5 irregularity bands and the bulk of the ionization
forming the patch had its origin in the dayside auroral oval which includes
both the solar photoionization and auroral precipitation. One argument that
has sometimes been used against the electron precipitation as a source of
polar patch densities is that soft precipitation in the cusp cannot produce
the required patch densities during a relatively short time interval of
several minutes during which a drifting "patch" (that is being formed) is
exposed to precipitation. This is one of the reason why models of
fragmentation of solar produced plasma (TOI) drawn into polar cap were
sought to explain patches [see, e.g., Bowline et al., 1996; and references
therein]. Such structuring of the polar ionosphere is a valid hypothesis that
is now supported by an extensive modeling work showing that patches can
be generated this way. However, as the latter authors have stated, "the
origin of patches could be equatorward of the cusp region, in the cusp
region, or a combination of both". The time-dependent ionospheric model
(TDIM) used by these authors was extended to include particle
precipitation on the statistical basis.
We suggest that FLR associated precipitation in auroral arcs (PMAFs)
drifting into the cusp region combined with the cusp precipitation (at the
time or just before a patch is formed in the cusp) can build up significant
absolute and/or relative (with respect to the background) densities in
addition to the existing solar EUV ionization. This plasma is then
restructured into enhancements and depletions by means of ionospheric
currents and enhanced recombination in the F region. Also, the arcs are
likely to be associated with latitudinally narrow, field-aligned depletions
possibly formed by vertical evacuation in the downward FACs adjacent to
an arc [Opgenoorth et al., 1990] which could then lead to greater relative
densities observed in patches as well as steep density gradients. The
density depletions between the patches could then be further enhanced if
such arcs drifting across the cusp were associated with magnetic merging
at the dayside magnetopause driven by the IMF configuration in the solar
wind but modulated by FLRs that created the arcs on the closed field lines
near the magnetopause. The elongated twin vortex ionospheric signatures
of such events (FTEs) at the foot of a pair of upward and downward
FACs [Southwood et al., 1988; Scholer, 1988] would then be associated
with a disruption of the flow at the polar cap boundary (FCEs). FCEs are
twin vortices whose single radar signature sometimes shows the return
flows adjacent to the flow channel (see, e.g., Figure 1 in paper 1). Some
indication of the return flow can be seen in Figure 15 in the present paper.
These signatures of FTEs are likely to be associated with cusp
precipitation [Sandholt et al., 1990]. The electron precipitation in the
region of upward FAC on the poleward (By>0) or equatorward (By<0)
boundary of a flow channel would further increase the electron densities
in a patch that is being formed. Here we refer to a patch being formed or
further enhanced in the region of upward FACs associated with the FCE
(in the cusp) as opposed to a patch being simply detached from existing
TOI by the density depletion associated with the enhanced flow. The
plasma density can be depleted by enhanced O+ recombination in the F
region [Valladares et al., 1994] and/or by vertical evacuation of electrons
in a strong downward FAC [Doe et al., 1995] increasing the relative
density in the patch with respect to the electron density trough formed
adjacent to it. Electron density depletions were observed associated with
regions of enhanced perpendicular electric fields adjacent to auroral arcs
[Opgenoorth et al., 1990]. The incoherent scatter radar measurements
[Doe et al., 1993] demonstrated that auroral ionospheric cavities (AICs)
with depletions of the F-region densities by 40% or more can occur on
time scales as short as 4 min.
In general agreement with the conclusions by Lockwood et al. [1993]
we suggest that patches are generated by FTEs during times of southward
IMF. The magnetic reconnection at the dayside magnetopause is pulsed by
the IMF By-driven FLRs and patches are a by-product in this process. The
FLR driven arcs and/or associated density structure of remanent F-region
density enhancements/depletions are convected into the ionospheric cusp
where this plasma structure is reformed by cusp precipitation and FCEs
producing even more pronounced density enhancements/depletions. This
process will eliminate some while it may reinforce other parts of the short
period FLR structure that was imposed on the ionosphere prior the entry
to the cusp. For example, during the first event presented here (1800-1830
UT) the reconnection associated flows in the cusp was modulated by low
frequencies (1-2 mHz) thus reshaping the FLR imposed high frequency (4-
6 mHz) structure into patches that were more irregular. After 1930 UT,
the amplitudes of the latter high frequency components weakened and the
patches were produced by a (more harmonious) combined action of 1- and
2-mHz FLR modulation of flows and densities near the cusp which
resulted in patches that were identified with individual flow bursts and
FLR backscatter bands. 
Valladares et al. [1994] presented data that show further evidence in
support of the latter observations. We now discuss this evidence in some
detail because it was not discussed in these terms by the latter authors.
Some of the density contour maps [Valladares et al. 1994; their Figures
4b and 5b] obtained by the Sondrestrom incoherent scatter radar prior the
onset time of the fast plasma jet which they associated with enhanced ion
temperature and recombination clearly show poleward drifting local
density enhancements (~5x1011 m-3) a few hundred kilometers apart in
north-south direction of the elevation scans. These density enhancements
were most likely produced by FLR associated arcs. Their Figure 9 shows
the ground magnetic data which reveal 6- and 12-min pulsations in the Z-
component at the SKT and STF stations around 1300 UT while their
Figure 2 shows that at least the 6-min periodicity is quite distinct in the
IMF data. For the event of 19 Feb 1990 Valladares et al. [1994]
concluded that the density structure moved poleward with an average
velocity of 700 m/s. Before 1300 UT, as their Figure 15 shows, this
velocity was only ~600 m/s and this estimate agrees with the value of
620 m/s [Valladares et al., 1996] for the poleward displacement of the
plasma jet related Hall current observed by the ground magnetometers.
Assuming this latter value for the poleward drift and the spatial separation
of 450 km between two density enhancements in the elevation scan (their
Figure 4b) taken before the onset time of the plasma jet one can obtain a
time separation of 12 min between the overhead passages of these two
density enhancement which agrees well with the ~12-min magnetic
pulsation (Z-component) observed on the ground at STF and also SKT
(before 1300 UT). About 5.5 min later, the next elevation scan showed
that the density structure which was overhead in the previous scan had
shifted ~200 km poleward (at ~600 m/s) and another density
enhancement was approaching the zenith. This latter structure clearly
suggests a freshly formed auroral arc with the density enhancement and
a sharp horizontal density gradients still penetrating to the E-region.
Similarly, one can argue that the radar also observed 6-min/200-km F-
region structure (their Figure 8b) suggesting precipitation into arcs
produced by FLRs which appeared to be observed by the ground
magnetometers. It should be noted that the fast eastward plasma jet
[Valladares et al., 1994] was observed during conditions of the IMF
By<0. In contrast, our observations of westward FCEs (positioned
equatorward of patches that were formed in the region of upward FACs
possibly associated with cusp precipitation) were during times of By>0.
Because an FCE (ionospheric signature of an FTE) is an antisunward
moving twin vortex the simulations by Valladares et al. [1996] could
provide a very useful tool to model the ionospheric response to various
IMF conditions favorable to magnetic merging. However, this particular
modeling effort so far suggested that there is a preferential local time
(restricted to a period of 1000-1200 LT) at which vortices can generate
patches. The pulsed magnetopause reconnection events that are discussed
in the present paper and paper 1 generated patches well past the magnetic
local noon.
Also, we have associated the patch formation with poleward
progressing DPY currents (Hall currents associated with FCEs). A large
statistical data base of DPY current observations was accumulated by
Stauning et al. [1994]. The longitudinal and latitudinal extents of the DPY
Hall currents that are most intense in the region between the downward
and upward field-aligned current sheets are comparable with the observed
separations between the patches. The horizontal electric fields associated
with the convection enhancements (FCEs) result in the E-region electron
heating which can explain the radio wave absorption near the cusp
[Stauning et al., 1995]. A gradual fading of DPY currents as they traverse
the dayside polar cap could explain the riometer absorption associated with
polar patches [Rosenberg et al., 1993]. The absence of DPY events for
northward IMF is consistent with the absence of patches during such
periods [Rodger et al., 1994a]. If the link between the poleward
progressing DPY currents and polar patches is confirmed, this mechanism
of patch formation could be identified as a dominant source of patches,
considering the abundance of the By-modulated DPY events [Stauning et
al., 1994].
We have refered to several possible sources of ionization and/or
mechanisms for patch formation. Now we briefly summarize their relative
importance. The electron density given by the International Reference
Ionosphere (IRI) 1990 model for an altitude of 250 km above Gillam at
2000 UT near solar mininum (mean sunspot number 60) is 7x1011 m-3 and
drops to 4x1011 m-3 at 400 km. With the cusp and auroral oval sunlit the
solar EUV ionization is the main source of the large background density
on which auroral and cusp ionization is superposed [Roble and Rees,
1977]. Sojka and Schunk [1986] considered the effect of discrete auroral
precipitation on the density profile and found large density enhancements
(see their Figure 3) within a few minutes of a plasma flux tube entering
a region of auroral precipitation. However, their model was for solar
maximum conditions and moderate geomagnetic activity and thus not
directly applicable to the observations presented here. We have proposed
(see above) that patches are produced as a by-product of the FLR
modulated reconnection at the dayside magnetopause with the FLR
associated precipitation significantly contributing to patches. This is
consistent with the incoherent radar measurements [Valladares et al.,
1994] showing that the FLR precipitation structure the ionospheric density
before it is segmented into patches. The F-region density enhancements
over the "background" (trough density between the enhancements) that can
be inferred from the elevation scans [see Valladares et al., 1994; their
Figures 4, 5 and 7] range up to 50%. Furthermore, the latter authors
attributed the excess (35%) F-region densities in another patch being
formed (presumably in the cusp) to soft precipitation. 
The second mechanism we consider is redistribution of ionization by
currents. To estimate the electron densities we assumed electric field ±50
mV/m and a standard value for ion mobility of 50 (MKS units) at an
altitude of 250 km. From BARS data showing bands associated with FCEs
we estimated the wavelength of the electric field variation to be ~150 km
(2 cycles in 300 km). Also, we assume that it takes ~15 min for the
bands to traverse a 300-km wide cusp. Then the relative plasma
"relocation" by currents can be estimated from the charge continuity
equation to be ~10%. This is comparable to the auroral precipitation,
because it would cause both an enhancement and a depletion of density
(i.e., relative enhancement of 20%). Doe et al. [1995] modeled the
formation of auroral ionospheric cavities (AICs) showing that FACs are
very efficient at modifying the polar ionosphere. They showed that
moderate downward FACs (0.2 µA m-2) can create a 40% depletion in
time scales of 30 s, while the diffusion and chemistry operate on longer
time scales (hundreds of seconds). These authors suggested that "cavities
are created by closure of downward magnetospheric field-aligned currents
and that AIC-arc pairs are the imprint of an oppositely directed pair of
FACs". This mechanism of density depletion appears to be directly
applicable to FLRs which are associated with a pair of oscillatory FACs.
Similarly, the signature of an FTE footprint (FCE) implies oppositely
directed FACs [Southwood et al., 1988]. While the horizontal dimensions
of FCEs are about one order of magnitude larger than those of AICs
modeled by Doe et al., [1995], the latter authors showed that the narrow
and fast growing depletions can grow to larger widths by allowing more
time to build them up.
Finally, as discussed above, Valladares et al. [1996] modeled the
observed depletions in the local density by a factor of 6 [Valladares et al.,
1994] and showed that plasma jets (FCEs) associated with poleward
progressing twin vortices can carve out such density depletions. For a
given simulation, they attributed 38% of this depletion to the O+
recombination loss and 62% to transport of low density plasma from
earlier local times. It should be noted that their model applies to a TOI
that is already extending into the polar cap. The position (with respect to
the day/night terminator) of the ionospheric footprint of the cusp (an
important source of moving twin vortices, e.g., FTE signatures) vary
throughout the year. However, during the event presented in this paper,
the cusp was sunlit and the day/night terminator was well poleward of the
cusp. Also, the westward as opposed to eastward flow bursts were
observed, thus the latter source of plasma depletion (transport of low
density plasma from earlier local times) is not readily applicable for this
event.

5. Conclusions
Poleward drifting F-region backscatter bands observed with
SuperDARN HF radars, and patches observed with CADI ionosondes and
an all-sky imager in the polar cap were associated with the Pc5 field line
resonances (FLRs) near the dayside polar cap boundary. BARS VHF and
SuperDARN HF radars observed FLR signatures - poleward drifting E-
region irregularity bands that were associated with alternating Doppler
velocities and the riometer absorption enhancements. These FLR
signatures were extended to the cusp footprint as enhanced flows (flow
channel events, FCEs). The F-region ionosphere was structured into polar
patches by FCEs carving out electron depletions while the soft cusp
precipitation inferred from the riometer data augmented the density of the
patch formed poleward (IMF By>0) of the flow channel. The density
patches structured by irregularities (causing the radar backscatter)
produced at the equatorward edge of the patches had a large longitudinal
extent spanning ~3 hours of local time (~2500 km) but only ~200 km
wide. In agreement with the results presented in paper 1, it is suggested
that the FCEs are the ionospheric signatures of flux transfer events (FTEs)
modulated by FLRs at the dayside magnetopause. It is shown that large
amplitude (~100 nT) FLRs that were excited on L shells adjacent to the
magnetopause were directly driven by the solar wind IMF By oscillations.
These multi-mode IMF oscillations coupled to closed field lines in the
magnetosphere driving FLRs and modulating the DPY (Hall) currents
which intensified in the cusp footprint (FCEs). The DPY currents
progressed further poleward as they continued to be driven by large
amplitude and low frequency IMF oscillations. It is suggested that the
FLRs quasiperiodically perturbed the magnetopause and modified the
magnetic and electric field configurations near the reconnection region
thus modulating the reconnection into pulses. Polar patches were a by-
product of this process. It is concluded that the solar EUV ionization
significantly augmented by the auroral/cusp ionization was reformed into
bandlike patches by FCEs redistributing the ionization into enhancements
and depletions by means of ionospheric/field aligned currents, while the
field lines were being reconnected to the IMF and convected into the polar
cap. Also, as previously reported, it is likely that the F-region O+
recombination loss associated with the enhanced flows further depleted the
trough densities between patches.

Acknowledgements. The CANOPUS program is a project of the
Canadian Space Agency. The CADI and polar camera installations and
operations were funded by the Canadian Network of Space Research. The
Saskatoon-Kapuskasing SuperDARN pair of radars is a collaboration of
Canada, France and United States. The Kapuskasing radar is supported by
NASA Grant NAG5-1099. The Saskatoon radar is funded as an NSERC
CSP. We acknowledge the contributions by R. Lepping (NASA, Goddard
Space Flight Center) who made available the IMP-8 IMF data and by F.
Rich (Phillips Laboratory, Hanscom Air Force Base) who provided us
with DMSP satellite data and helped with their interpretation. We
gratefully acknowledge use of the Arctic Stratospheric Ozone Laboratory
(Eureka, NWT), a facility of Environment Canada's Atmospheric
Environment Service, and instrument operation support provided by the
University of Saskatchewan. One of the authors (PP) acknowledges the
contribution by F. Fenrich (University of Alberta) making available the
FLR analysis software for SuperDARN data, A. McNamara and D. Wallis
(National Research Council of Canada) who assisted in evaluation of the
BARS and riometer data, respectively. We also benefited from discussions
with D. McDiarmid, T. Hughes, G. James, R. Jenkins, A. Rodger, J.
Samson, and A.D.M. Walker. 











References
Anderson, D.N., J. Buchau, and R.A. Heelis, Origin of density
enhancements in the winter polar cap ionosphere, Radio Sci., 23, 513-519,
1988.
André, D., A.G. McNamara, D.D. Wallis, B.A. McIntosh, T.J. Hughes,
G.J. Sofko, J.A. Koehler, D.R. McDiarmid, P. Prikryl, and J.
Watermann, Diurnal Radio Aurora Variations at 50 MHz Measured by the
Bistatic Auroral Radar System Radars, J. Geophys. Res., 93, 8651-8661,
1988.
Baker, K.B., J.R. Dudeney, R.A. Greenwald, M. Pinnock, P.T. Newell,
A.S. Rodger, N. Mattin, and C.-I. Meng, HF radar signatures of the cusp
and low-latitude boundary layaer, J. Geophys. Res., 100, 7671-7695,
1995.
Banks, P.M., R.W. Schunk, and W.J. Raitt, NO+ and O+ in the high
latitude F-region, Geophys. Lett., 1, 239-242, 1974.
Banks, P. M., T. Araki, C. R. Clauer, J.-P. St-Maurice, and  J. C.
Foster, The interplanetary electric field, cleft currents, and plasma
convection in the polar caps, Planet. Space Sci., 32, 1551-1557, 1984.
Basu, S., Su. Basu, P.K. Chaturvedi, and C.M. Bryant, Jr., Irregularity
structures in the cusp/cleft and polar cap regions Radio Sci., 29, 195-207,
1994.
Basu, S., Su. Basu, J.J. Sojka, R.W. Schunk, and E. MacKenzie,
Macroscale modelling and mesoscale observations of plasma density
structures in the polar cap, Geophys. Res. Lett., 22, 881-884, 1995.
Bowline, M.D., J.J. Sojka, and R.W. Schunk, Relationship of theoretical
patch climatology to polar cap patch observations, Radio Sci., 31, 635-
644, 1996. 
Buchau, J., and B.W. Reinisch, Electron density structures in the polar
F region, Adv. Space Res., 11(10), 29, 1991.
Buchau, J., B.W. Reinisch, E.J. Weber, and J.G. Moore, Structure and
dynamics of the winter polar cap F region, Radio Sci., 18, 995-1010,
1983. 
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,
325-338, 1985.
Buneman, O., Excitation of field aligned sound waves by electron
streams, Phys. Rev. Lett., 10, 285-287, 1963.
Chen, L., and A. Hasegawa, A theory of long-period magnetic pulsations,
1, Steady state excitation of field line resonance, J. Geophys. Res., 79,
1024-1032, 1974.
Christensen-Dalsgaard, J. D.O. Gough, and J. Toomre, Seismology of the
sun, Science, 229, 923-931, 1985.
Clauer, C.R., and P.M. Banks, Relationship of the interplanetary electric
field to the high-latitude ionospheric electric field and currents:
Observations and model simulation, J. Geophys. Res., 91, 6959-6971,
1986.
Clauer, C.R, P.M. Banks, A.Q. Smith, T.S. Jorgensen, E. Friis-
Christensen, S. Vennerstrom, V.B. Wickwar, J.D. Kelly, and J. Doupnik,
Observation of interplanetary magnetic field and of ionospheric plasma
convection in the vicinity of the dayside polar cleft, Geophys. Res. Lett.,
1, 891-894, 1984.
Clauer, C.R., P. Stauning, T.J. Rosenberg, E. Friis-Christensen, P.M.
Miller, and R.J. Sitar, Observations of a solar-wind-driven modulations
of the dayside ionospheric DPY current system, J. Geophys. Res., 100,
7697-7713, 1995.
Cowley, S.W.H, and M. Lockwood, Excitation and decay of solar-wind
driven flows in the magnetosphere-ionosphere system, Annales Geophys.,
10, 103-115, 1992.
Cowley, S.W.H., J.P. Morelli, and M. Lockwood, Dependence of
convective flows and particle precipitation in the high-latitude dayside
ionosphere on the X and Y components of the interplanetary magnetic
field, J. Geophys. Res., 96, 5557-5564, 1991.
Crooker, N.U., Dayside merging and cusp geometry, J. Geophys. Res.,
84, 951-959, 1979.
Crooker, N.U., F.R. Toffoletto, and M.S. Gussenhoven, Opening the
cusp, J. Geophys. Res., 98, 263, 1991.
Doe, R.A., M. Mendillo, J.F. Vickrey, L. Zanetti, and R. Eastes,
Observations of nightside auroral cavities, J. Geophys. Res., 98, 263,
1993.
Doe, R.A. and J.F. Vickrey, Electrodynamic model for the formation of
auroral ionospheric cavities, J. Geophys. Res., 100, 9683-9696, 1995.
Farley D.T., A plasma instability resulting in field-aligned irregularities
in the ionosphere, J. Geophys. Res., 68, 6083-6097, 1963.
Farley D.T. and B.G. Fejer, The effect of the gradient drift term on the
type 1 electrojet irregularities, J. Geophys. Res., 80, 3087-3090, 1975.
Fasel, G.J., Dayside poleward moving auroral forms: A statistical study,
J. Geophys. Res., 100, 11891-11905, 1995.
Fasel, G.J., J.I. Minow, R.W. Smith, C.S. Deehr, and L.C. Lee,
Multiple brightenings of transient dayside auroral forms during oval
expansions, Geophys. Res. Lett., 12, 2429-2432, 1992.
Fenrich, F.R., J.C. Samsom, G. Sofko, and R.A. Greenwald, ULF high
and low-m field line resonances observed with SuperDARN, J. Geophys.
Res., 100, 21535-21547, 1995.
Foster, J.C., Storm time plasma transport at middle and high latitudes, J.
Geophys. Res., 98, 1675-1689, 1993.
Foster, J.C., and J.R. Doupnik, Plasma convection in the vicinity of the
dayside cleft, J. Geophys. Res., 89, 9107-9113, 1984.
Friis-Christensen, E., and J. Wilhjelm, Polar cap currents for different
directions of the interplanetary magnetic field in the Y-Z plane, J.
Geophys. Res., 80, 1248-1260, 1975.
Grant, I.F., J.W. MacDougall, J.M. Ruohoniemi, W.A. Bristow, G.J.
Sofko, J.A. Koehler, D. Danskin, and D. André, Comparison of plasma
flow velocities determined by the ionosonde Doppler drift technique,
SuperDARN radars, and patch motion, Radio Sci., 1537-1549, 1995.
Greenwald, R.A., K.B. Baker, J.M. Ruohoniemi, J.R. Dudeney, M.
Pinnock, N. Mattin, J.M. Leonard, and R.P. Lepping T.B., Simultaneous
conjugate observations of dynamic variations in high-latitude dayside
convection due to changes in IMF By, J. Geophys. Res., 95, 8057-8072,
1990.
Greenwald, R.A., K.B. Baker, J.R. Dudeney, M. Pinnock, T.B. Jones,
E.C. Thomas, J.-C. Villain, J.-C. Cerrisier, C. Senior, C. Hanuise, R.D.
Hunsucker, G. Sofko, J. Koehler, E. Nielsen, R. Pellinen, A.D.M.
Walker, N. Sato, and H. Yamagishi, DARN/SUPERDARN A global view
of the dynamics of high-latitude convection, Space Sci. Rev., 71, 761-796,
1995.
Hughes, T.J., and G. Rostoker, Current flow in the magnetosphere and
ionosphere during periods of moderate activity, J. Geophys. Res., 82,
2271-2282, 1977. 
Leontyev, S.V., G.V. Starkov, V.G. Vorobjev, V.L. Zverev, and Ya.I.
Feldstein, Dayside aurorae and their relation to other geophysical
phenomena, Planet. Space Sci., 40, 621-639, 1992.
Lockwood, M., and H.C. Carlson, Jr., Production of polar cap electron
density patches by transient magnetopause reconnection, Geophys. Res.
Lett., 19, 1731-1734, 1992. 
Lockwood, M., P.E. Sandholt, S.W.H. Cowley, and T. Oguti,
Interplanetary magnetic field control of dayside auroral activity and the
transfer of momentum across the dayside magnetopause, Planet. Space
Sci., 67, 1347-1365, 1989. 
Lockwood, M., W.F. Denig, A.D. Farmer, V.N. Davda, S.W.H.
Cowley, and H. Luhr, Ionospheric signatures of pulsed reconnection at the
Earth's magnetopause, Nature, 361, 424-428, 1993.
MacDougall, J.W., I.F. Grant, and X. Shen, The Canadian Advanced
Digital Ionosonde: Design and Results, in "Report UAG-14: Ionospheric
Networks and Stations", World Data Center A for Solar-Terrestrial
Physics, 21, 1995. 
McHarg, M.G., and J.V. Olson, Correlated optical and ULF magnetic
observations of the winter cusp - Boundary layer system, Geophys. Res.
Lett., 19, 817-820, 1992.
McNamara, A.G., D.R. McDiarmid, G.J. Sofko, J.A. Koehler, P.A.
Forsyth, and D.R. Moorcroft, BARS - A dual bistatic auroral radar
system for the study of electric fields in the Canadian sector of the auroral
zone, Adv. Space Res., 2, 145-148, 1983.
Newell, P.T. and Ch.-I. Meng, Ionospheric projections of magnetospheric
regions under low and high solar wind pressure conditions, J. Geophys.
Res., 99, 273-286, 1994. 
Newell, P.T., D.G. Sibeck, and Ch.-I. Meng, Penetration of the
interplanetary magnetic field By and magnetosheath plasma into the
magnetosphere: Implications for the predominant magnetopause merging
site, J. Geophys. Res., 100, 235-243, 1995.
Opgenoorth, H.J., I. Haggstrom, P.J.S. Williams, and G.O.L. Jones,
Regions of strongly enhanced perpendicular electric fields adjacent to
auroral arcs, J. Atmos. Terr. Phys., 52, 449-458, 1990.
Papitashvili, C.R. Clauer, A.E. Levitin and B.A. Belov, Relationship
between the observed and modeled modulation of the dayside ionospheric
convection by the IMF By component, J. Geophys. Res., 100, 7715-7722,
1995.
Pinnock, M., A.S. Rodger, J.R. Dudeney, K.B. Baker, P.T. Newell,
R.A. Greenwald, and M.E. Greenspan, Observations of an enhanced
convection channel in the cusp ionosphere, J. Geophys. Res., 98, 3767-
3776, 1993.
Pinnock, M., A.S. Rodger, J.R. Dudeney, F. Rich, and K.B. Baker, High
spatial and temporal resolution observations of the ionospheric cusp, Ann.
Geophys, 13, 919-925, 1995.
Prikryl, P. and L.L. Cogger, Statistical analysis of the spatial relationship
between the radio and optical aurora: Further evidence for refraction,
Radio Sci., 27, 469, 1992.
Prikryl, P., G.J. Sofko, J.A. Koehler, J. Mu, and D. André, Steep
electron density gradients associated with discrete aurorae explaining the
diversity of auroral VHF doppler spectra, Annales Geophysicae, 13,
25-37, 1995. 
Roble, R.G. and M.H. Rees, Time-dependent studies of the aurora:
Effects of particle precipitation on the dynamic morphology of ionospheric
and atmospheric properties, Planet. Space. Sci., 25, 991-1010, 1977.
Rodger, A.S., M. Pinnock, J.R. Dudeney, K.B. Baker, and R.A.
Greenwald, A new mechanism for polar patch formation, J. Geophys.
Res., 99, 6425-6436, 1994a. 
Rodger, A.S., M. Pinnock, J.R. Dudeney, J. Watermann, O. de la
Beaujardiere, and K.B. Baker, Simultaneous two hemisphere observations
of the presence of polar patches in the nightside ionosphere, Ann.
Geophysicae, 12, 642-648, 1994b. 
Rosenberg, T.J., Z. Wang, A.S. Rodger, J.R. Dudeney, and K.B. Baker,
Imaging riometer and HF radar measurements of drifting F region electron
density structures in the polar cap, J. Geophys. Res., 98, 7757-7764,
1993. 
Rostoker, G., J.C. Samson, F. Creutzberg, T.J. Hughes, D.R.
McDiarmid, A.G. McNamara, A. Vallance Jones, D.D. Wallis, L.L.
Cogger, CANOPUS - A ground-based instrument array for remote sensing
the high-latitude ionosphere during the ISTP/GGS program, Space Sci.
Rev., 71, 743-760, 1995.
Ruohoniemi, J.M., R.A. Greenwald, K.B. Baker, and J.C. Samson, HF
radar observations of Pc5 field line resonances in the midnight/early
morning MLT sector, J. Geophys. Res., 96, 15697-15710, 1991.
Samson, J.C., R.A. Greenwald, J.M. Ruohoniemi, T.J. Hughes, and
D.D. Wallis, Magnetometer and radar observations of
magnetohydrodynamic cavity modes in the earth's magnetosphere, Can.
J. Phys., 69, 929, 1991. 
Samson, J.C., B.G. Harrold, J.M. Ruohoniemi, and A.D.M. Walker,
Field line resonances associated with MHD waveguides in the
magnetosphere, Geophys. Res. Lett., 19, 441-444, 1992a.
Samson, J.C., D.D. Wallis, T.J. Hughes, F. Creutzberg, J.M.
Ruohoniemi, and R.A. Greenwald, Substorm intensifications and field line
resonances in the nightside magnetosphere, J. Geophys. Res., 97,
8495-8518, 1992b. 
Sandholt, P.E., M. Lockwood, T. Oguti, S.W.H. Cowley, K.S.C.
Freeman, B. Lybekk, A. Egeland, and D.M. Willis, Midday auroral
breakup events and related energy and momentum transfer from the
magnetosheath, J. Geophys. Res., 95, 1039, 1990.
Sandholt, P.E., C.J.Farrugia, L.F. Burlaga, J.A. Holtet, J. Moen, B.
Lybekk, B. Jacobsen, D. Opsvik, A. Egeland, R. Lepping, A.J. Lazarus,
T. Hansen, A. Brekke, and Friis-Christensen, Cusp/cleft auroral activity
in relation to solar wind dynamic pressure, interplanetary magnetic field
Bz and By, J. Geophys. Res., 99, 17323-17342, 1994.
Scholer, M., Magnetic flux transfer at the magnetopause based on single
X line bursty reconnection, Geophys. Res. Lett., 15, 291-294, 1988. 
Smith, M. F., M. Lockwood, and S. W. H. Cowley, The statistical cusp:
A flux transfer event model, Planet. Space Sci., 40, 1251-1268, 1992.
Sojka, J.J., and R.W. Schunk, A theoretical study of the production and
decay of localized electron density enhancements in the polar ionosphere,
J. Geophys. Res., 91, 3245-3253, 1986.
Sojka, J.J., M.D. Bowline, R.W. Schunk, D.T. Decker, C.E. Valladares,
R. Sheehan, D.N. Anderson, and R.A. Heelis, Modeling polar cap F-
region patches using time varying convection, Geophys. Res. Lett., 20,
1783-1786, 1993. 
Southwood, D.J., Some features of field line resonances in the
magnetosphere, Planet. Space Sci., 22, 483-491, 1974.
Southwood, D. J., C. J. Farrugia, and M. A. Saunders, What are flux
transfer events?, Planet. Space Sci., 36, 503-508, 1988.
Stauning, P., Absorption of cosmic noise in the E-region during electron
heating events. A new class of riometer absorption events, Geophys. Res.
Lett., 11, 1184, 1984.
Stauning, P., E. Friis-Christensen, O. Rasmussen, and S. Vennerstrom,
Progressing polar convection disturbances: Signature of a open
magnetosphere, J. Geophys. Res., 99, 11303-11317, 1994.
Stauning, P., C.R. Clauer, T.J. Rosenberg, E. Friis-Christensen, and R.
Sitar, Observations of solar-wind-driven progression of interplanetary
magnetic field By-related dayside ionospheric disturbances, J. Geophys.
Res., 100, 7567-7585, 1995.
Steele, D. P. and L.L. Cogger, Polar patches and the "tongue of
ionization", Radio Sci., in press, 1996.
St.-Maurice, J.-P., P. Prikryl, G.J. Sofko, A. Kustov, J.A. Koehler, D.
Danskin, A. Hamza, and J. Chen, On the origin of narrow non-ion-
acoustic coherent radar spectra in the high-latitude E region, J. Geophys.
Res., 99, 6447-6474, 1994.
Taguchi, S., M. Sugiura, J.D. Winningham, and J.A. Slavin,
Characterization of the IMF By-dependent field-aligned currents in the
cleft region based on DE 2 observations, J. Geophys. Res., 98,
1393-1407, 1993.
Tsunoda, R.T., High-latitude F region irregularities: A review and
synthesis, Rev. Geophys., 26, 719-760, 1988.
Valladares C.E., S. Basu, J. Buchau, and E. Friis-Christensen,
Experimental evidence for the formation and entry of patches into the
polar cap, Radio Sci, 29, 167-194, 1994.
Valladares C.E., D.T. Decker, R. Sheehan, and D.N. Anderson,
Modeling the formation of polar cap patches using large plasma flows,
Radio Sci, 31, 573-593, 1996.
Vorobjev, V.G., G. Gustafsson, G.V. Starkov, Y.I. Feldstein, and N.F.
Shevnina, Dynamics of day and night aurora during substorms, Planet.
Space Sci., 23, 269-278, 1975.
Walker, A.D.M., R.A. Greenwald, and K.B. Baker, HF radar
observations of pulsations near the magnetospheric cusp, J. Geophys.
Res., 91, 8919-8928, 1986.
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. 
Ziesolleck, C.W.S., and D.R. McDiarmid, Auroral latitude Pc 5 field line
resonances: Quantized frequencies, spatial characteristics, and diurnal
variation, J. Geophys. Res., 99, 5817-5830, 1994.
Ziesolleck, C.W.S., and D.R. McDiarmid, Statistical survey of auroral
latitude Pc5 spectral and polarization characteristics, J. Geophys. Res.,
100, 19299-19312, 1995.






Figure captions

Plate 1. Range versus time plot (top) and radar maps (bottom) of spectral
widths. Radar cusp can be identified by large widths (particularly between
1900 and 2000 UT) while the region of small widths (<100 m/s) just
equatorward of it can be associated with low latitude boundary layer
(LLBL). The spectral widths in the patches drifting antisunward well
inside the polar cap are also significantly reduced.
    
Figure 1. The locations and fields of view of the SuperDARN
Kapuskasing-Saskatoon and BARS Nipawin-Red Lake radars projected
onto a map that is also showing some of the CANOPUS and CADI sites.
The polar all-sky imager was located in Eureka.

Figure 2. The IMP-8 interplanetary magnetic field.

Figure 3. (a) The X-component ground magnetic perturbations for the
CANOPUS North-South magnetometer array. The bottom trace is the
smoothed (solid) and unsmoothed (broken) IMP-8 IMF By component.
Poleward progressing disturbances (DPY currents) are indicated. Thick
upward arrows indicate corresponding phases of long-period variations of
IMF By component and ground magnetic perturbation (X component). Thin
downward arrows indicate times of flow bursts (FCEs) observed with
SuperDARN and BARS.  (b) The Z-component ground magnetic
perturbations for the CANOPUS North-South magnetometer array. The
horizontal bars indicate two field line resonance pulsation periods
discussed in the text.

Figure 4. The latitude profiles of magnetic perturbation used to estimate
the position of the poleward edge of the eastward DPY (Hall) current
system (indicated by an arrow).

Figure 5. (a) Range-versus-time plot of the Nipawin radar backscatter
power. The slanting arrows indicate poleward progressing Pc5 FLR
bands. The vertical downward arrows indicate observations of flow
channel events (FCEs). The IMP-8 IMF By component shifted by 5 min
is shown. (b) An approximate position of the poleward edge of the
eastward DPY (Hall) currents is mapped along the Nipawin radar beam
#1 (leftmost beam).

Figure 6. The riometer absorption (increasing downward relative to the
undisturbed background indicated by zero) at Gillam. The arrows indicate
the times of the overhead passage of the BARS backscatter bands over
Gillam.

Figure 7. (a) The Red Lake radar backscatter power. Poleward drifting
backscatter bands a-h are clearly resolved at small ranges while the
strongest radar backscatter is side-lobe contaminated. (b) FCHU and
BACK riometer absorption (detrended) superposed on the RTI plot of the
Nipawin radar backscatter power (enlarged portion of Figure 5a). The
corresponding ranges of FCHU and BACK are shown by horizontal
arrows and dotted lines. (c) ESKI riometer absorption (detrended) and
IMP-8 IMF By are superposed on the Doppler velocity plot for the
Nipawin radar velocities showing flow bursts near the cusp associated with
riometer absorption and flow channel events observed with SuperDARN
in the cusp.

Figure 8. BARS Red Lake radar maps of the mean backscatter power and
line-of-sight velocities for the Red Lake (top panels). Location of GILL
and FCHU CANOPUS sites are shown and the individual bands are
labeled. Broken lines represent likely positions of auroral arcs indicated
by riometer absorption observed when the structures moved overhead in
GILL. BARS spectra (bottom panels; add 600 km to the aliased range to
obtain the true slant range) near the cusp (band D). Both the frequency
and range aliasing can be resolved by examining the mean power and
Doppler obtained 30 s earlier (top panels).

Figure 9ab. Normalized 16- and 32-point FFT power spectra of IMP-8
IMF, ground magnetometer and BARS backscatter data for event 1 (a)
and event 2 (b). BARS backscatter power at range gates near BACK (left
panel) and GILL (right panel) is used.

Figure 9cd. The 64-point FFT power spectra of the IMP-8 IMF (c) and
the ground magnetic components at ISLL (d). The low resolution FFT (an
average of four 16-point FFTs of By time series) envelopes the
corresponding high resolution spectrum.

Figure 10. Time delays between the IMF By component and the
ionospheric/ground signals obtained from CCFs computed for 30-min
windows stepped by 15 min are plotted. The vertical lines represent the
zero delays and/or the start time for the corresponding window. The
delays for the BARS power (open symbols) and CANOPUS magnetometer
X components (solid symbols) are plotted for two levels of the correlation
coefficient. The largest values of the cross-correlation coefficient for
BARS (first row) and MARIA (second row) are printed at the top for each
window. The first set of points (squares) is for the interval of 1800-1830
UT, the second set of points (triangles) is for the interval of 1815-1845
UT, etc. A generic scale for time lags is shown (bottom-right). Also, the
minimum and average values of BARS delays are printed.

Figure 11. SuperDARN RTI plots for Kapuskasing (a) and Saskatoon (b)
radars before and after 2000 UT, respectively. The E-region backscatter
bands and corresponding F-region patches are labeled.

Figure 12. Contour plots of 32-point fast Fourier transform (FFT) power
spectra (top panel) of the line-of-sight velocity time series at several range
gates for beam #11 of Saskatoon SuperDARN radar. Also, latitude
profiles of the spectral power and phase at 2 mHz are shown (bottom
panel). 

Figure 13. The Saskatoon SuperDARN radar maps of backscatter power
showing the location of patches (G-K) at (a) 2108 UT and (b) 2114 UT.
The positions of the BARS E-region backscatter bands are shown
superposed in the BARS field of view.

Figure 14. Polar cap convection observed by the Saskatoon-Kapuskasing
pair of radars.

Figure 15. The Kapuskasing SuperDARN radar maps of line-of-sight
velocities. (a) Flow channel events (FCEs) is associated with the BARS
E-region backscatter bands f and g. Symbols in the BARS field of view
indicate the position of the latter band. (b,c) Two other FCEs are
associated with the E-region backscatter bands C and D.

Figure 16. The fixed frequency ionograms (virtual height vs time)
obtained in Rabbit Lake, Resolute Bay and Eureka. The arrows in the
bottom two panels show polar patches drifting over Rabbit Lake and
Resolute Bay identified with SuperDARN backscatter bands. The arrows
in the top panel indicate times when optical (630-nm) patches passed
overhead in Eureka (Figure 17).

Figure 17. The 630-nm emission images from the polar camera in Eureka
showing patches/bands associated with CADI and SuperDARN
observations.

Figure 18. The Resolute Bay CADI ionograms associated with polar cap
patches and troughs between patches.