Irregularities have been observed on the plasmapause by radio wave techniques and by in situ density, electric field, and magnetic field measurements. We review these data sets and show that Alfvén waves are present and could lead to the observed fluctuations through mixing of the gradient in plasma density at the plasmapause.
Jacobson and Erickson used the Very Large Array to study small variations in the total electron content (TEC) using radio star emissions. By studying how the scintillation pattern moved across the array, they were able to measure the trace velocity of the irregularities. They found velocities typical of gravity waves, implying that the waves generate irregularities with much smaller wavelengths than the waves themselves exhibit. This is a nice result but they also discovered structures moving at a much higher velocity. They eventually concluded that they were detecting irregularities at the edge of the plasmasphere which, to first order, were co-rotating with the Earth. The large distance to the irregularities accounts for the rapid transit across the array. TEC variations the order of 1014 m−2 were found.
LeDocq et al.  reported irregularities that were also near the plasmapause using the upper hybrid frequency measured with a wave receiver on the CRRES satellite. In this paper, we include measurements of the electric and magnetic field fluctuations [see Wygant et al., 1992] during the same event. We believe that these CRRES observations are of the same phenomenon reported by Jacobson and Erickson . Magnetic activity was high at this time with the AE index reaching 200 nT. The coordinate system is GSM with the x-axis toward the sun, the y-axis dawn-to-dusk, and the z-axis roughly parallel to the magnetic field. The x-axis is the satellite spin axis. Thus, the electric field in the spin plane (the only components measured) has components parallel and perpendicular to the plasmapause at the MLT. The magnetic field is in the z-direction.Menietti and Yoon have also reported density striations and electrostatic waves in a plasmapause-related density cavity.
2. Data Presentation
 The time series in Figure 1 shows a crossing of the plasmapause with irregularities between L = 3 and L = 4 on 20 September 1990, corresponding to a magnetic local time between 8–10 MLT. The largest structure had a δn/n of over 50%. Typical scale sizes are 0.1 Re or about 600 km. Figure 2 is an enlargement of the data and Figure 3 is its power spectrum. At scales less than 600 km, there is a power law with roughly a −5/3 slope. Typical density fluctuations were 20%. Note that these figures are slightly modified from LeDocq et al. , which has a more complete instrument description.
 In Figure 4we present the two components of the electric field measured at the same time in the y-z plane. Typical field strength is 0.5 mV/m. The magnetic field was also determined and is plotted inFigure 5 with fluctuations of 5 nT in the same region as the largest irregularity (near 21.45 UT). All three components of δB are found, indicating both transverse and compressional modes. For reference, the Alfvén speed is plotted in Figure 6, assuming a hydrogen plasma.
 The ratio of electric (Ey) and magnetic field strength (δB) is presented as a function of frequency in the top panel of Figure 7. For an Alfvén wave precipitating parallel to Bo, this should equal the Alfvén speed. This is in good agreement with Figure 6 and strongly argues for the existence of such a wave mode, at least in the time period near 21.45. However, since the ion thermal speed is comparable to the Alfvénic speed outside the plasmapause, this test cannot rule out the compressional mode. The lower panel gives the coherence, which is 0.75, and the phase, which is 90 degrees at 0.01 Hz and 180 at 0.014 Hz. A phase difference of 90 degrees implies a standing wave. The 90 s periodicities seem to be an Alfvén wave.
 Alfvén waves with a zonal electric field will cause radial drifts that can mix the plasmapause gradient. The growth rate in that case can be deduced from the continuity equation
where V is the E × B drift and ∇ is the radial gradient in plasma density with length scale R. Here we ignore production and loss and set ∇0V = 0, which is appropriate for E × B drifts [Kelley, 1986]. Then the time scale, τ, for mixing is τ = R/VR. Here VR is the radial component of the E × B drift. Since V≅ 1000 m/s and R≅ 0.5 Re, τ = 3 ms−1, which is fast enough for mixing to be a plausible notion. From the continuity equation, we can also estimate δn/n from
where ω is the wave frequency of about 0.05 rad/s, so δn/n = 5 × 10−3. This is lower than observed but is only an estimate. This is the fluctuation in one wave period. However, if zonal drifts are included and some randomness is included to make the physics nonlinear, the fluctuations will build up in time.
 In the local time studied by LeDocq et al. , the Kelvin-Helmholtz instability is stable, unlike the dusk-to-midnight period [Kelley, 1986; Viñas and Madden, 1986]. This mixing is likely creating the observed irregularities. For the observed fluctuations, a mean density of 108 m−3, a 20% fluctuation over 0.5 Re yields a TEC of 0.6 × 1012 m−2, which is comparable to the observations of Jacobson and Erickson .
 Research at Cornell University was supported by the National Science Foundation under grant ATM-0551107 and by the Office of Naval Research under grant N00014-09-1-0975.
 The Editor thanks two anonymous reviewers for their assistance in evaluating this paper.