Energy derived from the solar wind is coupled to ULF waves and transients by a number of mechanisms, first at the Earth's bow shock, where reflected ions generate ``upstream waves,'' and subsequently in the magnetosheath region in which the flow of the shocked solar wind is diverted away from the Earth's magnetospheric cavity. Locally generated magnetosheath waves and variations with a wide range of periods also impinge on the magnetopause. Some of these drive perturbations inside the magnetosphere, and the resulting ULF waves can be detected both in space and on the ground.
Our understanding of the waves in these upstream regions has taken a near quantum leap during these past four years. A considerable number of observational studies, based primarily on observations from the Active Magnetospheric Particle Tracer Explorer (AMPTE) and International Sun-Earth Explorer (ISEE) satellites, have for the first time provided a fairly complete characterization of the subsolar regions, and have stimulated considerable theoretical work as well. Readers are referred to reviews by Le and Russell [1994] for coverage of observations of ULF waves in the foreshock and by Krauss-Varban [1994] for coverage of the microscopic kinetic aspects of upstream wave generation, transmission and conversion at the bow shock, propagation in the magnetosheath, and interactions at the magnetopause, using both simulations and theory.
We begin here with ULF waves in the subsolar magnetosheath and at the dayside magnetopause (reviewed by Song [1994] and Anderson [1994b]). Identification of the modes of magnetosheath waves has proved quite difficult, and a variety of observational parameters have been suggested for empirical discrimination between especially the various compressional wave modes (e.g., Gary and Winske [1992]; Song et al. [1994b]; Anderson [1994b]; and Denton et al. [1994c]). Controversy continues at lower frequencies; for example, an event near the magnetopause identified as ``slow mode'' using the criteria of Song et al. [1994b] was identified as ``mirror mode'' using the criteria of Denton et al. [1994c].
Mode identification is critically dependent on adequate theoretical
descriptions of the waves, and exact formulations of these are in
themselves difficult to develop. Krauss-Varban et al. [1994]
clarified the differences to be expected between particle-based
kinetic (Vlasov) and fluid-based Hall-MHD (magnetohydrodynamic)
descriptions of waves with frequency 
(the proton
gyrofrequency) and showed that the fluid and kinetic descriptions do
not correspond well when the plasma beta (the ratio of particle
pressure to magnetic field pressure) is of order unity or larger.
This was confirmed by Orlowski et al. [1994], using data from
Pioneer-Venus, who found good agreement between the observations and
linear kinetic theory over all ranges of plasma beta, but found
fluid theory valid only for beta < 1.4. Detailed studies of the
microphysics of mirror waves, typically with frequencies well below
, were presented by Southwood and Kivelson
[1993a], who presented a physical picture of the kinetics of the
mirror instability, and by McKean et al. [1993], who used
numerical hybrid simulations.
Anderson et al. [1994], Denton et al. [1994a], and
Gary et al. [1994a,b] presented coordinated observations,
theoretical studies, and kinetic simulations of magnetosheath wave
modes that have made it possible to understand the conditions
governing the onset of two major and competing instabilities. As
shown in Figure 1, when observed magnetosheath proton temperature
anisotropy and parallel plasma beta values are plotted against each
other their values never exceed a certain threshhold, which closely
matches that given by proton cyclotron mode and mirror mode growth
rates of 0.01 
. The above papers demonstrate that
when magnetosheath conditions produce an instability growth rate
that approaches the values shown, the instability is excited and
wave-particle interactions return the parameters to values near
threshold; if the magnetosheath conditions do not exceed these
limits, there is no instability growth, little or no wave activity
or particle scattering, and no change in the parameters. These
instabilities act to regulate the macroscopic distribution of ion
pressures in the subsolar magnetosheath; this newly understood
correlation of macroscopic parameters can provide modelers with
empirical equations useful for future analytical models and computer
simulations of magnetosheath energy transport [ Denton et al.,
1994b].
Studies of the transmission of ``upstream waves'' with frequencies of
15--50 milliHertz (mHz, or 10
Hz) into the magnetosphere
were presented by Engebretson et al. [1991b] and Lin et
al. [1991a,b], using coordinated observations from three
satellites: ISEE 2 in the upstream solar wind, AMPTE IRM (Ion
Release Module) in the subsolar magnetosheath, and AMPTE CCE (Charge
Composition Explorer) in the dayside outer magnetosphere.
Engebretson et al. [1991b] noted that very high levels of broadband
magnetic field turbulence in the magnetosheath (with power both
parallel and perpendicular to the magnetic field, denoted B, but
with lowest frequency of enhanced power in the same 15--50 mHz
range) were associated with azimuthally polarized magnetospheric
pulsations in the 15--50 mHz range, denoted Pc 3-4 pulsations, and
with low interplanetary magnetic field (IMF) cone angle
, defined as cos
|B
/
B|< 45, where
B
is the component of the IMF along the Earth-Sun line. Such
an IMF orientation allows quasi-parallel shock conditions to hold at
the subsolar bow shock, from which point turbulent magnetosheath
plasma can be convected toward the nose of the magnetosphere. They
also noted that no magnetospheric Pc 3-4 activity occurred during
times when mirror-like (purely compressional) waves were observed in
the subsolar magnetosheath. Song et al. [1993a], in a case
study of a magnetopause crossing of the ISEE 1 and 2 satellites
during a time when strong slow mode (compressional) waves were
observed in the magnetosheath, showed that most of the wave energy
incident on the subsolar magnetopause was either reflected or was
diverted along magnetopause/boundary layer magnetic field lines
toward the ionosphere.
In studies focusing more on how magnetosheath variations drive ULF
waves inside the magnetosphere, Lin et al. [1991a,b] showed
that the high levels of three-dimensional magnetosheath turbulence
associated with low IMF cone angles and magnetospheric Pc 3-4 waves
were also accompanied by greatly increased thermal beta and
perturbation energy, and that a significant part of the fluctuation
energy of the magnetosheath consistently propagated toward the
magnetopause in the form of perturbation Poynting flux
(electromagnetic energy flux) and kinetic energy flux. In addition,
Engebretson et al. [1991b] noted that during these periods the
B
(north-south) component of the subsolar magnetosheath field
repeatedly reached values of 
30 or even 
50 nanoTesla (nT, or
10
T), possibly driving short-lived (and highly localized?)
reconnection processes.