ARTICLE IN PRESS Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 www.elsevier.com/locate/jastp Upstream whistler-mode waves at planetary bow shocks: A brief review C.T. Russell Institute of Geophysics and Planetary Physics and Department of Earth and Space Sciences, University of California, Los Angeles, CA 90095-1567, USA Received 9 August 2006; received in revised form 15 November 2006; accepted 2 February 2007 Available online 4 July 2007 Abstract Upstream whistler-mode waves appear to be present in front of all collisionless shocks. Because the whistler-mode group velocity exceeds its phase velocity over the frequency range in which the phase velocity increases with frequency, interesting alterations of polarization and frequency spectrum occur in the observer’s reference frame. Landau resonance also plays a role in the wave properties. The source of these waves is the shock but the mechanism for wave generation is not yet understood. r 2007 Elsevier Ltd. All rights reserved. Keywords: Upstream waves; Planetary bow shocks; Whistler-mode waves 1. Introduction Upstream whistler-mode waves on magnetic field lines connected to the bow shock were discovered by OGO-5, one of the first spacecraft to obtain high cadence magnetometer data upstream of the Earth’s bow shock (Russell et al., 1971). At 1 AU these waves are close to 1 Hz in the satellite (and Earth’s) reference frame, i.e. the Pc-1 frequency band. They are often called simply 1 Hz waves. However, as we will see in this review, these narrow-band waves do not pass through the Earth’s magnetosheath and do not enter the magnetosphere. Thus they do not contribute to the complex morphology of Pc-1 waves as observed in the magnetosphere and on the Tel.: +1 310 8253188; fax: +1 310 2063051. E-mail address: ctrussell@igpp.ucla.edu 1364-6826/$ - see front matter r 2007 Elsevier Ltd. All rights reserved. doi:10.1016/j.jastp.2006.11.004 ground. Instead magnetospheric Pc-1 waves appear to be associated with ion cyclotron resonance with magnetospheric ions. As we shall see, the upstream whistlers are clearly right-handed waves, generated at the bow shock, damping as they propagate upstream in the solar wind. Since they were discovered there have been several proposed hypotheses for their generation. Fairfield (1974) was one of the first to undertake a comprehensive study of these waves. He demonstrated that the waves propagated in the oblique whistler mode, and that Doppler shifting significantly affected the observed frequency of the waves, their polarizations and their spectral densities. He hypothesized that the waves were generated at the bow shock by some unspecified mechanism. Later studies by Rodriguez and Gurnett (1975) and Greenstadt et al. (1981) suggested that these waves ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 1740 Bz -20 By 10 0 -10 0 Bx Venus July 12, 1980 Bz By Bx Mercury March 29, 1974 20 -9.0 -10.5 6.0 4.5 3.0 0 10.5 0 |B| |B| 6.0 20 3.0 0 2102:00 2103:00 -0.8 -2.4 Bx 5.6 By 1139:00 0.2 0.0 -0.2 0.8 0 0.2 |B| 5.6 |B| 1138:00 Saturn August 30, 1981 0.0 4.0 4.0 2.4 0 2126:00 1137:00 Bz By Bx Earth November 7, 1977 Bz 0 2104:00 0.0 0.2 0.0 2127:00 2128:00 20:55 21:45 Fig. 1. Time series of components, Bx, By, Bz and jBj in solar orbital coordinates with X toward the Sun, Y in the direction opposite planetary motion and Z along the (northern) orbital pole illustrating the occurrence of upstream whistler mode waves at Mercury, Venus, Earth and Saturn. are subject to strong damping in the shock foot and unlikely to reach deep into the foreshock where they are observed. Sentman et al. (1983) addressed this problem by proposing that large pitch-angle elections backstreaming from the shock could amplify the waves in the solar wind. However, this mechanism could not predict all the properties of the upstream whistlers. In a series of papers, D. Orlowski and co-workers (Orlowski et al., 1990, 1993; Orlowski and Russell, 1991) reopened the study of these waves and concluded that the Fairfield hypothesis was indeed correct. An intriguing feature of the upstream whistlers is their variable spectral shape and polarization. Sometimes the waves have an extremely sharp upper cutoff frequency. Orlowski et al. (1993) proposed that the observed spectral shape could be explained by a large Doppler shift and the dispersive properties of a broadband (Do/oE1) whistler-mode wave. Orlowski et al. (1995) followed this suggestion with a detailed analysis of the spectral properties of the waves and the nature of the growth and damping that various particles distributions would produce. They found that the wave properties were inconsistent with those predicted by the Sentman et al. (1983) hypothesis and that the waves damped with distance from the shock rather than growing as Sentman et al. (1983) predicted. The end result of this analysis was that the waves must be generated at the shock and propagate upstream but they did not identify the instability that generates the whistler-mode waves. This problem is yet unsolved today. Our instruments have progressed much in the last decade and we now have new missions that could contribute to the solution of this problem such as Cluster, THEMIS, STEREO (in interplanetary space), Venus Express (at Venus) and in the more distant future the Magnetosphere Multiscale Mission. In this brief review, we present an overview ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 of the observations and analysis that are the underpinning of our present understanding of this phenomenon. Upstream whistlers appear to be a universal phenomenon at collisionless shocks. They have been observed at Mercury (Fairfield and Behannon, 1976; Orlowski et al., 1990); Venus (Orlowski and Russell, 1991; Orlowski et al., 1993); Saturn (Orlowski et al., 1992); as well as by many authors at Earth, including Russell et al. (1971), Fairfield (1974), and Orlowski et al. (1995). Fig. 1 shows the time series of these small amplitude waves at four planets: Mercury, Venus, Earth and Saturn. In this figure, the upstream whistlers are the highest frequency waves displayed. Since these four bodies have bow shocks of varying strengths and form the obstacle to the flow in different ways, we conclude that the creation of these whistler-mode waves is a generic shock process and is not specific to a small range of boundary conditions. We see from Fig. 1 102 Mercury Venus Trace Power (nT2/Hz) 100 Earth 10-2 Saturn 10-4 10-2 100 Frequency (Hz) 102 Fig. 2. Trace of the power spectral matrix as a function of frequency for upstream whistlers at Mercury, Venus, Earth and Saturn. 4 Frequency (Hz) 2. Observations 1741 3 f (Hz) = 0.196 BT(nT) 2 Mercury Venus Earth 1 0 0 5 10 15 Magnetic Field (nT) 20 25 Fig. 3. The frequency of upstream whistler mode waves at Mercury, Venus and Earth as a function of the magnetic field strength at the time of observation. that the wave amplitude is modulated during an occurrence, forming wave packets. The spectrum in the spacecraft frame is similar at the different planets as demonstrated in Fig. 2, but the frequency does vary with distance from the Sun. Three of the 4 power spectra shown in Fig. 2 exhibit the steep upper frequency cutoff that is often present in these waves. The Saturn spectrum is weak here and the sample rate is low so the waves appear close to the Nyquist frequency. Thus, it is not possible to judge the sharpness of the upper frequency cutoff. The frequency of the waves is in part controlled by the interplanetary magnetic field. This is illustrated in Fig. 3 that shows the wave frequency in the satellite frame versus the interplanetary magnetic field strength. To our knowledge no such study has been done on the frequency in the plasma frame. In fact such a study would be quite difficult, if only because, as will be shown below, the Doppler shifting of these signals by the solar wind is quite complex and hard to deconvolve. The sensitivity to Doppler shifting by the solar wind flow is illustrated by Fig. 4. Our reference to these waves as upstream whistlers clearly indicates that the waves are right-hand polarized in the plasma rest frame. However, in Fig. 4 the waves are left-hand polarized for all directions of propagation that are within 501 of the direction of the solar wind flow. These are the directions of propagation that are most greatly Doppler-shifted by the solar wind. As mentioned in the Introduction, if the waves are generated at the shock we expect them to be damped by Landau resonance. When a whistler mode wave travels obliquely to the magnetic field it ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 1742 50 0 Left Frequency (Hz) 1 -1 -2 0 30 60 90 Fig. 4. The frequency (with sign) of upstream whistlers as the angle between the wave normal and the solar wind direction changes. Mercury (Mariner 10) Number of Wave Events Normalized by Anglular Area of Bin Right 2 40 30 20 10 Venus (PVO) Wave Amplitude (nT) 1.0 Earth (ISEE-1) Saturn (Voyager) 0 0 20 40 60 80 Fig. 6. The number of wave events in 101 bins of the angle between the magnetic field and the wave normal. The rate is normalized for the solid angle subtended by the bin. The narrow distribution of wave normal angles about the field direction is consistent with Landau damping that attenuates the wave more as the angle to the field increases. 0.1 0.01 10000 20000 30000 40000 50000 60000 70000 Distance from the Shock (km) Fig. 5. The amplitude of upstream whistlers at Mercury, Venus, Earth and Saturn as a function of distance from the bow shock. has an electric field parallel to the magnetic field that can resonate with the drift of thermal electrons along the magnetic field. Unless there is a bump on the tail of the electron distribution, the usual solar wind distribution (Maxwellian) damps the wave. We can easily test this prediction. Fig. 5 shows the amplitude of upstream whistlers versus distance from the shock for Mercury, Venus, Earth and Saturn. Perhaps surprisingly the wave amplitudes do not exhibit differences that we can attribute to each planet. Rather there is one single curve that fits all the data, albeit it is a semilog plot. Fig. 6 supports the Landau damping hypothesis by showing that most events propagate along the magnetic field and fewer events propagate at a large angle. Here we have binned the data in 101 bins and normalized the rate for the solid angle subtended by the bin. Fig. 7 shows yet a third test. Here we show the wave amplitude versus the propagation angle to the field at two distances from the shock. Closer to the shock the waves are stronger and have a broader angular range of propagation. The waves clearly damp with distance from the shock and damp more strongly at greater angles of propagation to the magnetic field as predicted by Landau resonance theory. Thus, the concerns of Rodriguez and Gurnett (1975) and Greenstadt et al. (1981) were partially correct. Landau damping does affect any waves propagating at an angle to the magnetic field. However, the waves that propagate parallel to the ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 field are not affected by Landau damping and penetrate deep into the foreshock. 3. The effects of Doppler shifting As illustrated in Fig. 4, the polarization of the wave in the spacecraft frame depends strongly on the direction of propagation relative to the solar wind flow. However, this is not the only effect on the waves. Fig. 8 shows a time series and a power spectrum of upstream whistlers about 3.5 Wave Amplitude (nT) 3.0 d<0.3 R v 1<d<10 R v 2.0 1.0 0 0 30 60 Propagation Angle (deg.) 90 Fig. 7. The amplitude of upstream whistlers as a function of the propagation angle to the magnetic field direction at two distances from the shock. The solid lines show least square fits to the data points. The different amplitudes and slopes are consistent with Landau damping as the waves propagate away from the shock. 1743 Venus radii (Rv) upstream from the shock as observed by Pioneer Venus on July 12, 1980 (Orlowski and Russell, 1991). The power spectrum shows a very distinct peak with an intensity of 0.7 nT2/Hz and a frequency of 1.4 Hz. The waves are left-hand elliptically polarized with an ellipticity of 0.97 and are propagating at an angle of 361 to the magnetic field. The wave normal is 201 from the solar wind flow direction. A remarkable feature of this spectrum is the strong decrease in power spectral density near 1.5 Hz with a falloff of about 160 dB/decade. In contrast to the July 12 waves, Fig. 9 shows upstream whistlers observed on July 3, 1980 about 1 Rv upstream of the Venus shock. As would be expected for this closer distance the waves are stronger. However, they are right-hand polarized and have a much different spectrum. In fact, there is no sharp break in the spectrum here and the highfrequency right-handed whistler power joins smoothly with the more linear low-frequency power. The whistler waves here are propagating at 341 to the magnetic field, similar to the 361, in the previous example, but their propagation angle to the solar wind flow is 641, much larger than the previous 201. Thus, these whistler waves are not subject to as much Doppler shift as in Fig. 8. The weakness of the Doppler shifting can explain both the different polarization and the difference in the spectral shape. Fig. 8. Upstream whistlers seen 3.5 Rv upstream from the Venus shock. (Left) Time series in VSO coordinates. (Right) Trace of the power spectrum for these waves. These waves were strongly Doppler shifted, propagating at about 201 to the solar wind flow direction. ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 1744 102 Bx (nT) 8 Orbit 576 July 3, 1980 1427:26-1428:44 UT Trace Power (nT2/Hz) 0 By (nT) -8 8 0 Bz (nT) 0 BT (nT) -8 8 -8 8 0 1427:26 :45 1428:05 Universal Time :25 1428:44 100 10-2 10-4 10-2 100 Frequency (Hz) Fig. 9. Upstream whistlers seen 1 Rv from the Venus bow shock. (Left) Time series in VSO coordinates. (Right) Trace of the power spectrum for these waves. These waves were only weakly Doppler shifted, propagating at 641 to the solar wind flow direction. 105 Waves Swept Downstream Velocity [km/s] 104 Waves Propagate Upstream Polarization Reversal 103 No Polarization Change Vph Vg 102 101 Vg Vph f1 VSW k RH Branch (Whistlers) f2 LH Branch 100 0.001 0.01 0.1 1 10 100 1000 10.000 Frequency [Hz] Fig. 10. The phase and group velocities for right-handed (whistler) and left-handed waves in a cold plasma for parallel propagation in a 5 nT magnetic field and a 5 cm3 proton–electron plasma. The horizontal line shows the Doppler-shift velocity imposed by the solar-wind flow. We can best explain the effects we see with a plot of the phase and group velocities of whistler mode waves versus frequency in the plasma frame. We use the space physics education software (Russell et al., 1995) that incorporates the cold plasma dispersion relation of Stix (1962). Here we chose typical upstream values and do not try to precisely duplicate either event. Fig. 10 shows such a plot for parallel propagation in a cold plasma with a density of 5 protons and electrons per cubic centimeter and a magnetic field of 5 nT. The electron gyro frequency is 140 Hz and the proton gyro frequency is 0.076 Hz. The solid lines show the whistler mode and the left-hand mode phase velocities in the plasma frame. The dashed lines show the associated group velocities. In the portion of the frequency range in which the phase velocity increases with frequency, the group velocity exceeds the phase velocity and vice versa. A horizontal line marked Vsw k represents the speed of the solar wind moving antiparallel to the wave normal. This speed controls the Doppler shifting effects. As the frequency increases from zero, eventually at the plasma-frame frequency marked f1, the group velocity exceeds the solar-wind Doppler-shift velocity along the wave normal and wave energy can propagate upstream. However, at this frequency the Doppler-shifting velocity still exceeds the phase speed and the phase fronts are blown backwards over the observer. Thus, the intrinsically righthanded waves appear left-handed in the spacecraft frame. This situation continues until the plasmaframe frequency f2, which marks the point at which the phase speed matches the Doppler-shifting speed. Above this frequency, the group and phase speed exceed the Doppler-shift speed until close to the electron gyro frequency where we might expect there to be much weaker signals if at all. If the spacecraft is upstream from the shock no energy will reach the spacecraft at frequencies below f1. Between f1 and f2, waves will have their ARTICLE IN PRESS C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 10 1 100 Power (nT2/Hz) 10-1 10-2 10-3 10-4 10-5 10-6 0.1 2 3 4 5 6 1 2 3 4 5 6 10 2 3 Frequency (Hz) Fig. 11. Dark spectrum is the power spectral density of upstream whistlers in the solar wind. Light spectrum is the power spectral density of the same whistlers in the spacecraft frame. polarizations reversed and at f2 the Doppler-shifted frequency will be zero. Thus, f1 will be Doppler shifted to the highest frequency in the spectrum observed in the spacecraft frame. If this frequency is f1*, then there is no wave energy above f1* except energy arriving at much higher frequencies than f2 that reach the spacecraft with weaker Doppler shifts, so that their polarization are not reversed and their frequencies are only slightly altered. These higher frequencies should have much weaker signal strengths in most geophysical situations. Fig. 11 illustrates the effect on the power spectrum. Assume that the dark-shaded spectrum between 2 and 5 Hz, say, is the natural power spectrum in the solar-wind frame. The lower frequency of this shaded power-law spectrum is the lowest frequency whose energy can propagate upstream. It Doppler-shifts to the uppermost shaded frequency in the spacecraft frame (lightly shaded spectrum). The upper end of the dark solar wind frame spectrum maps to zero frequency, creating between the two frequencies the characteristic, upward-sloping power with a sharp, upperfrequency cutoff. Any power below 2 Hz generated at the shock is swept downstream and never reaches the spacecraft. Any power above about 5 Hz in this example does not have its polarization reversed and can be found in the spacecraft frame at positive frequencies, beginning at 0 Hz adding to the power 1745 from the negative frequency Doppler-shifted waves because the power spectrum sums the contributions from the negative frequencies and positive frequencies. As illustrated here we expect this power originally at higher frequencies in the plasma frame to be weaker than in the Doppler-shifted band. Since the spectra we see do very much resemble our hypothetical spectrum, we believe our assumption is justified. Finally, we can understand the dependence seen for the frequency peak in Fig. 3. There are several effects occurring simultaneously. The Alfven velocity decreases with distance from the Sun as does the magnetic field so that the curves in Fig. 10 slip to the left and down as one moves outward from the Sun. The Doppler-shifted frequency of the peak amplitude depends on the wavelength and the phase velocity at f1. The spiral angle of the interplanetary magnetic field that guides the waves also affects Doppler shifting. These several effects change the frequency somewhat but not as much as simply the overall shift to the left of both velocity curves with decreasing field strength. The net result is a general decrease of the frequency of the 1 Hz wave peak with magnetic field strength but with some modification of the dependence at each planet due to the specific conditions of its solar wind interaction. 4. Conclusions A very simple model of bow shock generation of a power-law spectrum of whistler waves can explain all the observed properties of the upstream whistler waves including their very odd spectral peak. This model includes the effects of Doppler shifting on whistler-mode signals, whose group velocity exceeds their phase velocity, with Landau damping of the propagating waves. Simply a segment of the powerlaw wave spectrum is cut and inverted in frequency by the Doppler shifting. Some waves are carried downstream toward the magnetosphere but are broadband and probably do not contribute to the Pc-1-wave spectrum seen in the magnetosphere This phenomenon still has one important mystery, what is the process at the shock that generates the spectrum of whistler mode waves? Clues may be found in the presently unknown dependence on the shock normal angle and on the Mach number. We encourage those groups of space plasma physicists with the high-resolution shock data that are now available from modern instrumentation to address these questions. ARTICLE IN PRESS 1746 C.T. Russell / Journal of Atmospheric and Solar-Terrestrial Physics 69 (2007) 1739–1746 Acknowledgments This work was supported by the National Science Foundation under Grant ATM04-02213 and by the National Aeronautics and Space Administration under Grant NNG06GC62G. References Fairfield, D.H., 1974. Whistler waves observed upstream from collisionless shocks. Journal of Geophysical Research 79, 1368–1378. Fairfield, D.H., Behannon, K.W., 1976. Bow shock and magnetosheath waves at Mercury. Journal of Geophysical Research 81 (22), 3897–3906. Greenstadt, E.W., Fredricks, R.W., Russell, C.T., Scarf, F.L., Anderson, R.R., Gurnett, D.A., 1981. Whistler mode wave propagation in the solar wind near the bow shock. Journal of Geophysical Research 86, 4511–4516. Orlowski, D.S., Russell, C.T., 1991. ULF waves upstream of the Venus bow shock: properties of one-Hertz waves. Journal of Geophysical Research 96, 11,271–11,282. Orlowski, D.S., Crawford, G.K., Russell, C.T., 1990. Upstream waves at Mercury, Venus and Earth: comparison of the properties of one Hertz waves. Geophysical Research Letters 17, 2293–2296. Orlowski, D.S., Russell, C.T., Lepping, R.P., 1992. Wave phenomena in the upstream region of Saturn. Journal of Geophysical Research 97, 19,187–19,199. Orlowski, D.S., Russell, C.T., Krauss-Varban, D., 1993. On the source of upstream whistlers in the Venus foreshock. In: Gombosi, T.I. (Ed.), Plasma Environments of Non-Magnetic Planets. Pergamon Press, New York, pp. 217–227. Orlowski, D.S., Russell, C.T., Krauss-Varban, D., Omidi, N., Thomsen, M.F., 1995. Damping and spectral formation of broadband upstream whistlers. Journal of Geophysical Research 100 (A9), 17,117–17,128. Rodriguez, P., Gurnett, D.A., 1975. Electrostatic and electromagnetic turbulence associated with Earth’s bow shock. Journal of Geophysical Research 80, 19–27. Russell, C.T., Childers, D.D., Coleman, P.J., 1971. OGO-5 observations of upstream waves in the interplanetary medium: discrete wave packets. Journal of Geophysical Research 76 (4), 845–861. Russell, C.T., Le, G., Luhmann, J.G., Littlefield, B., 1995. Educational software for the visualization of space plasma processes. In: Szuszczewicz, E.P., Bredekamp, J.H. (Eds.), Visualization Techniques in Space and Atmospheric Sciences. NASA SP-519, Washington, DC, pp. 263–270. Sentman, D.D., Thomsen, M.F., Gary, S.P., Feldman, W.C., Hoppe, M.M., 1983. The oblique whistler instability in the Earth’s foreshock. Journal of Geophysical Research 88 (A3), 2048–2056. Stix, T.H., 1962. The Theory of Plasma Waves. McGraw-Hill, New York.