International Journal of Geomagnetism and Aeronomy
Published by the American Geophysical Union
Vol. 1, No. 1, April 1998

Plasma turbulence spectrum in the lower ionosphere at the heights 80-100 km

A. V. Gurevich *, K. Rinnert and K. Schlegel

Max-Planck-Institut fur Aeronomie, Katlenburg-Lindau, Germany
Comparison With Theory


The results of observations of plasma density and electric field fluctuations measured in situ at the ionospheric heights 80-100 km are presented. The comparison of the observational data with the theory shows an agreement with the shape of plasma turbulence spactra. A remarkably good agreement with the plasma spectra anisotropy is demonstrated. Significant amplification of the low frequency plasma fluctuations in strongly disturbed ionospheric conditions is established.


To study the auroral E region structure and dynamics as well as plasma instabilities excited therein, four rockets have been launched during the "ROSE" campaign in 1988/1989 (F1 and F2 from the Andoya rocket range and F3 and F4 from ESRANGE) [Rose et al., 1992]. In the following investigation, data collected by the rocket-borne retarding potential analyzer and the electric field measuring instrument from Flights 1 and 4 are analyzed with emphasis on low-frequency electric field fluctuations.

Density fluctuations were measured using a retarding potential analyzer (RPA) operated in the saturation mode [Schlegel, 1992]. The relative density changes Dn/n were considered to be proportional to the saturation current Dj/j. The absolute values of the electron density are difficult to estimate with this technique, but relative density changes are easy to measure. The measured current in the saturation mode (10-10 to 10-6 A) was amplified by a fast electrometer and then sampled every 0.5 ms in order to retrieve the density fluctuation spectrum up to 1 kHz. For more details, see Schlegel [1992].

The electric field measuring instrument was a floating Langmuir double-probe system primarily designed to measure the dc electric field. Two pairs of booms were utilized which had gold-plated spherical probes (50 mm diameter) mounted on the ends. Each deployed boom pair placed the probes in opposite directions at a distance of 1.8 m from the payload axis. Hence the probe separation of each boom pair was 3.6 m. One boom pair was mounted at the lower level of the payload, and the other pair was mounted at an upper level 1.8 m apart from the lower level and rotated by 90o. Thus the four probes on the deployed booms formed the corners of a tetrahedron. All six differential voltages between any two probes were samples simultaneously at a rate of 4 kHz. The corresponding six electric field components are transformed into a three-dimensional electric field vector in the payload frame of reference and finally into an Earth-fixed frame of reference. For more details, see Rinnert [1992].

Link to Fig. 1 As an example, Figure 1 exhibits both the dc and the ac electric field data obtained during Flight 4. The lower frame of Figure 1 shows the dc electric field component perpendicular to the Earth's magnetic field versus flight time. It shows a fairly stable electric field primarily southward directed and slightly increasing during the flight from about 35 mV m-1 to about 50 mV m-1. The dc electric field in magnitude and direction is confirmed by STARE measurements and EISCAT measurements of the plasma drift at 150-km altitude just above the apogee of the payload trajectory [Kohl et al., 1992]. The dots indicate the payload altitude versus flight time. The apogee was reached at 123 km about 180 s after launch. In the upper frame of Figure 1 an ac signal is presented. Plotted are the contours of constant amplitude (voltage between the upper two probes) versus frequency and flight time. The spectra are 1-s averages. The fluctuations occur in a well-defined altitude region, and the payload passed this instability region twice, during upleg and during downleg. The ac electric field activity started at about 80-km altitude with low frequencies and at about 95 km with higher frequencies and stopped just above 115 km. The width of the instability region as well as the intensity is somewhat larger during downleg than during upleg because of the larger dc electric field.

An important feature is the occurrence of intense fluctuations in the frequency band below 50 Hz in the lower part of the instability region between about 80- and 100-km altitude. The high-frequency oscillations are practically absent in this region, but they dominate at higher heights. This is an indication that two different mechanisms are responsible for these oscillations in the lower and upper part. The fluctuations in the higher region are certainly related to the Farley-Buneman  (FB) or "two-stream instability" [Farley, 1963]. In the lower region the "gradient drift instability" is usually considered as a source of plasma oscillations [Fejer and Kelley, 1980], but it strongly depends on the local gradients of plasma density. On the other hand, the fact that the lower-altitude range is dominated by low-frequency fluctuations was found for all four flights under quite different conditions. This finding is therefore interesting to evaluate with regard to the turbulence of the neutral atmosphere, which is another possible source of low-frequency plasma fluctuations in the ionosphere.


Link to Fig. 2 Examples of the density spectra for different heights are presented in Figure 2 (Flight 1, upleg). Significant changes of the spectrum with altitude can be seen. At low heights (z \le 95 km, upper panel) the density oscillations are mostly in the low-frequency range (f \le 100 Hz). The high-frequency part is very weak here or is not seen at all. At higher altitudes (z ~ 100 to 110nbsp;km, middle and lower panels) the high-frequency part of the spectrum f \ge 100 Hz becomes significant and even dominates (lower panel). Link to Fig. 3 So we see that the density fluctuations agree well with the electric field fluctuations. The fluctuations at low heights have generally the same structure for the different flights and for the upleg and downleg motion of the rocket. Figure 3 shows results from ROSE Flight 4.

Link to Fig. 4 Figure 4 exhibits some relevant height dependences during Flight 4: vr is the rocket velocity and vr \bot B is the rocket velocity perpendicular to B, vDrift is the electron drift velocity deduced from the measured dc electric field according to vDrift = ([E, B]/B2) [1/(1 + y)], Cs is the speed of sound and is assumed to be constant (Cs= 300 m s-1 for the altitudes below ~110 km), fc is the critical frequency according to (1), and l0 is the characteristic scale of neutral turbulence:


Here ni is the collision frequency of ions, t is the lifetime of electrons, y is a nondimensional factor, determining the collisional damping of the drift velocity [Fejer and Kelley, 1980]. If the drift velocity exceeds the sound velocity (vDrift > Cs), which happens quite suddenly at about 95 km, then the FB instability is excited for the frequencies f > fc. Thus below 95 km, FB waves are not excited and the observed fluctuations must have another source.

On the downleg the payload flew almost parallel to B, and therefore vr \bot B is then much smaller than during the upleg. Assuming that the atmospheric conditions in both altitude regions were the same during the upleg and downleg (during the downleg the probed region was about 40 km further north and it happened about 3 min later), this allows some information on the directional dependence of these structures with respect to B.

The three ac components dEeast, dEnorth, and dEparallel with respect to the Earth's magnetic field have been Fourier analyzed with a frequency resolution of 3.906 Hz using overlapping time windows of 256 ms. The spectra were combined to provide an estimate of the total spectral energy density dE2.

Link to Fig. 5 Figure 5 shows some measured power spectra from Flight 4. The energy density is plotted versus a normalized wave number x = kl0 = 2p fl0/v; v is rocket velocity, and l0 is the characteristic scale of neutral turbulence and is a function of altitude. According to Tatarski [1961] it can be expressed as


(in meters). Here Nm is the number density of the neutral atmosphere (in cm-3), T is the neutral gas temperature (in kelvins), and e the turbulent energy dissipation parameter. We used the value e = 0.1 W kg-1 in our estimates [e.g., Rüster}, 1984]. In the range of interest, l0 is of the order of 1-10 m (see  Figure 4). (The heavy dotted curves at Figure 5 represent the theoretical power spectrum after formula (3) below. The constant C in (3) is chosen to fit the low-frequency part of the observed spectra.)

The transformation from the frequency scale to the kl0 scale leads to an altitude dependent position of the measured spectra with respect to the kl0 scale.

Plotted in Figure 5 are spectra from various height regions. The top frames are from the upleg, and the bottom frames are from the downleg. At the lower altitudes the measured spectra are steeper than the theoretical spectrum, and at higher altitudes they are flatter. The sudden change in the measured spectra at the heights higher then 95 km (upleg) is clearly seen. The additional contribution at high frequencies (or large k) is attributable to the FB instability. During the downleg this change happens at a somewhat lower altitude. As the dc electric field is larger during the downleg, the height range of the FB instabilities is extended (see Figures 1 and 4).

Link to Fig. 6 Figure 6 represents the same measured spectra that are now plotted versus k\botl0 = 2pl0 f/vr \bot B. Here one can see better agreement with the form of the theoretical spectrum at lower altitudes.

No data exist from below about 77 km because the booms were not released during the downleg and the payload became unstable. Unfortunately, there are no neutral wind data. Neutral winds, of course, have an impact on the transformation from the frequency into the k domain as the wind velocity adds to the rocket velocity vr, but as vr is large enough (see Figure 4), the role of neutral winds is clearly unimportant.

Link to Fig. 7
Qualitatively, the same behavior of turbulent spectra is found for the data from Flight 1. The dc electric field or the drift velocity was lower, and consequently the destabilization of FB waves was less. The trajectory of F1 was such that there was not a big difference in vr \bot B for the upleg and downleg, and therefore the differences in k, k\bot B up and k\bot B down is less pronounced. Link to Fig. 8 Figures 7 and 8 present the F1 data corresponding to Figures 5 and 6, respectively. Data analogous to those presented in Figures 5, 6, 7, and 8 were obtained during other flights, which demonstrates the general stability of the shape of the electric field fluctuations spectra.

Comparison With Theory

Experimental data of both the electric field and plasma density fluctuations definitely show that the region of low heights (z \le 95 km) is unique. In this region the low-frequency (f \le 100 Hz) part of the fluctuations dominates. Higher-frequency fluctuations are strongly excited at higher altitudes, in agreement with the existing theory of the Farley-Buneman instability [Fejer and Kelley, 1980].

On the other hand, fluctuations observed at lower heights (z \lesssim 95 km) cannot be excited by the FB instability, and so another process must be responsible for these oscillations. A theory describing such a process was developed recently which considered the electric field and plasma density fluctuations induced at low altitudes (z ~ 80 to 100 km) due to the turbulence of the neutral atmosphere [Gurevich et al., 1997]. We will compare the observational data with the predictions of this theory.

It should be emphasized that the theory of induced turbulence gives a quantitative description of plasma turbulence. It means that one is able to compare with the theory not only the wave vector k (or frequency f) dependence of the observed spectrum, but also its anisotropy (which is determined mostly by the Earth's magnetic field B) and, what is especially remarkable, its absolute values.

The averaged frequency spectrum of electric field fluctuations according to Gurevich et al. [1997] has the form


Here v is rocket velocity, l0 is the characteristic length of neutral turbulence (see (2)), and C is a normalizing constant. The spectrum (3) is compared with the experimental data at different heights in Figures 5 to  8. There is general agreement between the theory and observational data at low heights. At higher altitudes (z > 95-100 km) there is no agreement, as could be expected owing to excitation of the FB instability. Note that the 9 observed spectrum has a tendency to become steeper than (3) at low heights (z ~ 80 km).

It should be emphasized that the theory predicts the existence of an asymmetry of the spectrum with respect to the magnetic field direction: part of the inhomogeneities are strongly elongated along the Earth's magnetic field B. In the observational data there seems to be an indication of such an anisotropy. If the inhomogeneities are strongly elongated along B, only the component of the rocket velocity perpendicular to B} (vr \perp B) should be used in (3) instead of the full velocity v. Especially for lower altitudes there is better agreement with observations with respect to k\bot} = 2pf/vr \bot B ( Figures 6 and  8). In the same time that the anisotropy is not too strong, what can be seen from comparison of downleg spectra at Figure 6 and Figure 5 for the heights 90.66-94.48 km: note that velocity vr \bot B was very small in this case, and so the role of the kparallel component became important.

It is significant that the theory predicts a universal dependence of directional distribution of electric field fluctuations with respect to B: the dependence on the angle between k and B is the same for all values of k and varies only slightly with altitude. It allows us to average the observed angular dependence over k values and over the heights in the lower ionosphere (80 \lessim z \perp 100 km). Link to Fig. 9 The result of this averaging is compared with the predictions of the theory in Figure 9. One can see quite good agreement between the theory and the observations.

Analysis of the density observations leads to analogous results. To compare the experimental data with the theory, the following formula for d | Dn/n | /df was obtained using the density spectrum d á | Dn/n |2 ñ/dk, presented by Gurevich et al. [1997]:


where v again is rocket velocity, l0 is the characteristic length of neutral turbulence, and C1 is a normalization constant. The results of (4) for appropriate heights have been included in Figure 2 (top panel) and Figure 3 as a dashed line.

So we deduce that at low heights, shape and asymmetry of the spectra of the observed electric field and plasma density fluctuations are in agreement with the theory of neutral atmosphere induced plasma turbulence. A discrepancy with the theory is seen, however, when we compare the normalization constant C for electric field fluctuations. The observed values of this constant are approximately 103 times larger than those obtained in the theory.

It is important to note here that the theory by Gurevich et al. [1997] in fact establishes only direct connection between the spectrum of induced plasma turbulence and the spectrum of neutral turbulence. The latter (according to Carter and Balsley [1982], Rüster [1984], and Manson et al. [1987]) can be regarded as Kolmogoroff's, which means that it is fully determined by the only parameter e that characterizes the energy dissipation in neutral turbulence and is generally proportional to v03, where v0 is the characteristic value of neutral wind velocity fluctuations. The intensity of fluctuations v0 has significant seasonal, daily, and temporal variations (depending, for example, on gravity waves). Because of this even the time-averaged values of e can differ by an order of magnitude. As follows from observations, the averaged value of parameter e grows slightly with height in the upper atmosphere [Manson et al., 1981]; it can also be essentially amplified (up to 1 order of magnitude and even more) in the region of gradient temperature inversion in the height profile (z ~ 80-85 km). In our analysis here we do not take into account all these variations of neutral turbulence, using for parameter e its characteristic value e = 0.1 W kg-1.

Link to Fig. 10
This normalization is denoted at C = 1. It follows from the previous discussion that when the rocket measurements of the turbulent spectra in given concrete conditions are compared with the theory, the changes of normalization constant C at least within 1 order of magnitude should be considered as quite normal fluctuations. For example, the change of fluctuations in one and the same realization of turbulent spectra are usually more than 1 order of magnitude, which is directly seen from Figures 5, 6, 7, 8, 10, and  11. Link to Fig. 11 Regardless of these fluctuations, the established strong amplification of the observed turbulent spectra C ~ 103 is regular and definitely has real significance.


In the lower ionosphere the densities of electrons (ne) and ions (ni) are several orders of magnitude less than the density of neutral particles (nm). That means that the charged components are the minor species and have no influence on both the average motion and turbulent pulsations of the neutral gas. Rather, the motion of the neutral gas determines the motion of the charged species. This fact is the background of the discussion of the effect of neutral turbulence on radio wave scattering and plasma density fluctuations in previous works [Rastogi and Bowhill, 1976; Royrvik and Smith, 1984].

We further emphasize that as the velocity of turbulent fluctuation v0 in atmospheric conditions is much smaller than the sound velocity Cs, the compression and heating effects, which determine the temperature and density fluctuations in the neutral atmosphere, are small [Tatarski, 1961].

It is important that not only the neutral gas motion, but also the Earth's magnetic field B and electric field E, in the ionospheric plasma, have a strong influence on the motion of charged particles. Therefore the charged components at the ionospheric heights 70-100 km are not only transported by the turbulent motion of neutrals, but accomplish their own motion. This motion becomes compressible owing to the action of the magnetic field B and is thus essentially different from the motion of the neutral component. Compressibility leads to the creation of plasma density fluctuations and to the generation of random electric fields and currents. In this sense the plasma turbulence in the lower ionosphere induced by the noncompressible turbulence of the neutral gas is considered in the theory of Gurevich et al. [1997].

The interaction with neutral particles has a crucial influence on the motion of the charged components. The interaction between charged particles and excited plasma waves is much smaller. Because of these facts, the plasma turbulence in the lower ionosphere can be described as the superposition of the linear low-frequency plasma waves, excited by turbulent motions of the neutral gas. Neutral turbulence acts as a driving force for the natural ionospheric plasma modes. The key role in the induced turbulence is played by the drift mode. Note that exactly the same mode is excited in the Farley-Buneman instability, in the gradient-drift instability [Fejer and Kelley, 1980], and in the thermodiffusive instability [Gurevich and Karashtin, 1984].

The resulting k spectrum of induced plasma turbulence obtained by Gurevich et al. [1997] has some interesting peculiarities, depending on the angles between k and the directions of the electron drift velocity ve and the magnetic field B. The dependence on the angle between k and B is shown in Figure 9. One can see that the spectrum of induced plasma turbulence is essentially anisotropic.

The integration over the angles gives the dependence of the spectrum on the absolute value of k. The spectrum for the density fluctuations Dn at the low height of z \lesssim 95 km is


Here the characteristic scale l0 and the function E(x) are determined according to equations (2) and (3). The parameter Q0 depends substantially on height z and is given in Table 2 of Gurevich et al. [1997]. We emphasize that in the inertial interval kl0 <<1, the spectrum (5) is proportional to k-8/3. A comparison of the spectrum with density fluctuations at mesospheric heights obtained by Royrvik and Smith [1984] during quiet ionospheric conditions is presented in Figure 10. One can see good agreement between the theory and the observations (for 86.5 km, Q0 ~ 3 x 10-7, l0 3.3 m).

This agreement is present not only in the shape of the spectrum, but also in the absolute values of the fluctuations. The discrepancy is small enough and could be corrected with the chosen atmospheric model and the actual local value of the turbulence dissipative factor as indicated with the dashed line in Figure 10a. One can notice also that the amplification of Figure 10a, can be attributed to the region of the temperature gradient inversion in the height profile. As was already mentioned before, the amplification of fluctuations is often observed in the mesopause near this region (see, for example, Karashtin et al., [1997a]), and it is also clearly seen in Figure 2 of Royrvik and Smith [1984]. Note that gradual growth of the averaged value of | Dn/n | with height up to 95-98 km,which is seen in this figure is also in agreement with the theory of Gurevich et al. [1997].

Let us return to the comparison of the theory with our experimental data. We have seen that in the lower ionosphere the shape of the spectra of the density and of the electric field fluctuations presented in Figures 2, 3, 5, 6, 7, and 8 is in agreement with the theory. We especially emphasize the good agreement in the predicted and observed asymmetry of the spectra ( Figure 9): the spectrum is not isotropic, as it should be for neutral turbulence; nor is it strongly anisotropic, as in the case of a plasma-excited instability.

The absolute values of á |Dn/n | ñ were not measured in our experiment. The absolute values of the electric field fluctuations d á |DE |2 ñ df, however, were measured, and we have seen that the observed values are amplified by the factor C ~ 103 in comparison with the predictions in the theory.

One can suggest the following main reasons for the observed amplification of the low-frequency oscillations:

1.  The theory is developed for a homogeneous medium. The real ionospheric plasma, however, is height inhomogeneous, and this inhomogeneity leads to WKB amplification of the oscillations in the reflection region at the lower boundary of the ionosphere (see Gurevich et al., [1995]; Karashtin et al., [1997b]).

2.  Our measurements were made during strongly disturbed conditions, when the FB instability was excited. This instability, being excited at higher altitudes, possibly could amplify the oscillations at lower heights also. For example, in the equatorial electrojet under conditions when the FB instability was excited, the electric field fluctuations at low heights compared with the theory of induced turbulence have the same spectral form and the same values (C ~ 103) as in our experiments [Pfaff, 1991; Pfaff et al., 1987]. Quite analogous comparison with measurements in the absence of or with very weak FB instability gives a C value of 10-30 times less (see Figure 11, where a direct comparison of formula (3) with observations is presented; the exact value C = 1 from Gurevich et al. [1997] is used here to calculate the theoretical curve).

3.  The vertical electric currents, which usually exist in disturbed conditions, could also lead to the amplification of electric field oscillations (see Karashtin et al., [1997b]).

We have to add also that on the basis of the observations [Carter and Balsley, 1982; Manson et al., 1981; Rüster, 1984], it was supposed in the theory that the neutral turbulence is isotropic and has Kolmogoroff's spectrum. This statement for the real upper atmosphere does not seem to be proven well enough and should be studied more.

One can conclude that there is general agreement of the theory of induced turbulence with observational data in the lower ionosphere. It is significant that the observed shape of low-frequency spectra in the equatorial and auroral region are alike. On the one hand, this is indicative of a common mechanism for these phenomena and could easily be understood if the excitation of plasma oscillations at low heights is mainly determined by neutral plasma turbulence, which has a unique spectrum. On the other hand, it must be emphasized that in disturbed conditions the amplitudes of the observed electric field fluctuations are strongly amplified in comparison with the one predicted by the theory of induced turbulence.


The work of one of the authors (A. V. G.) was partially supported by grant 96-02-16465a from the Science Foundation of Russia Federation (RFFI).


Carter, D. A., and B. B. Balsley, The summer wind field between 80 and 93 km observed by MST radar at Poker Flat, Alaska (65oN), J. Atmos. Sci., 39, 2905, 1982.

Farley, D. T., A plasma instability resulting in field-aligned irregularities in the ionosphere, J. Geophys. Res., 68, 6083, 1963.

Fejer, B. G., and M. C. Kelley, Ionospheric irregularities, Rev. Geophys., 18, 401, 1980.

Gurevich, A. V., and A. N. Karashtin, Small-scale thermal diffusion instability in the lower ionosphere, Geomagn. Aeron., 24, 733, 1984.

Gurevich, A. V., N. D. Borisov, A. N. Karashtin, and K. P. Zybin, Plasma density perturbations in the ionospheric E region, Phys. Lett., A208, 214, 1995.

Gurevich, A. V., N. D. Borisov, and K. P. Zybin, Ionospheric turbulence induced in the lower part of the E region by the turbulence of neutral atmosphere, J. Geophys. Res., 102, 379, 1997.

Karashtin A. N., Yu. V. Shlyugaev, V. I. Abramov et al., First HF radar measurements of summer mesopause echoes at SURA. Ann. Geophysicae, 15, 935-941, 1997a.

Karashtin A. N., A. V. Gurevich, K. P. Zybin, Density irregularities in the weakly ionized nonuniform plasma. Phys. Plasma, 4, 4090-4102, 1997b.

Kohl, H., E. Nielsen, K. Rinnert, and K. Schlegel, EISCAT results during the ROSE campaign and comparison with STARE measurements, J. Atmos. Terr. Phys., 54, 733, 1992.

Manson, A. H., C. E. Meek, and J. B. Gregory, Winds and waves in the mesosphere and lower thermosphere, J. Geophys. Res., 86, 9615, 1981.

Pfaff, R. F., Rocket observations in the equatorial electrojet current status and critical problems, J. Atmos. Terr. Phys., 53, 709, 1991.

Pfaff, R. F., M. C. Kelley, E. Kudeki, B. G. Fejer, and K. D. Baker, Electric field and plasma density measurements in the strongly driven daytime equatorial electrojet, 1, The unstable layer and gradient drift waves, J. Geophys. Res., 92, 13, 578, 1987.

Rastogi, P. K., and S. A. Bowhill, Scattering of radiowaves from the mesosphere, J. Atmos. Terr. Phys., 38, 449, 1976.

Rinnert, K., Plasma waves observed in the auroral E region - ROSE campaign, J. Atmos. Terr. Phys., 54, 683, 1992.

Rose, G., et al., The ROSE project: Scientific objectives and discussion of first results, J. Atmos. Terr. Phys., 54, 657, 1992.

Royrvik, O., and L. G. Smith, Comparison of mesospheric VHF radar echoes and rocket probe electron concentration measurements, J. Geophys. Res., 89, 9014, 1984.

Rüster, R., Winds and waves in the middle atmosphere as observed by ground-based radars, Adv. Space Res., 4, 3, 1984.

Schlegel, K., Measurements of electron density fluctuations during the ROSE rocket flights, J. Atmos. Terr. Phys., 54, 715, 1992.

Tatarski, V. I., Wave Propagation in Turbulent Medium, Dover, New York, 1961.

Go to the top
 Load file for printing and local use.

* A. V. Gurevich - On leave from P. N. Lebedev Institute of Physics, Moscow, Russia