Satellite Observation of Low-Latitude VLF Radio Noises and Their

Transcription

1 J. Geomag. Geoelectr., 41, ,1989 Satellite Observation of Low-Latitude VLF Radio Noises and Their Association with Thunderstorms Masashi HAYAKAWA Research Institute of Atmospherics, Nagoya University, 3-13 Honohara, Toyokawa 442, Japan (Received October 1,1988; Revised March 6, 1989) VLF data from the low-altitude Ariel 4 satellite are used to investigate the characteristics and generation (propagation) mechanism of low- and equatoriallatitude VLF radio noises. Low-latitude emissions are defined as VLF noises whose data (December 11, 1971 to March 19, 1972) have been used to obtain a full local time coverage. It is then found that the noises are impulsive at that frequency as based on the high values of the observed mean/minimum and peak/mean ratios, and also that they are localized to the longitudes of thunderstorm centers. Hence, low-latitude VLF emissions, as observed in the southern hemisphere for our season, have been found to be well explained by the propagation of sferics to the lowaltitude satellite as short-fractional-hop whistlers; those observed in the northern hemisphere are attributed to the interhemisphere propagation of those sferics in the whistler mode. Furthermore, diurnal variations of the occurrence rate, intensity, spectral property and occurrence latitude are studied, and many of these characteristics are found to be compatible with the ionospheric D-region absorption of whistler-mode propagation in the southern hemisphere. Finally, the satellite VLF data as presented here are suggested as being useful for the study of the global distribution of lightning activity. 1. Introduction Radio noises at VLF and ELF ranges (so-called VLF/ ELF emissions) are known to take place in two major latitude ranges: high- and medium-latitudes. The high-latitude VLF emissions occur mainly in the auroral zone, and the so-called auroral hiss emissions are likely to be generated by Cerenkov instability by while other emissions are generated by different mechanisms (SAZHIN, 1982). The high-latitude emissions have been studied extensively based on ground observations (KOKUBUN et al., 1972; TANAKA et al., 1976) and also in situ measurements (GURNETT and FRANK, 1972; BULLOUGH et al., 1975; LAASPERE and HOFFMAN, 1976). Then, medium-latitude VLF emissions are found to be generally generated at geomagnetic latitudes just around the plasmapause. Many of these emissions are 573

2 574 M. HAYAKAWA closely associated with geomagnetic storms and are excited by energetic electrons in the outer radiation belt in terms of cyclotron instability (BULLOUGH et al., 1969; HAYAKAWA et al., 1977; ONDOH et al.,1983). GURNETT (1968) identified a third region of VLF emissions based on the limited VLF data from the Injun 3 satellite at low latitudes, and later the Ariel 3 satellite has observed those emissions based on data from three months with full local time coverage (LEFEUVRE and BULLOUGH, 1973). The conclusion by those authors is that those emissions are of less intensity than the medium-latitude emissions and that their detection at low latitudes means that they are only observable from low-altitude satellites. In this paper, we study, in greater detail, the morphology of these low- and equatorial-latitude VLF emissions by means of Ariel 4 satellite data, including the geographical location, local time variation in activity and spectrum, nature of the emissions, and so on. Then, we investigate the generation mechanism of those emissions. In conclusion, it is found that although we are interested in the emissions due to the wave-particle interactions, most lowlatitude VLF emissions originate in lightning discharges, and hence the present paper based on the satellite VLF measurement yields the global distribution of lightning activity, which is complementary to the similar satellite observation with optical lightning sensors by VORPHAL et al. (1970), ORVILLE and SPENCER (1979) and TURMAN and EDGAR (1982). 2. Selection of Low-Latitude Emission Data A detailed description of VLF experiments made with the Ariel 4 satellite is given in BULLOUGH et al. (1975), and we present only its essential points in the following. The intensity of the magnetic field of VLF radio noises is measured at several frequencies of 3.2, 9.6 and 16.0kHz with a bandwidth of 1kHz. An assessment of the type of emissions, such as the unstructured hiss type or structured discrete type, can be obtained by the simultaneous measurement of the peak, mean and minimum signals in each 28sec sampling interval, with the time constants of the peak, mean and minimum reading circuits being 0.01, 30 and 0.1sec, respectively. This kind of simultaneous observation of radio noise intensity by different kinds of detectors is known to be very useful in the general study of the properties of radio noises (HAYAKAWA, 1989). We use mainly the VLF data observed at 3.2kHz, and adopt the following selection criteria for identification of a low-latitude emission: Initially we have aimed principally at the study of low-latitude VLF emissions as the consequence of wave-particle interactions, such as in the case of high- and mediumlatitude emissions, and so we have adopted the magnetic latitude as a measure of latitude. Similarly to the case of the Ariel 3 satellite, we can obtain full local time coverage for each three month period (BULLOUGH et al., 1969), and we have selected the period of December 11, 1971 to March 19, 1972 for the present study.

3 Satellite Observation of Low-Latitude VLF Radio Noises Morphology of Low-Latitude Emissions and Their Generation Mechanism 3.1 Occurrence probability According to the criteria given in Section 2, we have first investigated the occurrence probability of low-latitude emissions. Figure 1 illustrates the relationship between the occurrence probability (not the occurrence number) and the geomagnetic activity expressed by the Kp index at the time of observation of the emission. The occurrence probability for each Kp index is defined as the ratio of the orbits along which we could observe the equatorial emission events satisfying the criteria in either the northern or southern hemisphere to the total number of orbits. The vertical rods indicate the standard deviation of the occurrence probability for each Kp index range. The figure may indicate that the occurrence probability seems to exhibit a slight increase with Kp index, and so in the following we discuss different properties for different Kp indices. We have obtained an average occurrence probability of about 5% over the whole Kp range, indicating that low-latitude emissions are not so frequent in occurrence as compared to high- and mediumlatitude emissions as given in BULLOUGH et al. (1969, 1975). Fig. 1. The relationship between the occurrence probability and the Kp index.

4 576 M. HAYAKAWA 3.2 The character of the emissions In order to determine the spectral character of the low-latitude emissions, the peak/mean and mean/minimum ratios at the frequency of 3.2kHz are examined for all low-latitude emission events. Figures 2(a) and 2(b) illustrate the local time variations of the above two ratios for two different Kp ranges. The dots and solid the average values of each ratio every two hours in MLT (magnetic local time) and those with circles indicate that they are not so reliable due to the small number of events. On the right of each figure one finds the occurrence histogram of the ratio. It is apparent from the figures that the peak/mean as well as mean/minimum ratios depend significantly on neither the local time nor the geomagnetic activity. The occurrence frequency of the peak/mean ratio on the right in Fig. 2(a) increases abruptly from 20dB, shows a maximum in the range 25-30dB, and then decreases above 35dB. The average values of that ratio (shown by arrows) range from db for both of two different Kp ranges. Correspondingly, in Fig. 2(b) the occurrence frequency shows a nearly linear increase with the mean/minimum ratio in the range up to 10dB, has a maximum around 12dB, and is followed by a gradual decrease over a wide range in the ratio from 4dB to 20dB, but the average value is also 11.5 db for the disturbed period. By comparing these average values for the peak/mean and mean/ minimum ratios with the corresponding values for the medium-latitude VLF emissions in HAYAKAWA et al. (1977), it is found that the low-latitude noises indicate a very impulsive nature. Although some emissions having smaller values of the mean/minimum ratio (such as less than 5dB) are probably more white noise, such as the unstructured hiss found by GURNETT (1968), when compared with the value of medium-latitude VLF hiss in HAYAKAWA et al. (1977), it seems likely that a great number of low-latitude emissions are very impulsive. 3.3 Geographical location Figure 3(a) shows the geographical location of the low-latitude emissions with geographical longitudinal interval is shown. Any quantity observed at a point is composed of a combination of the component having a local time variation and one with other characteristic variations such as geographical dependence. After having longitude interval, we have found that the average value over the L.T.'s of the occurrence number in a longitude interval exhibits a significant difference from those in other longitude intervals. Hence, Fig. 3(b) is considered to reflect the geographical distribution, and it reveals that the low-latitude emissions are longitudinally localized. Low-latitude emissions are seen to principally occur in two

5 Satellite Observation of Low-Latitude VLF Radio Noises 577 occurrence number. With the increase of geomagnetic activity, the concentration in those longitudes becomes less definite, as seen from the lower panel in Fig. 3(b). Between these regions there are large zones relatively free from emissions, corresponding to the Pacific and Atlantic Oceans. The impulsive nature of the emissions obtained in Subsection 3.2 and their geographical concentration in this subsection suggest that they have their origin in atmospherics (lightning discharges), such that they are the whistler-mode signals of lightning. The geographical regions of low-latitude emissions are all centers of thunderstorm activity as indicated by the WORLD METEOROLOGICAL ORGANIZA- TION (1956) and HAYAKAWA (1989). Southeast Asia is characterized by its high thunderstorm activity, and parts of Indonesia often report thunderstorms more than 350 days per year. In the second region, Africa, the rain forests of the equatorial regions and the heavy rainfall experienced in West Africa can both be expected to be associated with large numbers of thunderstorms. There exists a third region of thunderstorms over the West Indies and South America; however, we did not observe a corresponding noticeable region of emissions over America in the satellite data, as shown in Fig The local time dependence of the equatorial VLF emissions Diurnal variation of occurrence number Figures 4(a) and 4(b) indicate the diurnal variation of the emission activity for the two different Kp ranges at the time of the emission events. The occurrence number in the figures indicates the average over all longitudes when the satellite has surveyed all longitudes uniformly. During quiet periods there is a marked preponderance of emissions during the evening (17 and 18 h MLT) and night (0-5h MLT), and very few emissions occur during daylight hours. Moreover, during increased geomagnetic activity the emission activity is enhanced during the night compared with during the day, and we again observe few emissions around noon Diurnal intensity variation Figures 5(a) and 5(b) illustrate the diurnal variations of the emission intensity measured by the mean reading circuit at 3.2kHz for quiet and moderately disturbed periods, respectively. During quiet times we expect a highly enhanced intensity during 16-18h MLT, when the occurrence rate is also very increased as seen in Fig. 4(a), while the intensity during moderate disturbances seems to show no significant variation with local time Diurnal variation of occurrence region in latitude Figures 6(a) and 6(b) show the magnetic latitude versus magnetic local time for is no remarkable tendency in the latitudinal distribution of radio noises during the nighttime (MLT between 20h and 6h).

6 578 M. HAYAKAWA

7 Satellite Observation of Low-Latitude VLF Radio Noises 579

8 580 M. HAYAKAWA

9 Satellite Observation of Low-Latitude VLF Radio Noises 581 (b) Fig. 3. (continued).

10 582 M. HAYAKAWA

11 Satellite Observation of Low-Latitude VLF Radio Noises 583

12 584 M. HAYAKAWA

13 Satellite Observation of Low-Latitude VLF Radio Noises 585

14 586 M. HAYAKAWA Diurnal variation of spectral properties To investigate diurnal variation in the spectral characteristics of the observed emissions, the mean signal intensity ratio between 3.2 and 9.6kHz was examined for all the emission events satisfying the criteria in Section 2. The difference in db between 3.2 and 9.6kHz is plotted against magnetic local time in Fig. 7 for different amplitude dependence during the night and a 1/12 dependence during the day. An attempt will be made to interpret this great difference in frequency spectrum between day and night later on. 4. Summary of Observational Results and the Association of Low-Latitude Radio Noises with Thunderstorms It appears that the association of low- and equatorial-latitude VLF emissions (radio noises) with thunderstorms is well established, by summarizing the observational results presented in the previous sections and by indicating the following features which link the two:

15 Satellite Observation of Low-Latitude VLF Radio Noises 587 (1) the impulsive nature of emissions, (2) the geographical location, (3) the diurnal variation in activity (occurrence and intensity), (4) the diurnal variation in the spectrum 3.2kHz/9.6kHz. We first summarize the first point, the impulsive nature of the low-latitude emissions. A comparison of the high peak/ mean and mean/ minimum ratios of the emissions observed at 3.2kHz (shown in Fig. 2) with the corresponding values of medium-latitude VLF emissions (hiss and chorus) in HAYAKAWA et al. (1977) has strongly indicated that the observed signals are very impulsive. A further support for this impulsive nature has been provided by the study of BULLOUGH et al. (1975) based on the data from impulse counters at the three frequencies from the same satellite. They have concluded that the duration of the low-latitude emissions has an average of several tens of msec and that many low-latitude emission events are found to be fitted to a log-normal distribution, the average standard deviation of 7 db being in good agreement with the ground-based measurements of atmospherics. One more thing we have to consider here is the effect of other wave phenomena on our measurement at 3.2kHz. The probable and plausible candidate is plasmaspheric ELF hiss (ONDOH et al., 1983; HAYAKAWA et a!., 1985a, 1985b), but its upper cutoff frequency is considered to only rarely extend up to 3.2kHz. However, we have found very few emissions with small mean/ minimum ratios less than 5dB, which might be either the effect of relatively wide-banded plasmaspheric ELF hiss, as found in ONDOH et al. (1983) and HAYAKAWA et al. (1985a), or wide-band VLF The second point, the geographic location, is discussed next. Figure 3 indicates the clear association of the active regions of low- and equatorial-latitude emissions with the main centers of thunderstorms as given in HAYAKAWA (1989). The measurement by optical satellite sensors by TURMAN and EDGAR (1982) has indicated that the shift from northern to southern dominance of lightning is completed by November. Taking into account both this implication and our study period of December to March, the lightning activity seems to be dominant in the southern hemisphere. Hence, the distribution of radio noises observed by this satellite seems to strongly reflect the distribution of lightning in the southern hemisphere, while a considerable part of the radio noises in the northern hemisphere are probably due to whistler-mode propagation in the magnetosphere. We next discuss the third point, the diurnal variation in activity. Both the intensity and occurrence rate (see Figs. 4 and 5) exhibit a conspicuous day-night asymmetry, such that the occurrence of low-latitude emissions is relatively concentrated at night and we notice very low activity during the daytime. As it is closely connected with the third point, we discuss the important property of the fourth point, diurnal spectral variation. Figure 7 illustrates a large difference in the frequency spectrum as defined at the ratio of the intensities at 3.2 and 9.6kHz between night and day. Since the source is located beneath the ionosphere, a large diurnal variation in the emission intensity, activity and spectral property is expected because the power transmitted through the D region is very

16 588 M. HAYAKAWA small during the day at these low latitudes. We have calculated the absorption of upgoing whistler waves through the ionosphere by means of a numerical full-wave method similar to that of PITTEWAY and JESPERSEN (1966). Figure 8 shows the profiles of electron density and collision frequency of the ionospheric models used for the computations. The latitudinal effect is included such that the gyrofrequency is due to a centered dipole model. The result is illustrated in Fig. 9, which takes the form of the ionospheric transmission loss as functions of magnetic latitude and frequency for both daytime and nighttime models. In these calculations we have assumed a vertical incidence. During the night the transmission loss increases with decreasing latitude, but the ratio of decrease is not very high even at low latitudes. The difference in absorption between 3 and 10kHz is 3 db at a magnetic latitude of We cannot account for the diurnal variations of intensity and occurrence rate in Figs. 4 and 5 by using only this large theoretical difference in ionospheric absorption between day and night. The intensity and occurrence rate at MLT=16-20h is very much enhanced despite the large absorption loss in Fig. 9, which implies that the source activity at those LT's is extremely enhanced compared with the nighttime activity, as suggested by VONNEGUT (1982) and TURMAN and EDGAR (1982). However, the observed spectral shape in Fig. 7 is satisfactorily explained by the theoretical calculation for the vertical incidence only, which supports the interpretation that low- and equatorial-latitude radio noises have propagated through the ionosphere from beneath the ionosphere. The last point we have to discuss is the relationship of the low-latitude VLF radio noises with the Kp index as given in Fig. 1. The previous work on mediumlatitude VLF emissions has indicated their strong correlation with geomagnetic activity (Kp)(BULLOUGH et al., 1969; HAYAKAWA et al., 1977), and HAYAKAWA et Fig. 8. The profiles of electron density (day (a) and night (b)) and collision frequency (c) used for the full-wave calculations.

17 Satellite Observation of Low-Latitude VLF Radio Noises 589 Magnetic Latitude (degrees) Fig. 9. The latitudinal dependence of the ionospheric transmission loss of upgoing whistler waves with wave frequency as a parameter. al. (1975) have found that the occurrence probability of mid-latitude VLF emissions exhibits a sharp increase with the Kp index. However, in Fig. 1 the data on the low-latitude emissions are found to indicate a sharp contrast with the case of midlatitude emissions, such that the low-latitude emissions exhibit a slight increase with Kp index. The same kind of relationship between the occurrence of thunderstorms and Kp has never been studied, but MARKSON (1971) and FREIER (1978) have studied the correlation of thunderstorm activity with solar activity, both based on long-term data, which has indicated a slight dependence. As the solar activity seems to be considerably correlated with geomagnetic activity, we can consider it reasonable that the association of thunderstorm activity with Kp is not so strong, just as in Fig. 1. Based on the above conclusion, when the satellite receives VLF radio noises from below the ionosphere as short fractional-hop whistlers as defined in HELLIWELL (1965), we can map the global distribution of lightning activity on the reasonable assumption of quasi-longitudinal propagation. However, the problem is the accuracy (or uncertainty) in locating the lightning activities. The coverage area

18 590 M. HAYAKAWA for whistler transmission into the ionosphere is about 500km radius at high latitudes (HELLIWELL,1965), because the nearly vertical magnetic field results in no azimuthal dependence of the coverage range. However, the situation is complicated at low latitudes by the effects of incidence and magnetic azimuth and, in general, transmission is best when the incident radiation is field-aligned. A southward propagating wave in the northern hemisphere will suffer much less attenuation on transmission through the lower ionosphere than a northward propagating wave whose wave normal makes a very large angle with the local magnetic field. The situation is reversed in the southern hemisphere, and a southward propagating wave will have a very small transmission coefficient south of the equator (PITTEWAY and JESPERSEN, 1966; HAYAKAWA, 1974). High absorption loss as found in Fig. 9 suggests that the coverage range for coupling to short fractional-hop whistlers becomes much smaller at our low latitudes than the value at high latitudes, and that the coverage region is extremely azimuth-dependent as discussed above; therefore, the uncertainty of locating lightning seems to be on the order of a few hundred kilometers. As seen from Figs. 3(a), 4, 5 and 6, despite the inhibiting effect of D region absorption on low-latitude emissions during the day, after MLT=16h at 3.2kHz a band of emissions appears which merges into the nighttime emissions as time progresses. It is expected from the diurnal variation of thunderstorms that morning hours are characterized by little activity, and after mid-day the sun has heated the ground sufficiently for convection and subsequent thunderstorm development to occur (VONNEGUT, 1982; TURMAN and EDGAR, 1982). Whistler occurrence at low latitudes shows a similar diurnal variation (HAYAKAWA and TANAKA, 1978), and this is generally attributed to the combined effect of thunderstorm occurrence and magnetospheric propagation. The optical measurement by TURMAN and EDGAR (1982) has indicated that the shift from northern to southern lightning dominance is completed by November. Taking into account both their implication and our study period, lightning activity seems to be dominant in the southern hemisphere; the dusk activity especially shows such an enhanced north-south asymmetry (TURMAN and EDGAR, 1982). However, their result does not imply that there are no thunderstorms in the northern hemisphere. Hence, most low-latitude noises are well explained by assuming a sub-ionospheric source and propagation up to the satellite in the southern hemisphere in the whistler mode, and the distribution of radio noises can be mapped to that of lightning activity with the location uncertainty mentioned above. However, the low-latitude radio noises observed in the northern hemisphere require further careful attention. When we look at Figs. 3(a), 6(a) and 6(b), the noise intensities observed in both hemispheres are not so different, and hence we are obliged to take into account the interhemispheric propagation of atmospherics from the southern hemisphere in the whistler mode in order to interpret the northern radio noises, while some others have propagated up to the satellite from beneath the ionosphere. There is no way to estimate the exact percentage of the two contributions. The propagation of whistlers at low latitudes has been comprehensively

19 Satellite Observation of Low-Latitude VLF Radio Noises 591 reviewed by HAYAKAWA and TANAKA (1978), who have shown that the situations for whistler propagation during day and nighttime are different, depending on the different ionospheric and magnetospheric plasma conditions. As for the propagation of evening whistlers at MLT=16-20h when we have observed enhanced radio emissions in Fig. 6, the equatorial anomaly in electron density is known to play an important role in whistler propagation (HASEGAWA and HAYAKAWA, 1980). The diurnal variations as in Figs. 4, 5 and 6 seem to result from the joint effects of source intensity, ionospheric absorption and magnetospheric propagation. The equatorial anomaly develops during daytime until MLT=20h and the ionospheric absorption loss maximizes around noon, followed by a decrease with MLT (see Fig. 9). Then, the source activity is likely to be most enhanced around sunset, resulting in the observed diurnal variations in Figs. 4, 5 and 6. Raytracing studies by HASEGAWA and HAYAKAWA (1980) have yielded that a large-scale field-aligned equatorial anomaly acts as a one-sided duct, such that field-aligned ducting is possible in the inner and outer flanks of the anomaly in the whispering-gallery mode. Furthermore, the fact that highly enhanced small-scale irregularities capable of trapping lowlatitude whistlers (HASEGAWA et al., 1978) are possible is associated with the fine structure within the anomaly (BULLOUGH et al., 1969; HAYAKAWA and TANAKA, 1978; TANAKA and HAYAKAWA, 1985). Hence, some of the intense radio noises observed in the northern hemisphere at MLT=16-20h might be the consequence of whistler-mode ducting of atmospherics in conjugate latitudes. As for whistler propagation at night, it seems probable that ground-based propagation, while satellite observations have suggested a non-ducted propagation at L<1.7 (CERISIER, 1973). However, the whistler-mode propagation at latitudes and needs further study. The equatorial anomaly developing during the day dies out at night, and raytracing studies of non-ducted propagation for a typical nighttime electron density profile for the lower exosphere, as shown in Fig. 10(a), are summarized in Fig. 10(b). The rays starting at 300km altitude with vertical wave plasmasphere taking the magnetospheric reflection paths as initially found by KIMURA (1966) and as seen in Fig. 10(b). Hence, some nighttime noises in the northern hemisphere may be due to this kind of non-ducted whistler-mode propagation. However, ONDOH et al. (1979) have suggested that the propagation trapped by field-aligned irregularities associated with sporadic E layers is plausible hemisphere propagation in either the non-ducted or ducted mode has also been identified in ground-based multi-station measurement for global distribution of lightning by the use of VLF atmospherics (VoLLAND et al., 1983).

20 592 M. HAYAKAWA (a) (b)

21 Satellite Observation of Low-Latitude VLF Radio Noises Conclusion The low- and equatorial-latitude VLF emissions as observed by the Ariel 4 satellite have been found to be well explained by assuming thunderstorms as their source. They are localized in longitude to the centers of thunderstorm occurrence, and the character of the emissions is very impulsive because the observed mean/ minimum and peak/ mean ratios are high. Many of the characteristics of the emissions can be attributed to D region absorption and whistler-mode propagation in the lower exosphere. Taking into account the present study period of December to March, the lightning activity is predominant in the southern hemisphere over the northern hemisphere, especially at dusk, according to TURMAN and EDGAR (1982). Hence, the observed global distribution of radio noises in the southern hemisphere is reasonably considered to be the global distribution of lightning. In contrast, the radio noises observed in the northern hemisphere result from two contributions; one is the atmospherics coming up to the satellite from below in the whistler mode, and the other is atmospherics propagated from the opposite hemisphere in the whistler mode. Even for whistler-mode waves from the opposite hemisphere, knowledge of the propagation mechanism in the magnetosphere enables us to locate the input region as causative lightning regions. In conclusion, this kind of VLF radio noise in satellites can be extensively used for study of the global distribution of lightning activity, complementary to other techniques such as optical measurements (VORPHAL et al., 1970; ORVILLE and SPENCER, 1979; TURMAN and EDGAR, 1982) and VHF radio noises (KOTAIU and KATOH, 1983). There are several important problems to be solved concerning these equatorial radio emissions. Although some discussions have been made on the propagation mechanism of whistler waves in the inner magnetosphere, further detailed investigation of magnetospheric propagation toward the northern hemisphere is required. Next, TATNALL (1978) has suggested, based on impulse counter data, that some amplification or storage process is taking place for about 10% of the low-latitude emissions, being very strong because the amplitude and frequency of atmospherics required to produce the observed mean reading are prohibitively high. Detailed study is sorely needed on this interesting implication. Then, the effect of these low-latitude VLF radio noises on other geophysical phenomena in the inner plasmasphere is a future subject of study, and we can show an example of this kind. LEFEUVRE and BULLOUGH (1973) have found a close correlation between the locations of these low-latitude emissions and suprathermal electrons, which implies that there may be some kind of interconnecting physics between these two phenomena. Finally, although our principal aim was to make a further study of VLF emissions which might be the consequence of wave-particle interaction in the inner magnetosphere, a considerable number of emissions observed are found to originate in lightning, and so we would like to investigate the contribution of the VLF hiss as found by GURNETT (1968).

22 594 M. HAYAKAWA This work was started while the author was at the Department of Physics, the University of Sheffield. The author wishes to express his sincere thanks to Prof. T.R. Kaiser and Dr. K. Bullough of Sheffield University, UK for their kindly providing him with the satellite data. Useful discussions with both of them and Dr. A.R.L. Tatnall are appreciated, and thanks are also due to Mr. M. Ashworth for his contribution to the computer data reduction. This work is, in part, financially supported by the Secom Foundation's Research Grant, to which the author expresses thanks. Finally, the author is grateful to Dr. A. Kimpara, Professor Emeritus of Nagoya University for his continued encouragement and to the two referees for their useful criticisms. REFERENCES BULLOUGH, K., A.R.W. HUGHES, and T.R. KAISER, Satellite evidence for the generation of VLF emission at medium latitudes by the transverse resonance instability, Planet. Space Sci.,17, , BULLOUGH, K., M. DENBY, W. GIBBONS, A.R.W. HUGHES, T.R. KAISER, and A.R.L. TATNALL, ELF/ VLF emissions observed on Ariel 4, Proc. Roy. Soc. London, A343, , CERISIER, J.C., A theoretical and experimental study of non-ducted VLF waves after propagation through the magnetosphere, J. Atmos. Terr. Phys., 35, 77-94,1973. FREIER, G.D., A 10-year study of thunderstorm electric fields, J. Geophys. Res., 83, , GURNETT, D.A., Observation of VLF hiss at very low L values, J. Geophys. Res., 73, , GURNETT, D. A. and L. A. FRANK, VLF hiss and related plasma observations in the polar magnetosphere, J Geophys. Res., 77, , HASEGAWA, M. and M. HAYAKAWA, The influence of the equatorial anomaly on the ground reception of non-ducted whistlers at low latitudes, Planet. Space Sci., 28, 17-28, HASEGAWA, M., M. HAYAKAWA, and J. OHTSU, On the duct trapping conditions of low-latitude whistlers, Ann. Geophys., 34, , HAYAKAWA, M., A study of the propagation of whistler waves in the magnetospheric and ionospheric plasmas, Ph. D. Thesis, Nagoya Univ., HAYAKAWA, M., Recent progress in the study of low-latitude whistlers, Proc Int'l. Symp. on Radio Propagation, Beijing, pp , HAYAKAWA, M., Radio noise theory and measurement, Proc. Int'l. College on Theoretical and Experimental Radiopropagation Physics, Int'l. Center for Theoretical Phys., Trieste, Italy, 1989 (in press). HAYAKAWA, M. and Y. TANAKA, On the propagation of low-latitude whistlers, Rev. Geophys. Space Phys., 16, , HAYAKAWA, M., Y. TANAKA, and J. OHTSU, On the morphologies of low-latitude and auroral VLF 'hiss', J. Atmos. Terr. Phys., 37, , HAYAKAWA, M., K. BULLOUGH, and T.R. KAISER, Properties of storm-time magnetospheric VLF emissions as deduced from the Ariel 3 satellite and ground-based observations, Planet. Space Sci., 25, , HAYAKAWA, M., T. OKADA, and Y. TANAKA, Morphological characteristics and the polarization of HAYAKAWA, M., F. LEFEUVRE, and J.L. RAUCH, The direction finding aboard Aureol-3 of elf waves at frequencies above and below the proton gyrofrequency, in Resultat du Project ARCAD 3 et des Programmes Recents en Physique de la. Magnetosphere et de l'ionosphere, Toulouse 84, pp , Cepadues Edit., Toulouse, 1985b. HELLIWELL, R.A., Whistlers and Related Ionospheric Phenomena, Stanford Univ. Press, California, JORGENSEN, T.S., Interpretation of auroral hiss measured on Ogo 2 and at Byrd Station in terms of incoherent Cerenkov radiation, J. Geophys. Res., 73, , 1968.

23 Satellite Observation of Low-Latitude VLF Radio Noises 595 KIMURA, I., Effects of ions on whistler mode ray tracing, Radio Sci., 1, ,1966. KOKUBUN, S., K. MAKITA, and T. HIRASAWA, VLF-LF hiss during polar substorm, Rep. Ionosph. Space Res. Japan, 26, ,1972. KOTAKI, M. and C. KATOH, Global distribution of atmospheric radio noise derived from distribution of lightning activity, J. Radio Res. Labs. Tokyo, 30, 35-57, LAASPERE, T. L. and R. A. HOFFMAN, New results on the correlation between low-energy electrons and auroral hiss, J. Geophys. Res., 81, , LEFEUVRE, F. and K. BULLOUGH, Ariel3 Evidence of Zones of VLFEmissions which Co-Rotate with the Earth, Space Res., 13, pp , Akademie Verlag, Berlin, MARKSON, R., Consideration regarding solar and lunar modulation of geophysical parameters, atmospheric electricity and thunderstorms, Pageoph, 84, , ONDOH, T., M. KOTAKI, T. MURAKAMI, S. WATANABE, and Y. NAKAMURA, Propagation characteristics of low-latitude whistlers, J. Geophys. Res., 84, ,1979. ONDOH, T., Y. NAKAMURA, S. WATANABE, K. AIKYO, and T. MURAKAMI, Plasmaspheric hiss observed in the topside ionosphere at mid- and low-latitudes, Planet. Space Sci., 31, , ORVILLE, R.E. and D.W. SPENCER, Global lightning flash frequency, Mon. Weather Rev., 107, , PITTEWAY, M.L.V. and J.L. JESPERSEN, A numerical study of the excitation, internal reflection and limiting polarization of whistler waves in the lower ionosphere, J. Atmos. Terr. Phys., 28, 17-41, SAZHIN, S.S., Natural Radioemissions in the Earth's Magnetosphere, 158 pp., Nauka, Moscow, TANAKA, Y. and M. HAYAKAWA, On the propagation of daytime whistlers at low latitudes, J. Geophys. Res., 90, ,1985. TANAKA, Y., M. NISHINO, and M. HAYAKAWA, Study of auroral VLF hiss observed at Syowa Station, Antarctica, Mem. Natl. Inst. Polar Res., Tokyo, No. 16, Ser. A., 58 pp., TATNALL, A.R.L., A satellite study of power line harmonic radiation, thunderstorm noise, and associated magnetospheric phenomena, Ph. D. Thesis, Sheffield Univ., TURMAN, B.N. and B.C. EDGAR, Global lightning distributions at dawn and dusk, J. Geophys. Res., 87, , VOLLAND, H., J. SCHAFER, P. INGMAN, W. HARTH, G. HEYDT, A.J. ERIKSSON, and A. MANES, Registration of thunder centers by automatic atmospherics stations, J. Geophys. Res., 88, , VONNEGUT, B., The physics of thunderclouds, in Handbook of Atmospherics, Edited by H. Volland, p. 4, CRC-Press, Boca Raton, Florida, VORPHAL, J.A., J.G. SPARROW, and E.P. NEY, Satellite observations of lightning, Science, 169, , WORLD METEOROLOGICAL ORGANIZATION, World distribution of thunderstorm days, Report TP 21, Geneva, XU, J.S., M. HAYAKAWA, M. TIAN, K. OHTA, C.C. TANG, and S. SHIMAKURA, Direction finding of nighttime whistlers at very low latitudes in China: Preliminary results, Planet. Space Sci., 1989 (in press).