Irregularities of the topside ionosphere observed by GPS/MET radio occultation

Transcription

1 RADIO SCIENCE, VOL. 37, NO. 6, 1101, doi: /2001rs002599, 2002 Irregularities of the topside ionosphere observed by GPS/MET radio occultation K. Hocke and K. Igarashi Communications Research Laboratory, Koganei, Tokyo, Japan A. Pavelyev Institute of Radio Engineering and Electronics, Fryazino, Moscow, Russia Received 25 December 2001; revised 23 May 2002; accepted 29 June 2002; published 27 November [1] Large-scale fluctuations (vertical scales km) of total electron content (TEC) are retrieved from phase path fluctuations of GPS radio occultation links of the GPS/MET experiment. The large-scale TEC fluctuations at observation heights km are maximal at local times from 20:00 to 24:00 h. The fluctuations mainly occur on the summer hemisphere at low magnetic latitudes 0 30 N and seem to be concentrated over Africa and Arabia. The derived global distribution of TEC fluctuations around solar minimum (June/July 1995) is assumed to be closely related to the occurrence of spread F and radio scintillations. The meridional distribution of the background electron density (averaged for 19:00 23:00 LT) has a maximum in the topside ionosphere at around 10 N, beyond an anomaly structure of the F 2 peak layer. The meridional location of the anomaly structure agrees with the location of enhanced TEC fluctuations. Both phenomena are possibly related to the enhancement of upward ion velocity in the postsunset ionosphere at equatorial latitudes. Large TEC depletions of 1 20 TEC-units are present at observation heights km in many TEC profiles of the equatorial ionosphere after sunset. The TEC depletions possibly indicate plasma bubbles on their way from the lower ionosphere to the topside ionosphere. GPS navigation signal loss due to spread F or highlatitude irregularities is not found in the GPS/MET data of June/July INDEX TERMS: 2415 Ionosphere: Equatorial ionosphere; 6979 Radio Science: Space and satellite communication; 2427 Ionosphere: Ionosphere/atmosphere interactions (0335); 2439 Ionosphere: Ionospheric irregularities; KEYWORDS: GPS radio occultation, topside ionosphere, spread F Citation: Hocke, K., K. Igarashi, and A. Pavelyev, Irregularities of the topside ionosphere observed by GPS/MET radio occultation, Radio Sci., 37(6), 1101, doi: /2001rs002599, Introduction [2] Analysis of phase and amplitude data of GPS receivers in low-earth orbit (LEO) is of interest for autonomous GPS navigation of LEO satellites, improvement of receiver and tracking software, recognition/ understanding of the variable morphology of ionospheric irregularities, and study of the impact of ionospheric irregularities on radio signals. An initial and detailed study has been carried out by Kramer and Goodman [2001] analyzing scintillations of GPS navigation signals received by the Space Shuttle. Enhanced GPS scintillations are observed when the radio links are perpendicular Copyright 2002 by the American Geophysical Union /02/2001RS to the flight direction of Space Shuttle. In this case the relative movement between receiver and the irregularity structures is maximal (around 8 km/s). According to the theory presented by Hinson and Tyler [1982] radio scintillations or TEC fluctuations depend on the ionospheric irregularities (axial ratio, angular orientation of the anisotropic irregularities, power spectrum of electron density fluctuations) as well as on the movement and direction of the transmitter-receiver link relative to the irregularities. [3] Radio communication and navigation depends also on the lengths of signal fade times. VHF radio signals, propagating from geostationary satellites to airborne receivers, show lengthened fade times when the movement of the receiver matches the ionospheric drift speed [Aarons et al., 1980]. GPS radio propagation effects

2 13-2 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE depending on the relative velocity between satellite, ionospheric irregularities, and ground-based receiver have been recently investigated by Kintner et al. [2001] using GPS data from a receiver network in Brazil. [4] Irregularities of the F-region and topside ionosphere mainly occur at high geomagnetic latitudes (ionospheremagnetosphere processes) and at low geomagnetic latitudes (thermosphere-ionosphere coupling). The tidal winds of the upper atmosphere lead to an electric current system between E and F region at low latitudes with downward ion velocity during nighttime and upward ion velocity during daytime (electrodynamic lifting). The present study is focused on the plasma irregularities of the postsunset phase when the thermosphere-ionosphere system becomes unstable. This is due to the sudden disappearance of ionization in the lower ionosphere after sunset (fast recombination processes and loss of photoionization). As a result strong eastward electric fields are induced in the E-region, and a sharp enhancement of upward ion velocity occurs, the so-called prereversal enhancement [Heelis et al., 1974]. Because of high ion velocities and dynamic processes, generation of ionospheric irregularities is favored during the postsunset phase. According to a review article by Abdu [2001] the prereversal enhancement of zonal electric fields and vertical ion velocity causes plasma bubbles (electron density depletions along the magnetic flux tubes) rising from the lower ionosphere into the topside ionosphere. The bubble irregularities grow under the generalized Rayleigh-Taylor instability and cascade into a wide spectrum of scale sizes from a few hundreds of kilometers to a few meters [Haerendel, 1973]. The small scale irregularities yield enhanced backscatter and spread F in radar and ionosonde observations [Tsunoda, 1985]. [5] The forces for rise and growth of plasma bubbles are gravity (buoyancy of depleted neutral density cells) and upward E B-drift due to induction of eastward electric fields in the lower ionosphere. Atmospheric gravity waves are sometimes regarded as seeding mechanism causing the initial density depletions at E-region heights [e.g., Anderson et al., 1982; Röttger, 1981], while meridional thermospheric winds may suppress the growth of plasma bubbles [Maruyama and Matuura, 1984]. Enhanced plasma bubble activity can be associated to a well-developed equatorial ionization anomaly [Raghavaro et al., 1988]. The morphology of equatorial irregularities is only known in fragments [Aarons, 1993]. Since various coupling processes of ionosphere and lower/ middle/upper atmosphere participate in generation of plasma bubbles/spread F, the morphology depends on longitude, season, solar cycle, day-to-day variability of F- region ionization, atmospheric winds and waves. Global observations by an ionosonde on the ISS-b satellite gave evidence for strong seasonal and longitudinal variability of spread F significantly differing from ground-based radar observations in the American and Pacific longitude sectors [Maruyama and Matuura, 1984; Tsunoda, 1985; Aarons, 1993]. Further worldwide monitoring of spread F from space and ground is certainly desirable for detailed description and understanding of the spread F morphology and for improved prediction of radio scintillation occurrence. [6] In the present study we analyze observations of the GPS/MET radio occultation experiment which has been a proof-of-concept for measurement of temperature in the lower atmosphere from space [Rocken et al., 1997]. The GPS radio occultation technique consists in limb sounding of the atmosphere and ionosphere using a radio link between a GPS satellite and a GPS receiver onboard of a LEO satellite. Ionosphere sounding by spaceborne GPS receivers is a rather new discipline. A topside ionosonde certainly collects more information on the state of the upper ionosphere than a single GPS receiver in low-earth orbit. The attractiveness of spaceborne GPS ionosphere sounding will consist in the analysis of a large amount of GPS navigation and occultation data provided by an increasing number of LEO satellites. This might enable a continuous monitoring of the ionosphere and its irregularities with a relative high spatial-temporal resolution and global coverage in the near future [Hajj et al., 2000]. [7] The TEC fluctuations (vertical scales km) observed by GPS/MET possibly occur in coincidence with small-scale fluctuations observable by ground-based radars or topside ionosondes. This is suggested by theory as mentioned above [Haerendel, 1973] and by numerous impressive radar soundings of spread F where enhanced small-scale irregularities are embedded in patches of larger scales. Characteristics and occurrence of largescale TEC fluctuations of GPS radio occultation are firstly investigated and discussed here. 2. GPS/MET Radio Occultation Experiment [8] The GPS/MET radio occultation experiment started in April 1995 with the launch of the satellite Microlab-1 in a low-earth orbit (h = 735 km, inclination 70 ). Using the phase-lock-loop technique, the space-qualified GPS TurboRogue receiver of Microlab-1 separately measures phases and amplitudes of the L1 and L2 radio signals of GPS satellites. The sampling rate of the receiver is 50 Hz for limb sounding of the atmosphere below 120 km height and 0.1 Hz (or sometimes 1 Hz) for the ionosphere beyond [Schreiner et al., 1999]. The TurboRogue receiver is a low-power, sparse-sampling receiver developed at NASA s Jet Propulsion Laboratory. In the 0.1 Hz sampling mode the receiver awakes every 10 seconds from a sleep mode and measures a period of 20 ms long. In the present study we only use 0.1 Hz GPS occultation data.

3 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE 13-3 Figure 1. Scheme of GPS limb sounding of the topside ionosphere and its irregularities. Radio rays (dotted lines) from a far-distant GPS satellite are received by a low-earth orbit satellite (LEO). The geometrical path of a radio occultation ray is maximal inside the tangent point layer (thick solid line indicates this ray path segment). [9] The measurement configuration is illustrated in Figure 1 for the topside ionosphere. The distribution of irregularities is sketched by light and dark patches corresponding to depletions and enhancements of electron density. Generally the electron density decreases with height in the topside ionosphere. The radio ray from GPS to LEO satellite is approximately a straight line through the ionosphere. The TEC contribution resulting from the ionospheric layer at the ray tangent point is usually maximal because of the decrease of electron density in the layers beyond and because of maximal geometrical path length of the signal within the tangent point layer (Figure 1). TEC is proportional to the phase path difference of the simultaneously measured L1 and L2 GPS signals and can be represented as a function of tangent point height. [10] The present study favors radio occultation links with tangent points in the topside ionosphere (beyond 400 km height) for sounding of topside irregularities. Recently an alternative approach has been suggested for retrieval of high-altitude ionospheric irregularities from occultation links with tangent points less than around 120 km height (50 Hz high rate data). The method consists in back propagation of the GPS radio field measured along the LEO orbit and mainly provides information on the locations of irregularities somewhere between GPS and LEO satellite [Sokolovskiy et al., 2002; Gorbunov et al., 2002]. Application of this radio holographic method to a large amount of occultation events, and geophysical interpretation of the results would be interesting. [11] For estimation of the sink velocity or vertical velocity of the tangent point the slow angular movement of the GPS satellite can be neglected since the revolution time of a GPS satellite is around 12 h, while the revolution time of Microlab-1 is just around 100 min. The vertical velocity of the tangent point is estimated by v w r sin j (Figure 1). For r = km, w =2p/ 100 min, and j =15 at h = 500 km the vertical velocity of the tangent point is around 1.9 km/s. In case of a sampling frequency of 0.1 Hz the vertical distance of successive tangent points is around 19 km. According to the Nyquist theorem a measurement with a sampling rate of 0.1 Hz only provides information on structures with vertical scales larger than 38 km. So we roughly estimate a height step of 20 km for the 0.1 Hz data, and a cut off for ionospheric structures smaller than 40 km. [12] A limitation to vertical scales greater than 40 km (for 0.1 Hz sampling rate) is certainly not optimal for a study of ionospheric fluctuations but it is shown later that low rate GPS occultation data contain valuable information on topside irregularities with vertical scales of km. Higher sampling rates (up to 100 Hz) of the GPS receiver are feasible and would provide improved recordings of GPS scintillations induced by the upper iono-

4 13-4 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE sphere (but with the penalty of increasing data rate and workload of the receiver). [13] The GPS/MET radio links are always selected in backward direction, and the azimuth angle between GPS- LEO connection line and the LEO orbit plane is less than 30. Statistical analyses of the performance of GPS receivers and occurrence of ionospheric irregularities are partly influenced by signal aquisition thresholds and tracking software of the receiver selecting the GPS radio links. The radio occultation measurements of the GPS/ MET experiment inform about vertical ionospheric structures since the tangent point mainly moves in vertical direction. [14] The GPS radio scintillations are caused by the natural plasma fluctuations of the ionosphere with fluctuation scales from 10 cm to 1000 km. An investigation of aliasing effects due to GPS scintillations at frequencies >0.1 Hz (resulting from plasma fluctuations with scales less than 20 km across to the occultation ray) requires 50 Hz or 100 Hz occultation links in the topside ionosphere. To our knowledge such an investigation has not yet been performed for GPS radio occultation of the Earth s ionosphere. Analyzing scintillations in radio occultation measurements of Jupiter s ionosphere, Hinson and Tyler [1982] found power law spectra with decreasing power towards high frequencies. [15] A rough estimate of aliasing effects can be obtained by a simple simulation assuming that electron density fluctuations are proportional to GPS phase path fluctuations and neglecting diffraction effects. For this aim an artificial spectrum of sine waves with random phases has been generated following a k 2 power law where k the vertical wave number is. This power law of electron density fluctuations has been observed by rockets in the equatorial ionosphere [Hysell et al., 1994]. At high wave numbers (k = 100/km) a knee of the spectrum occurs, and fluctuations begin to follow a power law of k 5. The normalized power spectrum is shown by the dotted line in Figure 2. For illustration the knee of the spectrum is shifted from 100/km to 50/km, otherwise the knee would not appear in the plot and simulation range. By means of FFT a series of electron density fluctuations as function of height is calculated from the model spectrum. This fluctuation series (5 m height resolution) is now sampled with a height step of 20 km (corresponding to 0.1 Hz GPS sampling rate) and 40 m (corresponding to 50 Hz GPS sampling rate). Then power spectra are calculated from the resampled fluctuation series and are depicted as solid lines in Figure 2. In both cases an average of 10 spectra has been taken for a better illustration of the aliasing effects. The filter band width which is applied to the 0.1 Hz GPS data of the present study is indicated by the two dash-dotted vertical lines in the lower panel of Figure 2. Aliasing effects produce in average a positive bias of magnitudes and hide the spectral slope. However the Figure 2. Rough estimation of aliasing effects in radio occultation data. The model spectrum is shown by the dotted line. The solid lines are the power spectra which are obtained when the fluctuation series (derived by FFT from the model spectrum) is resampled with 50 Hz (upper panel) and 0.1 Hz (lower panel).the applied band pass filter for fluctuations with vertical scales of km is indicated by the two vertical dash-dotted lines in the lower panel for 0.1 Hz sampling rate. aliasing bias will be present in all 0.1 Hz occultation measurements, and if we concentrate on the discussion of the relative fluctuation power (e.g., difference of power in the filter band for quiet and disturbed ionosphere), then the low rate 0.1 Hz GPS data are valuable. [16] The potential of radio occultation for measurement of fluctuation scales down to 100 m (wave number 10/km) in the F-region can be anticipated by the power spectrum of the 50 Hz GPS data in Figure 2. However it should be borne in mind that occultation is an integral measurement of the electron content along the ray. So the electron density fluctuations with small scales (across to the moving ray) should be extended along the ray (e.g., km) in order to generate a recognizable ionospheric phase path variation of the GPS signal. Such a situation occurs if the GPS-LEO occultation ray becomes parallel to the geomagnetic field lines (e.g., GPS profiling of electron density depletions stretched along magnetic field lines in the equatorial ionosphere). [17] A significant handicap of the GPS/MET mission was the GPS signal encryption (anti spoofing) leading to

5 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE 13-5 a low signal to noise ratio of the recorded L2 signal. Thus the following data analysis can be only applied to the short time intervals of several weeks of high quality data of the GPS/MET experiment when GPS anti spoofing was switched off by the US Department of Defense. Meanwhile new GPS antenna and receiver systems have improved signal to noise ratios, and high quality measurements of phase path can be obtained during GPS anti spoofing on, so that a continuous monitoring of ionospheric irregularities is now feasible for upcoming GPS radio occultation missions [Wickert et al., 2001]. The present study is further limited to the ionospheric data of June/July 1995 which has been kindly provided by University Corporation for Atmospheric Research (UCAR) in Boulder, leading and operating the GPS/ MET experiment. 3. Data Analysis and TEC Fluctuation Profiles [18] Vertical profiles of phase path difference, TEC, and electron density are analyzed. An individual profile contains temporal and spatial variations of the topside ionosphere, since GPS/MET requires a few minutes for profiling the height range h = km. The present study favors this height range since 400 km is well beyond the F 2 peak layer, and the occultation recordings of GPS/MET begin at least in 600 km height of the ray tangent point (often at km). For a statistical analysis the assumption of a frozen distribution of plasma irregularities can be applied here, that means, the vertical fluctuations of the TEC and electron density profiles are representing in average the spatial structures of ionospheric irregularities. [19] Electron density profiles are provided by UCAR and have been derived by means of the Abel transformation from TEC gradient profiles [Schreiner et al., 1999, 1998]. In total 3495 ionospheric profiles have been observed from 19 June to 10 July TEC is calculated from the phase path difference of the L1 (l 1 =19 cm, f 1 = MHz) and L2 (l 2 = 24.4 cm, f 2 = MHz) GPS signals TECðr o Þ ¼ f1 2f :3 f ðs 1 ðr o Þ S 2 ðr o ÞÞþk: ð1þ [20] Units are [electrons/m 2 ] for TEC, [Hz] for f 1 and f 2, and [m] for phase paths S 1 and S 2. r o denotes the geocentric radial distance of the ray tangent point. The constant k is related to the signal phase integer ambiguity. k is of no further interest since only the fluctuations of TEC will be regarded in the following. [21] Because of the low sampling rate of the GPS measurements high frequency fluctuations with vertical wavelengths less than around 30 km are not present in the data. The ionospheric profiles largely differ with local time and place of observation and require a robust filter method for an extraction of the fluctuations dtec(r o ) from the profile TEC(r o ). The filter method must be flexible enough to handle all TEC profiles at all local times and geographic locations. This is the reason why a sliding window average has been selected as reliable low-pass filter. [22] The TEC(r o ) profile is filtered by means of a sliding window average with a window length of 60 km. This yields a low-pass filtered profile htec(r o )i containing fluctuations with vertical wave lengths greater than around 60 km. The difference of TEC(r o ) and htec(r o )i gives the fluctuation profile dtec b (r o ) which is shown by the dotted line in the middle column of Figure 3. The filtering process has no sharp cut off at 60 km, and components with vertical wave lengths around km may pass the filter with some damping of their amplitudes. We also tested sliding window lenghts of 50 and 70 km. The main results of the present study do not change by variation of the window size. A window size exceeding 70 km is not desirable, since the height variation of the background ionosphere at h = km could be partly interpreted as fluctuation. [23] The exponential decrease of electron density in the topside ionosphere can cause systematic biases in the dotted TEC fluctuation profiles (e.g., at h = km). These biases are removed by repeating the previous filtering procedure. The fluctuation profile (dotted line) is averaged by a 60 km-sliding window, and the average is subtracted from the dotted line, yielding the final and relatively unbiased fluctuation profile dtec(r o ) as shown by the solid line in the middle column of Figure 3. Obviously dtec(r o ) mainly contains fluctuations with vertical wave lengths around km. [24] The TEC fluctuations (dtec) are directly related to variations of the GPS ray bending angles causing focusing and defocusing of the GPS signal. Thus the TEC fluctuations (dtec) are more appropriate than the relative TEC fluctuations (dtec/hteci) for the present study. In addition relative TEC fluctuations are depending too much on the background, in particular if the background electron density vanishes. For analysis of neutral-wind induced, ionospheric perturbations and in regional areas with small background variation the relative TEC fluctuations should be favorable. [25] Figure 3 depicts two examples of a disturbed topside ionosphere at low latitudes during evening/nighttime (top panels) and two examples of a quite ionosphere (bottom panels). The TEC profiles (left-hand-side) can be compared to the retrieved electron density profiles (right-hand-side). Obviously, shapes and structures of the profiles agree well. This indicates a significant influence of the electron density of the lowest layer (tangent point

6 13-6 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE Figure 3. Selected profiles of four radio occultation events of the GPS/MET experiment (axis notation height means the height of the ray tangent point).the TEC profiles are on the left-handside, TEC fluctuations (vertical scales km) are in the middle, and the corresponding electron density profiles (retrieved by Abel inversion) are on the right-hand-side. Locations and times are shown as figure inlets on the left-hand-side (negative longitude corresponds to west). layer) on the total electron content integrated along the radio ray. We favor the analysis of the TEC fluctuations rather than of electron density fluctuations since the electron density profiles could be biased by retrieval errors of the Abel inversion (e.g., errors due to spherical symmetry assumption). 4. Results 4.1. Distribution of TEC Fluctuations [26] An average amplitude s of TEC fluctuations is calculated for each dtec profile using dtec values from 400 to 600 km height of the ray tangent point sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 1 X n s ¼ ðdtecþ 2 : ð2þ n 1 i¼1 [27] Figure 4 shows the locations of the observed radio occultations in June/July 1995 (tangent points at h = 500 km are marked). Each dot radius is linearly proportional to the fluctuation amplitude s of the corresponding dtec profile. The dot size (pr 2 ) is proportional to the variance s 2. The largest dot radius in Figure 4 corresponds to a value of electrons/m 2 or 1.10 TEC-units. The figure depicts occultation events as function of geomagnetic latitude and magnetic local time. Magnetic local time has been selected because of occultations at higher latitudes. However the figure differs only marginal if solar local time is used instead of magnetic local time. The S curve formed by the dots corresponds to the orbit trajectory of Microlab-1 (tangent points of the occultations are less than 15 apart from Microlab-1 trajectories). In June/ July 1995 the sun was within the orbit plane of Microlab- 1. Each day the orbit plane precesses by 3.3 relative to the sun, this means the S trajectory of Microlab-1

7 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE 13-7 Figure 4. Occultation events in June/July 1995 in dependence on geomagnetic latitude and geomagnetic local time. Negative latitude or longitude corresponds to south or west respectively. The dot radius is linearly proportional to the average TEC fluctuation amplitude s (vertical scales km) of the observation height range h = km. The largest dot radius corresponds to 1.10 TEC-units. The crosses denote interrupted, failed occultation measurements. moves each day by around 15 minutes to earlier times in Figure 4 [Schreiner et al., 1998]. [28] At the end of the observation time interval some measurement failures occur which are denoted by crosses in Figure 4. The failures correspond to measurement interruptions (around 2% of all occultations). Since the failures only occur at the end of the observation time interval in July 1995 and since they do not appear together with large dots of TEC fluctuations (e.g., crosses in the morning sector), these failures are possibly not related to ionospheric irregularities but to the tracking software or performance of the experiment. Moreover geomagnetic activity was minimal when the failures occurred, and the crosses are worldwide equally distributed (Figure 5). So the failures have possibly not a geophysical reason such as enhanced particle precipitation from the magnetosphere. [29] In Figure 4, enhanced TEC fluctuations (large dots) are obviously observed between 20 and 24 LT (or MLT) at low northern geomagnetic latitudes from 0 to 30 N. The TEC fluctuations are possibly related to the onset of equatorial spread F at around 21 LT or at least to unstable ionospheric conditions of the postsunset phase. An extended occurrence of spread F from equatorial to mid latitudes of the summer hemisphere during years of low sunspot numbers has been noted by Tao [1965] analyzing data of the worldwide ionosonde network in A higher activity of mid-latitude spread F and equatorial plasma depletions has been also noted in more recent observations during solar minimum years [Kelley and Miller, 1997; Fejer et al., 1993]. These observations are in qualitative agreement with our result for June/July 1995 (low sunspot number). Tao [1965] further found that spread F occurrence was restricted to the equatorial region and high latitudes during the International Geophysical Year (1958) which has a high sunspot number. [30] The spaceborne GPS/MET observations of enhanced TEC fluctuations during the postsunset phase at low latitudes qualitatively agree with long-term observations of GPS scintillations by ground-based GPS receivers. By means of theory, results of ground-based observations at transmitting frequencies from VHF to L- band have been summarized into a wideband ionospheric scintillation model (WBMOD) providing worldwide and regional maps of occurrence percentage of scintillations [Secan et al., 1995]. Though the observation geometry is different than in case of GPS radio occultation, it is likely that the climatological worldwide maps of the WBMOD scintillation model are useful for spaceborne GPS navigation and remote sensing. [31] On the other hand an improvement of the scintillation maps might be possible by using the spaceborne GPS data in addition to the ground-based GPS data. The limb sounding technique (radio occultation) can in particular separate topside irregularities from irregularities in the E- region. Lower ionospheric irregularities are excluded by taking occultation links with tangent points beyond the E- region. In the present work a direct comparison between ground- and space-based GPS data is not carried out because of the relative small amount of GPS/MET data. Ground-based scintillation measurements and the climatological maps of WBMOD possibly play an important role for validation of spaceborne GPS measurements of ionospheric irregularities. Figure 5. Same as Figure 4, but for TEC fluctuations in dependence on geographic latitude and longitude. Contours are drawn at geomagnetic dip angles of 0, ±30, ±70.

8 13-8 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE [32] Figure 5 shows the geographic positions of the occultation events. The TEC fluctuations seem to be maximal over Africa-Arabia. However, statistics of the 3-week observation interval of GPS/MET are not sufficient for detailed geophysical interpretations. Enhanced GPS scintillations over Africa have been observed in June 1999 by Kramer and Goodman [2001] analyzing GPS navigation data received by the Space Shuttle. Contrary to the present study the radio navigation links of Space Shuttle have positive elevations and cover all azimuths. Space Shuttle might be inside the height range km which has been sounded by GPS/MET Microlab-1 from outside, in a orbit height of 735 km. [33] The dependence of topside irregularities on geomagnetic longitude and magnetic local time is depicted in Figure 6. GPS/MET covered all geomagnetic longitude sectors during June/July Geomagnetic longitude around 0 corresponds to the American sector. Enhanced irregularities are between 20 and 24 MLT at 90 geomagnetic longitude which is the Africa-Europe sector. There might be also a slight enhancement of fluctuations over the Pacific sector (dots around 120, MLT) Background Ionosphere [34] Electron density profiles of the local time interval from 19:00 to 23:00 h are sorted according to their geomagnetic latitude and averaged by a sliding window of 15 in latitude. This yields a meridional slice of average electron density of the postsunset ionosphere around 21:00 LT. Figure 7 depicts the electron density slice together with the number of occultation events available for the sliding window average. As expected the electron density is higher in the summer (northern) hemisphere. The anisotropic distribution of the electron density around Figure 6. Same as Figure 4, but for TEC fluctuations in dependence on geomagnetic longitude and geomagnetic local time. Geomagnetic longitude 0 corresponds to the American sector while the Africa-Europe sector is at 90. Figure 7. Meridional slice of electron density averaged for occultation events of the local time interval h. The upper panel shows the number of averaged electron density profiles within a latitude window of 15. the equator is in agreement to topside ionosonde measurements [Maruyama and Matuura, 1984, Figure 7]. The F 2 peak layer is around 300 km height from 30 Sto60 Nin Figure 7. Around 10 NtheF 2 peak layer shows upwelling which is associated with maximal electron density in the topside ionosphere beyond this latitude region. Such an upwelling of the ionosphere can be caused by an eastward electric field with maximum at 10 N, since the associated E B plasma drift will result in upward ion velocity. The phenomenon of strong enhanced vertical ion drift just after sunset is regularly observed by incoherent scatter radars at equatorial latitudes. Calculations of Anderson and Haerendel [1979] suggest the importance of strong eastward electric fields in the E-region for development of plasma bubbles and spread F. The largest TEC fluctuations (Figure 4) occur around 10 N geomagnetic latitude consistent with the upwelling of the F 2 peak layer and the electron density maximum in the topside ionosphere. However, the experiences with ionospheric limb sounding are just in the beginning phase. Data quality and quantity of the GPS/MET mission are not appropriate for definite conclusions here. The initial results (Figures 4 7) show in particular the capability of radio occultation for comprehensive study of spatial and temporal structures of the background ionosphere and its fluctuations. 5. Profiling of Electron Density Depletions in the Equatorial Ionosphere [35] The statistical analysis indicated that GPS radio occultation observed enhanced topside irregularities

9 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE 13-9 Figure 8. Selection of TEC profiles at equatorial latitudes after sunset. The TEC profiles show TEC depletions of 1 20 TEC-units around the F 2 peak possibly caused by arising plasma bubbles from the lower ionosphere. mainly at equatorial latitudes after sunset (20 24 LT). Thus we inspected all TEC profiles from 20 S to 20 N geomagnetic latitude and between 20:00 and 24:00 LT. The result is that many TEC profiles show clear depletions at observation heights km, just below and around the F 2 peak. A selection of disturbed equatorial TEC profiles is depicted in Figure 8. The depletions are around 1 20 TEC-units and possibly indicate plasma bubbles on aeronomic or large scales (1000 km to 20 km) or intermediate scales (20 km to 100 m), referring to a irregularity classification by Kelley [1985]. So the enhanced topside irregularities are possibly (partly) caused by these electron density depletions arising from the lower ionosphere at equatorial latitudes after sunset (the irregularities of the topside ionosphere are almost not visible in Figure 8 since their TEC depletions are less than 1 TEC-unit). [36] The new aspect of this finding is the application of GPS radio occultation to profiling of plasma bubbles. In spite of the low sampling rate of 0.1 Hz and the bad height step of 20 km, the information content of the depletions in the TEC profiles of Figure 8 looks promising. If the depletions are limb-sounded with 50 Hz sampling rate (40 m height step of the ray tangent point) the TEC profiles will resolve the fine structure of the wedge-shaped, field-aligned plasma bubbles which have a strong depletion gradient at their topside [e.g., Steigies et al., 2002]. As mentioned before radio occultation may provide information on bubble structures down to 100 m if the structures are stretched by km (or more) along the field line and if the occultation ray is parallel to the field line. Having an electron density depletion of the order of m 3 over a distance of 100 km, this results in a TEC variation of m 2 or one TEC-unit which is no measurement problem for radio occultation. [37] The present study is concentrated on the equatorial irregularities since their effects in radio occultation observations seem to be easier for analysis and interpre-

10 13-10 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE tation than the effects of mid- and high-latitude ionospheric irregularities. Nevertheless an application of radio occultation to mid-latitude spread F and high latitude plasma bubbles [e.g., Fukao et al., 1991; Fejer, 1996] should be also feasible. In any case investigations of the interplay between plasma irregularity structure, geomagnetic field orientation, and GPS occultation ray are necessary. This can be performed by improved analysis of huge occultation data sets and by radio propagation simulations of ionospheric occultations. Further a combined analysis of spread F and sporadic E characteristics is desirable since both can be derived from occultation data [Hocke and Tsuda, 2001a, 2001b; present study]. 6. Summary and Outlook [38] GPS radio occultation can sound background electron density and irregularities of the topside ionosphere and allows a combined interpretation of both. A statistical estimate of occurrence of topside irregularities at solar minimum (June/July 1995) gave reasonable results which are in agreement with previous, observational studies and theory of spread F. The capability of GPS radio occultation for global observation of topside irregularities is due to several facts, a) clear variation of spread F with local time and geographic location, b) exponential decrease of electron density profiles within the topside ionosphere, c) ray path segment of the tangent point layer is maximal (Figure 1). The facts b) and c) support the interpretation of fluctuations of GPS phase path differences (or TEC) by ionospheric irregularities located around the ray tangent points. In average the phase path differences are mainly effected by the electron density values of the tangent point layers. [39] Our study is limited to GPS/MET radio occultation data of June/July By means of a filtering process TEC fluctuations with vertical scales of around km at ray tangent point heights km are extracted from the 0.1 Hz low rate GPS data. These fluctuations are maximal at geomagnetic latitudes around 10 N of the summer hemisphere. Though the GPS/MET data set has no full coverage in local time and latitude, the observations indicate a concentration of large TEC fluctuations at local times from 20:00 to 24:00 LT. The maximal TEC fluctuations seem to occur over Africa Arabia. The meridional slice of the average electron density of the postsunset ionosphere (19 23 LT) shows an upwelling of the F 2 peak layer at around 10 N and an associated maximum of electron density in the topside ionosphere beyond. All these observations indicate that the large-scale TEC fluctuations are possibly related to the onset of spread F and plasma bubbles in the unstable equatorial ionosphere after sunset. Further we find large TEC depletions (1 20 TEC-units) just below and around the F 2 peak in the TEC profiles of the equatorial ionosphere after sunset. The TEC depletions possibly indicate plasma bubbles arising from the lower ionosphere. A GPS receiver sampling rate of 50 Hz may allow the limb sounding of plasma bubble structures with vertical scales down to 100 m at equatorial latitudes. [40] Statistical and direct comparisons between spaceborne GPS data and ground-based observations from GPS networks, ionosondes, and radars are desirable as well as comparison to other spaceborne techniques such as topside ionosondes or in situ plasma measurements. For the GPS/MET data it is difficult to find coincidences in space and time to other observations because of the relative small amount of occultation events collected within the short observation time intervals of the GPS/ MET mission. A comparison between worldwide scintillation maps of GPS ground receivers and spaceborne GPS receivers is envisaged as well as the use of climatological scintillation models such as WBMOD [Secan et al., 1995] for validation. At the moment the Argentine USA satellite SAC-C collects radio occultations in a sun-synchronous orbit (10:30 LT am pm) in 705 km height. These continuous measurements should be suitable to derive a detailed climatology of the topside ionosphere and its large-scale irregularities at premidnight and prenoon. [41] Acknowledgments. We are grateful to Dr. C. Rocken and Dr. W.S. Schreiner from University Corporation for Atmospheric Research (UCAR, Boulder) for retrieval and provision of phase path and electron density profiles of the GPS/MET experiment in June/July We thank Dr. G. Hajj for informations on SAC-C and discussion. We are grateful to the reviewers for essential improvements. K.H. thanks the Telecommunications Advancement Organization (TAO) of Japan for a research fellowship. References Aarons, J., The longitudinal morphology of equatorial F-layer irregularities relevant to their occurrence, Space Sci. Rev., 63, , Aarons, J., J. P. Mullen, H. E. Whitney, and E. M. MacKenzie, The dynamics of equatorial irregularity patch formation, motion, and decay, J. Geophys. Res., 85, , Abdu, M. A., Outstanding problems in the equatorial ionosphere-thermosphere electrodynamics relevant to spread F, J. Atmos. Sol.-Terr. Phys., 63, , Anderson, D. N., and G. Haerendel, The motion of depleted plasma regions in the equatorial ionosphere, J. Geophys. Res., 84, , Anderson, D. N., A. D. Richmond, B. B. Balsley, R. G. Roble, M. A. Biondi, and D. P. Sipler, In-situ generated gravity waves as a possible seeding mechanism for equatorial spread F, Geophys. Res. Lett., 9, , Fejer, B. G., Natural ionospheric plasma waves, in Modern

11 HOCKE ET AL.: IRREGULARITIES OF THE TOPSIDE IONOSPHERE Ionospheric Science, editedbyh.kohl,r.rüster, and K. Schlegel, pp , EGS, Katlenburg-Lindau, Germany, Fejer, B. G., W. B. Hanson, and R. A. Heelis, Satellite observations of low latitude ionospheric plasma depletions, in Low-Latitude Ionospheric Physics, Proceedings of COSPAR Colloquium, Taipei, Taiwan, edited by F. S. Kuo, pp , Elsevier Sci., New York, Fukao, S., M. C. Kelley, T. Shirakawa, T. Takami, M. Yamamoto, T. Tsuda, and S. Kato, Turbulent upwelling of the midlatitude ionosphere, J. Geophys. Res., 96, , Gorbunov, M. E., A. S. Gurvich, and A. V. Shmakov, Backpropagation and radio-holographic methods for investigation of sporadic ionospheric E-layers from Microlab-1 data, Int. J. Remote Sens., 23, , Haerendel, G., Theory of equatorial spread F, report, MPI für Extraterrestrische Phys., München-Garching, Germany, Hajj, G. A., and L. J. Romans, Ionospheric electron density profiles obtained with the Global Positioning System: Results from the GPS/MET experiment, Radio Sci., 33, , Hajj, G. A., L. C. Lee, X. Pi, L. J. Romans, W. S. Schreiner, P. R. Straus, and C. Wang, COSMIC GPS ionospheric sensing and space weather, Terr. Atmos. Oceanic Sci., 11, , Heelis, R. A., P. C. Kendall, R. J. Moffet, D. W. Windle, and H. Rishbeth, Electrical coupling of the E and F regions and its effect on the F region drifts and winds, Planet. Space Sci., 22, , Hinson, D. P., and G. L. Tyler, Spatial irregularities in Jupiter s upper ionosphere observed by Voyager radio occultations, J. Geophys. Res., 87, , Hocke, K., and T. Tsuda, Using GPS satellites to study plasma irregularities, GPS World, 7, 34 36, 2001a. Hocke, K., and T. Tsuda, Gravity waves and ionospheric irregularities over tropical convection zones observed by GPS/ MET radio occultation, Geophys. Res. Lett., 28, , 2001b. Hysell, D. L., M. C. Kelley, W. E. Swartz, and D. T. Farley, VHF radar and rocket observations of equatorial spread F on Kwajalein, J. Geophys. Res., 99, 15,065 15,085, Kelley, M. C., Equatorial spread F: Recent results and outstanding problems, J. Atmos. Terr. Phys., 47, , Kelley, M. C., and C. A. Miller, Mid-latitude thermospheric plasma physics and electrodynamics: A review, J. Atmos. Sol.-Terr. Phys., 59, , Kintner, P. M., H. Kil, T. L. Beach, and E. R. de Paula, Fading timescales associated with GPS signals and potential consequences, Radio Sci.., 36, , Kramer, L. and J. L. Goodman, Global scale observations of ionospheric instabilities from GPS in low earth orbit, paper presented at Space 2001 Conference, Am. Inst. of Aeronaut. and Astronaut., Albuquerque, N. M., Aug Maruyama, T., and N. Matuura, Longitudinal variability of annual changes in activity of equatorial spread F and plasma bubbles, J. Geophys. Res., 89, 10,903 10,912, Raghavaro, M., N. Rao, J. H. Sastri, G. D. Vyas, and M. S. Rao, Role of equatorial ionization anomaly in the initiation of equatorial spread F, J. Geophys. Res., 93, , Rocken, C., et al., Analysis and validation of GPS/MET data in the neutral atmosphere, J. Geophys. Res., 102, 29,849 29,866, Röttger, J., Equatorial spread-f by electric fields and atmospheric gravity waves generated by thunderstorms, J. Atmos. Terr. Phys., 43, , Schreiner, W. S., D. C. Hunt, C. Rocken, and S. V. Sokolovskiy, Precise GPS data processing for the GPS/MET radio occultation mission at UCAR, in Proceedings of the Institute of Navigation-Navigation 2000, pp , Inst. of Navig., Alexandria, Va., Schreiner, W. S., S. V. Sokolovskiy, C. Rocken, and D. C. Hunt, Analysis and validation of GPS/MET radio occultation data in the ionosphere, Radio Sci., 34, , Secan, J. A., R. M. Bussey, and E. J. Fremouw, An improved model of equatorial scintillation, Radio Sci., 30, , Sokolovskiy, S. V., W. Schreiner, C. Rocken, and D. Hunt, Detection of high-altitude ionospheric irregularities with GPS/MET, Geophys. Res. Lett., 29(3), 1033, doi: / 2001GL013398, Steigies, C. T., M. Hirt, and A. Piel, Electron density and temperature measurements obtained in the DEOS campaign, Adv. Space Res., 29(6), , Tao, K., World-wide maps of the occurrence percentage of spread F in years of high and low sunspot numbers, J. Radio Res. Lab. Jpn., 12, , Tsunoda, R. T., Control of the seasonal and longitudinal occurrence of equatorial scintillation by longitudinal gradient in integrated Pedersen conductivity, J. Geophys. Res., 90, , Vorob ev, V. V., A. S. Gurvich, V. Kan, S. V. Sokolovskii, O. V. Fedorova, and A. V. Shmakov, Structure of the ionosphere based on radio occultation data from GPS Microlab-1 satellites: Preliminary results, Earth Obs. Remote Sens., 15, , Wickert, J., C. Reigber, G. Beyerle, R. König, C. Marquardt, T. K. Meehan, W. G. Melbourne, and K. Hocke, Atmosphere sounding by GPS radio occultation: First results from CHAMP, Geophys. Res. Lett., 28, , K. Hocke and K. Igarashi, Communications Research Laboratory, Nukui-Kita, Koganei, Tokyo, Japan. (hocke@crl.go.jp) A. Pavelyev, Institute of Radio Engineering and Electronics, Fryazino, Vvedenskogo sq. 1, Moscow, Russia.