Letter to the editor: The ionospheric response during an interval of Pc5 ULF wave activity

. A preliminary analysis of Pc5, ULF wave activity observed with the IMAGE magnetometer array and the EISCAT UHF radar in the post midnight sector indicates that such waves can be caused by the modulation of the ionospheric conductivity as well as the wave electric ﬁeld. An observed Pc5 pulsation is divided into three separate intervals based upon the EISCAT data. In the ﬁrst and third, the Pc5 waves are observed only in the measured electron density between 90 and 112 km and maxima in the electron density at these altitudes are attributed to pulsed precipitation of electrons with energies up to 40 keV which result in the height integrated Hall conductivity being pulsed between 10 and 50 S. In the second interval, the Pc5 wave is observed in the F-region ion temperature, electron density and electron temperature but not in the D and E region electron densities. The analysis suggests that the wave during this interval is a coupled Alfven and compressional mode.


Introduction
Observations of ULF waves by coherent radars such as STARE and SABRE are many (e.g. Walker et al., 1979;Allan et al., 1982;Yeoman et al., 1990). From such observations, the electric ®eld of the wave in the ionosphere could be measured for the ®rst time (Walker et al., 1979). Observations of waves with large azimuthal wave numbers, previously dicult to observe with ground based magnetometers, could also be made with these radars (Allan et al., 1982). Furthermore, a method for identifying the``compressiveness'' of ULF waves was discussed by Yeoman et al. (1990). There are, however, few simultaneous observations of the variations in the ionospheric electron density, electron temperature and ion temperature during ULF wave activity (Doupnik et al., 1977;Crowley et al., 1985Crowley et al., , 1987Lathuillere et al., 1986;Buchert et al., 1999). Such observations are important to determine the ULF wave mode and the nature of the damping mechanism for ULF waves amongst other things. In this letter we present a preliminary analysis of observations by EISCAT during ULF waves in the interval 0200±0500 UT on April 21, 1993.

Magnetometer data
Band-pass ®ltered data, between 20 and 500 s, from four stations of the IMAGE array, Tromsù, Kilpisjarvi, Masi and Kevo, for the interval of interest are given in Fig. 1. Between 0230 and 0450 UT there appears to be a single Pc5 ULF wave, which is composed of several dierent wave packets, all of a similar frequency. The interval can be divided into three distinct parts (see also later), between 0230 and 0315 UT, 0315±0410 UT and 0410± 0500 UT. The Y, geographic east-west, component ( Fig. 1b) is the strongest component in the ®rst interval, while the X, geographic north-south, (Fig. 1a) and Y components are comparable during the second interval and the X component is the largest in the third interval. Spectral analysis of the data from all IMAGE stations indicates that the dominant frequency in both X and Y components is 4.8 mHz, 5.2 mHz and 5.3 mHz in the ®rst, second and third intervals, respectively. Note that the resolution of these frequencies is 0.4 mHz. Comparison of phase across the four stations in Fig. 1 provides an estimate of the azimuthal wave numbers, or m values. The outermost stations, TRO and KEV, have the smallest separation in latitude, 0.3°, thereby reducing any eects caused by latitudinal phase gradients, and are separated by 6.3°in longitude. Based upon these stations, the X and Y component m values for the dominant spectral component in each interval range between 13 and 18 (i.e. westward).

EISCAT data
During the interval, the EISCAT radar was operating in common mode CP-1-K which is a simple mode whereby the Tromsù UHF radar points in a direction which is approximately aligned along the magnetic ®eld direction in the F-region. The two remote receiver beams intersect the Tromsù beam at an altitude of 278 km. The mode and experimental codes are described in detail elsewhere (e.g. Davies et al., 1997). In this brief work we present both long pulse estimates of electron density and temperature and ion temperature at altitudes above 150 km, with a resolution of 22 km along the beam, and height integrated Hall and Pedersen conductivities derived from raw electron density measurements, or power pro®le, between 70 and 400 km at a resolution of 4.5 km. The Hall and Pedersen conductances have been calculated following the method outlined in Lester et al. (1996). The tristatic measurements of ion velocity at 278 km allow the calculation of the electric ®eld.
Representative time series of electron density, N e , electron temperature, T e , ion temperature, T i , at an altitude of 212 km, the east-west, V E , and north-south, V N , components of the ion velocity at 278 km, and the height integrated Hall conductivity, S H , for the interval 0200 to 0500 UT are presented in Fig. 2a. The electron density, electron temperature, ion temperature and S H are plotted with a 10 s integration, while the ion velocity data are plotted with a 20 s integration. Vertical lines are drawn to indicate the three separate intervals identi®ed. The F-region measurements only respond to the wave activity between 0315 and 0410 UT, while there is periodic behaviour in S H during both the ®rst and third intervals but not the second.
In interval 1, 0230±0315 UT, N e at 212 km is typically 1´10 11 m )3 . At this altitude T e varies between 1200 and 1400 K and T i between 800 and 1200 K. The larger ion temperatures are related to the enhanced ion velocities and are caused by ion frictional heating (e.g. Davies et al., 1997). The lower E-region electron density (data not shown) is 10 11 m )3 between 0210 and 0230 UT, but during the following 45 min it varies periodically between 5´10 10 m )3 and 5´10 11 m )3 . A dominant spectral peak near 5.2 mHz is evident in the spectral analysis of the E-region electron density leads to the periodic variation in S H between 10 and 50 S, shown in Fig. 2a, while the height integrated Pedersen conductivity, S p , varies between 5 and 10 S.
During interval 2 there are clear periodic variations in F-region electron density, electron temperature and ion temperature. The maxima in N e , about 3´10 11 m )3 , are the highest values at these altitudes During the ®nal wave packet observed by the magnetometer data, the F-region parameters are fairly constant throughout the interval with no sign of the periodicities seen earlier. There are, however, clear periodic variations in S H between 0420 and 0500 UT with no similar variations in S P . The dominant spectral peak in the E-region electron density is 5.3 mHz.
The EISCAT measurements during the second of the wave packets demonstrate the relative variations in the F-region ionospheric parameters during the ULF wave activity. Figure 2b is in the same format as Fig. 2a but covers the interval 0315 UT to 0410 UT. There are two separate packets of pulsed behaviour in electron density and temperature. Initially there are four peaks in N e and T e between 0320 UT and 0334 UT, and a further six pulses in the second packet. On the other hand, T i appears to be pulsed throughout the period up to 0400 UT.
The ®rst signature of the wave activity occurs at 0319 UT when T i increases from 800 K to more than 1000 K. The pulse of enhanced T i lasts for approximately 120 s and is followed by an interval of unperturbed ion temperature. This behaviour continues and the ion temperature has the appearance of a square wave. During this pulse of enhanced ion temperature, N e and T e are low and become enhanced as T i decreases. Normally T e peaks just before or at the same time as N e and also begins to decrease before N e . In contrast to T i , the pulses in N e and T e have rapid rise times to a peak value with subsequent slower decay times. A cross correlation analysis of the three times series indicates that the ion temperature and electron density are exactly 180°out of phase while the electron temperature leads the electron density by about 40°. This latter phase dierence is a result of the decrease in electron temperature leading that in electron density.

Discussion
The Pc5 wave activity observed in the magnetometer data can be divided into three separate intervals as a result of the analysis of the EISCAT data. In the ®rst and third intervals, the only response in the EISCAT observations is in the lower E-region and D-region conductivity Hall conductivity. a covers the time interval 0200 to 0500 UT and b the interval 0315 to 0410 UT. The vertical lines in a identify the three separate intervals of ULF wave activity as described in the text electron density, while in the second interval the response is seen in the F-region but not in the lower E-or D-regions. For the purpose of this work we discuss separately the wave activity in the ®rst and second intervals.
A careful study of the electron density as a function of altitude during the ®rst interval (data not shown) indicates that the wave occurs in the electron density measured between 90 km and 112 km altitude, with the largest variations in density at 99 km. At this altitude the peak density is 5´10 11 m )3 , with a minimum density of 5´10 10 m )3 . The cause of such pulsed behaviour in electron density is most likely pulsed particle precipitation since such densities are higher than expected for this local time, 0430 LT, in April. It is instructive to consider the energy of the particles which would cause such ionisation at these altitudes. For a monoenergetic, ®eld-aligned beam of electrons of 20 keV the peak ionisation rate occurs near 100 km (Rees, 1989), while a beam of electrons of 40 keV has a peak ionisation rate near 90 km. A detailed inversion of the ionisation pro®le would provide a more accurate energy spectrum but is not attempted here. This variation in D-region and lower E-region densities leads to the conductance varying between 10 and 50 S.
Recently, Buchert et al. (1999) have reported observations of a ULF wave in the afternoon sector during which the conductance varied by a factor of 2. These authors concluded that the cause of the conductance variations was pulsed electron precipitation in the energy range 2±5 keV. The electrons which caused the conductance variations presented in Fig. 1 are of much higher energy than those reported by Buchert et al. (1999). The m value of the event discussed by Buchert and colleagues is 30, larger than that of the wave in Fig. 1, while the direction of phase propagation is westward for both events. However as the local time of the event discussed here is pre-dawn, a westward phase propagation is consistent with the wave being driven by the solar wind as it passes over the magnetopause and by ions drifting in the magnetosphere. Thus, it is not possible to distinguish between these two driving mechanism, as in the case of Buchert et al. (1999). It is also worth noting that no signi®cant solar or substorm activity took place at the time of either observation, although a substorm did occur a few hours before the case reported here.
Moving to interval two, the ®rst point to discuss is the change in the E-region density. The density at 99.4 km varies between 2 and 5´10 11 m )3 , starting at the higher value and gradually decreasing, although not monotonically. Again this density is higher than expected at these altitudes at this local time and is also higher than the density at higher altitudes. For example, at 112 km the density has a maximum of 3´10 11 m )3 and is always lower than the values at 99 km. Secondly, there is no clear signature in the Eregion density of the ULF wave activity seen in the magnetometer data or in the F-region ionospheric parameters. As discussed already the likely cause of the enhanced D region and lower E region ionisation is particle precipitation and on this occasion it appears to be more continuous, if variable, rather than pulsed. The high electron densities lead to values in R H which vary between 20 and 50 S.
The phase relationships between N e , T e and T i represent the other intriguing aspect of this interval. The simultaneous enhancements in electron density and temperature are likely to be caused by particle precipitation, while the ion temperature enhancements are caused by enhancements in the ion velocity, most easily seen in the eastward component (Fig. 2). The latter enhancements are possibly caused by the ULF wave electric ®eld. Of the ULF wave modes, the Alfven mode has a signi®cant electric ®eld in the ionosphere (e.g. Yeoman et al., 1990). Observations by SABRE demonstrated that ULF waves are found to be increasingly compressional as the aziumthal wave number increased, at least up to a value of 12. The magnetometer observations discussed earlier indicate that the wave numbers are >13, suggesting that the wave observed during the second interval had a signi®cant compressional component as well as an Alfvenic component. Furthermore, the wave frequency of 5.2 mHz during this interval is close to the expected resonant frequency for the latitude of Tromsù (e.g. Mathie et al., 1999).
The high R H indicates continuing precipitation of high energy particles which is not pulsed as in the ®rst and third intervals. The particles causing the enhanced D and lower E-region electron densities are of much higher energy than those which are believed to be causing the enhancements in the F-region electron density and temperature. Why the higher energy part of the spectrum is not pulsed is unclear, however.

Conclusions
We report joint observations of Pc5 pulsations by the IMAGE magnetometer array and EISCAT. Such observations are rare, although with improvements to the receiver sensitivity at Tromsù we anticipate that more events will be observed. The interval of ULF wave activity examined here has two dierent responses in the ionosphere. In the ®rst part of the interval, only enhancements in lower E region and D region electron densities are observed. We suggest these are caused by pulsed particle precipitation at energies of order 20± 40 keV. Furthermore, the ground magnetic signature may simply be a response to the changing Hall conductivity. In the second interval a complicated phase relationship between the ion temperature and electron density and temperature exists. There are two processes occurring, one associated with the wave electric ®eld and the other with the pulsed precipitation of low energy particles. The joint observations of the ULF wave electric ®eld and high wave number indicates that the wave consists of both Alfven and fast modes. Simultaneously, there continues to be precipitation of high energy particles, 20±40 keV, in order to maintain the high Hall conductance. The preliminary analysis under-taken here cannot distinguish whether this is solely a temporal variation or solely a spatial variation. The interval does clearly respond in two dierent ways during the period of ULF wave activity, without any clear dierence in the ground magnetic signatures at the time. Further detailed modelling and analysis of this event is required and will be presented in a future study.