A test of the magnetospheric source of traveling convection vortices

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 109,, doi: /2003ja010214, 2004 A test of the magnetospheric source of traveling convection vortices M. M. Lam and A. S. Rodger Physical Sciences Division, British Antarctic Survey, Cambridge, UK Received 27 August 2003; revised 5 November 2003; accepted 11 November 2003; published 6 February [1] Traveling convection vortices (TCVs) are a powerful tool for probing the nature of the coupling between the solar wind, the magnetosphere, and the ionosphere. There is no reliable model of the plasma concentration in the magnetosphere, resulting in uncertainties about the factors controlling the scale size, the motion, and the numbers of field-aligned currents associated with TCV events. There is also uncertainty about whether TCV generation is current driven, voltage driven, or even driven by some more complex source. We use conjugate ground-based magnetometer data from the Greenland magnetometer chain and Antarctica to test the nature of the magnetospheric source of 18 TCV events associated with changes in the magnetopause dynamic pressure. This is achieved by statistically comparing two groups of TCV events: for one group the conjugate ionospheres are of similar conductivity, and for the other group the conductivities of the conjugate ionospheres differ by an order of magnitude. Statistically, we find that conjugate TCV events are of similar intensity in both hemispheres regardless of any difference in conductivity between the two hemispheres. We propose that this is evidence in favor of a constant current source for TCVs where the amplitude of a TCV is controlled by the local plasma concentration, the magnetic field strength, and the acceleration of the plasma. INDEX TERMS: 2708 Magnetospheric Physics: Current systems (2409); 2784 Magnetospheric Physics: Solar wind/magnetosphere interactions; 2736 Magnetospheric Physics: Magnetosphere/ionosphere interactions; 2730 Magnetospheric Physics: Magnetosphere inner; 2409 Ionosphere: Current systems (2708); KEYWORDS: traveling convection vortices, conjugate, magnetosphere-ionosphere coupling, solar wind pressure pulses, current generator, voltage generator Citation: Lam, M. M., and A. S. Rodger (2004), A test of the magnetospheric source of traveling convection vortices, J. Geophys. Res., 109,, doi: /2003ja Introduction [2] Traveling convection vortices (TCVs) were first identified from impulses in ground magnetometer data as pairs of antiparallel field-aligned currents (FACs) which close in the ionosphere, giving rise to ionospheric current vortices traveling from near noon around the auroral oval [Friis- Christensen et al., 1988]. They are now known to be the ionospheric signatures of interactions between the solar wind and the magnetosphere, caused by, for instance, solar wind pressure pulses [Friis-Christensen et al., 1988]. Their current systems are frequently (on average once per day) detected by ground-based magnetometers, making their observational and theoretical study a powerful tool for probing the nature of the coupling between the solar wind, the magnetosphere, and the ionosphere. In this paper we use simultaneous conjugate ground-based magnetometer measurements of TCVs to test existing hypotheses of the mechanism for their generation. [3] TCVs are localized in the dayside, high-latitude ionosphere; peak event amplitudes and occurrence rates Copyright 2004 by the American Geophysical Union /04/2003JA have been found to occur prior to local noon [Moretto and Yahnin, 1998]. More recent work by Moretto et al. [2003] suggests that this may in fact be due to a dependence of the latitudinal distribution of events in local time. TCVs typically travel at a few kilometers per second, last between 10 and 15 min, have a horizontal scale of 1000 km, and are most frequently observed during periods of low geomagnetic activity [Glassmeier et al., 1989; Sibeck and Korotova, 1996]. An upward current of a TCV pair carried by energetic electrons has been inferred by Sitar et al. [1998] from Polar ultraviolet imager (UVI) data. Such energetic electron particle precipitation may enhance the local ionospheric conductivity significantly and be the cause of the observed asymmetry in the ground signature of many TCVs [Zhu et al., 1999]. [4] Sibeck and Korotova [1996] find that TCV events are associated with abrupt changes in the interplanetary magnetic field (IMF), and they suggest that changes in the IMF orientation modulate the fraction of the solar wind pressure applied to the magnetosphere. Sitar et al. [1998] have suggested that their observations show that a change in the orientation of the IMF may produce a hot flow anomaly at the bow shock which can interact with the magnetopause to intensify an observed TCV pair. 1of11

2 [5] There is now a large body of work, such as the studies by Lysak and Lee [1992], Araki [1994], Yahnin and Moretto [1996], Lühr et al. [1996], Moretto and Yahnin [1998], Slinker et al. [1999], and Lam and Rodger [2001], which supports the following theory for the source of TCVs: The solar wind interacts with the magnetosphere to generate a sharp change in the dynamic pressure at the magnetopause. This change launches a fast mode hydrodynamic wave into the magnetosphere, causing the movement of magnetospheric plasma, which converts to an Alfvén wave. Assuming the validity of magnetohydrodynamics for this situation, the inertial limit of the current continuity equation gives the perpendicular current density as J? ¼ rb dv B 2 dt ; ð1þ where B is the magnetic field, r is the plasma density, and v is the plasma velocity. At present we are still lacking in situ measurements of these three quantities; such observations would establish the relationship between solar wind parameters and the ionospheric response. We can postulate that the amplitude maxima of the FACs lie some distance earthward of the magnetopause because although plasma density increases with decreasing distance from the Earth, the amplitude of the fast mode wave decreases because of the generation of FACs. [6] Existing theoretical models of TCV generation do not properly represent the coupling between the magnetosphere and the ionosphere. There is, at present, no reliable model of the plasma concentration in the magnetosphere. This means that there is still uncertainty about the factors controlling the scale size, motion, and numbers of FACs generated during TCV events; observed signatures sometimes indicate complicated current structures [Iijima and Potemra, 1976]. How the intensity of TCVs is related to the nature of their source is also not certain. [7] In sections 2.2 and 2.3 we examine how the relationship between the ratio of the amplitudes of conjugate ground-based magnetometer signals and the ratio of the conjugate ionospheric conductivity depends upon the nature of the magnetospheric source. In section 2.4 we propose a test that can differentiate between a current source and a voltage source for TCVs which depends upon the fact that the ratio of the high-latitude ionospheric conductivities in the two hemispheres varies with season; therefore we first review in section 2.1 findings relating to the ionospheric conductivity in order to arrive at an estimate of this ratio. [8] In sections 3.1 and 3.2 we describe the method of selection of TCV events and of measurement of the amplitude ratio of the ground magnetometer signal in the Northern Hemisphere to that in the Southern Hemisphere. In section 3.3 we describe a statistical test to differentiate between a current source and a voltage source for TCVs. We present detailed data from a case study event in section 4.1. In section 4.2, we then give some statistical results from 18 TCV events, including those for the test of the nature of the source of the TCVs. Our conclusions are in section Scientific Rationale 2.1. Ionospheric Conductivity [9] In this section we estimate, to an order of magnitude, the ratio of the Northern and Southern Hemisphere highlatitude ionospheric conductivities associated with a TCV event observed at conjugate points. Let us first consider the conductivity due to auroral precipitation and solar radiation. The ionospheric conductivity is proportional to the electron concentration N. According to Rishbeth and Garriott [1969, Table III, p. 131], the ratio of the day to night values of N is 17. Data from Hardy et al. [1987], presented by Sibeck et al. [1996], suggest that the dayside Hall conductivity varies latitudinally between 10 and 13 mho in the summer and between 1 and 8 mho in the winter before 1400 magnetic local time (MLT) at 75. The Hardy et al. [1987] data, however, do not represent the cusp well; there is no contribution from particles with energies below 1 kev (which do not contribute particularly to the Hall conductivity but are significant for the Pedersen conductivity), and the data binning smooths all peaks and boundaries. Therefore the actual values of the ratio of the sunlit to nonsunlit dayside conductivity are higher than those suggested by the Hardy et al. [1987] data, which are between 2 and 10 during periods of moderate geomagnetic activity. [10] The total conductivity during a TCV event also has a contribution from the particle precipitation associated with TCVs. In their model of the effect of such particle precipitation on the ground signature of TCVs, Zhu et al. [1999] find a spatially localized (100 km) increase of 1 mho in the Hall conductivity associated with the upward FAC for a peak characteristic energy of 1 kev and for a circular geometry. All-sky camera data from the South Pole for a TCV event on 25 July 1997 at 1843 universal time (UT) [Murr et al., 2002] reveal the geometry of the TCV on this occasion to be long and thin, not circular, with intense emission present at 630 nm and less intense emission at nm, indicating F region precipitation (<300 ev). We speculate that the increase in conductivity associated with upward FAC particle precipitation is 1 mho. Since this is a relatively small amount and it is not appropriate to simply add this contribution to the conductivity due to auroral precipitation and solar radiation, we shall ignore it. For our purposes it is enough to make an order of magnitude estimate of the ratio of dayside sunlit to nonsunlit conductivity appropriate to a TCV event, which we assume is 10. [11] Horizontal gradients in ionospheric conductivity are certainly present; however, using their theoretical model of the ionosphere, Zhu et al. [1997] demonstrate that the precipitation associated with the upward FAC of a TCV, although certainly causing a distortion in the ground magnetic signature, produces much less than an order of magnitude change. We are still able to derive the horizontal ionospheric currents from the Hall currents to within an order of magnitude, which is sufficient for our purpose. We will therefore neglect the effect of horizontal gradients in ionospheric conductivity in the following treatment and assume that we can, in principle, derive the Hall current from the ground magnetometer signal [Fukushima, 1969; Zhu et al., 1997; Lühr et al., 1993]. We also treat the ionosphere as a two-dimensional layer with height-integrated Hall and Pedersen conductivities, H and P. Southwood and Hughes [1978] have shown that the minimum horizontal wavelength of the magnetospheric magnetic field that is detectable at the ground is limited by the ionospheric screening effect to 120 km; therefore the magnetic signature of TCVs is spatially large enough to be detected by ground magnetometers. The 2of11

3 amplitude of the magnetic field on the ground is related to its amplitude in the magnetosphere by the ratio of the Hall to Pedersen height-integrated conductivities, which is always close to unity. In the following treatment we assume that the magnetospheric driver is equatorial and that it is therefore reasonable to assume that whatever the form of the driver, it will map to both hemispheres Voltage Generator [12] Assuming that the Hall current pattern is similar in both hemispheres and that there is a temporally constant voltage source, the ratio of the magnitude of a groundbased magnetometer signature of a TCV in the Northern Hemisphere, jb n j, to that at a conjugate point in the Southern Hemisphere, jb s j, is equal to the ratio of the Hall conductivities at those conjugate points (Ohm s law), i.e., j R ¼ B nj j j ¼ Hn : Hs B s When the conjugate ionospheric regions concerned are either both ionized by solar radiation or both in darkness, then R 1. When the region of the ionosphere in one hemisphere is sunlit but the conjugate region of the ionosphere in the other hemisphere is not, then we estimate, using the values mentioned in section 2.1, that R Current Generator [13] Constant current sources provide a fixed current I that maps down magnetic field lines and closes in the ionosphere via Pedersen currents. The strength of the ionospheric electric field required to drive the Pedersen current depends on the amplitude of the current density and the Pedersen conductivity P, i.e., E ¼ I P : The Hall currents are given by I H ¼ I H P : We assume that the ratio of the Hall conductivity to the Pedersen conductivity remains approximately the same regardless of season and hemisphere [Rishbeth and Garriott, 1969], so that the ratio of the ground magnetometer signals at conjugate points in each hemisphere is always approximately unity: j R ¼ B nj j j ¼ Hn Ps Pn B s Hs 2.4. TCVs: Voltage or Current Source? [14] We can see from sections 2.2 and 2.3 that it is possible to differentiate between a voltage and a current source. This can be achieved by comparing the value of the ratio R of the ground magnetometer signal in the Northern Hemisphere to that in the Southern Hemisphere ð2þ ð3þ ð4þ ð5þ in the following situations: (1) when conjugate regions of the ionospheres in each hemisphere have similar conductivities (both hemispheres sunlit or neither hemisphere sunlit) and (2) when conjugate regions of the ionospheres in each hemisphere have dissimilar conductivities (the region of the ionosphere of interest in one hemisphere is sunlit but the ionosphere at the conjugate point in the other hemisphere is not). If TCVs have a voltage source, then the values of R in situations 1 and 2 will be different by at least an order of magnitude, and if TCVs have a current source, then the ratios will be approximately the same in both cases. 3. Method 3.1. Selecting TCV Events [15] To test the nature of the magnetospheric source of TCVs, we used ground-based magnetometer data from two near-conjugate high-latitude chains; from the Southern Hemisphere we used the U.S. Automated Geophysical Observatories (AGOs) Polar Experiment Network for Geophysical Upper-Atmosphere Investigations (PENGUIN) 2 and 3 (i.e., P2 and P3) [Rosenberg and Doolittle, 1994], the British Antarctic Survey (BAS) AGO A84 [Dudeney et al., 1998], and the U.S. fluxgate magnetometer at the South Pole. From the Northern Hemisphere we made use of the Greenland West magnetometer chain [Wilhjelm and Friis-Christensen, 1976]. In order to make estimates of TCV azimuthal phase velocities, we also used the BAS AGO A81, magnetometer data from the Greenland East magnetometer chain, the Magnetometer Array on the Greenland Ice Cap (MAGIC), the International Monitor for Auroral Geomagnetic Effects (IMAGE) [Lühr, 1994], and the Magnetometer Array for Cusp and Cleft Studies (MACCS) stations [Engebretson et al., 1995]. The details of all ground magnetometer stations used in this study are presented in Tables 1 (Southern Hemisphere) and 2 (Northern Hemisphere). [16] TCVs are located in the auroral oval, so we confine the study to between 70 and 77 altitude-adjusted corrected geomagnetic (AACGM), i.e., the lowest and highest extents of the dayside statistical auroral oval [Holzworth and Meng, 1975]. There are no ground-based magnetometer stations located between 74 S (location of the South Pole) and 77 S AACGM, so it is important that data from the South Pole are available for each event. Also, data coverage is better in the first half of the year from the PENGUIN AGOs; therefore we used a table of data availability for A84, P2, P3, and the South Pole to select intervals in the first 6 months of the years in which to look for TCV events. We examined the summary plots for signatures of TCVs, in particular, unipolar or bipolar deflections of between 25 and 300 nt in either H (local magnetic north) or D (local magnetic east), which lasted between 5 and 15 min in the time interval UT, i.e., ±6 hours about magnetic noon. If good data from the South Pole and at least one station in the Northern Hemisphere were available for the period in question, then we investigated the event further. [17] In order to make measuring the amplitude of the TCVs easier we preferred impulsive events over semicontinuous ones. Where an event was semicontinuous, we checked that it was a TCV event rather than a field line 3of11

4 Table 1. Location of Southern Hemisphere Magnetometers a Magnetometer Geographic Coordinates, deg. AACGM, deg. Name Code Latitude Longitude Conjugate Latitude Conjugate Longitude Noon MLT UT t res,s British Antarctic Survey Automated Geophysical Observatories (AGOs) AGO 84 A AGO 81 A U.S. AGO Network PENGUIN 2 P PENGUIN 3 P South Pole SP a Columns 1 4 are as follows: full station name, abbreviation of the name, geographical latitude, and geographical longitude. Columns 5 7 contain the conjugate altitude-adjusted corrected geomagnetic (AACGM) latitude, the conjugate AACGM longitude, and the universal time (UT) at which the point conjugate to the station is at noon magnetic local time (MLT), all calculated using the International Geomagnetic Reference Field/Definitive Geomagnetic Reference Field (IGRF/DGRF) geomagnetic field models (available from the National Space Science Data Center (NSSDC) at at an altitude of 100 km and for the year Column 8 contains the time resolution of the data in seconds. resonance (FLR) by examining the variation of the phase characteristics with latitude. For FLRs the wave phase changes by 180 about the latitude of the resonance [Samson et al., 1971]. For TCVs, there is a change in the sense of the bipolar deflection in H across the center of the vortex and typically a 90 phase difference between the H and D components [Friis-Christensen et al., 1988]. [18] We looked for sharp changes in solar wind dynamic pressure using data from the Advanced Composition Explorer (ACE) [McComas et al., 1998], Wind [Ogilvie et al., 1995], and the Interplanetary Monitoring Platform (IMP) 8[Lepping et al., 1992] satellites (time resolution for these data sets is 1 2 min). We used data from the magnetic field investigation (MFI) instrument (resolution 3 s) aboard the Wind satellite [Lepping et al., 1995] to estimate the IMF associated with the TCV events. We also checked for evidence of an increase in the pressure applied to the magnetosphere by examining geostationary satellite magnetic field data from Geostationary Operational Environmental Satellites (GOES) 8 and 9 [Singer et al., 1996]. For each event we calculated a value for the TCV azimuthal phase velocity in both hemispheres by comparing the occurrence times of the maxima of deflections in the ground magnetometer data from pairs of longitudinally separated Table 2. Location of Northern Hemisphere Magnetometers a Geographic Coordinates, Magnetometer deg. AACGM, deg. Name Code Latitude Longitude Latitude Longitude Noon MLT UT t res,s Greenland West Coast Nuuk GHB Maniitsoq SKT Kangerlussuaq STF Attu ATU Qeqertarsuaq GDH Uummannaq UMQ Upernavik UPN Greenland East Coast Ittoqqortoormiit SCO Tasiilaq AMK Magnetometer Array on the Greenland Ice Cap (MAGIC) MAGIC-1 West MCW MAGIC-2 Gisp MCG Magnetometer Array for Cusp and Cleft Studies Cape Dorset CD Igloolik IG Coral Harbour CH Repulse Bay RB International Monitor for Auroral Geomagnetic Effects Ny Ålesund NAL a Columns 1, 2, 3, 4, and 8 are as for Table 1. Columns 5 7 contain the altitude-adjusted corrected geomagnetic (AACGM) latitude, the AACGM longitude, and the universal time (UT) at which the station location is at noon magnetic local time (MLT). 4of11

5 stations. This enabled us to check that the direction of travel of each observed magnetic impulse event is toward the nightside in both hemispheres as one would expect for TCVs (except possibly close to magnetic noon). In this way we collected the 18 TCV events with which this study is conducted. [19] The voltage generator hypothesis predicts that the amplitude of TCV events at conjugate stations is a factor of 10 different for cases where one hemisphere is sunlit and the other is not. We browsed for TCV events using Southern Hemisphere data, so while a signal of reasonable amplitude in the Southern Hemisphere would always be spotted, a signal of reasonable amplitude in the Northern Hemisphere could correspond to a signal in the Southern Hemisphere which is too small to be spotted. We checked that no bias was introduced into our study using the following simple test: We identified 10 TCV events in the ground magnetometer data from the Northern Hemisphere station SKT during January We found that for eight of these events a TCV could also be identified at the approximately conjugate station A84. Then, without referring to the list of events identified at SKT, we identified 10 TCV events in the A84 ground magnetometer data from January For each of these events a TCV event could be identified in the SKT ground magnetometer data. These test TCV events have no further part in this study; they are used to reassure us that we do not have a list of TCV events biased toward one type of magnetospheric driver Finding the Ratio of Hall Currents [20] For each TCV event we filtered the ground magnetometer data in the magnetic latitude range AACGM using a 30-min high-pass filter and identified, by eye, deflections associated with FAC pairs in the data for all available stations in both hemispheres. For each station we estimated the maximum amplitude of the horizontal ground magnetic field during the first deflection, jb 1 j, from the peak values in H and D during the first deflection, H 1 and D 1, using jb 1 j =(H D 2 1 ) 1/2. In each hemisphere we found the station which had the greatest value for jb 1 j, then identified the stations at which jb 1 j was at least half the value of the maximum value of jb 1 j, that is, those located within the full width at half amplitude. For these stations we found the ratio R of jb 1 j in the Northern Hemisphere to its value at a conjugate station in the Southern Hemisphere. A station in the Northern Hemisphere was considered to be conjugate with a station in the Southern Hemisphere if the two stations laid within 3 of magnetic latitude of each other, i.e., conj =3. The latitudinal extent of TCVs varies but is of the order of 6 10 of magnetic latitude. We therefore consider the choice of conj = 3 to be an acceptable compromise given the relatively low numbers of ground magnetometers in the Southern Hemisphere. A sensitivity study, presented in section 4, shows that the results are little changed by altering conj by ±1. We imposed no criterion for conjugacy on the longitudinal separation, but all stations involved in the calculation were separated by <2 hours of MLT. A pair of values of jb 1 j from two conjugate stations were assumed to have a ratio R 1 if one value is no more than twice the other value (R lower R R upper where R lower = 0.5 and R upper = 2.0). [21] In this way we obtained a set of conjugate pairs of stations for which the value of B 1 in both hemispheres lies within the full width at half amplitude. A percentage p of these pairs had a ratio R of the ground magnetic signal that is close to unity. This process was repeated for the second deflection associated with the TCV event. For each TCV event j we calculated the average of this percentage p for the pair of deflections. This quantity is referred to frequently in the rest of the paper as P j (R 1); it gives a measure of how similar the amplitudes of the TCV perturbations are at conjugate points in the Northern and Southern Hemisphere ionospheres. In section 4 we illustrate this method using a case study event and examine the sensitivity of the similarity P j (R 1) to the values of R lower and R upper Testing the Nature of the Source of TCVs [22] Student s t test uses the concept of standard error to measure the significance of a difference in the means of two data sets. When two distributions have the same variance, then the t statistic for sample populations x and y, with means x and y, is computed by estimating the standard error of the difference of means from the pooled variance using the following formula: t ¼ x y SE ; where SE is the standard error and is equal to vffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi h ðn 1 þ M 1 Þ X N 1 ðx i¼0 i xþ 2 þ X i u M 1 t ðy i¼0 i yþ 2 SE ¼ : N þ M 2 ð7þ [23] N and M are the number of points in the first and second samples, respectively. The Student s distribution probability function (not given here) A(tjn) gives the probability that two distributions with equal means would not generate this t statistic by chance. Two means are considered to be significantly different if the Student s distribution probability function is >0.99. The significance level is equal to 1 A(tjn); therefore a small numerical value of the significance (0.01) indicates that the observed difference is highly significant. [24] We can test whether the two sample populations x and y have significantly different variances using a similar test called the F test. A significance of 0.01 generated by the F test indicates that x and y have significantly different variances and that therefore the t test may not be applied. [25] We divided 18 TCV events into two groups (a and b) defined as follows: Group a contains events for which the conjugate ionospheric regions in each hemisphere have similar conductivities (both hemispheres are sunlit at E region altitudes or neither hemisphere is sunlit), and group b contains events for which the ionosphere in one hemisphere is sunlit but the ionosphere at the conjugate point in the other hemisphere is not. We used the Interactive Data Language (IDL) routine Student s t statistic and statistical mean (TM) test and F-statistic and variance (FV) test to compute the t test and F test significances, respectively. The F test was used to check that the variances of the data sets ð6þ 5of11

6 were similar enough for the t test to be applicable. The t test was then used to determine if there is a significant difference between the mean values of P j (R 1) (defined in section 3.2) for the two groups. If TCVs have a voltage source, then we would expect a significant difference between P j ðr 1Þ a and P j ðr 1Þ b, and if TCVs have a current source, then we would expect there to be no significant difference between these means. 4. Results 4.1. A Case Study [26] We illustrate our method using the TCV event which occurred at 1310 UT on 15 June The magnetometer chain in the Northern Hemisphere was sunlit, and the chain in the Southern Hemisphere was in darkness in the magnetic latitude range AACGM. The positions of all magnetometers used in this paper are plotted as a function of magnetic latitude and MLT for this day in Figure 1. The Figure 2. Solar wind data as provided by the Wind satellite. Shown are the following: the three components of the interplanetary magnetic field in GSM coordinates, B x, B y, and B z ;thex component of the solar wind ion velocity v x ; and the ion number density n i. The vertical dashed line marks the time at which a series of oscillations commence in the solar wind dynamic pressure P sw = m i n i jv x j 2. Figure 1. A map of the locations of the magnetometer stations used in this study as a function of altitude-adjusted corrected geomagnetic (AACGM) latitude and magnetic local time (MLT) at 1310 UT 15 June The details of the magnetometer stations are in Tables 1 and 2. The triangles and squares denote Northern and Southern Hemisphere stations, respectively. triangles and squares denote the position of stations in the Northern and Southern Hemispheres, respectively. A bipolar signature is observed in the magnetic field data of the GOES 8 satellite (not shown), position (3, 6, 1) R e in GSM coordinates at 1308 UT (0800 MLT), when a similar deflection in the magnetic field is observed at the Earth s surface. This suggests that the TCV event is associated with a change in the magnetopause dynamic pressure [Sibeck and Korotova, 1996]. The fields in Figure 2 from the Wind satellite are as follows: the three components of the IMF (GSM coordinates), B x, B y, and B z, respectively, the x component of the solar wind ion velocity v x, and the ion number density n i. The TCV event is associated with the commencement of a series of oscillations in the solar wind dynamic pressure P sw = m i n i jv x j 2 for which (P sw /P sw ) 1. The oscillations in P sw (and in the IMF) are present in the data from the Wind satellite at 1200 UT (marked by the vertical dashed line in Figure 2). The Wind satellite is located at (205, 22, 8) R e (GSM). Using v x = x/t, we derive a rough estimate of 72 ± 8 min for the time t which the initial changes in solar wind dynamic pressure and IMF 6of11

7 take to reach the magnetosphere, a distance x from the Wind satellite. [27] Figure 3 shows 30-min high-pass-filtered data for the magnetic field components H and D from ground-based magnetometers in the northern (Figure 3a) and southern (Figure 3b) chains for 15 June Features typical of TCVs which may be observed in the data are the change in the sense of the bipolar deflection in H between stations UMQ and UPN in the Northern Hemisphere and between stations A84 and SP in the Southern Hemisphere and the roughly 90 phase difference between the H and D components. The azimuthal phase speeds obtained from longitudinally displaced stations are 14 km/s (at 74 N using data from STF and CH) and 8 km/s (at 73 S using data from A84 and SP), and as expected for a TCV event, the direction of travel is away from noon in both hemispheres. The error in the phase velocity for two longitudinally separated stations that yield a finite phase velocity is at least 20%, which is too high to use this as a method for comparing the magnitudes of the velocities in each hemisphere. [28] The Southern Hemisphere stations with data in the required latitude range for this event are A84 and SP. The maximum in jb 1 j occurs at SP (86 nt). The amplitude at A84 (21 nt) is less than half that at SP; therefore SP is the only Southern Hemisphere station which was used in the subsequent calculation. In the Northern Hemisphere, data in the magnetic latitude range N AACGM are available from stations GHB, SKT, STF, GDH, and UMQ. The maximum value of jb 1 j occurs at GDH (65 nt). The value of jb 1 j at stations STF and UMQ is greater than or equal to half the value of jb 1 j at GDH. These three Northern Hemisphere stations contribute to the calculation, as they are within 3 magnetic latitudinal separation of SP. The ratio R of jb 1 j in the Northern Hemisphere to that in the Southern Hemisphere for each of the pairs SP-UMQ, SP-GDH, and SP-STF lies in the range 0.5 R 2.0, and so the percentage of pairs of stations for which 0.5 R 2.0 is 100%. We repeated the calculation for the second deflection, jb 2 j, and again found that 100% of the pairs of stations satisfied the condition 0.5 R 2.0. We define P j (R 1) as the average percentage for both deflections, which for this TCV event is 100% Statistical Study [29] Table 3 contains a list of TCV events found using the method outlined in section 3.1. For each event, Table 3 lists the following: TCV event j; date; UT; factor by which the solar wind dynamic pressure changes as observed using ACE, Wind, or IMP 8 data (a factor of 2 means that the dynamic pressure has approximately doubled or halved); whether a bipolar pulse in the magnetospheric magnetic field data is observed using geostationary satellites; magnitude of TCV speed in the Northern Hemisphere with the corresponding latitude in parentheses; and magnitude of TCV speed in the Southern Hemisphere. An identifiable change in the solar wind dynamic pressure was found for 10 events, and a bipolar signature in the magnetospheric magnetic field was found for all but one event. Having checked that the 18 candidate events exhibit typical TCVlike signatures in the ground magnetic field, we are satisfied that we are dealing with TCVs caused by pressure changes to the magnetopause. Figure 3. The 30-min high-pass filtered data for the magnetic field components H (solid line) and D (dotted line) from magnetometer stations in the magnetic latitude range altitude-adjusted corrected geomagnetic (AACGM) for our case study TCV event which occurred at 1310 UT 15 June The offset between the plot for each station is 100 nt. Data from the (a) northern and (b) southern magnetometer chains are shown. For one station in the Northern Hemisphere and for one in the Southern Hemisphere we have marked H 1 and H 2, the first and second deflections in the H component of the ground magnetic field associated with the TCV, and we have also marked D 1 and D 2, the two deflections in the D component of the ground magnetic field. 7of11

8 Table 3. Evidence to Support the Classification of the Events as TCVs a Event j Date UT Change in SW Signature in MS N S 1 19 Jan (W) y 12 (76 ) 3 (73 ) 2 21 Jan (W) y Inf. (70 ) 5 (73 ) 3 23 Jan (W) y 12 (73 ) 3 (71 ) 4 9 Feb (W) y 9 (74 ) Inf. (72 ) 5 16 Feb (A, W) y nd nd 6 22 Feb (W) y 19 (75 ) 5 (73 ) 7 2 March (I) y 8 (75 ) Inf. (71 ) 8 5 March (W) y 6 (75 ) 8 (71 ) 9 28 March nd y nd nd 10 3 April (I, W) y 9 (72 ) 14 (73 ) 11 4 April (I, W) n 7 (77 ) 13 (71 ) April (A, W) y 6 (78 ) 2 (73 ) 13 2 May (W) y 30 (73 ) Inf. (73 ) 14 3 May (A, W) y 71 (75 ) Inf. (73 ) 15 5 May (A, W) y nd nd 16 9 June (I) y 4 (77 ) 5 (73 ) June (I, W) y 14 (74 ) 8 (73 ) June (A, I, W) y 13 (74 ) 2 (73 ) a Columns contain the following: TCV event j; date; universal time (UT); factor by which the solar wind (SW) dynamic pressure changes as observed using ACE (A), Wind (W), or IMP 8(I) data; whether a pulse in the magnetospheric magnetic field was observed (MS means magnetosphere, y means yes, and n means no); and TCV speed with corresponding latitude in brackets for the Northern Hemisphere (N) and the Southern Hemisphere (S). Inf. means infinite, and nd means there are no suitable data. Speed [30] We obtained an infinite value for the azimuthal phase velocity in the Southern Hemisphere for four events. The stations used in the Southern Hemisphere to calculate azimuthal phase velocity have a significant latitudinal displacement with respect to each other. Friis-Christensen et al. [1988] observed that TCVs can be elliptically shaped rather than circular, with the major axis forming a significant angle with the direction of travel; so it is possible that under certain conditions the phase front of the TCV may reach the pair of magnetometers being used to calculate the velocity at the same time. For all events (except event 18) for which we are able to extract a finite phase velocity, the direction of travel of the TCV is toward the nightside in both hemispheres. [31] Table 4 gives the properties of the TCV events used in this study. The columns in Table 4 are as follows: TCV event j; date; MLT; ionospheric conductivity level in the Northern Hemisphere, low (L) if the magnetometers concerned are in darkness and high (H) if they are sunlit; conductivity level in the Southern Hemisphere; estimated values of the initial and final IMF associated with the event; and average percentage of conjugate pairs of Table 4. Properties of the TCV Events Used in This Study a IMF (nt) Event j Date MLT N S Initial Final P j (R 1), % 1 19 Jan H H 7, 12, 1 7, 3, Jan L H 7, 1, 11 6, 3, Jan L H 2, 2, 3 3, 2, Feb H H 3, 2, 2 0, 5, Feb H H 5, 0, 3 6, 0, Feb H H nd nd March H H 2, 3, 2 2, 4, March H H 4, 0, 2 4, 4, March H H nd nd April H H nd nd April H H 2, 2, 3 2, 0, April H L 0, 6, 7 7, 6, May H L 4, 7, 12 1, 5, May H L 2, 2, 1 2, 2, May H L 1, 2, 1 6, 8, June H L 5, 2, 2 5, 2, June H L 0, 3, 2 2, 3, June H L 5, 0, 10 5, 6, a Columns contain the following: TCV event j; date; magnetic local time (MLT); ionospheric conductivity level in the Northern Hemisphere (N) and conductivity level in the Southern Hemisphere (S), where L is low and H is high; estimated values of the initial and final IMF (B x, B y, and B z ) associated with the event; and the average percentage of conjugate pairs of ground magnetometer measurements which are similar in amplitude, P j (R 1) (defined in section 3.2). Here nd means there are no appropriate data. 8of11

9 ground magnetometer measurements which are similar in amplitude, P j (R 1) Current Versus Voltage Source [32] The nine TCV events for which the ionospheric conductivities in the relevant regions of the two hemispheres are different by an order of magnitude are the following: 2, 3, 12, 13, 14, 15, 16, 17, and 18. The corresponding values of P j (R 1) to the nearest integer are 67, 75, 100, 86, 0, 50, 71, 100, and 86%, with a mean value over all nine events of 71%. The remaining nine TCV events (1, 4, 5, 6, 7, 8, 9, 10, and 11) for which the ionospheric conductivities in the two hemispheres are similar to an order of magnitude have the following values of P j (R 1): 100, 50, 30, 100, 60, 100, 38, 100, and 95%, with a mean value over all nine events of 75%. The standard deviations for these two groups of events (31 and 30%, respectively) are similar in value according to the F test, so it is valid to proceed with the Student s t test. The significance of the difference between these two means is 0.78 (indicating no significance); the similarity in the intensity of a TCV at conjugate points is not correlated with the ratio of the conductivities at conjugate points. Following the arguments presented in sections 2.4 and 3.3, we propose that this is evidence in support of the hypothesis that TCVs have a current source in the magnetosphere Sensitivity Study [33] The average value, over all 18 events, of the percentage of conjugate pairs of ground magnetometer measurements for a TCV event which are similar in magnitude P j ðr 1Þ is 73%. There is a spread in the values of P j (R 1) over the 18 TCV events between 0 and 100%; most events can be observed by eye to be of similar intensity in both hemispheres, whereas some, most notably event 14, are distinctly different in intensity. We investigated the sensitivity of the method to the way in which we define similar, that is, the values of R lower and R upper as defined in section 3.2. If the range is changed from 0.5 R 2.0 to (1) 0.4 R 2.5 or (2) 0.6 R 1.7, then the overall percentage of readings in the Northern and Southern Hemisphere which are similar changes from 73% to (1) 88% or (2) 59%. [34] In both cases 1 and 2 we find that there is no significant difference between the values of P j ðr 1Þ a and P j ðr 1Þ b (defined in section 3.3). Our conclusion that TCVs have a current source can still be maintained. It is not possible to change the latitude range by much as there are not enough stations in the Southern Hemisphere to do so. Changing the value of the maximum separation of conjugate stations conj from 3 to 4 does not alter the overall percentage for all 18 events at all and only slightly alters the percentages for individual events. Using a value of conj =2 rather than conj =3 only changes P j ðr 1Þ by 2%, to 71%. [35] We investigated whether variations in other factors might be the cause of the spread in the values of P j (R 1). We used the same statistical method as before; for a proposed cause we divided the 18 events into two groups and tested to see if there was a significant difference between the mean value of P j (R 1) for the two groups. We examined whether any indication of a correlation exists between P j (R 1) and (1) the factor by which the solar wind dynamic pressure is observed to change, (2) the MLT at which we observe the TCV event, (3) the orientation of the IMF at the assumed time of relevance, or (4) the observed azimuthal phase speed of the TCV. [36] The mean and the standard deviation of P j (R 1) for the group of events associated with an observed change in solar wind dynamic pressure of a factor of two or above (see column 4 of Table 3) are 71 and 31%, respectively, and for the group where no change in solar wind dynamic pressure is observed, the mean and the standard deviation are 80 and 26%, respectively. The standard deviations are similar according to the F test, so we may apply Student s t test to the two groups of data. The significance of the t test is 0.55; that is, we find no significant correlation between an observed change in solar wind dynamic pressure of a factor of two or greater and the similarity of intensity of conjugate TCV events, and we conclude that variability of the dynamic pressure change in the solar wind between TCV events is not associated with the spread in the values of P j (R 1). [37] We also find that the value of P j (R 1) is not associated with the proximity of the MLT of observation of a TCVevent to noon. Using column 3 of Table 4, we compare events observed in the noon sector ( MLT) with events observed closer to the flanks ( and MLT). According to the F test, the two groups have similar standard deviations (28 and 32%). The mean values of P j (R 1) are 68 and 77%, respectively, and the significance of the t test is The prenoon occurrence peak of TCVs observed by Glassmeier et al. [1989] is also observed in our data set, making it difficult to compare TCV events in the prenoon and postnoon sectors. [38] We compared events where IMF B x is the largest or equal largest component just before the assumed time of relevance to those where it is not the largest component using the values from column 6 in Table 4. The standard deviations of P j (R 1) for these two groups (38 and 18%) are fairly dissimilar (with a significance of 0.06 according to the F test), but for completeness we present them; the mean values of P j (R 1) are 50 and 82%, with an intriguing t test significance for their difference of We also compared events where the change in the value of IMF B x associated with an event is nonzero to those where the change is zero. According to the F test, the standard deviations of P j (R 1) are similar enough to conduct a t test (20 and 39% for the two sets, respectively); the mean values of P j (R 1) are 74 and 69%, with a t test significance for their difference of We are able to conclude that there is no correlation between the change in IMF B x and the spread in the values of P j (R 1), but we are unable to quantify how the degree of radiality of the IMF is correlated with P j (R 1). [39] The mean values of P j (R 1) for the groups of events with finite observed speed in both hemispheres and an infinite observed azimuthal phase velocity in at least one hemisphere (column 7 in Table 3) are 93 and 53%, respectively. This difference has an interesting significance (0.003), but we cannot presume any correlation between the observed azimuthal velocity and variability in P j (R 1) since according to the F test, the standard deviations of the two groups (11 and 32%) are dissimilar with a significance of [40] We find that the variation in the values of P j (R 1) for our 18 TCV events cannot be correlated with the 9of11

10 observed change in solar wind dynamic pressure associated with the TCV event, nor can it be associated with proximity of the MLT of observation to noon or the change in IMF B x. We are unable to comment on the roles that the variation in the degree of radiality of the IMF or the azimuthal phase speed of the TCV play in determining the variation in the values of P j (R 1) for this data set. 5. Conclusions [41] Averaged over the 18 TCV events presented, the intensity at conjugate points is similar in value, regardless of any difference in the conductivity of the two hemispheres, strongly indicating a current generator mechanism. This result provides a new and important part of our description of the magnetospheric-ionospheric coupling that gives rise to TCVs. It is consistent with the amplitude of TCV events being controlled by the local plasma density, the magnetic field strength, and the acceleration of the plasma (see equation (1)). We need in situ measurements of these three quantities in order to establish the relationship between the solar wind parameters and the ionospheric response. It is also consistent with the maximum perturbation amplitude of TCVs being well within the magnetosphere and is a result of the dissipation of fast mode compressional waves by the generation of FACs which close in the ionosphere. [42] We find that the differences between individual TCV events of the ratio of TCV intensity in the Northern Hemisphere to that in the Southern Hemisphere are not correlated with the factor by which the solar wind dynamic pressure changes, with changes in IMF B x at the time of the event, or with the MLT at which we observe the TCVs. The cause of the dissimilarity in amplitude between the two hemispheres for some TCV events is still unresolved. [43] Acknowledgments. We would like to acknowledge the Coordinated Data Analysis Web (CDAWeb) for providing access to ACE, Wind, IMP 8, and GOES satellite data and the following people for providing satellite magnetometer data: N. Ness at the Bartol Research Institute (ACE), R. Lepping at NASA/GSFC (Wind), A. Szabo and R. P. Lepping at NASA/ GSFC (IMP 8), and H. Singer at NOAA SEC (GOES 8 and GOES 9). We thank K. Ogilvie at NASA/GSFC for Wind plasma data, D. J. McComas at Southwest Research Institute for ACE plasma data, and A. Lazarus (MIT) for IMP 8 plasma data. We thank J. Watermann at the Danish Meteorological Institute for the Greenland West and East magnetometer data and SPRL and NSF for the provision of MAGIC data. We thank A. T. Weatherwax at Maryland University for providing fluxgate magnetometer data from South Pole and the PENGUIN AGOs, and we thank the principal investigator for the fluxgate magnetometers, L. J. Lanzerotti of Bell Labs/Lucent Technologies. We acknowledge the institutes that maintain the IMAGE magnetometer array, the principal investigative institute for IMAGE, the Finnish Meteorological Institute, and the institutes who run the MACCS array of magnetometers in arctic Canada, namely, Boston University and Augsburg College. We thank Mervyn Freeman for useful comments on the original manuscript. [44] Arthur Richmond thanks David G. Sibeck and another reviewer for their assistance in evaluating manuscript 2003JA References Araki, T. (1994), A physical model of the geomagnetic sudden commencement, in Solar Wind Sources of Magnetospheric Ultra-Low-Frequency Waves, Geophys. Monogr. Ser., vol. 81, edited by M. J. Engebretson, K. Takahashi, and M. Scholer, pp , AGU, Washington, D. C. Dudeney, J. R., R. I. Kressman, and A. S. Rodger (1998), Automated observatories for geospace research in polar regions, Antarct. Sci., 10(2), Engebretson, M. J., W. J. Hughes, J. L. Alford, E. Zesta, L. J. Cahill Jr., R. L. Arnoldy, and G. D. Reeves (1995), Magnetometer array for cusp and cleft studies observations of the spatial extent of broadband ULF magnetic pulsations at cusp/cleft latitudes, J. Geophys. Res., 100(A10), 19,371 19,386. Friis-Christensen, E., M. A. McHenry, C. R. Clauer, and S. Vennerstrøm (1988), Ionospheric traveling convection vortices observed near the polar cleft: A triggered response to sudden changes in the solar wind, Geophys. Res. Lett., 15(3), Fukushima, N. (1969), Equivalence in ground geomagnetic effect of Chapman-Vestine s and Birkland-Alfvén s electric current systems for polar magnetic storms, Rep. Ionos. Space Res. Jpn., 23(3), Glassmeier, K.-H., M. Hönisch, and J. Untiedt (1989), Ground-based and satellite observations of traveling magnetospheric convection twin vortices, J. Geophys. Res., 94(A3), Hardy, D. A., M. S. Gussenhoven, R. Raistrick, and W. J. McNeil (1987), Statistical and functional representations of the pattern of auroral energy flux, number flux, and conductivity, J. Geophys. Res., 92(A11), 12,275 12,294. Holzworth, R. H., and C.-I. Meng (1975), Mathematical representation of the auroral oval, Geophys. Res. Lett., 2(19), Iijima, T., and T. A. Potemra (1976), Field-aligned currents in the dayside cusp observed by Triad, J. Geophys. Res., 81(34), Lam, M. M., and A. S. Rodger (2001), A case study test of Araki s physical model of geomagnetic sudden commencement, J. Geophys. Res., 106(A7), 13,135 13,144. Lepping, R. P., A. J. Lazarus, L. J. Moriarty, P. A. Milligan, R. S. Kennon, R. E. McGuire, and W. H. Mish (1992), IMP-8 solar wind magnetic field and plasma data in support of Ulysses-Jupiter encounter: January 1992, report, Lab. for Extraterr. Phys., NASA Goddard Space Flight Cent., Greenbelt, Md., Dec. Lepping, R. P., et al. (1995), The Wind magnetic field investigation, Space Sci. Rev., 71, Lühr, H. (1994), The IMAGE magnetometer network, in STEP International, vol. 4, Sci. Comm. on Sol.-Terr. Phys., Boulder, Colo. Lühr, H., W. Blawert, and H. Todd (1993), The ionospheric plasma flow and current patterns of traveling convection vortices: A case study, J. Atmos. Terr. Phys., 55(14), Lühr, H., M. Lockwood, P. E. Sandholt, T. L. Hansen, and T. Moretto (1996), Multi-instrument ground-based observations of a travelling convection vortices event, Ann. Geophys., 14, Lysak, R. L., and D. Lee (1992), Response of the dipole magnetosphere to pressure pulses, Geophys. Res. Lett., 19(9), McComas, D. J., S. J. Bame, P. Barker, W. C. Feldman, J. L. Phillips, P. Riley, and J. W. Griffee (1998), Solar wind electron proton alpha monitor (SWEPAM) for the advanced composition explorer, Space Sci. Rev., 86, Moretto, T., and A. Yahnin (1998), Mapping travelling convection vortex events with respect to energetic particle boundaries, Ann. Geophys., 16, Moretto, T., D. G. Sibeck, and J. F. Watermann (2003), Occurrence statistics of magnetic impulsive events, Geophys. Res. Abstr., 5, Murr, D. L., W. J. Hughes, A. S. Rodger, E. Zesta, H. U. Frey, and A. T. Weatherwax (2002), Conjugate observations of traveling convection vortices: The field-aligned current system, J. Geophys. Res., 107(A10), 1306, doi: /2002ja Ogilvie, K. W., et al. (1995), SWE, a comprehensive plasma instrument for the Wind spacecraft, Space Sci. Rev., 71, Rishbeth, H., and O. K. Garriott (1969), Introduction to Ionospheric Physics, Int. Geophys. Ser., vol. 14, edited by J. van Mieghem, Academic, San Diego, Calif. Rosenberg, T. J., and J. H. Doolittle (1994), Studying the polar ionosphere and magnetosphere with Automatic Geophysical Observatories: The U. S. program in Antarctica, Antarct. J. U. S., 29(5), Samson, J. C., J. A. Jacobs, and G. Rostoker (1971), Latitude-dependent characteristics of long-period geomagnetic micropulsations, J. Geophys. Res., 76(16), Sibeck, D. G., and G. I. Korotova (1996), Occurence patterns for transient magnetic field signatures at high latitudes, J. Geophys. Res., 101(A6), 13,413 13,428. Sibeck, D. G., R. A. Greenwald, W. A. Bristow, and G. I. Korotova (1996), Concerning possible effects of ionospheric conductivity upon the occurrence patterns of impulsive events in high-latitude ground magnetograms, J. Geophys. Res., 101(A6), 13,407 13,412. Singer, H. J., L. Matheson, R. Grubb, A. Newman, and S. D. Bouwer (1996), Monitoring space weather with the GOES magnetometers, in GOES-8 and Beyond, edited by E. R. Washwell, Proc. SPIE Int. Soc. Opt. Eng., 2812, Sitar, R. J., J. B. Baker, C. R. Clauer, A. J. Ridley, J. A. Cumnock, V. O. Papitashvili, J. Spann, M. J. Brittnacher, and G. K. Parks (1998), Multi-instrument analysis of the ionospheric signatures of a hot flow anomaly occurring on July 24, 1996, J. Geophys. Res., 103(A10), 23,357 23, of 11