Relative contribution of ionospheric conductivity and electric field to ionospheric current

Transcription

1 JOURNAL OF GEOPHYSICAL RESEARCH, VOL. 107, NO. A10, 1330, doi: /2001ja007545, 2002 Relative contribution of ionospheric conductivity and electric field to ionospheric current Masahiko Sugino, Ryoichi Fujii, Satonori Nozawa Solar-Terrestrial Environment Laboratory, Nagoya University, Nagoya, Japan Stephan C. Buchert and Hermann J. Opgenoorth Uppsala Division, Swedish Institute of Space Physics, Uppsala, Sweden Asgeir Brekke Faculty of Science, University of Tromsø, Norway Received 30 July 2001; revised 21 January 2002; accepted 1 April 2002; published 31 October [1] Based on 10 years of European Incoherent Scatter (EISCAT) radar data, we have investigated the variation of ionospheric conductivities, electric fields, and currents over magnetic local time (MLT). Pedersen and Hall currents are generally enhanced especially on the nightside, and the MLT dependencies of these two types of ionospheric currents is similar, as previously found for field-aligned currents and auroral electrojets, respectively. We readdress the question to what extent the conductivity or the electric field contributes to the ionospheric current in different periods of MLT. On average, higher Pedersen conductivities are seen at MLT in comparison with other MLT intervals, and these conductivities crucially contribute to Pedersen currents over two MLT sectors, around midnight and in the late morning. However, when the Pedersen current is stronger, not only the Pedersen conductivity but also the electric field becomes higher statistically on the nightside. Hall conductivities are also higher at about MLT, showing two maxima, around midnight and in the late morning, and they increase more strongly than Pedersen conductivities on a statistical basis. Thus the nightside electrojets are mainly due to high Hall conductivities. INDEX TERMS: 2409 Ionosphere: Current systems (2708); 2407 Ionosphere: Auroral ionosphere (2704); 2431 Ionosphere: Ionosphere/magnetosphere interactions (2736); KEYWORDS: auroral ionosphere, current systems, electric fields, ionosphere/ magnetosphere interactions, particle precipitation, plasma convection Citation: Sugino, M., R. Fujii, S. Nozawa, S. C. Buchert, H. J. Opgenoorth, and A. Brekke, Relative contribution of ionospheric conductivity and electric field to ionospheric current, J. Geophys. Res., 107(A10), 1330, doi: /2001ja007545, Introduction [2] The current system connecting the Earth s ionosphere with space consists of magnetospheric currents, fieldaligned currents (FACs), and closure currents in the ionosphere. The distributions in time and space of these currents depend not only on the driving mechanism of the FACs but also on the conditions in the ionosphere, particularly on the ionospheric conductivity. A change of the conductivity may alter the electric current as well as other parameters such as the electric field. Therefore the ionospheric conductivity and the electric field are related to each other in various current systems. [3] Large-scale distribution in time and space of FACs has been determined using magnetometers on board polar orbiting satellites. In the auroral zone, region 1 and region 2 currents [Iijima and Potemra, 1976] form roughly two concentric ovals. The FACs are part of a system which Copyright 2002 by the American Geophysical Union /02/2001JA transfers energy and momentum from the Earth s magnetosphere to the ionosphere [see e.g., Iijima, 2000]. While it is clear that the system is driven ultimately by solar wind, the direct physical mechanisms providing energy and momentum and initiating the transfer of these into the ionosphere are still not so well understood. Also, detailed knowledge is lacking as to whether and how these driving physical mechanisms are influenced by ionospheric conditions. Ionospheric conditions vary in space and time, and provide feedback to the magnetosphere. Several experimental studies addressing the relation between ionospheric conditions, electric fields, and currents have been performed. Under geomagnetic quiet conditions, when solar illumination produces higher ionospheric conductivities, the intensity of large-scale region 1 currents increases [Fujii and Iijima, 1987]. This behavior is consistent with magnetospheric plasma convection and the corresponding electric field which are controlled by other factors rather than the amount of mechanical braking by the ionosphere and its corresponding electrical load. In other words, the region 1 currents are likely driven by a voltage generator SIA 20-1

2 SIA 20-2 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD [e.g., Lysak, 1985; Vickrey et al., 1986]. On the other hand, the generator of the region 2 currents shows the characteristics of mixture of the voltage and current generators. It might be worth to point out that the current density (latitudinal width) of the region 1 (region 2) current tends to increase with an increase of the ionospheric conductivity [Fujii and Iijima, 1987]. [4] On the other hand, incoherent scatter (IS) radar observations show that ionospheric conductivities and electric fields are often anti-correlated [e.g., Robinson, 1984]. However, the electric field does not always show anticorrelation with the conductivity, but rather it can show a variety of relations [e.g., Marklund, 1984]. Sometimes both the ionospheric conductivity and the electric field are enhanced. IS radar observations, however, are not capable of detecting the direction of FACs. Both upward and downward FACs are carried mostly by electrons due to their larger mobility in comparison to ions, while the ionospheric closure current is mostly the Pedersen current carried by ions. This implies that an enhancement of the ionospheric electron density or the Pedersen conductivity is generally seen in an upward FAC region, and depletion is seen in a downward FAC region. A downward FAC is expected to require a larger ionospheric electric field than the corresponding upward FAC to maintain the same amount of closure current connecting different Pedersen conductivity regions. [5] As we have seen above, all of the electrodynamic parameters are closely interrelated, and the relation between the ionospheric conductivity and the electric field can shed some light on the nature of the magnetospheric generator and energy source. Thus it is desirable to investigate statistically and quantitatively the relations between these parameters simultaneously measured, using databases as large as possible, covering the entire magnetic local time (MLT) regions. IS radars in the polar region, such as the EISCAT radar, enable us to measure simultaneously the electron density, the ion velocity, the ion or electron temperature with a good altitude resolution, and hence to derive the ionospheric conductivity, the electric field, and the ionospheric current distributions [e.g., Senior et al., 1990; Davies and Lester, 1999; Ahn et al., 1999]. Based on Chatanika IS radar data, Kamide and Vickrey [1983] studied the relative contribution of the ionospheric conductivity and the electric field to the ionospheric current. Kamide and Vickrey [1983] concluded that the auroral electrojet around midnight was mainly caused by enhanced conductivity, while in the afternoon and morning sectors, the enhanced electric field mainly drove the electrojet. Davies and Lester [1999] showed that the times of peak conductivity did not correspond to the time at which the electric field maximized on a statistical basis. [6] The purpose of this study is to evaluate the relative contribution of the ionospheric conductivity and the electric field to the ionospheric current in order to further understand the generation mechanisms of the current systems and the energy coupling between the ionosphere and the magnetosphere. The refined analysis method we have presently adopted enables us not only to evaluate more quantitatively the relative contribution than previous studies but also to determine how strongly these two physical parameters contribute relatively to the ionospheric current when the current intensity changes. 2. Data Set [7] In this study we have used Common Program One and Two (CP-1 and CP-2) mode data obtained from the tristatic European Incoherent Scatter (EISCAT) radar, the so-called Kiruna-Sodankylä-Tromsø (KST) UHF radar system [Folkestad et al., 1983; Rishbeth and Williams, 1985]. The CP-1 mode is one of the seven EISCAT radar common program modes [see e.g., Collis, 1995], and the transmitting antenna is fixed along the magnetic field direction at Tromsø (69.6 N, 19.2 E in geographic coordinates). Scattered signals of the radar beam transmitted at Tromsø are received at three stations; Tromsø itself, Kiruna, and Sodankylä, and thus the CP-1 mode provides a threedimensional ion velocity at a specific altitude (usually 278 km) in the F region. It also provides the electron density and ion or electron temperature with a height resolution of about 3 km in the altitude range from 86 to 268 km, and of about 22 km from 146 to 586 km. The half width of the radar beam is about 0.7. The time resolution of the postintegrated parameters is 2 5 min for the electron density and temperatures, and 2 10 min for the electric field. In the CP- 2 mode, the transmitting antenna is sequentially pointed to four positions. One of the positions is the field-aligned direction (remaining for 45 s), and the other three positions form a triangle. Six minutes are required to complete the antenna cycle. Tristatic measurements are made in the F region (usually 278 km) in each of the four positions. Here we have used only data from the field-aligned beam, which are the same as for the CP-1 mode. In this study we used 56 sets of CP-1 observations from 138 days and 33 sets of CP- 2 observations from 90 days during a 10-year period (November 1986 to November 1996). [8] The electric field can be determined from the threedimensional ion velocity in the F region, where ions move by E B drift since the ion gyrofrequency is much greater than the ion-neutral collision frequency in this location. The perpendicular electric field E is thus expressed as E ¼ v B where v is the ion velocity measured in the F region, and the magnetic field B is derived from the International Geomagnetic Reference Field (IGRF1995) model [Barton, 1997]. Since the magnetic field-aligned electric field is much smaller than the perpendicular electric field, the magnetic field lines are assumed to be equipotential down to around 90 km in height [Whalen et al., 1975]. [9] As in previous studies, we introduce our data in diurnal distribution patterns of the electric field. Figure 1 shows the MLT variations of the electric field; north-south component (left panels), east-west component (middle panels), and magnitude (right panels) with different ordinate scales. In order to compare our data with other data [e.g., Senior et al., 1990; Ahn et al., 1999] using the similar method in terms of sorting data, we present our data according to the Kp index; Kp <1,2<Kp <3,4<Kp < 5, and 6 < Kp. The abscissa is MLT from midday through midnight to midday. Here we define MLT at Tromsø as ð1þ

3 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA 20-3 Figure 1. Northward electric field (left), eastward electric field (middle), and magnitude of electric field (right) over MLT for different ranges of the Kp index (top to bottom). In each interval in MLT, the thick lines show mean values, and the vertical bars indicate the standard deviations. UT + 3 hours 15 min. The thick line in each of the panels connects the mean values of the electric fields which are calculated for every 1 hour in MLT. The vertical bar indicates the standard deviation of the electric fields for each one hour. As is well known, the north-south electric field is generally much greater than that in the east-west direction at least on a statistical basis. Except for 6 < Kp, the northward electric field in the evening sector is higher than the southward one in the morning sector, as reported by Ahn et al. [1999]. North-south electric field reversals occur around midnight, and this reversal time shifts to an earlier time with the increase of Kp and does not coincide with the reversal of the east-west electric field in time. The strongest westward component is recorded around midnight, consistent with previous measurements [e.g., Senior et al., 1990]. Senior et al. [1990] have pointed out that there is a particularly good agreement between electric field models derived from EISCAT and Millstone Hill radar observations. [10] The Pedersen and Hall conductivities are calculated using formulas by Brekke and Hall [1988] and the MSISE- 90 neutral atmospheric model [Hedin, 1991]. The heightintegrated Pedersen and Hall conductivities are calculated

4 SIA 20-4 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD by integrating the conductivities over the height range from 90 to 300 km. The Pedersen conductivity is typically largest around 120 km in height, while the Hall conductivity is largest around 110 km in height [Brekke and Hall, 1988]. The adoption of the lower height limit of 90 km is justified except under extremely energetic precipitation which could produce as much as 15% of the height-integrated Hall conductivity below 90 km [Schlegel, 1988]. [11] Figure 2 shows the MLT variations of the heightintegrated conductivities according to the Kp index. The ordinate is the height-integrated Pedersen (left panels) or Hall (right panels) conductivity with different scales. Two major sources of the ionospheric conductivity can be noticed; one associated with the solar EUV radiation varying smoothly and reaching maximum near the local noon sector and the other associated with auroral particle precipitation. [12] The magnitudes of both the electric field and the conductivity increase with a higher Kp value. Particularly for the lower Kp, there are no significant differences between our data and the Chatanica radar measurements shown by Ahn et al. [1999], although they do not include the extreme cases such as 6 < Kp. Our data particularly show enhancements of the conductivity due to particle precipitation in the morning sector. 3. Analysis and Results [13] Based on the electric field (E) and the heightintegrated Pedersen or Hall conductivity ( P or H ), we calculate the height-integrated Pedersen or Hall current (J P or J H ) as: J H J P ¼ P jej ¼ H jb Ej where b is a unit vector along the geomagnetic field. Current continuity requires that j jj ¼ E grad P þ P dive þ ðb E ð2þ Þ grad H þ H divðb EÞ ð3þ where j k is a field-aligned current density. Assuming that a curl-free potential electric field exists in the ionosphere, the last term is omitted. [14] Figure 3 shows the diurnal distribution patterns of the height-integrated Pedersen (left panels) and Hall (right panels) currents sorted by the Kp index with different ordinate scales. The thick line with vertical bars in each panel denotes the current intensity averaged over each 1-hour MLT interval along with its standard deviation. The MLT distribution of the Pedersen current, particularly that of the averaged current intensity, has the following two characteristics while the current intensity scatters greatly, ranging from near zero to a few hundreds, occasionally even 1000 ma m 1. First, the Pedersen current intensity is enhanced in the dusk and dawn sectors rather than around the midnight sector for all Kp ranges. The two peaks at dawn and dusk tend to move toward the dayside with the increase of Kp. For the lowest Kp range, enhanced currents are concentrated more around midnight, while currents are enhanced over a wide range of MLTs for higher Kp values. Second, the current intensity increases with a higher Kp value. Iijima and Potemra [1976] have reported that FACs also increase with a higher Kp value. It is mainly the Pedersen current which closes FACs in the ionosphere [e.g., Sugiura et al., 1982]. Hence the characteristics of the Pedersen current are expected to provide information not only on the electrodynamic processes in the ionosphere, such as how the polarization field is produced due to conductivity inhomogeneity, but also on the driving mechanisms of FACs. Therefore we will mainly discuss the Pedersen current in the present study. [15] On the other hand, the height-integrated Hall current (right panels) also shows nearly the same tendency as the Pedersen current. Generally the intensity of the Hall current is greater than that of the Pedersen current due to relatively high values of the Hall conductivity compared to the Pedersen conductivity. The Hall current is, by definition, dissipation free and is considered to play a lesser important role in the closure current of FACs than the Pedersen current. The Hall current generally contributes more to the magnetic field variations measured on the ground than the Pedersen current does [Fukushima, 1969]. Previously the Hall current has been used instead of the Pedersen current for similar studies [e.g., Kamide and Vickrey, 1983; Kamide and Kokubun, 1996]. [16] Next we proceed with assessing the relationships between the ionospheric conductivity and the electric field quantitatively, focusing on how strongly the ionospheric conductivity or the electric field contributes to the ionospheric current. As we have mentioned above, since the ionospheric conductivity and the electric field closely interact with each other, it is necessary to analyze these parameters together. [17] When we compare two sets of the ionospheric conductivity and the electric field obtained at two different times, it is obvious that a higher conductivity (electric field) does not necessarily mean that the conductivity (electric field) plays a more important role in the ionospheric current than the electric field (conductivity) does, since the electric field (conducti-vity) can also be higher. One of the best ways to evaluate the relative contribution of the conductivity and the electric field E is to compare these two parameters for a fixed value of current intensity J = E. In this situation, a higher conductivity (electric field) always means that the conductivity (electric field) make a larger relative contribution. [18] Therefore we sort all available data according to the current intensity, and then we compare the conductivity with the electric field only within a data set where the current has a certain fixed value of intensity. In other words, we compare with E for a data set where all data have approximately the same constant, J(= E ). Also, this method enables us to determine quantitatively how strongly the ionospheric conductivity or the electric field contributes to the ionospheric current for different levels of the current intensity. In practical terms, we have sorted the data into three levels; relatively strong, medium, and weak current levels, each of which has a finite range (±50 ma m 1 for J P and ±100 ma m 1 for J H ) of the current intensity. Obviously this method can provide significant statistical characteristics only when it is applied to a large number of data, such as the EISCAT radar long-term database. [19] First, we show the relative contribution of the Pedersen conductivity and the electric field to the Pedersen

5 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA 20-5 Figure 2. Height-integrated Pedersen (left) and Hall (right) conductivity over MLT for different ranges of the Kp index (top to bottom). The format is the same as in Figure 1. current (JP = PE) as a function of MLT and also how the Pedersen conductivity or the electric field contributes relatively to the Pedersen current for different levels of the current intensity. Each of the three left panels in Figure 4 is the scatterplot of the height-integrated Pedersen conductivity and the electric field strength using all the available data. The total number of data is The data points are not distributed homogeneously, but appear to have envelopes, as reported by Robinson [1984]. [20] In each scatterplot two hyperbolic curves are drawn. Their equations are C1 = PE and C2 = PE with C1 < C2. All data points between the two hyperbolic curves have the current intensity JP where C1 < JP < C2. The data points between the two hyperbolas in each panel are considered in the present study to have the same current intensity. Our three levels of the current intensity are: a strong current level (top panel) of 400 < J P < 500, a medium current level (middle panel) of 250 < JP < 350, and a weak current level (bottom panel) of 100 < JP < 200 in ma m 1. The center current intensities are 450, 300, and 150 ma m 1, and the numbers of the selected data are 660, 1591, and 4098, respectively. [21] The corresponding right panels in Figure 4 show the MLT dependence of the Pedersen conductivity for the data

6 SIA 20-6 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD Figure 3. Height-integrated Pedersen (left) and Hall (right) current over MLT for different ranges of the Kp index (top to bottom). The format is the same as in Figure 1. selected in the left panels by the inequality C 1 < J P < C 2 The ordinate is the height-integrated Pedersen conductivity and the abscissa is the MLT. A high conductivity implies that at the same time the electric field is low, since all the data points in one panel are within a narrow range of the current intensity. Thus, if a period of MLT shows a high average conductivity compared to other periods, this indicates that the current flows due to the high conductivity rather than due to the high electric field, or that the conductivity is relatively more important for the current in this MLT period. [22] At about MLT in comparison with other MLT intervals, the mean curve in the right panels shows a period of a significant high Pedersen conductivity and thus a corresponding low electric field. Also, the average conductivity over MLT has two peaks; one is a broad peak around midnight (around 2300 MLT) and the other one is a narrow peak in the late morning (around 0600 MLT). These characteristics can be seen regardless of the current intensity. Therefore how the values of the current intensity for strong, medium, and weak current levels are selected is not so important. [23] At periods of the high conductivity, data scatter greatly from the mean values. As an example, at MLT for J P = 150 ma m 1, the conductivities are spread up to 20 S,

7 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA 20-7 Figure 4. (Left) Scatterplot of the height-integrated Pedersen conductivity ( P ) and the electric field strength (E) for all the available data used in the present analysis. All data points between two hyperbolic curves, C 1 = P E and C 2 = P E (C 1 < C 2 ), have the Pedersen current intensity C 1 < J P < C 2. Three levels of the current intensity are selected: a strong current level (top) of 400 < J P < 500, a medium current level (middle) of 250 < J P < 350, and a weak current level (bottom) of 100 < J P < 200 in ma m 1. (Right) P over MLT for the data with C 1 < J P < C 2 which are selected in the left panels. while the mean values are 7 10 S. On the other hand, the scatter of data in the daytime is smaller and comes mainly from the change of the solar zenith angle depending on the season. [24] In order to show how the data scatter from the mean values on the nightside, we present histograms of conductivity distribution for the different current levels and MLT sectors in Figure 5. From left to right, the current levels (J P ) are 150, 300, and 450 ma m 1, corresponding from bottom to top in Figure 4, respectively. From top to bottom in Figure 5, MLT changes from dusk through midnight to dawn. The data are binned according to the difference between the Pedersen conductivity ( P ) and its mean value (Ave. P ). The ordinate is the percentage of data. Bins with more than 10% are darkly shaded. Mean values and numbers of data

8 SIA 20-8 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD Figure 5. Histograms of the height-integrated Pedersen conductivity distribution for 2-hour MLT periods. From left to right, J P = 150, 300, and 450 ma m 1. From top to bottom, the MLT changes from dusk through midnight to dawn. In each column, the abscissa is the difference between the Pedersen conductivity ( P ) and its mean value (Ave. P ), and the ordinate is the percentage of data. Mean values and numbers of data points are also noted. Bins with more than 10% of all data are shaded. points are also shown. Most of the data in each column are concentrated around the mean values ( P Ave. P =0). Thus, in Figure 4 the curve connecting the mean values represents the MLT dependence of the bulk of the data which are in the darkly shaded bins in Figure 5. Data are more concentrated around the mean values especially in dusk sectors (before 2000 MLT) than in other MLT regions. After 2000 MLT, more scattered data appear as a tail at higher conductivities. Especially at MLT, the scattering of the conductivity data increases. Presumably this is an effect of substorm activity which is associated with particle precipitation causing high conductivity values. Also, the electric fields at these scattered data areas with high conductivities become considerably lower than the electric fields for most of the data in the darkly shaded bins. The conductivity distribution has a more pronounced tail around

9 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA 20-9 Figure 6. Same as Figure 4 but for the height-integrated Hall conductivity and current. Three levels are selected: a strong current level (top) of 800 < J H < 1000, a medium current level (middle) of 500 < J H < 700, a weak current level (bottom) of 200 < J H < 400 in ma m 1. midnight than in the late morning. After 0400 MLT, data concentrate more again around the mean values, however, the distributions are more flat around the peaks compared with those in dusk sectors. [25] We summarize our observations so far as follows: (1) For a certain fixed value of the current intensity, the conductivity is higher in the MLT interval. This must be at the expense of the electric field strength. (2) Two peaks of the high conductivity are seen within this interval; one is a broad peak around midnight and the other one is a narrow peak in the late morning. (3) Especially around midnight, there are events with much higher than average conductivities and with correspondingly very low electric fields. [26] Next we investigate the Hall conductivity and current (J H = H E) using the same approach as was used for the Pedersen. Figure 6 shows the Hall conductivity and current in the same format as in Figure 4. Again, we define three levels: a strong current level (top panel) of 800 < J H < 1000, a medium current level (middle panel) of 500 < J H < 700,

10 SIA SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD Figure 7. Ratio of P for a J P of 300 ma m 1 to P for a J P of 150 ma m 1 is plotted over MLT (top left). The same ratio is shown for the electric field (top right). Ratio of H for a J H of 600 ma m 1 to H for a J H of 300 ma m 1 is plotted over MLT (bottom left), as well as the ratio of the electric field (bottom right). and a weak current level (bottom panel) of 200 < J H < 400 in ma m 1. The center current intensities are 900, 600, and 300 ma m 1, and the numbers of the selected data are 357, 947, and 3684, respectively. [27] Plotting the Hall conductivity over MLT for these three levels of the current intensity reveals similar features as before for the Pedersen conductivity, that is, high Hall conductivities implying corresponding low electric fields at about MLT. Within this interval, two peaks of the high conductivity are seen; one around midnight and the other in the late morning. The peak in the late morning is clear even for the weak current level, and the peaks are generally more distinct for the Hall current than for the Pedersen current. Also, events with scattered, very high conductivities (such over 40 S) can be seen mainly at MLT. [28] The results above show that the conductivity is higher and the electric field lower at certain MLT compared to other MLT regions for fixed ranges of the current intensity. Next we address the question of whether and how much, on average, the conductivity or the electric field changes between different levels of the current intensity. In Figure 7 the conductivity or electric field ratios of medium to weak current levels are plotted over MLT. [29] First we discuss the ratios for the Pedersen current levels, J P = 300 and J P = 150 ma m 1, which are shown in the upper two panels. The left panel shows the ratio of the Pedersen conductivity P [J P = 300]/ P [J P = 150], and the right panel shows the same of the electric field E[J P = 300]/ E[J P = 150]. On average, the ratios of both the Pedersen conductivity and the electric field are larger than 1. On the nightside, higher Pedersen conductivities and higher electric fields contribute almost equally from weak to medium levels of the Pedersen current intensity. Here we use the term equally in the sense that, for example, both the conductivity pffiffi and the electric field are higher by a factor of 2 when the current intensity is stronger by 2 (from 150 to 300 ma m 1 ). As a reference, p the ffiffiffi centered dotted line in each panel in Figure 7 denotes 2. At MLT, as shown in Figure 5, the conductivity distribution becomes more scattered, reaching sometimes rather high values presumably due to substorm activity. However, it should be noted that the electric field also has a higher value by a factor of 1.4 to 1.6 on average when the Pedersen current intensity is stronger by a factor of 2. [30] The lower two panels in Figure 7 show the ratio of the Hall conductivity (left panel) and the ratio of the electric

11 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA Figure 8. Same as Figure 7 but for different levels of current intensity, 450 to 150 ma m 1 for J P, and 900 to 300 ma m 1 for J H. field (right panel) for the Hall current. The format is the same as in the upper panels, and the Hall current between medium and weak levels also differ by a factor of 2 (from 300 to 600 ma m 1 ). The ratio of the Hall conductivity is 1.3 to 1.9 at MLT in the left panel, and the ratio of the electric field is 1.0 to 1.5 as seen in the right panel. Thus a stronger level of the Hall current intensity is due more to higher Hall conductivities than to higher electric fields. Similar conclusions can be drawn from the ratios of the strong to low current levels which are plotted in Figure 8. In this case the ratio of the current levels is 3, although the results are not based on numerous data. [31] As shown in Figures 4 and 6, both the Pedersen and Hall conductivities are high in the late morning [e.g., Brekke et al., 1974]. Figure 9 shows the distributions of the Pedersen (left panels) or Hall (right panels) current density in both altitude and MLT according to current intensity levels. A variation of the current density in altitude is caused by the altitude variation of the conductivity, since the magnetic and electric fields can be assumed constant at a height down to about 90 km [Whalen et al., 1975]. The distributions of the Pedersen current density are not changed so significantly with the current intensity levels. The Hall current and conductivity vary in altitude mainly according to the relatively high energy of incident electrons. For any current intensity levels, the height of the lower border of the Hall current density is considerably lower just between the evening and the morning sectors, implying that the electron precipitation energy becomes harder. This descent of the lower edge of the Hall current is especially noticeable around 0800 MLT for all the three current intensity levels, and the lower border goes down to below 90 km. According to Rees [1963], rather hard precipitation with energies of up to 50 kev is required to ionize at such a low altitude. Hard precipitation occurs around 0800 MLT regardless of the level of the current intensity. Also around midnight there are high Hall conductivities and currents at low altitudes as well, however, these conductivities and currents are clearly seen only for the strong current level. 4. Discussion [32] We have investigated the variations of ionospheric conductivities and of electric fields, their relations, and their distributions in different MLT sectors. We developed a refined method to investigate a number of IS radar measurements of the conductivity and the electric field. The present method of analysis addresses the division of the magnetosphere-ionosphere system into states of strong, medium, and weak levels of the current intensity, and how strongly the conductivity or the electric field relatively contributes to the current in different MLT regions. This method of analysis

12 SIA SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD Figure 9. Contour plot of the Pedersen (left) or Hall (right) current over altitude and MLT for the three levels of the current intensity. does not address whether or which temporal correlations exist between the conductivity and the electric field. Rather, the question is asked how strongly the conductivity or the electric field relatively contributes to the different levels of the current intensity and different MLT regions. [33] We have classified the data into two regions: the region where ionospheric current intensity is characterized by a relatively high conductivity, and the region where intensity is characterized by a relatively strong electric field. From the point of view of energy inputs from the magneto-

13 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA sphere into the ionosphere, differences between the two regions exist. Assuming that the ionospheric current (J = E) is constant, the magnetic energy dissipation (described as J E = J P E = P E 2 = J P 2 / P ) is larger in the strong electric field region than in the high conductivity region. On the other hand, precipitating plasmas, mainly electrons, also bring the energy into the ionosphere. Very roughly, the conductivity represents the effect of the precipitating electrons, thus the energy due to particle precipitation is brought significantly into the high conductivity region. [34] Also differences between the two regions can be inferred from a FAC driving mechanism. In the region characterized by strong electric fields, the current system is driven considerably by the plasma convection, and a FAC is much related to P dive (related to rotv ). On the other hand, in the region characterized by high conductivities, the current system is expected to be more driven by other mechanisms that enhance particle precipitation, and a FAC may be related significantly to the divergence of the pressure gradient current. [35] Our results show that, on average, a region characterized by high conductivities is seen at MLT in comparison with other MLT intervals, and the higher conductivities crucially contribute to the ionospheric currents over two MLT sectors, around midnight and in the late morning. Around midnight, the conductivity is higher especially for the strong current level. In the late morning, even for the weak current level, the conductivity is high compared with other MLT intervals (Figures 4 and 6). [36] First of all, we discuss whether the relative contribution obtained above should be ascribed to the relative location of the observation site in the auroral zone. In the CP-1 mode the EISCAT KST radar always looks at the same corrected magnetic latitude (L-value = 6.2, invariant latitude = 66.2 ), while the auroral zone diurnally (and temporarily) changes its location. The EISCAT radar is thus expected to sample the ionosphere normally in the auroral zone. The EISCAT radar is sometimes in the polar cap around midnight, and at the equatorward region of the auroral zone or in the subauroral zone around midday [Feldstein and Starkov, 1967]. Therefore it is possible that the MLT variations shown in Figures 4 and 6 are partially due to a change in the location of the Tromsø field line relative to the auroral zone in latitude. As we concluded from the data in Figure 9, rather high energy electron precipitation up to 50 kev causes the Hall conductivity enhancements in the late morning. McDiarmid et al. [1975] have showed that a broad maximum of high energy precipitating electrons are centered on the morning side around 65 invariant latitude (see their Figures 8 and 9) which is lower in latitude than the auroral zone at this MLT. Thus the Hall conductivity enhancements in the late morning shown in our Figure 9 appear to come from the EISCAT radar beam moving crossing this lower latitude region, while the variations around the midnight most likely are rather longitudinal inside the auroral zone. [37] The relative contribution of the ionospheric conductivity and the electric field obtained in the present study is related to precipitation from the magnetosphere to the ionosphere. Hardy et al. [1985, 1987] determined statistical distributions of the electron precipitation and the ionospheric conductivities from particle observations. In agreement with our data, their distributions show considerably high Pedersen and Hall conductivities in the dawn and dusk sectors. These conductivity enhancements are produced by electrons precipitating from the Central Plasma Sheet (CPS) [Winningham et al., 1975], implying that the CPS is seen at this MLT. The CPS is considered to correspond to regions of the upward region 2 current in the morning sector [Senior et al., 1982] and the downward region 2 current in the afternoon sector [Sugiura et al., 1984]. On the other hand, the region 1 current is located in the Boundary Plasma Sheet (BPS) and is associated with very low conductivities in the morning sector. We expect that the EISCAT radar is under lower conductivities, downward region 1 in the early morning, and under higher conductivities, upward region 2 in the late morning. This could explain why our data showed lower conductivities in the early morning and increased conductivities in the late morning. [38] We have shown the MLT distribution of the Pedersen current for several Kp value ranges in Figure 3. This MLT and Kp dependence of the Pedersen current seem similar as has been reported by Iijima and Potemra [1976] for FACs. The Pedersen current is the main closure current of FACs in the ionosphere. In particular, the large-scale region 1 and region 2 currents are distributed widely at almost all MLTs and have higher currents in the dawn and dusk sectors. Thus the average Pedersen currents should share similar characteristics as FACs. Possible current closures are a meridional current connecting adjacent region 1 and 2 currents, a crosspolar cap current, and a longitudinal current in the auroral zone. If the cross-polar cap and longitudinal currents are considerably smaller than the meridional current, then the Pedersen current should have nearly the same MLT and Kp dependence as that of the region 1 and 2 currents balancing each other meridionally. [39] Based on magnetic field variations measured on the ground, Ahn et al. [2000] have pointed out that the Alaska meridian chain shows a magnetic maximum disturbance at around 0300 MLT, however, the IMAGE chain in northern Scandinavia does not show this characteristic. The maximum disturbance in the IMAGE data appears at 0100 MLT. During disturbed periods, it appears at 0600 MLT (B.-H. Ahn, private communication, 2000). Also Senior et al. [1990] have pointed out that the average electric field derived with EISCAT radar data in northern Scandinavia differs from the one derived with the Chatanica radar data in Alaska. As suggested by Sojka et al. [1980], the discrepancies between the two electric field patterns may well be due to the offset between the geographic and geomagnetic poles and differences in the solar illumination. Our EISCAT radar data show enhancements on the morning side in not only ionospheric currents (Figure 3) but also in conductivities (Figure 2). We also speculate that the relative strength of the Earth s magnetic field between the northern and southern conjugate points may alter the mirror points in altitude and vary distributions of the particle precipitation. [40] Kamide and Vickrey [1983] assessed the relative contribution of the conductivity and the electric field in the driving of the eastward and westward electrojets. They have actually looked at the Hall current rather than the Pedersen current, since the Hall current generally contributes more to the equivalent current estimated on the ground than the Pedersen current does. Kamide and Vickrey [1983]

14 SIA SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD found that the strength of the eastward electrojet was mainly controlled by the magnitude of the northward electric field in the evening sector. They also found that the westward electrojet current appeared to have two distinct components; in the midnight and early morning sectors (before 0300 MLT) a strong electrojet was due to a weak southward electric field and a high Hall conductivity, whereas its late morning portion (after 0300 MLT) was caused by a strong southward electric field. [41] The present method, sorting all available data according to the current intensity and comparing the conductivity with the electric field strength for a data set where the current has a certain fixed value of the current intensity, enables us to evaluate more quantitatively the relative contribution of the conductivity and the electric field than previous studies. In Figures 4 and 5, for all current levels, we see most of the data concentrated around the mean values and few data points greatly scattering from the mean values. The latter data, called scattering data, are data obtained during intense and short-lived precipitation periods around midnight, and hence they are most likely associated with DP1 type disturbances, namely substorms. On the other hand, the former data, called background data, show smooth enhancements with MLT, and hence they are likely associated with DP2 type enhancements [Kamide and Kokubun, 1996]. The conductivity dominant region proposed by Kamide and Vickrey [1983] may correspond to our conductivity dominant region around midnight. Most of the Hall conductivity data shown by Kamide and Vickrey [1983] exceed 20 S before 0300 MLT in the westward electrojets (see their Figure 3a), while the number of their data is not so large. Therefore we think that Kamide and Vickrey [1983] show mainly what we would call scattering data in Figures 4 and 6. For these data the conductivity contributes relatively more to the current around midnight than in other MLT regions. The conductivity data scattering greatly from the mean value are presumably associated with exceptionally intense precipitation under disturbed geomagnetic activity. Since the number of our scattering data is small, we have not attempted to check the characteristics for these data separately in the same way as we did for the bulk, background data. [42] Comparing different levels of the Pedersen current intensity in Figures 7 and 8, we have shown that not only the Pedersen conductivity but also the electric field plays an equally important role for the background Pedersen current on the night side. In other words, both the Pedersen conductivity and the electric field become higher when the Pedersen current is stronger. This becomes obvious only by using our method of analysis. Therefore background Pedersen currents and associated FACs must be driven by a mechanism that produces both high electric fields and high conductivities for strong currents. The left panels in Figure 4 show qualitatively that many of the data points with stronger Pedersen currents have both higher Pedersen conductivities and higher electric fields. While the anti-correlation between the Pedersen conductivity and the electric field may also apply to the bulk background data, as we have shown, it does not mean that a strong Pedersen current is produced by either a high Pedersen conductivity or a high electric field, but it is both that contribute on the night side. [43] The Pedersen and Hall currents are driven by the electric field E 0 which is the sum of the potential electric field E (derived from the plasma motion in the F region) and the neutral dynamo electric field U B, where U is the neutral wind velocity [Brekke and Rino, 1978; Nozawa and Brekke, 1995]. The dynamo electric field U B is highly height dependent, and the amplitude of the dynamo electric field occasionally becomes comparable to that of the potential electric field [Fujii et al., 1998, 1999]. In this analysis we have not been able to calculate the dynamo electric field for some of the data, and hence we have ignored the effect of this field. This disregard may make the evaluation of the ionospheric current intensity imprecise and hence can be a cause of the fluctuations of the conductivity and the electric field. Further studies considering neutral wind effects are needed for a better understanding. 5. Summary [44] Based on EISCAT radar data, we have determined the relative contribution of the ionospheric conductivity and the electric field to the ionospheric current. We have refined a method in this paper in order to evaluate quantitatively how these two parameters behave in situations of strong or weak current intensity in different MLT regions. We have sorted all available data according to the current intensity and then compared the conductivity with the electric field strength for a data set where the current has a certain fixed value of current intensity. [45] In agreement with previous studies, we see a conductivity dominant region around midnight, where the conductivity is comparatively high for the same level of the current intensity as compared to other MLT intervals. However, we find that this region extends at about MLT for all levels of the current intensity, and two peaks of the conductivity dominance are seen; one around midnight and the other in the late morning. The peak around midnight becomes more apparent with higher geomagnetic activity, while the other peak in the late morning is always evident even for the weak current level. [46] Comparing different levels of the current intensity, we see that a stronger level of the Pedersen currents is caused by an almost equally important contribution of higher Pedersen conductivities and higher electric fields. [47] In this study, we have not directly considered seasonal effects [e.g., Fujii et al., 1981; de la Beaujardiere et al., 1991], geomagnetic activity, nor substorm phases, but rather we used the current intensity to sort out the data. Future study considering substorm effects are necessary to improve our understanding of the relation between FACs, ionospheric conductivities, and electric fields all of which are important aspects of the magnetosphere-ionosphere coupling. [48] Acknowledgments. We are indebted to the director and staff of EISCAT for operating the facility and supplying the data. EISCAT is an international association supported by Finland (SA), France (CNRS), the Federal Republic of Germany (MPG), Japan (NIPR), Norway (NFR), Sweden (NFR), and the United Kingdom (PPARC). We thank S. Oyama, T. Hagfors, and B.-H Ahn for their valuable suggestions. This research was financially supported by the grant-in-aid for scientific research A ( ), B ( ), and C ( ) by the Ministry of Education, Science, Sports and Culture, Japan.

15 SUGINO ET AL.: RELATIVE CONTRIBUTION OF CONDUCTIVITY AND ELECTRIC FIELD SIA References Ahn, B.-H., B. A. Emery, H. W. Kroehl, and Y. Kamide, Climatological characteristics of the auroral ionosphere in terms of electric field and ionospheric conductance, J. Geophys. Res., 104, 10,031 10,040, Barton, C. E., International geomagnetic reference field: The seventh generation, J. Geomagn. Geoelectr., 49, , Brekke, A., and C. Hall, Auroral ionospheric quiet summer time conductances, Ann. Geophys., 6, , Brekke, A., and C. L. Rino, High-resolution altitude profiles of the auroral zone energy dissipation due to ionospheric currents, J. Geophys. Res., 83, , Brekke, A., J. R. Doupnik, and P. M. Banks, Incoherent scatter measurements of E region conductivities and currents in the auroral zone, J. Geophys. Res., 79, , Collis, P. N., EISCAT data base for ionospheric modelling: F-region and topside ionosphere, Adv. Space. Res., 16, (1)37 (1)46, Davies, J. A., and M. Lester, The relationship between electric fields, conductances and currents in the high-latitude ionosphere: A statistical study using EISCAT data, Ann. Geophys., 17, 43 52, de la Beaujardiere, O., D. Alcayde, J. Fontanari, and C. Leger, Seasonal dependence of high-latitude electric fields, J. Geophys. Res., 96, , Feldstein, Y. I., and G. V. Starkov, Dynamics of auroral belt and polar geomagnetic disturbances, Planet. Space. Sci., 15, , Folkestad, K., T. Hagfors, and S. Westerlund, EISCAT: An updated description of technical characteristics and operational capabilities, Radio Sci., 18, , Fujii, R., and T. Iijima, Control of the ionospheric conductivities on largescale Birkeland current intensities under geomagnetic quiet conditions, J. Geophys. Res., 92, , Fujii, R., T. Iijima, T. A. Potemra, and M. Sugiura, Seasonal dependence of large-scale Birkeland currents, Geophys. Res. Lett., 8, , Fujii, R., S. Nozawa, N. Matuura, and A. Brekke, Study on neutral wind contribution to the electrodynamics in the polar ionosphere using EIS- CAT CP-1 data, J. Geophys. Res., 103, 14,731 14,739, Fujii, R., S. Nozawa, S. C. Buchert, and A. Brekke, Statistical characteristics of electromagnetic energy transfer between the magnetosphere, the ionosphere, and the thermosphere, J. Geophys. Res., 104, , Fukushima, N., Equivalence in ground geomagnetic effect of Chapman- Vestine s and Birkeland-Alfvén s electric current-systems for polar magnetic storms, Rep. Ionos. Space Res. Jpn., 23, , Hardy, D. A., M. S. Gussenhoven, and E. Holeman, A statistical model of auroral electron precipitation, J. Geophys. Res., 90, , Hardy, D. A., M. S. Gussenhoven, R. Raistrick, and W. J. McNeil, Statistical and functional representations of the pattern of auroral energy flux, number flux, and conductivity, J. Geophys. Res., 92, 12,275 12,294, Hedin, A. E., Extension of the MSIS thermosphere model into the middle and lower atmosphere, J. Geophys. Res., 96, , Iijima, T., Field-aligned currents in geospace: Substance and significance, in Magnetospheric Current Systems, Geophys. Monogr. Ser., vol. 118, edited by S. Ohtani et al., pp , AGU, Washington D. C., Iijima, T., and T. A. Potemra, The amplitude distribution of field-aligned currents at northern high latitudes observed by Triad, J. Geophys. Res., 81, , Kamide, Y., and S. Kokubun, Two-component auroral electrojet: Importance for substorm studies, J. Geophys. Res., 101, 13,027 13,046, Kamide, Y., and J. F. Vickrey, Relative contribution of ionospheric conductivity and electric field to the auroral electrojets, J. Geophys. Res., 88, , Lysak, R. L., Auroral electrodynamics with current and voltage generators, J. Geophys. Res., 90, , Marklund, G., Auroral arc classification scheme based on the observed arcassociated electric field pattern, Planet. Space. Sci., 32, , McDiarmid, I. B., J. R. Burrows, and E. E. Budzinski, Average characteristics of magnetospheric electrons (150 ev to 200 kev) at 1400 km, J. Geophys. Res., 80, 73 79, Nozawa, S., and A. Brekke, Studies of the E region neutral wind in the disturbed auroral ionosphere, J. Geophys. Res., 100, 14,717 14,734, Rees, M. H., Auroral ionization and excitation by incident energetic electrons, Planet. Space. Sci., 11, , Rishbeth, H., and P. J. S. Williams, The EISCAT ionospheric radar: The system and its early results, Q. J. R. Astron. Soc., 26, , Robinson, R. M., Kp dependence of auroral zone field-aligned current intensity, J. Geophys. Res., 89, , Schlegel, K., Auroral zone E-region conductivities during solar minimum derived from EISCAT data, Ann. Geophys., 6, , Senior, C., R. M. Robinson, and T. A. Potemra, Relationship between fieldaligned currents, diffuse auroral precipitation and the westward electrojet in the early morning sector, J. Geophys. Res., 87, 10,469 10,477, Senior, C., D. Fontaine, G. Caudal, D. Alcayde, and J. Fontanari, Convection electric fields and electrostatic potential over 61 < <72 invariant latitude observed with the European incoherent scatter facility, 2, Statistical results, Ann. Geophys., 8, , Sojka, J. J., J. C. Foster, W. J. Raitt, R. W. Schunk, and J. R. Doupnik, High-latitude convection: Comparison of a simple model with incoherent scatter observations, J. Geophys. Res., 85, , Sugiura, M., N. C. Maynard, W. H. Farthing, J. P. Heppner, B. G. Ledley, and L. J. Cahill Jr., Initial results on the correlation between the magnetic and electric fields observed from the DE-2 satellite in the field-aligned current regions, Geophys. Res. Lett., 9, , Sugiura, M., T. Iyemori, R. A. Hoffman, N. C. Maynard, J. L. Burch, and J. D. Winningham, Relationships between field-aligned currents, electric fields, and particle precipitation as observed by Dynamics Explorer-2, in Magnetospheric Current Systems, Geophys. Monogr. Ser., vol. 28, edited by T. A. Potemra, pp , AGU, Washington, D. C., Vickrey, J. F., R. C. Livingston, N. B. Walker, T. A. Potemra, R. A. Heelis, M. C. Kelley, and F. J. Rich, On the current-voltage relationship of the magnetospheric generator at intermediate spatial scales, Geophys. Res. Lett., 13, , Whalen, B. A., H. J. Verschell, and I. B. McDiarmid, Correlations of ionospheric electric fields and energetic particle precipitation, J. Geophys. Res., 80, , Winningham, J. D., F. Yasuhara, S.-I. Akasofu, and W. J. Heikkila, The latitudinal morphology of 10-eV to 10-keV electron fluxes during magnetically quiet and disturbed times in the MLT sector, J. Geophys. Res., 80, , A. Brekke, Auroral Observatory, University of Tromsø, Tromsø, Norway. S. C. Buchert and H. J. Opgenoorth, Uppsala Division, Swedish Institute of Space Physics, Uppsala, Sweden. R. Fujii, S. Nozawa, and M. Sugino, Solar-Terrestrial Environment Laboratory, Nagoya University, Nagoya, Japan. (msugino@stelab.nagoya-u. ac.jp)