Comparison of auroral electrojet indices in the Northern and Southern Hemispheres
Abstract
[1] The auroral electrojet (AE) index is traditionally calculated from a set of about 10 to 13 ground magnetometer stations located around the typical northern auroral oval location. Similar coverage in the Southern Hemisphere does not exist, so the AE calculation has only been performed using Northern Hemisphere data. In the present paper, we use seven southern auroral region ground magnetometers as well as their conjugate Northern Hemisphere data to calculate conjugate AE indices during the Northern Hemisphere winter using the standard method. The correlation coefficient between the northern and southern AE indices for many of the intervals is above 0.7, but in one interval, it is close to 0. We compare our conjugate AE indices with the standard AE index and find a number of asymmetries because of station coverage gaps both in our conjugate indices and in interplanetary magnetic field (IMF) conditions. When the southern AE index is compared with the standard AE index, we find that for most intervals the correlation is less than 0.7, and in one interval, the correlation is about 0. The correlation between the conjugate northern AE index and the standard AE index is somewhat better, suggesting true interhemispheric asymmetries. The mean difference between the southern and northern AE indices is largest during southward IMF and for large values of IMF ∣By∣ (>5 nT). This is most likely due to the increased activity levels during southward IMF and the greater twisting of the magnetic field lines during strong IMF By. IMF Bx seems to have no effect on the interhemispheric differences of the AE index. The mean differences between the southern and conjugate northern H component are of the order of ∼35 nT, with the largest differences occurring in the midnight magnetic local time (MLT) sector. We suggest that these differences may be related to seasonal effects, ionospheric effects, and MLT effects.
1. Introduction
[2] The auroral electrojet (AE) index [Davis and Sugiura, 1966] is used as an indicator of the level of global geomagnetic activity and it is well correlated with the strength of the auroral electrojet currents in the Northern Hemisphere ionosphere. The AE index is the difference between the auroral electrojet upper (AU) and lower (AL) indices, which are derived from the H component (pointing toward magnetic north) from relatively well spaced ground magnetometer stations in the Northern Hemisphere auroral zone. The AU index provides information on the strength of the eastward auroral electrojet and the AL index indicates the intensity of the westward auroral electrojet. The AE index, as well as the AU and AL indices, has been traditionally produced by the World Data Center (WDC) for Geomagnetism at Kyoto University, Kyoto, Japan. Unfortunately, lack of data from some of the ground magnetometers has left the community without formal AE, AU, and AL indices for the last several years. Temporary or quick-look AE indices are used instead.
[3] Because of the limited availability of land in the Southern Hemisphere an equivalent southern AE index is not regularly produced. However, previous studies have demonstrated a good correlation for conjugate auroral observations [Belon et al., 1969; Stenbaek-Nielsen et al., 1972; Craven et al., 1991; Weygand, 1998] and conjugate ground magnetometer data [e.g., Maclennan et al., 1991; Saroso et al., 1992; Ballatore et al., 1998; Ballatore and Maclennan, 1999]. Differences between the northern and southern AE indices are most likely the results of hemispheric or seasonal asymmetries in the ionosphere or could be due to interplanetary magnetic field (IMF) conditions. The study of Maclennan et al. [1991] employed 22 ground magnetometers in Antarctica to calculate a southern AE index that was then compared to the northern WDC AE index for 7 days in 1982. The selected Southern Hemisphere stations covered a range from 50° to 90° magnetic latitude. However, many of the stations used in the study of Maclennan et al. [1991] were not within the auroral region and there were two significant gaps in the station longitudinal coverage, from 140° to 235° and 250° and 330° magnetic longitude (i.e., gaps of about 95° and 80° in magnetic longitude) in West Antarctica. Maclennan et al. [1991] found that the magnitude of the northern standard AE index generally exceeds that of the southern AE (SAE) index. They also found that the difference between AE and SAE was a strong function of universal time (UT), their correlation being the highest between 00 to 11 UT, and during primarily southward IMF (Bz < 0.5 nT). Similarly, Hajkowicz [1998] determined using the standard AE index that there was a UT effect in the onset of auroral disturbances, which tend to peak between 09 and 18 UT and reach a minimum between 03 and 06 UT. A possible explanation for the minimum in the auroral disturbances between 03 and 06 UT has been examined by Ahn et al. [2000] who found that that the AL index often underestimated the disturbance conditions between 02 and 08 UT. Hajkowicz also found that the UT effect in the Northern Hemisphere was strongest in the winter and weakest in the summer and acknowledges that auroral disturbances occur simultaneously in both hemispheres, but the UT effect in the summer hemisphere may be obscured by the higher ionospheric conductivity in the summer hemisphere. This would result in additional interhemispheric asymmetries between the indices derived from the two hemispheres.
[4] The work of Saroso et al. [1992] is similar to that of Maclennan et al. [1991], but the Saroso et al. [1992]. study compared the standard northern AE index with a southern polar cap AE index calculated from four evenly spaced Antarctic ground magnetometers at latitude greater than 77° geomagnetic latitude. Saroso et al. examined data from 1966, during solar minimum, and from 1980, during solar maximum. One of the aims of their study was to derive indices that monitor the ionospheric currents over the polar cap during northward IMF when the auroral oval contracts. However, at times when the auroral oval significantly contracts and the southern stations would more accurately represent the auroral oval, the standard northern AE stations were likely too far equatorward of the auroral oval and may no longer accurately reflect the activity in the ionospheric electrojet. Hence a comparison between the northern standard AE and a southern polar cap AE index was in the best of cases inconclusive. Nevertheless, Saroso et al. [1992] determined that the correlation between the southern polar AE and the standard AE index is highest during the winter season.
[5] The phenomenon of substorms is one of the most dynamic in the field of Space Physics and has been extensively studied and modeled. All of today's models of substorms are based on observations from the Northern Hemisphere and assume conjugacy for the Southern Hemisphere. The study of the onset and development of substorms at conjugate auroral latitudes is quite limited and has been conducted mostly by conjugate ground auroral imagers and magnetometers [e.g., Belon et al., 1969; Stenbaek-Nielsen et al., 1972; Hajkowicz, 2006]. Some recent works have used fortuitous conjugate auroral observations from spacecraft [e.g., Østgaard et al., 2004, 2007; Sato et al., 1998; Frank and Sigwarth, 2003] and reported on significant asymmetries between the two hemispheres during the substorm onset and its development. Frank and Sigwarth [2003] presented the first high-resolution observations of the development of a substorm observed by the Polar VIS camera at both hemispheres and found significant differences in the auroral images from the different hemispheres. Frank and Sigwarth [2003] found that there was a 1–2 minute delay in the occurrence of the onset between the two hemispheres. Such time delays have also been reported by Hajkowicz [2006] and Sato et al. [1998]. Most important, Frank and Sigwarth showed that mapping of the onset location to the tail resulted in very different locations from the two hemispheres. The study of Kivelson et al. [1996] suggests that interhemispheric conjugacy will exhibit predictable variations dependent mostly on IMF By. Østgaard et al. [2004, 2007] also found that substorm onsets exhibit a systematic displacement from one hemisphere to the other and the amount of this displacement is controlled primarily by the IMF cone angle.
[6] It is clear from the above works that our lack of conjugate observations of substorms may have led to insufficient understanding of the phenomenon. The AE index is routinely used for identifying the occurrence and/or onset of substorms. Considering the recent evidence for significant differences in the location or timing of the substorm onset in the two hemispheres, it seems likely that the AE index calculated from the two hemispheres will also exhibit similar differences.
[7] In the present paper we use the most complete, to date, local time coverage of stations from the Southern Hemisphere to calculate a SAE index and compare it with both the standard AE index and a NAE index calculated from stations conjugate to the stations used for SAE. We test whether NAE is applicable for use in the Southern Hemisphere and visa versa. We also see if the NAE index can be used instead of the standard AE or not, thus further validating our use of a SAE index.
[8] For this study we have calculated a southern AE index and a conjugate AE index for 7 days worth of data recorded during the first part of December 2005. We will examine the differences and correlations between the southern and conjugate northern AE, AU, and AL indices as a function of magnetic local time (MLT) and various IMF condition.
2. Data and Methodology
[9] For this study we use fluxgate ground magnetometer data from seven stations located in the southern auroral region, most of them in Antarctica, during the period of 3 to 10 December, 2005. Note that these seven stations are all the available Southern Hemisphere stations within auroral latitudes. For 10 December 2005, only 12 hours of data are simultaneously available from all the stations, while on 3 December 2005 only 18 hours of data are simultaneously available. Table 1 gives, in columns 2–8, the geographic (GEO) and corrected geomagnetic (CGM) coordinates of these seven stations, as well as the geographic coordinates of the magnetically conjugate point of each station (magnetic longitude is assumed to be the same and magnetic latitude opposite of that of the original station). The magnetic conjugate location is determined using the online 2005 International Geomagnetic Reference Field (IGRF 2005) at (http://modelweb.gsfc.nasa.gov/models/cgm/cgm.html). The local declination and the Universal Time (UT) when the station crosses magnetic local midnight are also given in Table 1. It is apparent, from the last column of Table 1, that there are large gaps in the auroral coverage by the available Southern Hemisphere stations and, by extension, in the set of conjugate northern stations. For example, there are gaps for over 6.6 hrs in MLT from BET to PBQ, and over 10 hrs in MLT from BJN to BET in the northern auroral coverage.
Southern Stations | Geographic Latitude | Geographic Longitude | Corrected Geographic Latitude | Corrected Geographic Longitude | D, deg | Conjugate Geographic Latitude | Conjugate Geographic Longitude | UT of 0000 MLT |
---|---|---|---|---|---|---|---|---|
MCQ | −54.5 | 159.0 | −64.4 | 248.3 | 32.1 | 67.1 | 193.2 | 1154 |
MAW | −67.6 | 62.9 | −70.3 | 90.3 | −66.9 | 71.8 | 0.07 | 2241 |
SYO | −69.0 | 39.6 | −66.2 | 72.0 | −48.6 | 66.0 | 342.8 | 2355 |
SNA | −70.5 | −2.5 | −60.7 | 44.9 | −17.8 | 57.0 | 318.1 | 0142 |
NVL | −70.77 | 11.8 | −62.8 | 53.2 | −28.0 | 60.3 | 325.4 | 0110 |
HBA | −75.8 | −26.68 | −62.0 | 29.3 | −1.41 | 55.3 | 304.21 | 0244 |
WSD | −79.5 | −112.2 | −67.0 | 355.7 | 63.4 | 56.9 | 280.0 | 0508 |
Northern Stations | ||||||||
STJ | 47.6 | 307.3 | 53.2 | 31.1 | −20.5 | −64.3 | 332.5 | 0302 |
PBQ | 55.3 | 282.2 | 65.3 | 358.8 | −17.9 | −78.8 | 259.5 | 0505 |
NAQ | 61.2 | 314.6 | 65.9 | 42.8 | −26.8 | −76.9 | 4.4 | 0211 |
HLL | 63.8 | 339.4 | 64.3 | 67.2 | −16.4 | −68.3 | 30.2 | 0010 |
AMK | 65.6 | 322.4 | 69.0 | 53.3 | −26.8 | −76.9 | 28.6 | 0121 |
BET | 66.9 | 208.5 | 66.5 | 255.3 | 22.0 | −58.0 | 165.6 | 1129 |
SCO | 70.5 | 338.0 | 71.5 | 71.8 | −20.8 | −73.5 | 53.4 | 2350 |
SOR | 70.5 | 22.2 | 67.4 | 106.0 | 7.5 | −61.3 | 67.7 | 2111 |
BJN | 74.5 | 19.2 | 71.5 | 107.8 | 6.2 | −64.6 | 75.3 | 2105 |
[10] Figure 1 shows the distribution of the Southern Hemisphere stations we use in our study. In the map, the black dotted lines are lines of geographic latitude and longitude, while the solid blue lines are lines of constant geomagnetic latitude at −40°, −50°, −60°, and −70°. The green and magenta solid circles are the locations of the 7 available southern auroral stations. In magenta is the most recent station of the South American Meridional B-field Array (SAMBA) station, WSD, installed in the West Antarctic Ice Sheet Divide. Stations MAW, SYO, SNA, NVL, HBA, and WSD provide a closely spaced coverage of the auroral zone, while MCQ station is farther away. The installation of WSD extends the prior auroral coverage from 4 hrs in MLT to 6.5 hrs in MLT.
[11] The standard AE index is calculated from 10 to 13 standard magnetic observatories in the Northern Hemisphere that cover the full MLT range, although some gaps do exist. For the 7 days used in this study, data from the standard magnetic observatories were available from most or all of the magnetometers. Table 2 displays the CGM location of those stations. Because the coverage of the available Southern Hemisphere auroral stations (see Figure 1) is not the same as the ones from the northern auroral stations used in the standard AE calculation, we used a set of nine Northern Hemisphere auroral stations that are near-conjugate to the southern stations for direct comparisons. The names, GEO, and CGM coordinates of these nine northern stations, as well as their declination and UT of 00 MLT for each station are given in the bottom half of Table 1. There are more northern magnetometer stations than southern stations because exact conjugate stations are not always available. To overcome this difficulty, data from northern magnetometers that “surround” the conjugate southern station location are averaged together.
Standard AE Station | Corrected Geomagnetic Latitude | Corrected Geomagnetic Longitude |
---|---|---|
ABK | 66.0 | 115.1 |
DIK | 63.0 | 161.6 |
CCS | 66.3 | 176.5 |
TIK | 60.4 | 191.4 |
PBK | 64.5 | 222.7 |
CWE | 61.8 | 237.1 |
BRW | 68.5 | 241.2 |
CMO | 64.6 | 256.5 |
YKC | 69.0 | 292.8 |
FCC | 68.7 | 322.8 |
PBQ | 66.6 | 347.4 |
NAQ | 71.2 | 36.8 |
LRV | 70.2 | 71.0 |
[12] Figure 2 displays the location and distribution of the Northern Hemisphere stations used for the conjugate AE calculation (green solid circles), the Southern Hemisphere stations (magenta solid circles), and the standard AE stations (black triangles). As in Figure 1, the black dotted lines are lines of geographic latitude and longitude and the solid blue lines are lines of constant geomagnetic latitude. It is apparent from Figure 2 why in some cases data from more than one northern station are used to create the conjugate signature for one of the southern stations. Table 3 indicates which northern station or stations are used to produce the conjugate signatures for the seven southern station magnetometer data. For example, the conjugate signature for HBA (magenta circle immediately to the right of PBQ) is produced by averaging the data from NAQ, STJ, and PBQ.
Southern Station | Conjugate Northern Stations |
---|---|
MCQ | BET |
MAW | SOR, SCO, BJN |
SYO | HLL |
SNA | NAQ, STJ |
NVL | NAQ, AMK |
HBA | NAQ, STJ, PBQ |
WSD | PBQ |
[13] The latitudinal distribution of the ground stations in Figure 2 raises another important issue. If the ground station and the station (or stations) that are used as its conjugate point are at different latitudes, then the stations may measure different signals relative to the position of the electrojet. For a ground station located directly below the peak of the electrojet most of the ground perturbation is in the H component with minimal or no Z perturbation (azimuthally stretched electrojets do not create D component perturbations). For a station located either poleward or equatorward of the electrojet peak but still underneath the bulk of it, there will be both H and Z component perturbations. The latitudinal distance of the ground station from the peak of the electrojet with respect to the latitudinal width of the electrojet will determine the ratio between the H and Z perturbations.
[14] For example, if the southern station and its conjugate northern both lie directly below their respective eastward electrojet, then they will both record an increase in the H component of the magnetic field and no change in the Z component. Note that in the H, D, Z coordinate system H points toward the magnetic north and Z toward the nadir (i.e., downward toward the center of the Earth) in both hemispheres. However, if the southern station lies directly below the electrojet peak but its conjugate northern station is poleward of the eastward electrojet, then the northern station will observe a decrease in the Z component and little or no change in the H component due to the electrojet. As a result the measurements of the two stations may appear to be uncorrelated because of the latitudinal difference in their location. Considering that some of our conjugate pairs are more than 2° of latitude apart, we must consider such a situation. In such a case we can use the magnitude and polarity of the Z perturbation in the conjugate stations to help determine their possible location with respect to the electrojet peak and whether their latitudinal separation creates some of the differences observed. The Z component will show no change in both the northern and Southern Hemisphere if both stations are directly below the peak of the electrojet, but the Z component will show the opposite changes in the two hemispheres if the two conjugate station are both above or below the electrojet. If the conjugate stations do not display zero or a mirrored Z perturbation then differences observed between the two conjugate stations could be in part due to their latitudinal difference.
[15] Data from all 7 southern and 9 northern ground magnetometer stations are converted, if necessary, into the HDZ coordinate system. We interpolate all the magnetometer data to 1 minute resolution and remove quiet levels, which are calculated from the five quietest days of December 2005. These days are the 6th, 7th, 8th, 15th, and 23rd of December 2005 and are identified based on the sum of the daily Kp index values (according to http://swdcwww.kugi.kyoto-u.ac.jp/qddays/index.html). Using the final processed H component perturbation we calculate the southern and the conjugate northern AU, AL, and AE indices based on the standard method outlined at the World Data Center (WDC) (http://swdcwww.kugi.kyoto-u.ac.jp/aedir/index.html). We also obtain the WDC quick-look AE index for comparison because final standard AE values are not available. In all the events described below we compare these three derived or downloaded indices.
3. Observations
[16] In the next subsections we will examine the magnetometer data from individual southern magnetometers stations, the conjugate northern magnetometer data, and the derived AE indices.
3.1. 10 December 2005
[17] Figure 3 displays the ground magnetometer data from the northern (gray curves) and southern (black curves) hemispheres for 10 December 2005 when the agreement between the northern and southern indices is of average quality. The code for the Southern Hemisphere stations is noted in the beginning of each trace pair. Next to the southern station codes are the codes, in gray of the northern station(s) conjugate to the southern one. Only 12 hours of data (0–12 UT) are analyzed, as this is the available time period from all stations involved. The open circles in the individual panels indicate local midnight for that station, while the solid circles indicate local noon. This is a relatively active day with ΣKp = 21+, which is also evident in the magnetograms of Figure 3. The solar wind speed, as measured by the Wind spacecraft at 235 RE in front of the Earth and 85 RE to the duskside, is relatively steady during this interval. The Vx component is about 300 km s−1 and increases to about 350 km s−1 after 0600 UT. The density peaks at ∼17 cm−3 at 0430 and 0630 UT. IMF Bx is approximately 0 nT at the start of the interval and decreases to −3 nT after 0600 UT, By fluctuates, and Bz switched from negative to positive at ∼0600 UT. Throughout the studied interval the magnitude of the IMF remains ∼13 nT. A substorm occurred just after 0600 UT, following the IMF northward turning, clearly seen in the WSD station and its northern conjugate PBQ located at 0100 MLT at the time. No station data are available from a premidnight MLT sector in the Southern Hemisphere (all of our stations are arranged from post midnight to post dawn with only MCQ at dusk), which is why this substorm is only observed clearly at WSD. Much weaker signatures are seen in HBA and NVL, which are located at approximately 03 and 05 MLT, respectively.
[18] The AU, AL, and AE indices for the southern (SAE/AU/AL) and conjugate Northern Hemisphere data (NAE/AU/AL), as well as the WDC indices for this day are shown in Figure 4. The Northern Hemisphere indices are given in the top panels as solid black lines, and the Southern Hemisphere indices are shown in the bottom panels, also in solid black lines. The gray lines in the top panels are the WDC quick-look indices that can stand in place of the standard AE, AU, and AL indices.
[19] There is good agreement between our northern indices and the WDC quick-look index and the substorm just after 0600 UT is clearly visible in both NAE and AE. The correlation coefficient between AE and NAE is 0.86. There is less agreement between the NAE and SAE with a correlation coefficient of 0.69. The three main peaks (including the 0600 UT substorm) observed in NAE and AE indices are also observed in SAE index although with much smaller magnitude. The main difference is the peak between 0430 and 0600 UT that is seen in the SAE index but not in either of the northern indices. This particular peak is the result of some strong perturbations observed at MAW located at approximately 0600 MLT at the time (see Figure 3). No similar perturbations are visible in any of the nearby northern stations, resulting in this significant asymmetry.
[20] For this particular event the evidence indicates the existence of significant asymmetries between the northern and Southern Hemisphere auroral activity. Such asymmetries, apparently, are not the result of dissimilar station coverage or MLT gaps in the north and south hemisphere, given the fact that the correlation between NAE and AE is so high. The asymmetry observed could be due to latitudinal differences between our conjugate points, the most obvious pair being MAW and the 3 northern stations, SOR, SCO, and BJN, averaged to produce the conjugate signal. First, we inspected the H component at the individual northern stations and in none of them significant variations exist that can match that seen at MAW. Inspection of the Z component in the conjugate points indicates opposite sign perturbations of similar magnitude around 06 UT, implying that the conjugate points are both on the same (equatorward) side of the peak westward electrojet. Thus we conclude that there are no significant latitudinal asymmetries between MAW and its conjugate point.
[21] We inspected in a similar fashion the Z perturbations in the remaining 6 conjugate pairs and only in NVL and its conjugate point we found same sign variations, implying that NVL is just equatorward of the westward electrojet while its northern conjugate point is just poleward of it. Nevertheless, the fact there are no significant asymmetries in the 04–06 UT period in the H component between NVL and its conjugate point (see Figure 3), we conclude that in this particular case latitudinal asymmetries were not a significant part of the asymmetries observed between NAE and SAE.
[22] Finally, the significantly different ionospheric conductivity in the two ionospheres at that time of the year could also certainly be a contributor, but this cannot be determined at this time.
3.2. 3 and 4 December 2005
[23] 3 December 2005 is an example of another active day (ΣKp = 23) when the agreement between the northern and southern indices is good (albeit not great). For this day only the 0600 to 2400 UT interval, when data are available from all stations, is analyzed. The solar wind speed decreases from ∼750 km s−1 to ∼675 km s−1, as measured by the Wind spacecraft at 230 RE in front of the Earth and 95 RE to the duskside. Vx is ∼720 km s−1 and the density is relatively constant at ∼1.5 cm−3. IMF By is mostly negative and frequently fluctuates around zero, and Bz fluctuates around zero throughout the studied interval. The IMF magnitude remains at ∼4.5 nT.
[24] Figure 5 shows the ground magnetometer data for 3 December 2005 from 0600 to 2400 UT in the same format as Figure 3. Similar to the 10 December interval, there are some significant differences between the southern and conjugate northern curves, particularly in the magnitude of the individual perturbations. Figure 6 shows the calculated AE indices from the southern and conjugate northern stations, as well as the WDC quick-look indices in the same format as Figure 4. The horizontal black bar from 1500 to 1940 UT in each panel indicates the period when no ground magnetometer stations in our SAE array are within the 21 to 3 MLT sector, thus most likely to miss any relevant substorm activity in SAE during that period. The differences between the southern (black lines in bottom panels) and conjugate northern (black solid lines in upper panels) indices appear to be more significant than in the 10 December 2005 interval. Nevertheless, the correlation coefficient between the SAE and NAE indices is 0.72, better than that in the 10 December 2005 interval. Furthermore, our NAE/NAL/NAU indices appear to consistently underestimate the WDC indices, although similar perturbations are present, for the most part, in both indices. The associated correlation coefficient between NAE and AE is 0.65, which is due to the significantly different amplitude of the perturbations between the two indices. Not surprisingly, the greatest amplitude differences between NAE and AE appear in the time period when none of the NAE stations were in the midnight region (indicated by the bar).
[25] We would like to discuss in more detail the onset at ∼1900 UT, which is seen in the WDC AE index, and in the SAE index, but not as strong in the NAE index. Examining the magnetograms of Figure 5, we see that it is the southern station of MAW that records the substorm activity at ∼19 UT. MAW is at that time located at 2020 MLT. No similar activity is recorded in the average from the stations SCO, SOR, and BJN (see map of Figure 2), encircling the location conjugate to MAW (indicated in the map of Figure 2 as the rightmost in longitude magenta circle). However, when we examine the individual station magnetograms from the north we find that a significant amount of activity is observed at BJN and some at SOR, but not as much at SCO. As a result the activity observed at BJN is averaged out. An examination of the Z component (not shown here) also supports this observation and further shows that BJN and SOR see the opposite signal in the Z component, while SCO measures very little. Furthermore, when we examine the individual stations for the WDC AE we find that the substorm is clearest at ABK, which is at about the same longitude as BJN (in Figure 2 it is the black triangle directly below BJN and SOR in latitude). At the time of the substorm activity in question, 19 UT, stations BJN, SOR and ABK were located at ∼22 MLT, while MAW is located at ∼2020 MLT in the Southern Hemisphere, and SCO at ∼1910 MLT. Considering that the strongest perturbations are at MAW and ABK, it appears that the substorm activity is localized and at a lower latitude in the north than in the south. In fact examination of the IMAGE FUV images over the Southern Hemisphere at that time (observations not shown here) indicate the onset of a substorm equatorward and just at a later MLT of the MAW location, which explains the smaller perturbations at BJN and stronger ones at ABK, which is located at the same longitude but lower latitude. Such asymmetries have been reported, especially during times of significant IMF By, as is the case for this event [Østgaard et al., 2004, 2007].
[26] A more interesting substorm occurs at about 1040 UT. This substorm is clearly observed in both the NAE and WDC AE, but not as well in the southern SAE. In this instance BET and BRW (see Figure 2), which are both at about the same magnetic longitude, record the substorm, but MCQ (magenta circle just west of BET) at about 7 degrees to the west does not measure as large a change in the magnetic field. When we examine the Z component of the magnetometers only MCQ shows any signal and this signal is the opposite of the BET Z component, which suggests both stations are at the same position relative to the electrojet. Again, we believe that this is the result of asymmetric substorm activations between the two hemispheres under strong IMF By, as reported by Østgaard et al. [2004, 2007].
[27] Figure 7 shows the AE, AU, and AL indices derived for 4 December 2005 from 1200 to 2400 UT. For this day, which is only moderately active with ΣKp = 12+, there is good agreement between the ground magnetometer data (not shown) and between the southern and conjugate northern AE indices. The correlation coefficient between NAE and SAE is 0.73. There is also good agreement between the NAE and AE indices, except around 1630 UT when the standard AE records a substorm at CCS (see Figure 2) while both NAE and SAE measure very little activity. At 1630 UT, local midnight is in the gap between MAW and MCQ in the Southern Hemisphere (see map of Figure 1); therefore it is not surprising that neither our SAE nor NAE record this particular substorm activity while several Siberia stations seen in the map of Figure 2 are present within the midnight sector. Another interesting observation is that the AU index has much higher amplitude perturbations than those in the NAU or even SAU index. The AU index effectively measures the strength of the afternoon/dusk eastward auroral electrojet. The greatest amplitude differences between NAU and AU occur at 20–24 UT. During that UT period most of the afternoon and dusk MLT sector is covered by the Antarctica magnetometers. Thus lack of coverage cannot account for the differences. North-South asymmetries can probably account for some of the differences, but the differences are also between NAU and AU, which include stations with similar MLT coverage in the same hemisphere. A close examination of the individual stations indicates that again the differences between NAU and AU is the results of averaging the BJN, SCO, and SOR. Both BJN and SOR record fluctuations of approximately similar magnitude as the WDC AE station ABK, but SCO measures only a moderate amount of activity. Excluding SCO data during this period results in similar NAU and AU magnitudes. During this interval SCO and SOR display the opposite signals in the Z component from about 18 to 20 UT. After about 20 UT the two stations look similar to MAW for about 2 hours, and then reverse again. BJN shows about the same shape as MAW from about 1800 to 1930 UT, and then reverses sign for the remainder of the day. This observation indicates that the electrojet is moving up and down and averaging of the northern stations and their latitudinal spread result in the lack of observed activity from 20 UT to 24 UT. In other words, latitudinal differences between the southern and conjugate northern stations and their relative location with respect to the electrojet do contribute, at least, to the amplitude asymmetries between NAE and SAE.
[28] For a day when the correlation between SAE, NAE, and AE is otherwise very good, we see the clear effect of the lack of coverage over significant MLT region in the Southern Hemisphere. This is a fundamental and ingrained in the geography of the southern high latitudes and a problem that cannot be remedied, at least not at this point in time.
3.3. 8 December 2005
[29] The three prior examples shown are mainly moderate to active period intervals. Figure 8 displays a quiet interval (ΣKp = 2) in the Southern Hemisphere on 8 December 2005 from 0000 to 2400 UT. Again, open circles in each panel indicate magnetic local midnight and solid circles indicate magnetic local noon. The solar wind speed as measured by the Wind spacecraft at 230 RE in front of the Earth and 90 RE to the duskside decreases from ∼320 km s−1 to ∼300 km s−1. The density is relatively constant at ∼1.6 cm−3 until about 1500 UT when it sharply increases to 3.7 cm−3. IMF By is initially −2 nT, then sharply turned to positive at 0430 UT and then sharply back again to −2 nT at 1500 UT. IMF Bz starts at about 0 nT, then increases to ∼2.2 nT at 0430 UT and drops back to ∼−2 nT at 1630 UT. Throughout the interval the magnitude of the IMF is ∼2.5 nT. In general, the southern ground magnetometer data are similar to their conjugate northern ground magnetometer data. The most significant difference appears in the MAW station and its northern conjugate data, 6th panel from the top, at ∼2030 UT, where the magnitude of the H component at MAW reaches ∼150 nT while average of the northern conjugate stations never exceeds a magnitude of 50 nT.
[30] Figure 9 shows the NAE, SAE, and AE indices, as well as the respective AU and AL indices. These are plotted in a more appropriate scale (half the scale used in the prior events) and significant differences between the SAE and our NAE indices are now clear. There is also significant difference between NAE and the AE indices from about 1800 to 2400 UT. The NAE activity is lower in magnitude than the WDC AE because of the averaging of data from BJN, SOR, and SCO. The bulk of the activity occurs at BJN, but little activity is observed at SCO and SOR. For the measurable perturbations the H component observations at SOR are in the opposite direction of BJN, which argues for small localized currents creating the observed signatures. An examination of the Z components (not shown here) does not add any additional insights, only indicating that the conjugate points are on the same side of the electrojet.
[31] The greatest difference between the three AE indices occurs at ∼2000 UT when the SAE displays an onset in the AE activity. This substorm in the SAE data is recorded by MAW and no exact conjugate northern station is available. The closest available stations measure some activity, but no substorm. The calculated correlation coefficient between SAE and NAE is 0.52, while the correlation coefficient for NAE and AE is 0.69 and the correlation coefficient between SAE and AE is 0.60.
[32] Interestingly, it appears that our SAE index agrees better with the standard AE index than the NAE index, especially in the late part of the day. Once again, even for a very quiet day, we see evidence for significant asymmetries between the southern and northern ovals, typically resulting from asymmetric substorm onsets under significant IMF By.
4. Analysis and Discussion
[33] In section 3 we gave several examples of the differences observed between the southern and northern conjugate AE indices. Figure 10 better demonstrates this with a summary of the differences between NAE and SAE for all 7 days of December 2005 that we are considering in this study. This figure also demonstrates that there is no systematic difference between the southern and northern AE related to UT as one would expect based on the work of Maclennan et al. [1991] and Hajkowicz [1998]. The reason our study using the NAE and SAE indices does not reproduce the results of the Maclennan et al. and Hajkowicz studies is not clear, but it may be related to the difference in the distribution of the ground magnetometers used in our study.
[34] To better quantify the differences between SAE and NAE we calculated from the available data the mean, standard deviation, and the correlation coefficients of the difference for each one of the intervals. The results are shown in Table 4, which also includes the interval of data available and the ΣKp for each day. Figure 10 and Table 4 indicate that the most active days we examined in section 3, which are 3 and 10 December, have the largest mean and standard deviation, but also relatively high correlation coefficients. 8 December, which is one of the 5 quietest days of the month, has the smallest mean and standard deviation, but also nearly the smallest correlation coefficient. In fact, from the small statistical sample in Table 4 it appears that days with moderate and high geomagnetic activity have significantly higher correlation coefficients than very quiet days. The correlation may be better during moderate and active periods because the electrojets would be wider and the probability of an AE station being located under an electrojet would be higher. However, we still question whether a simple correlation coefficient is the most accurate way of depicting the NAE to SAE agreement or disagreement. Nevertheless, Figure 10 and Table 4 do indicate that there are generally significant differences between the southern and the conjugate northern AE indices and even though the two indices may be better correlated in shape and the occurrence of individual perturbations during more active days, the differences in amplitude are largest during those active intervals. It is unclear from such a small statistical sample whether this is a general result.
Interval | Mean ΔAE (nT) | σ (nT) | Correlation Coefficient NAE SAE | Correlation Coefficient SAE AE | Correlation Coefficient NAE AE | ΣKp |
---|---|---|---|---|---|---|
0600–2400 UT 3 Dec | 52.8 | 71.0 | 0.72 | 0.64 | 0.65 | 23 |
0000–2400 UT 4 Dec | 34.5 | 49.0 | 0.73 | 0.57 | 0.67 | 12+ |
0000–2400 UT 5 Dec | 35.9 | 21.2 | −0.02 | 0.01 | 0.35 | 5 |
0000–2400 UT 6 Dec | 17.6 | 26.9 | 0.55 | 0.63 | 0.49 | 4 |
0000–2400 UT 8 Dec | 11.4 | 18.5 | 0.52 | 0.60 | 0.69 | 2 |
0000–2400 UT 9 Dec | 37.5 | 51.8 | 0.81 | 0.90 | 0.79 | 10− |
0000–1200 UT 10 Dec | 82.2 | 116.8 | 0.69 | 0.71 | 0.94 | 21+ |
- a The fourth column gives the correlation coefficient for the SAE and NAE index, the fifth column provides the correlation coefficient for the SAE and standard AE indices, and the sixth column shows the correlation coefficient between the NAE and AE indices. The last column displays the ΣKp values for the day.
[35] The fact that the differences between SAE and NAE are larger during active days suggests that local conjugate magnetometers record larger perturbations with significantly greater amplitude differences than more moderate and smaller perturbations. Since typically larger perturbations are observed around the magnetic local midnight region, one might expect conjugate differences to be greater near 00 MLT. Figure 11 explores this question for 10 December 2005, a relatively active day. The difference between the H component at each one of our seven southern stations and its conjugate northern H component are plotted with respect to MLT. The largest differences between the conjugate locations do indeed occur in the midnight sector between 21 and 03 MLT, while the smallest differences appear to occur in the noon sector between 9 and 15 MLT. This is more than likely the result of localized substorms occurring regularly near local midnight during active days. When we examine the remaining 6 days in plots similar to that of Figure 11 (not shown here), we find that the largest differences between the northern and southern H component occur in the 21 to 3 MLT sector and the smallest differences occur in the 9 to 15 MLT sector for both active and quiet days. This result is not surprising considering that nightside large-amplitude disturbances (substorms, pseudobreakups, or poleward boundary intensifications) occur during both active and quiet geomagnetic conditions.
[36] Kivelson et al. [1996] suggested that such nightside asymmetries would be present during substorm-associated flux ropes and the disagreement would be influenced by the IMF By. Østgaard et al. [2004, 2007] used simultaneous global imaging in the ultraviolet wavelengths by IMAGE and Polar spacecraft to examine auroral substorms in the conjugate hemispheres. Østgaard et al. [2004, 2007] found that a systematic displacement in the onset locations of nighttime activations exists in one hemisphere compared to the other, which is controlled primarily by the IMF clock angle, and secondarily by the dipole tilt angle. Hajkowicz [2006] found a seasonal dependence on the displacement of the onset location between hemispheres, while Sato et al. [1998] found similar displacement asymmetries for all types of auroral activations and arcs and they suggested the existence of nonsymmetric FACs.
[37] Our results are consistent with the studies of Østgaard et al. [2004, 2007] and Sato et al. [1998], and the suggestion of nonsymmetric FACs could explain the persistently lower amplitude perturbations recorded in the Southern Hemisphere AE index compared to those in the Northern Hemisphere index.
[38] An additional explanation for the differences between the SAE and WDC AE is the lack of SAE array ground stations in the 21 to 3 MLT sector during part of the day. As we noted in section 2 there are two large gaps in the SAE array. The midnight sector is within the largest gap from 1500 to 1940 UT. In Figures 6, 7, and 9 we have indicated this interval with a thick horizontal bar and in all three figures there are significant differences between the SAE and the standard AE in the structure and magnitude of the indices while the SAE and NAE indices are relatively similar to one another. The clearest difference is apparent in Figure 7 where the WDC AE records an isolated substorm at about 1630 UT while both the SAE and NAE measure little geomagnetic activity.
[39] A number of disagreements between the NAE index and the standard AE index have also been observed. These include differences in the magnitude of the NAE and the AE index, but sometimes also NAE and AE show quite different individual perturbations/activations. The differences in the magnitude in the two indices are most likely due to the averaging together of multiple ground stations that are spread over a range of latitudes for our NAE index, as we found for the description of the 4 December, 2005 period (section 3.2). Future studies may wish to include the stations in the derivation of the NAE index but not average the data. Another source for potential differences is the gaps in our northern NAE index. These gaps are due to the limited number of southern auroral stations. This means some activity could be missed by the northern stations used for our NAE index that the standard AE station are able to measure. Most interesting were the instances when there is better agreement between SAE and AE, in terms of individual perturbations/activations, than between NAE and AE, as in the time periods of 3 and 8 December 2005 (Figures 6 and 9, respectively). In these instances we deduced that the activations occurred at different MLTs in the two hemispheres, as determined by Østgaard et al. [2004, 2007].
[40] Another aspect worth exploring, which stems from the above discussion, is the differences between the SAE and the NAE under different IMF conditions. We would expect during IMF Bz > 0 that the differences between the southern and northern AE will be smaller than during IMF Bz < 0 because there tends to be more geomagnetic activity during IMF Bz < 0. To investigate the AE differences we propagated Wind solar wind measurements and resampled them to 1 min resolution from the original spacecraft position to the nominal subsolar bow shock at (17, 0, 0) RE in GSE coordinates with the pseudominimum variance technique outlined in Weimer et al. [2003] and Weimer [2004]. We then calculated the mean difference for all the available data between the absolute value of the difference between the southern and northern AE indices during IMF Bz > +1.0 nT and IMF Bz < −1.0 nT. The mean difference is determined to be 31.05 ± 0.01 nT during positive IMF Bz and 51.50 ± 0.04 nT during negative IMF Bz, where the uncertainty is the error of the mean. Approximately 3600 points went into the determination of the positive IMF Bz mean and about 1800 points into the negative IMF Bz mean.
[41] For large magnitudes of IMF By we would also expect the differences between the SAE and NAE to be large, because of the twisting of the magnetotail and the asymmetric reconnection sites in the two hemispheres (e.g., Kivelson et al., 1996). We similarly calculated the mean difference for all the available data of the absolute value of (NAE − SAE) during IMF ∣By∣ > +5.0 nT and IMF ∣By∣ < 5.0 nT. The mean difference is 81.81 ± 0.10 nT during IMF ∣By∣ > +5.0 nT and 31.35 ± 0.01 nT during IMF ∣Bz∣ < 5.0 nT, where again the uncertainty is the error of the mean. Approximately 7700 points went into the determination of IMF ∣By∣ > +5.0 nT mean, and about 800 points went into the IMF ∣By∣ < 5.0 nT mean. On the basis of additional analysis of different value of IMF By we find that there is less difference between NAE and SAE for small ∣By∣ and a larger difference between NAE and SAE for large values of ∣By∣.
[42] For completeness we also examined the differences in the SAE and NAE indices during IMF Bx > 0 and IMF Bx < 0. As before, we determined the mean difference for all the available data of the absolute value of (NAE – SAE) during IMF Bx > +1.0 nT and IMF Bx < −1.0 nT. The difference is determined to be 38.80 ± 0.02 nT during positive IMF Bx and 34.41 ± 0.01 nT during negative IMF Bx where the uncertainty is the error of the mean. Approximately 4100 points when into the determination of the positive IMF Bx mean and about 2300 points when into the negative IMF Bx mean. The difference between the SAE and NAE indices is approximately equal for all IMF Bx cutoff magnitudes.
[43] In summary, we find that the differences in magnitude between NAE and SAE are greater for negative IMF Bz and for IMF By greater than ±5 nT. IMF Bx seems to have no effect on the differences between Northern and Southern Hemisphere indices.
5. Summary and Conclusions
[44] In this study we have compared an AE index calculated from Southern Hemisphere ground magnetometer stations with both a conjugate northern AE index and the WDC quick-look AE index for 7 days from December 2005. The purpose of this study is to determine how accurately the standard AE index, calculated from observatories only in the Northern Hemisphere, can represent activity in the Southern Hemisphere.
[45] We have only used ground magnetometer stations within the standard oval location, 60° to 70° geomagnetic. The oval can move to lower or higher locations than this during more or less active times, respectively. We have not considered such times in this paper. Nevertheless, we used the most complete, to date, coverage of Southern Hemisphere auroral oval stations, even though there are two large gaps in the MLT coverage of stations because of the inherent limitation of limited landmass in the auroral zone in the Southern Hemisphere.
[46] On average, the SAE and NAE indices were relatively well correlated with a mean of 0.57 and a maximum correlation of about 0.81, although for one event the correlation coefficient was near 0. SAE and the WDC AE are equally well correlated with a mean of 0.58, a maximum of about 0.90 and a minimum of 0. However, the NAE was better correlated with the standard AE with a mean of 0.65, a maximum of about 0.65 and a minimum of 0.35. Despite the fact that the correlation between NAE and SAE, or SAE and AE can be moderate or good, we found significant asymmetries in the indices and individual magnetometer traces from the two hemispheres. Such differences can be well explained by geomagnetic activity in the midnight sector, IMF Bz, and IMF By conditions. In some cases the latitudinal difference between northern and southern stations and their likely different location with respect to the electrojet also contributed to the observed asymmetries. The correlation coefficient between NAE and SAE is greater during active days, but the amplitude differences between the two indices are also greater during active days. We were able to show that differences between NAE and SAE are mainly the results of strong perturbations/activations in the midnight (21–03 MLT) region, which occur during all activity levels but are more pronounced during higher geomagnetic activity. The differences in magnitude between NAE and SAE are greater for negative IMF Bz and for IMF ∣By∣ greater than 5 nT, while IMF Bx seems to have no effect on the differences between northern and Southern Hemisphere indices.
[47] While many of the differences between the SAE and standard AE can be explained by the gap in the SAE array of ground stations, we found significant north-south asymmetries that were not related to the gap of stations in the Southern Hemisphere. We deduced that those are likely the result from interhemispheric asymmetries rising under strongly negative IMF, or when IMF By is large, in agreement with the results of Østgaard et al. [2004, 2007]. Interhemispheric asymmetries in ionospheric conductivity and localized ground conductivity gradients may influence the signals in the ground magnetometers and may be a source of the SAE − NAE differences we recorded. A larger statistical investigation of these ideas is beyond the scope of this study.
[48] This study services as a reminder to exercise caution when using the northern AE indices to diagnose activity in the Southern Hemisphere. We suggest that SAE should be considered along with AE for all UT times except for 15–19 UT, when there are no southern stations in the midnight region.
Acknowledgments
[49] This work was supported by NASA grant NNG05GE00G. We would also like to thank GIMA (Geophysical Institute Magnetometer Array) for the data from the BET station; DMI (Danish Meteorological Institute) for data from AMK, NAQ, and SCO; BAS (British Antarctic Survey) for the data from the HBA station; SAMNet (Sub-Auroral Magnetometer Network) for the data from the HLL station; Geoscience Australia for the data from MCQ and MAW; Oleg Troshichev for the data from NVL station; INTERMAGNET (International Real-time Magnetic Observatory) for the data from PBQ; Pieter Stoker for the data from the SNA magnetometer; IMAGE (International Monitor for Auroral Geomagnetic Effects) for the data from SOR and BJN; and the WDC for Aurora in National Institute of Polar Research in Japan for the data from the SYO station. Finally, we want to give special thanks to Nose Masahito for providing us with some detailed plots of the magnetometer data that contributed to the quick-look AE index we used as a standard.
[50] Wolfgang Baumjohann thanks Gordon Rostoker and Viacheslav Pilipenko for their assistance in evaluating this paper.