Large impulsive geomagnetic field disturbances from auroral current systems have always been well understood as a concern for power grids in close proximity to these disturbance regions, predominantly at high-latitude locations. Magnetospheric shocks (SSCs) due to large-scale interplanetary pressure pulses are familiar from a geomagnetic disturbance perspective but have not been understood in the context as a potential driver for large geomagnetically induced currents (GICs). Observational evidence and analysis contained in this paper illustrate such events are capable of producing large geoelectric fields and associated GIC risks at any latitude, even equatorial locations. A large SSC disturbance on 24 March 1991 produced some of the largest GICs ever measured in the United States at midlatitude locations. The analysis methods and understanding of electromagnetic coupling processes at that time were unable to fully explain these observations. Electrojet-driven disturbances common at high-latitude locations during geomagnetic substorms cause large amplitude variations in locally observed B field, while SSC events are characterized as low-amplitude B field disturbance events. Disturbance amplitude only accounts for part of the electromagnetic coupling process. The attribute of spectral content of the disturbance is equally important and heretofore had not been well understood and was not well measured unless high-cadence observations were conducted. The deep-earth ground conductivity also provides an important enabling role at higher frequencies. Deep-earth ground response to geomagnetic field disturbances is highly frequency-dependent. Therefore for nearly all ground conditions the higher the spectral content of the incident magnetic field disturbance, the higher the relative geoelectric field response.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
 It is well known that large magnitude geomagnetic field disturbances can be generated at high-latitude locations due to the intensification of ionospheric electrojet currents during substorms associated with major geomagnetic storms. These large magnitude disturbances which are usually accompanied with a large rate-of-change in the local geomagnetic field will also produce large geoelectric fields and the flow of geomagnetically induced currents (GIC) in exposed technology systems like electric power grids and metallic-conductor portions communication lines and cables. However, this conventional wisdom does not explain the occurrence of large GICs and power system impacts that have been occasionally observed at low-latitude locations or at other midlatitude locations where no electrojet current intensification has occurred. In looking at the time specifics of these noted events, it was apparent that the magnetospheric shock associated with a sudden impulse or storm sudden commencement (SSC) was the disturbance source of these unexpected GIC events. These SSC or shock events are quite small in magnitude at low and midlatitude locations, especially in comparison to electrojet-driven disturbances at high latitudes. SSC events are generally less than 200 nT in total delta B change [Araki, 1994; Araki et al., 1997; Tsunomura, 1998]. Therefore using the perspective of larger geomagnetic field disturbances (>2000 nT) driven by electrojet intensifications, the observation of important GIC impacts is surprising.
2. An Overview of Large GIC Observations and Impacts due to Magnetospheric Shock (SSC) Events
 Due to the large number and serious nature of power system impacts that were observed during the 13–14 March 1989 superstorm, several electric utility companies across the United States began or reinstituted continuous observations of neutral GIC flows in transformers. While these measurements never constituted more than a handful of the many thousand possible locations on the U. S. power grid that GIC could flow and were usually carried out in the more northerly locations, this effort provided some important observations. During a SSC event on 24 March 1991 (23 March 1991 LT in North America), some of the largest GICs ever observed in the United States were recorded. In one of the transformers at the Limerick 500 kV Nuclear Plant near Philadelphia, a paper chart recording (Figure 1) measured both a large DC current (the GIC) and a correspondingly large AC current in the transformer neutral (D. Fagnan, plot, Philadelphia Electric, 1991). These paper chart recorders are usually set at ±100 amps full scale, and as noted by the arrows, the GIC went to full scale briefly and then rapidly decayed. The corresponding AC measurement at this site also reacted in the same manner. The AC current measurement is caused by half-cycle saturation in the transformer due to the GIC and therefore provides a high-quality measure of the degree of power system impact that was being observed. In western Virginia at the same time, two measurements of GIC were made at the Meadowbrook 500 kV substation on transformers T2 and T4; these measurements (Figure 2) taken at a average cadence of ∼20 seconds indicated GIC peaks of −47 and −66 amps each at time 0343:03 UT. If there had been only one transformer, the GIC flow would have been approximately ∼110 amps, as the two transformers were providing parallel paths for the GIC flow [Gattens, 1992]. At the same time, at the Pleasant Valley 345 kV substation in the lower Hudson River valley in New York, an even larger GIC was observed [Electric Research and Management Inc., 1995]. As shown in Figure 3, both the neutral GIC and a neutral AC third harmonic current were recorded. As in the case of the observations at Limerick, both the GIC and the AC current reached large values and then rapidly decayed. The GIC measurement reached a peak of 130 amps, while the third harmonic (due to transformer saturation) had a peak just short of 350 amps.
 More contemporary observations of GIC and GIC-related power system impacts indicate that even smaller intensity SSC events have the capacity to disrupt power system operations over wide geographic areas. For example, a SSC at the start of the 15 July 2000 storm caused simultaneous tripping of capacitor banks at time 1439:01 UT at two substations in the Tennessee Valley Authority (TVA) power grid, one at the Madison station in northern Alabama (geographic latitude of 34.79°) and the other at the New Albany station in northern Mississippi (geographic latitude of 34.48°) (F. Elmendorf, personal communication, 2000). On 31 March 2001 the NY-ISO reported the tripping of the East Fishkill capacitor bank at time 0000:54 UT due to GIC affects, a timestamp coincident with the SSC onset for that storm [Ingleson, 2001]. Meanwhile, observations at even lower latitudes confirm the presence of large GIC flows in power grids due to SSC events. For example at the Fukumitsu substation in central Japan (geographic latitude ∼34°, geomagnetic latitude ∼26°), a peak GIC of over 40 amps was observed at one station due to a moderate SSC event on 6 November 2001, and GIC was also simultaneously observed at another station as shown in Figure 4 [Erinmez et al., 2002]. For this event, nearly equally large GICs, as shown, were due to equatorial ring current intensifications over the region in the several hours period after the SSC onset. Also, during this same 6 November 2001 SSC event, the power grid in New Zealand reported observing GIC, operational problems, and the failure of a large power transformer coincident with the SSC event [Small, 2003].
 These observations clearly provide a compelling case that a new class of GIC risk is plausible and that these risks can occur on power grids at very low-latitude locations that are typically not concerned about or seldom in the proximity of large electrojet intensifications. The SSC disturbance conditions can extend to low-latitude and midlatitude locations, while most large electrojet disturbances are generally confined to high-latitude locations. It is also evident that the magnitude of the delta B geomagnetic field disturbance, in isolation, does not provide a good predictor of the relative severity of the resulting GIC flow in technology systems such as power grids. Therefore a detailed assessment was undertaken to consider the relative risk factors leading to GIC. These risk factors considered all attributes of the geomagnetic disturbance environments, caused by both an SSC event in contrast to the more commonly considered geomagnetic disturbance threat due to electrojet intensifications. To further examine the factors that contribute to GIC risks, it was also necessary to evaluate the electromagnetic coupling behavior between the geomagnetic disturbance and the deep-earth ground conductivity that creates the geoelectric field. The spectral content attributes of the impulsive disturbances associated with a SSC event along with the role of the deep-earth ground conductivity conditions provides the enabling conditions to produce large geoelectric field responses and large GIC flows in exposed infrastructures.
3. Example Ground Models for Geoelectric Field Response Evaluations
 While the focus of this paper regards the nature of geoelectric fields due to SSC events at low-latitude and midlatitude locations, a complimentary understanding of the role of deep-earth conductivity in determination of the geoelectric field is also needed. Considerable prior work has been done to model the geomagnetic induction effects in ground-based systems, however these approaches have primarily been focused upon the induction due to auroral current sources [Albertson and Van Baalen, 1970; Lanzerotti, 1983; Pirjola, 1984]. As an extension to this fundamental work, numerical modeling of ground conductivity conditions have been demonstrated to provide accurate replication of observed geoelectric field conditions over a very broad frequency spectrum [Kappenman et al., 1997]. Recent reviews of space weather events have briefly noted the effects due to SSC onsets of geomagnetic storms [Lanzerotti et al., 2001].
 Ground conductivity models need to accurately reproduce geoelectric field variations that are caused by the extensive range of frequency or spectral content of geomagnetic disturbance events from the large magnitude/low frequency content disturbances to the low amplitude but relatively high frequency content impulsive disturbances commonly associated with SSC events. This variation of electromagnetic disturbances therefore require models accurate over a frequency range from 0.3 Hz to as low as 0.00001 Hz. Over this range of frequencies of the disturbance environments, diffusion aspects of ground conductivities must be considered to appropriate depths. Therefore skin depth theory can be used in the frequency domain to illustrate the range of depths that are of importance. Ground conductivity measurements have routinely been made at 50 or 60 Hz frequencies and also at much higher frequencies; however, these measurements are not of great use since they only characterize the upper portion of the ground conductivity profile. Table 1 shows the depths to which conductivity information is needed for geomagnetic storms relative to 50 Hz measurements. Using these frequencies as a range, a comparison can also be made of the effective skin depths that high-frequency measurements cover versus the depth range that needs to be considered for various ground conductivities. It is clear that for constant Earth conductivities, the depths required are more than several hundred kilometers, although the exact depth is a function of the layers of conductivities present at a specific location of interest.
Table 1. Depths to Which Conductivity Information is Neededa
 It is generally understood that the Earth's mantle conductivity increases with depth. In most locations, ground conductivity laterally varies substantially at the surface over mesoscale distances, these conductivity variations with depth can range 3 to 5 orders of magnitude. While surface conductivity can exhibit considerable lateral heterogeneity, conductivity at depth is more uniform. Conductivities range from values of 1 to 10 S/m at depths from 600 to 1000 km [Campbell, 1987; Masse, 1987]. If sufficient low-frequency measurements are available to characterize ground conductivity profiles, models of ground conductivity can be successfully applied over mesoscale distances and can be accurately represented by use of layered conductivity profiles or models.
 For illustration of the importance of ground models on the response of geoelectric fields, a set of four example ground models have been developed that illustrate the probable lower to upper quartile response characteristics of most known ground conditions, considering there is a high degree of uncertainty in the plausible diversity of upper layer conductivities. Figure 5 provides a plot of the layered ground conductivity conditions for these four ground models to depths of 700 km. As shown, there can be as much a four orders of magnitude variation in ground resistivity at various depths in the upper layers. Models A and B have very thin surface layers of relatively low resistivity. Models A and C are characterized by levels of relatively high resistivity until reaching depths exceeding 400 km, while models B and D have high variability of resistivity in only the upper 50 to 200 km of depth [Campbell and Zimmerman, 1980; Rasmussen et al., 1987; Rasmussen, 1988].
Figure 6 provides the frequency response characteristics for these same four layered earth ground models of Figure 5. Each line plot represents the geoelectric field response for a corresponding incident magnetic field disturbance at each frequency. While each ground model has unique response characteristics at each frequency, in general, all ground models produce higher geoelectric field responses as the frequency of the incident disturbance increases. Also shown on this plot are the relative differences in geoelectric field response for the lowest and highest responding ground model at each decade of frequency. This illustrates that the response between the lowest and highest responding ground model can be as much as a factor of 13 different. Also because the frequency content of an SSC event can have higher frequency content, the disturbance is acting upon the more responsive portion of the frequency range of the ground models. Therefore, the same disturbance energy input at these higher frequencies produces a proportionately larger response in geoelectric field. For example, in most of the ground models the geoelectric field response is a factor of 50 higher at 0.1 Hz compared to the response at 0.0001 Hz. This frequency response characteristic explains why the SSC events though smaller in magnitude have been responsible for producing large GICs.
4. Parametric Analysis of Ground Model Response to SSC Events
 As previously noted, prior work on geoelectric field response has traditionally been focused upon the disturbance environments associated with high-latitude impulsive disturbances driven by electrojet intensifications. Variations in ground models have been shown to produce large differences in geoelectric field response for these disturbances. In one such study, a factor of 5 difference in peak geoelectric field response was established due to ground model differences alone, given the same impulsive geomagnetic field disturbance [Kappenman, 2001]. In order to assess the potential for large geoelectric fields and GICs in power grids posed by SSC shock events, it is necessary to assess the possible range of geoelectric field response differences that are due to natural variations from the differing ground models that have been presented.
 As with any other aspect of space weather, there can be a wide variation in the nature of ground-level impulsive magnetic field disturbances due to SSC events. Figure 7 provides a comparison of two extremes of observed Bx during SSC events at the MSR observatory on the 210 Longitude Chain of magnetic field observatories. While both these disturbances have similar magnitudes of delta Bx disturbance intensity, they have significantly different rates of rise to peak. The slower rise disturbance occurred during the SSC from the 15 July 2000 SSC event at around time 1538 UT, the MSR observatory was near local midnight. The faster rise disturbance was observed during the 24 March 1991 SSC event at around time 0341 UT. During this event, the MSR observatory was near local noon. Figure 8 provides a comparison of the rate of change of the disturbance field or dBx/dt and demonstrates the broad range of variation that is possible, given nearly identical peak amplitudes of Bx. The 15 July 2000 SSC had a peak dBx/dt of ∼4 nT/s while the 24 March 1991 SSC had a peak dBx/dt over 10 times larger at ∼40 nT/s. These differences in spectral content of the impulsive disturbances are important and are intimately coupled with the variations in ground conductivity conditions in determining the resulting geoelectric field response.
 Because SSC events can present at times complex impulsive disturbance waveforms that vary both in amplitude and spectral content, a parametric analysis using a uniform series of impulsive geomagnetic disturbances has been performed to more clearly illustrate the role of the ground model in the resulting geoelectric fields that can occur. In Figure 9 a parameterized SSC waveform is shown (blue time plot) compared to an actual Bx waveform from the 15 July 2000 SSC as measured at the MSR observatory. The parameterized waveform replicates closely both the peak delta B magnitude as well as the slope of the rise in delta B. While the MSR B field decays more slowly than the profile waveform, the high rate of change at the event onset will be the important driver in the formation of the geoelectric field. Figure 10 provides a comparison of the rate of change of the two B fields from Figure 9, and this comparison indicates that both B fields reach peak dB/dt of ∼4 nT/s.
 As will be demonstrated in a series of parametric simulations, the rate of change of the B field is one of the most important factors in determining the expected geoelectric field response. In general, the larger the dB/dt, the larger the resulting geoelectric field response will be for most ground conductivity conditions. SSC events can produce large delta B variations (>1000 nT) at auroral latitudes. However, at low and midlatitude locations, SSC events typically have much less variation in the peak magnitude of delta B. Therefore when an increase is dB/dt occurs at these locations, the increase in dB/dt primarily is generally due to enhancing the spectral content of the disturbance. To illustrate this principle, a set of four parameterized waveforms has been developed as shown in Figure 11. Each of these B fields represents plausible B fields from SSC events and have been constrained such that the peak delta B only increases to 100 nT, yet the dB/dt intensity differs, with values at 2, 8, 15, and 30 nT/s, respectively. Figure 12 illustrates the dB/dt characteristics of each of these four profile B field disturbances. Fast Fourier Transform (FFT) methods can be used to approximately estimate the spectral content of each of the profile B field disturbances. As shown in Figure 13, at low frequencies (below 0.05 Hz) the magnitude of the four B fields are similar; however, at higher frequencies the higher dB/dt disturbance conditions have more content of high frequency disturbance signal.
 These profile B field disturbances will be simulated with the four example ground models that have been previously discussed. From the frequency response plots of the ground models as provided in Figure 6, some of the expected geoelectric field response due to these profile B field characteristics can be inferred. For example, ground C provides the highest geoelectric field response across the entire spectral range; therefore it would be expected that the time domain response of the geoelectric field would be the highest for nearly all B field disturbances. At low frequencies, ground B has the lowest geoelectric field response while at frequencies above 0.02 Hz, ground A produces the lowest geoelectric field response. Because each of these ground models have both frequency-dependent and nonlinear variations in response, the resulting form of the geoelectric field waveforms would be expected to differ in form for the same B field input disturbance. In all cases each of the ground models produces higher relative increasing geoelectric field response as the frequency of the incident B field disturbance increases. Therefore it should be expected that a higher peak geoelectric field should result for a higher dB/dt disturbance condition, due to the higher frequency spectral content as dB/dt increases.
Figures 14 and 15 provide the geoelectric field responses for each of the four ground models for the 2 nT/s and 15 nT/s B field disturbances. In Figure 14, the geoelectric fields are compared for the 2 nT/s uniform geomagnetic field disturbance. As expected, the ground C model produces the largest geoelectric field, while ground A is next largest and the ground B model produces the smallest geoelectric field response. For this disturbance scenario the ground C geoelectric field peak is more than six times larger than the peak geoelectric field for the ground B model. It is also evident that significant differences result in the shape and form of the geoelectric field response. For example in the 2 nT/s disturbance, the peak geoelectric field for the ground A model occurs 15 s later than the time of the peak geoelectric field for the ground B model. In addition to the differences in the time of peak, the waveforms also exhibit differences in decay rates. For the higher impulsive disturbances (8, 15, and 30 nT/s), similar variations in the times of peak and decay rates of the geoelectric field also occur. In addition, there is considerable variation in the order of severity of the geoelectric field amongst the models. For example, while ground A had the second largest peak geoelectric field at 2 nT/s, at 15 nT/s (Figure 15), the ground A model had the lowest peak geoelectric field as both ground B and D models responded more readily to the higher spectral content of the B field disturbance. The ground C model produces the largest geoelectric field in all impulse disturbance scenarios, which was an expected outcome.
 For all the ground models, the larger dB/dt disturbance scenarios resulted in larger peak geoelectric fields as well. However, the rate of increase in peak geoelectric field was not uniform, hence the reason that the ground A model slipped from second highest geoelectric field at 2 nT/s disturbance conditions to lowest geoelectric field at 15 nT/s. Figure 16 provides a comparison of the peak geoelectric fields for each of the four ground models for all four of the B field disturbance scenarios. In the case of ground B, the geoelectric field increased by a factor of 5.5 between the 2 nT/s disturbance and the 30 nT/s disturbance, while for ground A the increase was much lower with only a increase of 1.6 times in peak geoelectric field. There are also important differences in the peak response between the ground models. As previously noted, the difference between the highest responding ground (ground C) and the lowest responding ground (ground B) was a factor of ∼6 for the 2 nT/s disturbance. For the 30 nT/s disturbance the ratio of max response (ground C) to min response (ground A) has increased to nearly a factor of 10 difference. Figures 17 and 18 provide a comparison of the spectral content of the geoelectric field respectively for the ground C and ground B ground models for the 2, 8, and 15 nT/s disturbances. In relative terms, the ground B spectral content is lower than the ground C model spectral content due to the much lower geoelectric field response of the ground B model. More importantly as shown in these two figures, the spectral content below frequencies of 0.05 Hz are essentially equal for the 2, 8, and 15 nT/s disturbances. However, at higher dB/dt disturbance levels, the geoelectric field content at frequencies above 0.05 Hz increases with increased dB/dt intensity. This higher geoelectric field spectral content is sufficient to cause the previously demonstrated increases in peak geoelectric field levels.
5. An Analysis of Peak Geoelectric Fields at Low Latitudes and Midlatitudes for the 24 March 1991 SSC Event
 The SSC disturbance scenario B field waveforms developed in Figure 11 have provided a set of simulations to undertake a parametric analysis of the importance of ground model response and the spectral content of the B field disturbance in the determination of the resultant geoelectric field response. These profile disturbances do not represent worst-case scenarios, in that the peak of the delta B disturbance was limited to only 100 nT, while large SSC events even at low latitude locations can exceed 200 nT. The 24 March 1991 SSC was one such large and known event that produced widespread impacts to systems at mid and low latitude locations as described in section 2. Only a limited number of high-cadence observatories were in operation at the time of this disturbance, but they are sufficient to provide a generally broad description of the B field disturbance intensities and propagation characteristics at low and midlatitudes. By extension and with the use of the example ground models it is also possible to estimate geoelectric field intensities that could have been possible in these locations. Key observatories with high cadence data for this event are along the 210 Chain of observatories (CBI, KAK, and MSR) in Japan, the AT and T observatories at Point Arena (PTA) in California and Tuckerton (TUK) in New Jersey. Also available are data from the VIC and OTT observatories in southern Canada.
 The onset time for this SSC event occurred at ∼0341 UT on 24 March 1991. Therefore local noon is located west of Japan, while local midnight was located in the mid-Atlantic region off the eastern coast of North America. Figures 19 and 20 provide time plots of the observed delta Bx and delta By for observatories running east to west from KAK (central Japan), PTA (central California), and TUK (New Jersey). As the plots of delta Bx and By suggest, a large coherent disturbance propagated across the low and midlatitude regions. At times across North America, the disturbance would have presented as a large-scale coherent plane wave. An example of the geographic scale of the disturbance can best be shown by a plot of the coherent horizontal geomagnetic field disturbance vectors over midlatitude portions of North America as shown in Figure 21 at time 0342:22 UT. The disturbance conditions observed at the KAK observatory produced a dBx/dt of ∼15 nT/s. Figure 22 provides a summary of the estimated geoelectric fields using the four ground models. Since the Kakioka Bx component is being used in this calculation, the resulting geoelectric field will have an east-west orientation. As with the prior set of example B field simulations, the results of the geoelectric field waveforms and peak geoelectric fields show a wide variation of results that are heavily influenced by the nature of the ground model. In this case, the peak geoelectric field again occurs for the ground C model, reaching a peak of about −3.4 V/km (for this calculation an eastward field is positive polarity, the polarity of the peak of this geoelectric field is westward oriented). The lowest geoelectric fields occur for both the ground A and ground B models reaching levels of only about −0.7 V/km, which provides a factor ∼5 difference between the peak geoelectric field for the example ground models considered. These levels are nearly twice as large as the peak geoelectric field magnitudes from the earlier profile B field disturbances.
 Because of the sparsity of observatories, a cross-sectional analysis both north to south and east to west in the low and mid latitude regions provides a broad perspective of the possible range of geoelectric fields that could have resulted from the SSC event. Figure 23 provides a cross-sectional summary of the geoelectric field peaks for the four example ground models from the MSR location to the CBI location, which spans north-to-south in the local noon dayside region of the SSC event. In the worst-case ground model, the peak geoelectric fields range from ∼5.5 V/km at the northern point on the chain and decrease to ∼3 V/km at the lower latitude CBI observatory. Since the geomagnetic disturbances in the Southern Hemisphere have a symmetrical behavior of the northern hemisphere, it would be expected that a similar variation of geoelectric field intensity would also occur for the Southern Hemisphere regions.
 This SSC event also produced a consistently large geoelectric field levels from the dayside to the nightside as illustrated by the summary of peak geoelectric field observations in Figure 24. The observations range from the MSR (37.6° geomagnetic latitude) observatory (dayside) to the PTA (44.6° geomagnetic latitude) and TUK (51.4° geomagnetic latitude) observatories at midlatitude locations in North America (nightside). While these three observatory locations are at differing geomagnetic latitude locations, they indicate a trend that relatively high levels of geoelectric field are being observed globally at low and midlatitude locations and not strictly limited to dayside regions. The geoelectric fields estimated to occur at MSR (midlatitude near local noon) are approximately equivalent to those estimated at TUK in eastern North America (midlatitude near local midnight). The large geoelectric fields estimated based upon the TUK observations would provide an explanation for the large GIC flows in these regions as noted earlier in the paper.
 In all cases and locations, the larger variation in geoelectric field levels is due to the nature of the ground model that is being considered and not as much due to local B field environment differences. For example the peak geoelectric field for the ground C example ground model varies from ∼4 to ∼5.5 V/km at midlatitude locations from dayside to nightside. For the other three less responsive ground models, the peak geoelectric fields vary between ∼0.75 to ∼2 V/km across this same broad geographic region.
 The implications of these findings on the role of SSCs and related GIC risks at low latitudes are potentially far reaching and could ultimately and dramatically transform the investigations of GIC and GIC risks to infrastructure operators. New facets of space weather risks have been established in that large geoelectric fields and associated GICs can and have been observed due to SSC events and at various latitude locations.
 Implications of GIC hazard also extend to power grids that have never considered the risk of GIC previously because they were not at high latitude locations. In contrast to prior assumptions that GIC hazards were only limited to infrastructures at high latitude locations, it is now apparent that low magnitude/high spectral content disturbances can provide large geoelectric fields that can trigger power system upsets comparable to those more commonly observed at high latitudes.
 This research was made possible by a grant from the National Science Foundation (award ATM-0207058). The author also appreciates the considerable assistance and inputs from Metatech coworkers William A. Radasky, James L. Gilbert, and Jeffrey Patrick, Jason Rauner and Khaled Ejaz. The author would also like to thank the numerous suppliers of observed magnetometer data, data providers include Lou Lanzerotti and Carol Maclennan at Lucent Technologies, David Boteler at the Geological Survey of Canada, the 210 Longitude Chain of magnetic field observatories, the World Data Center for Geomagnetism in Kyoto, Japan.