The path of maximum obscuration for the annular solar eclipse of January 15, 2010, crossed the magnetic equator at Trivandrum, India, in the early afternoon hours. A strong counter-electrojet was observed shortly after maximum obscuration. Moreover, as the eclipse passed overhead, the F region density peak underwent a large amplitude vertical oscillation. At the same moment, there was an oscillation in the zonal electric field inferred from the magnetometer data. The electric field turned westward after the time of maximum obscurity, reaching its largest westward value one hour before the end of the local eclipse. We show that these data are consistent with a fast eastward moving local neutral wind dynamo generated by a low pressure system postulated to have been triggered by the cold temperatures in the region of maximum obscuration.
Solar eclipses provide unique opportunities to study how the regional absence of solar radiation affects the upper atmosphere and ionosphere. Among the most recognizable ionospheric effects are the reduction in E and F-region electron densities as a function of the sun's degree of obscurity [e.g., Van-Zandt et al., 1960], and a decrease in electron temperature in the F-region [e.g., MacPherson et al., 2000]. Satellite beacon measurements have also revealed decreases in total electron content (TEC) of the ionosphere by as much as 20–30% [Cohen, 1984], followed by a quick recovery after the passage of an eclipse.
For the majority of cases studied until now, solar eclipses have taken place in mid and low latitude regions, where the production and distribution of plasma is governed mostly by photochemistry and diffusion [Van-Zandt et al., 1960]. A few equatorial latitudes studies have also been reported. Most relevant to the present work, Tomas et al.  have studied satellite observations of eclipses straddling the magnetic equator. They found clear evidence for counter-electrojet tendencies in ‘the wake’ of the eclipse in 4 of the 5 events that they could study. Similarly, we present here a rare case for which a solar eclipse with 84% obscuration passed, during the middle of the day, overhead of ground stations at the magnetic equator in Southern India. We could establish from these observations not only that a strong counter-electrojet was triggered by the eclipse, but also that the plasma was prone to additional unusual electrodynamics that we attributed to the neutral wind and conductivity perturbations that accompany the eclipse at the magnetic equator. We present the observations in Section 2 and discuss them in terms of plausible mechanisms in Section 3.
2.1. The Eclipse
The annular eclipse of Jan 15, 2010, started in central Africa at local sunrise at 8:25 Indian Standard Time (IST) or 3:55 UT. The eclipse moved east, straddling the geographic equator, first crossing it from the north before coming back from the south. The eclipse over Trivandrum, India (8.5°N, 76.9°E, 0.2°S dip-latitude) started at 11:05 IST (Local Time, LT, = IST − 22 min). In Trivandrum, at the peak of the eclipse, solar radiation was depleted by 84.3% in the F region during 8 minutes centered at 13:14 IST. The eclipse ended locally at 15:07 IST. The eclipsed region moved resolutely to the north-east after crossing southern India. The event ended at sunset in southern China, near 30° latitude. For more details, see http://eclipse.gsfc.nasa.gov/SEmono/ASE2010/ASE2010.html.
2.2. Ionospheric Densities
Ionospheric densities were obtained from the DPS-4D Digisonde installed at Trivandrum. We show the time variation of the electron density profiles retrieved from the ionosonde in Figure 1 (top). We note that while ionosondes cannot provide densities above an ionization maximum like the F region peak, a scale height estimation was used to infer the behavior of the plasma above the peak [Kutiev et al., 2009] in order to provide full electron density profiles estimates.
Below 200 km, the effect of the eclipse on the ion density was quite apparent, with a density minimum coinciding with the time of maximum obscuration: when molecular ions dominate the composition, chemical equilibrium applies and the density has to be proportional to . Applying this relation to 160 km reproduced the observed densities to near perfection. The story was different above 200 km. On eclipse day, the F region densities in the morning hours were large and at an elevated altitude, indicative of the role played by the electrodynamical uplift from zonal electric fields with an eastward component at this time of day. However, Figure 1 (middle) shows that the height of the F peak (hmF2) started to behave in an unusual manner as a function of time after 9 AM IST. In particular, after 10 AM IST, the F peak started to go down, indicating that the normal uplift had been replaced by a downward motion. Moreover, shortly after noon, during the passage of the eclipse, the F region above 280 km went up as the obscuration increased before quickly coming down as the obscuration decreased. After the passage of the eclipse and until 16:00 IST, the F region peak stayed near 270 km while being depleted compared to the morning hours. It started to move up again after 16:00 IST.
We verified that the upward motion was not the result of an apparent motion due to chemistry by reproducing the photo-production function and using the 11 IST profile as an initial condition. The calculations showed that, consistent with the chemical lifetimes in excess of 5 hours above 300 km, there should have been very little change in the F peak position during the eclipse without contributions from a vertical transport term. We furthermore verified that the upward motion of the peak between 12 and 13 IST could only be due to an upward plasma drift and the downward motion of the peak seen after 13 IST could only be due to a downward drift.
2.3. Magnetometer Data
Being under the equatorial electrojet (EEJ), Trivandrum records large magnetic perturbations on a daily basis along the geomagnetic north when strong east-west currents are flowing above head. An often-used procedure to extract the magnetic perturbations from the EEJ is to subtract the Alibag magnetic perturbations (10° farther to the north) from the Trivandrum ones in other to eliminate planetary scale contributions [Rastogi and Klobuchar, 1990]. This procedure assumes that the zonal electric field is uniform as a function of latitude, which we found was definitely not the case on eclipse day. Having verified that there was no magnetic activity during that day, we concluded that it was best to isolate the contribution of the electrojet from the ambient magnetic field and the ring currents simply by subtracting the current seen at 2 AM from the total current. We estimated the resulting accuracy on the local magnetic perturbations to be 5 nT.
We defined the resulting perturbation in the magnetic north direction as ΔH, which is displayed in Figure 1 (bottom) where, for reference, we also show the average behavior of ΔH for the month, together with the variability, as derived from the observations standard deviation. Figure 1 (bottom) shows that the evolution of ΔH was at first quite normal, with an early climb soon after 8 AM IST. That climb stopped at 9 AM IST, and was followed by an unusual steady decrease in the current intensity. This decrease continued until the start of the eclipse over Trivandrum when there was a short-lived increase in ΔH starting around 11:30 IST. This was followed by a sharp drop which continued until one hour before the eclipse ended, at which point ΔH started to increase again. The second drop was so sharp that a strong counter-electrojet (negative fluctuations) had fully developed by 14:15 IST.
We also sought to highlight the behavior of the electric field itself. Since currents are proportional to the product of the electric field times the plasma density in the EEJ, we estimated the relative changes in the electric field by dividing ΔH by the electron density in the E region. We took the average of the digisonde-derived densities between 100 and 110 km, since the densities in the equatorial region vary by as much as 30% from one day to the next or one time of day to the next, implying that transport processes of the kind associated with sporadic E layer generation had to play a role. The inferred zonal E region electric field variations are shown in Figure 1 (bottom right).
As shown above, there was an early weakening of the electrojet current, a temporary increase in the eastward electric field strength one hour after the start of the local eclipse, and a strong counter-electrojet after the period of maximum obscuration. Also, the F region plasma underwent a clear oscillation at eclipse time, due to an eastward electric field (E × B up) followed by a westward field (E × B down). We now propose explanations for these features.
3.1. Large Scale Context
The presence of E region neutral winds always triggers electric fields and/or currents. With magnetized electrons and weakly magnetized ions, either a polarization electric field is set-up by the ions in a direction antiparallel to the wind or the pushing action of the neutrals triggers a weak ion current in the Vn × Bv direction where Vn is the neutral wind and Bv is the vertical component of the geomagnetic field. Either way, a current is generated in the Vn × Bv direction. The neutral wind dynamo then drives a ‘Solar Quiet’ (SQ) current system that ultimately pushes electrons to the east and ions to the west at the magnetic equator. The closure of the currents through the magnetic equator generates the intense zonal currents known as the EEJ through a double Hall effect that starts with a vertical electric field produced by the electrons (see Kelley  for additional details). From a global perspective, the SQ current system is driven by a high pressure (HI) driven by in situ solar heating, and which can be described in terms of the (1,−2) tidal mode [Richmond et al., 1976]. Higher order corrections come mostly from the semidiurnal (2,4) modes with planetary waves affecting the magnitudes of the tides for example through latent heat release in the troposphere [Hagan and Forbes, 2002].
In December-January in southern India, the solar heating-generated HI is centered more than 25° to the south of the magnetic equator and the Indian sector EEJ is at its weakest. This HI is in addition not present in the northern hemisphere. This weakens the EEJ because (1) with equipotential magnetic field lines, the southern HI has to drive currents for both hemispheres and (2) much of the SQ system can close without having to pass through the equator. If we now postulate that the eclipse generated a LO through the absence of solar radiation in the eclipsed region, we are adding a visible perturbation to the already weakened EEJ produced by the HI 25° to the south. With the winds around a LO in a direction opposite to the winds around a HI, the LO had to drive with it a localized traveling ‘counter-SQ’ (CSQ) current system. This CSQ had to close either through a counter electrojet (CEJ) or a weakening of the EEJ at the magnetic equator. The local direction and strength of the current had to be determined by the competition between the HI-driven EEJ and the LO-driven CEJ. Furthermore, the strength of the equatorial closure of the HI-generated SQ or of the LO-generated CSQ depended not only on the strength of the HI or LO pressure systems but also on their distance to the magnetic equator. The further away from the equator the pressure centers are, the less current closure is needed at the magnetic equator, and the weaker the effects on the electrojet currents. In this light, the arrival of a full fledge CEJ in the afternoon hours can be explained by the already weak EEJ being overwhelmed by the LO-driven equator-centered CEJ in the afternoon hours.
Our take on the electrodynamics is illustrated in Figure 2. For the early and central phases we positioned the LO very near the magnetic equator, by contrast to the normal HI, centered substantially to the south of the magnetic equator. As discussed above, the SQ and CSQ loops have to close at the magnetic equator. In early afternoon, the contrast between the currents generated by the LO and HI centers is particularly strong, since the LO is to the north of the HI and squarely centered on the magnetic equator, so that a locally strong CEJ can be expected. In the later phases, the LO moves to the northeast, to disappear two hours before Indian sunset. By then, the electrojet has recovered. Finally, while we show a current reversal at the equator around the LO, we could have equally shown just a weakening of the eastward currents instead. However, this would have unnecessarily cluttered Figure 2.
Figure 2 should help picture what happened to the EEJ at Trivandrum. First, a strong EEJ growth was interrupted after 9 AM IST, one half-hour after the eclipse had started in Africa. By then the LO was approaching Trivandrum from the west, and the influence of the CEJ was already being felt. After maximum obscuration at 13:18 IST, Trivandrum was now on the west side of the LO, registering a strong CEJ. Finally, the LO moved to the north-east, the CEJ disappeared and was replaced with more normal EEJ conditions.
3.2. F Region Dynamo Around the Region of Maximum Obscurity
While the large scale context of a traveling CSQ system can be explained with the help of Figure 2, another important feature of Figure 1 requires an explanation. We refer here to the F peak oscillation between 12 and 14 IST, in conjunction with the short-lived but clear peak in the eastward EEJ near 12:10 IST. We also need to address why the largest negative current and/or electric field values were observed after a symmetric time interval, at around 14:15 IST, one hour after maximum obscuration, at 13:14. For these observations, we require an additional much more localized process, namely, a local electric field enhancement effect produced by the E region conductivity gradient at the eclipse terminators.
When the E region conductivities become small compared to the integrated F region conductivities, a dynamo field is often generated by F region winds moving towards a LO. In the present case the winds had to converge towards the center of the eclipsed region. We illustrate our mechanism in Figure 3. Focus first on the F region part of the diagram: the LO was driving winds (red arrows) towards its center. This meant westward winds on its east side and eastward winds on its the west side. With the eclipse moving from west to east, Trivandrum first encountered the east side of the LO before ending on its west side later on. The zonal winds Vn associated with the local LO did not drive a zonal electric field in the F region. Rather, the F region electric field driven by the winds (blue arrows attached to the red neutral wind arrows in the F region) is actually in the −Vn × B direction [Kelley, 1989], namely upward when the eclipse starts and downward when it ends. Furthermore, the F region vertical electric field tends to be eliminated by E region conductivities during the day. However, the electrodynamics gets more interesting near a terminator.
Our mechanism is a modification of a proposal by Farley et al.  to explain why the zonal equatorial electric field frequently becomes enhanced near sunset prior to its nighttime reversal. As illustrated in Figure 3 the upward F region electric field produced on the east-side of the F region LO (blue arrow at the top right) maps into a southward pointing electric field in the E region in the southern hemisphere (blue arrow tied to the east-most magnetic field trace). In response, the E region electrons E × B drift (double blue arrows pointing toward the terminator). They accumulate near the terminator, where the electron density is becoming small. On the west side of the eclipsed region, all signs reverse: the F region neutral wind changes sign, the F region electric field is driven downward, mapping into an E region northward electric field in the southern hemisphere, once again pushing the outside electrons towards the terminator. For a simple view of the E region charge arrangement, picture negative charges gathering around two 10 km thick regions (EEJ thickness), a few hundred km in horizontal extent (EEJ width) along the magnetic north-south meridian, with a separation between the two negatively charged regions (- signs shown in green) covering a couple of hours in local time, or 3800 km, implying that the electric field is not uniform between the charged layers. As a result, inside the eclipsed region, the zonal polarization field (green arrows pointing toward the terminators) points to the east over the first half of the eclipse (on the east side) and to the west during the second part of the eclipse (on the west side).
Our electric field pre/post-reversal mechanism explains why the F region peak went up prior to the strong reversal in the electric field, before coming down just as suddenly (Figure 1, middle). This feature is shown by the double green arrows associated with EpolE × B in Figure 3. It is completely consistent with the electric field signature inferred from the magnetometer if we allow for the fact that around 11 IST the influence of the LO, because of its position, would have been to weaken the EEJ without yet reversing its sign.
Both the magnetometer and ionosonde observations imply that the first of the two layers of negative charges shown by the green negative charges in Figure 3 had to be positioned around 12:15 IST, namely, one hour after the eclipse had started in Trivandrum and one hour prior to maximum obscuration. After having passed the line of symmetry at maximum obscuration, the sign of the electric field flipped. The resulting westward electric field was produced in part by the placement of more negative charges on the west terminator and in part by the CEJ current pattern that was imposed by the position of the LO. We note that the largest westward values were observed one hour after maximum obscuration, suggesting that the symmetric charged layers positioned one hour on each side of the region of maximum obscuration had a major influence on the overall electrodynamics of the eclipsed region.
The eclipse had two separate impacts on the electrodynamics of the magnetic equator. For five of its six hours, the LO that it generated produced a counter-SQ system and counter-electrojet that quickly moved eastward over the equatorial regions. Also, at the heart of the LO, the currents were interrupted by a decrease in E region conductivity, which produced electric field enhancements around the two terminators.
JPSTM gratefully acknowledges a ADCOS fellowship from the Indian Space Research Organization.