Cause of different local time distribution in the postsunset equatorial ionospheric irregularity occurrences between June and December solstices

Authors


Abstract

[1] Global averaged postsunset equatorial ionospheric density irregularity occurrences observed by ROCSAT during the moderate to high solar activity years of 1999 to 2004 indicate a different local time distribution between June and December solstices. The irregularity occurrences during the December solstice show a faster increase rate to peak at 2100–2200 local time, while the irregularity occurrences during the June solstice have a slower increase rate and peak one hour later in local time than that in the December solstice. The cause of such different local time distributions is attributed to a large contrast in the time of zonal drift reversal and the magnitude of postsunset vertical drift observed by ROCSAT at longitudes of large magnetic declination in the two solstices. That is, a delay in the zonal drift reversal in association with a smaller postsunset vertical drift observed at longitudes of positive magnetic declination has greatly inhibited the irregularity occurrences during the June solstice in contrast to an earlier zonal drift reversal together with a large vertical drift occurring at longitudes of negative magnetic declination to accelerate the irregularity occurrences during the December solstice. We think that the different geomagnetic field strengths that existed between the longitudes of positive and negative magnetic declinations have played a crucial role in determining the different local time distributions of irregularity occurrences for the two solstices.

1. Introduction

[2] It has been stated that an enhanced eastward electric field will be produced from the electrodynamical interaction of the eastward blowing thermospheric wind with the geomagnetic field around the dip equator at the sunset terminator where a large westward longitudinal gradient in the flux tube integrated E region Pedersen conductivity exists between the high-conducting dayside ionosphere and the low-conducting nightside ionosphere [Rishbeth, 1971, 1977, 1981; Heelis et al., 1974; Batista et al., 1986]. This enhanced eastward electric field will elevate the postsunset ionosphere to a high altitude to accelerate the growth of the generalized gravitational Rayleigh-Taylor (R–T) instability to result in the equatorial spread F (ESF) density irregularities [Farley et al., 1970; Haerendel, 1973; Zalesak and Ossakow, 1980; Ossakow, 1981; Abdu et al., 1982; Aarons, 1993; Fejer et al., 1999]. Since the longitudinal gradient of the Pedersen conductivity can be resulted from the relative sunset times at conjugate E regions which is calculated from the alignment of the sunset terminator with the local magnetic flux tube, the seasonal/longitudinal (s/l) distributions of ESF density irregularity occurrence rates have been explained by a relationship between the field line declination and the sunset terminator [Abdu et al., 1981, 1992, 2006; Tsunoda, 1985]. Therefore, we would expect a mirror symmetry existed in the longitudinal distribution of irregularity occurrences between June and December solstices if the dipole axis of the Earth is merely tilted with respect to the Earth rotation axis. However, the observed s/l distributions of irregularity occurrences indicate no such longitudinal mirror symmetry existed between the two solstices [Kil and Heelis, 1998; McClure et al., 1998; Hei et al., 2005; Su et al., 2006], because the Earth dipole axis is slightly displaced from the Earth rotation axis in addition to the tilt. This dissimilarity can also be noticed in the global averaged local time (LT) distribution of ESF density irregularity occurrences between the two solstices shown in Figure 1, which is replotted from a previously published report [Su et al., 2006, Figure 10].

Figure 1.

ROCSAT observation of local time distribution of equatorial density irregularities in 1999–2004 during magnetic quiet times.

[3] Figure 1 shows the global averaged density irregularity occurrence distribution observed by ROCSAT-1 at the 600-km altitude in the equatorial region between ±15° in dip latitudes during magnetic quiet times (Kp < 3) for the four seasons. Although Figure 1 shows the seasonal variations of irregularity occurrences in the LT distribution for four seasons, we will only focus on the different occurrence distribution between June and December solstices. The occurrence of postsunset irregularities is noticed to begin at 1800–1900 LT sector for every season, but it increases at a faster rate to peak at 2100–2200 LT sector during the December solstice (as well as in the equinoxes) in contrast to a slower increase rate for the June solstice in which the peak occurs at 2200–2300 LT sector, one hour later than the peak hour in the December solstice. The cumulative occurrence rate of irregularities is also smaller during the June solstice than in the December solstice. The different LT distribution of irregularity occurrences and the difference in the total number of irregularity occurrences between the two solstices imply that the postsunset ionospheric conditions are different for the two solstices in triggering the R–T instability and accelerating the instability growth rate.

[4] In this report, we present the observational facts to support a conjecture that a slower increase rate in the irregularity occurrence and the delay in the time of peak occurrence during the June solstice are related to a large geomagnetic field strength existed at the longitude region of positive magnetic declination to hinder the R–T instability process when the best postsunset ionospheric condition is supposed to exist at that longitude region during the June solstice. Using the evening plasma vortex as the perturbation seed to trigger the R–T instability [e.g., Kudeki and Bhattacharyya, 1999; Yokoyama et al., 2004; Su et al., 2008], we present the ROCSAT observations of zonal flow reversal in the dusk sector as the reference time of triggering the R–T instability during the two solstices. The observed vertical drifts at longitudes of large magnetic declinations are then used to refer the height of the postsunset ionosphere for the instability growth rate between the two solstices. The relationship between the time of zonal drift reversal, the vertical drift and the magnetic field strength at longitudes of large magnetic declinations in the two solstices is compared to reach a conclusion for the result of Figure 1.

2. ROCSAT Observations

2.1. Zonal Drifts

[5] The magnetic eastward plasma drift (or the zonal drift hereafter) observed by ROCSAT-1 is constructed from two measured horizontal flow components that are perpendicular to each other. One comes from the in situ measurement of the horizontal flow velocity across the satellite track in the horizontal Drift Meter (HDM), and the other is the derived ram flow velocity from fitting the current-voltage (I–V) data taken by the Retarding Potential Analyzer (RPA). The HDM sensor that measures the horizontal flow velocity has been cross-calibrated with the vertical DM (VDM) sensor that measures the vertical flow velocity. First, the VDM-measured vertical flows are calibrated regularly in the postflight analysis to remove the offset in such a way that the longitudinal average of the diurnal variation in the vertical drift velocities at the dip equator is zero during geomagnetic quiet times in one season. Then in four weeks in 2000, the VDM sensor and the HDM sensor were switched in operation so that HDM sensor that was originally measuring the horizontal drift velocity is changed to measure the vertical drift velocities. The bias in the vertical drift velocities measured by the HDM sensor is analyzed in the same procedure as were done for the measurements of the VDM sensor. The difference between the offsets in calibrating the VDM sensor and calibrating the HDM sensor is noted, and is assumed to be unchanged during the ROCSAT mission period so that the offset in the HDM-measured data is obtained as the vertical drift velocities measured by the VDM sensor are continuously calibrated.

[6] On the other hand, the derived ram flow velocity from the least squares fit to the I–V curve obtained by the RPA senor can have 0.5–1% uncertainty in the fitting parameters, and this uncertainty is equivalent to a possible bias of 37.8–75.6 m/s in the ram velocity measured in the Earth-fixed coordinate system based on the spacecraft velocity of 7.56 km/s. When this ram component is converted to the zonal drift using Vequation image = Vram cos i with i = 35° around the dip equator, a bias of 31–62 m/s could exist. We are uncertain if such a bias really exists in the zonal drifts. However, we can check the ROCSAT result taken at longitudes 280–300° with the radar measurements at Jicamarca Radio Observatory (JRO) near Lima, Peru (11°57′S, 76°52′W; magnetic dip 2°N).

[7] Before showing the comparison, the most important point of quiet time zonal drift characteristics from previous reports [Coley and Heelis, 1989; Fejer et al., 1991, 2005] is recapitulated in the following. The nighttime zonal drift peaks at about 100 m/s in contrast to the typical daytime westward drift of 40 m/s. That is, the nighttime eastward drift is about twice as large as the daytime westward drift, and this nighttime drift increases with solar activity. The reversal of zonal drift from westward to eastward occurs around the dusk sector but before the prereversal enhancement (PRE) of the vertical drift. A typical example of the zonal drift measurements from the JRO is copied in Figure 2a [Fejer et al., 2005] to compare with the ROCSAT observation at longitudes 280–300° in Figure 2b. Figure 2b shows the median values of all 1-s data points in one ROCSAT pass within ±5° in dip latitudes in a 1-h LT sector. The averages of these median values are connected with the solid lines, while the upper and lower bounds of the standard deviation of the mean values are connected with the dashed lines to show the envelope of possible variations in the zonal drift. Notice that possible bad data points exist from 0400 to 0800 LT when the ion density is low during the sunrise hours; therefore, this section of the data should be ignored and we use the dotted-line connections in this time period. For the rest of the other time periods, the data indicates that the averaged maximum of the westward drift during the day is about 40–60 m/s, while the averaged maximum of the nighttime eastward drift is about 150–200 m/s. These values seem to fall in line with the expected zonal drifts at the topside ionosphere and are in good agreement with the zonal drifts measured around the F peak at JRO during the high solar activity period (F10.7 = 160) in Figure 2a. Thus no systematic bias of the ROCSAT zonal drifts seems to exist in the averaged value. Even if it exists, the bias could be smaller than the expected uncertainty values.

Figure 2.

(a) Zonal drifts observed at Jicamarca Observatory Radar [Fejer et al., 2005]. (b) Zonal drifts observed by ROCSAT at 600-km altitude. The average and the variances are connected by the solid and dashed lines in Figure 2b, respectively. Data points from 0400 to 0800 local time in Figure 2b are ignored and are connected by the dotted lines.

[8] The ROCSAT observed zonal drifts in Figure 2b indicate one difference from the results of the JRO measurements in Figure 2a. That is, the ROCSAT data indicate some seasonal variation of zonal drift reversal in local time, while the JRO observations indicate none. Although a large fluctuation exists in the ROCSAT observed zonal drifts as indicated by the upper and lower dashed lines, the averaged zonal drift reversal in local time during the December solstice is at 1520 LT, in the September equinox at 1610 LT, in the March equinox at 1620 LT, and in the July solstice at 1640 LT. The exact time of reversal may vary as indicated by scatters of the data points in each plot (season) of Figure 2b, but it seems that, on the average, the time of zonal drift reversal occur earlier in the December solstice than that in the June solstice. The difference in the solstitial variations of zonal drift reversal observed by JRO and by ROCSAT will be discussed in section 3.

2.2. Global Distribution of Zonal and Vertical Drifts in Local Time

[9] The solstitial variations of the average equatorial vertical and zonal drifts in local time at several longitude regions observed by ROCSAT in 2000–2004 are plotted in Figure 3. We have chosen six longitude regions to represent the global ionospheric density variation and electrodynamical effect. The six longitude regions are chosen to be at longitudes 20–40° (N, δ∼0), 100–120° (N, δ∼0), 150–170° (N, δ > 0), 190–210° (E, δ > 0), 280–300° (S, δ < 0), and 310–330° (E, δ < 0). The letters N, E and S in the parentheses represent the location of the dip equator in the northern hemisphere, at the geographic equator, and in the southern hemisphere, respectively. The geographic location of the dip equator either in the northern or southern hemisphere will modify the solar zenith angle effect to alter the ionospheric property in a season. The symbol δ represents the magnetic declination angle which will affect the interhemispheric ion flow due to neutral winds to shape different hemispherical ionospheric structures. It will also alter the longitudinal gradient in the postsunset ionospheric conductivity to affect the postsunset plasma drifts and irregularity occurrences. Thus these six longitude regions should adequately represent the seasonal variation of global ionospheric plasma and electrodynamical effects (see Figure 5).

Figure 3.

Solstitial variation of quiet time averaged vertical and zonal drifts in local time observed by ROCSAT at six longitude regions. Note that the magnitude of the vertical drift velocity has been multiplied by a factor of 3.

[10] The averaged zonal drift in the plots of longitudes 280–300° in Figure 3 is the same as the one shown in Figure 2b. For clarity of the results in Figure 3, we did not plot the error bars for either vertical or zonal drift. Furthermore, because the vertical and zonal drifts are quite different in magnitude, we have multiplied by a factor of 3 in the vertical drift in Figure 3 to enhance the contrast of the local time variation in the vertical drift. Figure 3 clearly indicates the local time variations of two equatorial drift velocities in the two solstices at every longitude region. However, it also indicates less smoothness in the averaged vertical drift compared to a recent result reported by Fejer et al. [2008] in which same ROCSAT data set is used. The difference from the report of Fejer et al. [2008] occurs because their results are the model derivations from the ROCSAT data, whereas our report shows the averaged 1-h ROCSAT data. We did not adopt the vertical plasma drift model from the report of Fejer et al. [2008] for the reason that the seasonal/local time variations of the zonal drifts are needed in this study and are not available from the model yet. Thus for the consistency of our presentation, both vertical and zonal drifts are obtained from the same averaging process of the ROCSAT data. Although the averaged vertical drift at 1800–1900 LT shown in Figure 3 may miss the true peak of the postsunset PRE obtained in the model calculation of Fejer et al. [2008], the s/l variations of the vertical drifts shown in Figure 3 has the similar variations with that in the report of Fejer et al., and they are adequate for the study in the current report.

[11] The local time variations of zonal drift reversal from westward to eastward in the two solstices at six longitude regions are obtained from Figure 3 and are plotted in Figure 4. The averaged time of zonal drift reversal and the range of the variation in the time of reversal are indicated, respectively, by a capital letter and the error bar over it in Figure 4. We use the capital letter to represent the season, such as J for the June solstice and D for the December solstice. The errors in the time of zonal drift reversal are obtained between the times of zonal drift reversal indicated by the envelope of the variations shown in Figure 2b. The local hours of zonal drift reversals are given in the left-hand side coordinate. In Figure 4, the solstitial variations of the averaged vertical drift velocity at 1800–1900 LT sector are identified by the script capital letter equation image for the June solstice and equation image for the December solstice. Possible variations in the hourly averaged vertical drift are also indicated by the error bars. The magnitudes of the vertical drift are given in the right-hand side coordinate.

Figure 4.

Averaged zonal drift reversals in local time at different longitudes are indicated by the letters J for the June solstice and D for the December solstice. The averaged vertical drifts at different longitudes are indicated by the script letters equation image for the June solstice and equation image for the December solstice.

[12] The solstitial variation of zonal drift reversal in Figure 4 is examined first. It is noticed that the averaged zonal drift reversal in the June solstice always occurs at a local time that is later compared to the time of zonal drift reversal in the December solstice at every longitude region. Most of the zonal drift reversals during the June solstice occur around or after 1800 LT, with the earliest one occurring at 1730 LT at longitudes 280–300°. In contrast, the averaged reversal time of the zonal drift in the December solstice occurs earlier than 1700 LT. Even taking all the possible variations in the time of zonal drift reversal into consideration for the two solstices, we can still state that except at longitudes 190–210°, the zonal drift reversal at all other longitude regions in the June Solstice occurs later compared to that in the December solstice. At longitudes 190–210°, the large error bar in the time of zonal drift reversal for the December solstice overlaps most of the times of zonal drift reversal observed in the June solstice so that no definitive conclusion can be reached.

[13] As for the longitudinal variation of the vertical drift, no contrasting observation is noticed between the two solstices when possible errors in the vertical drifts are included for the comparison, except at longitudes 280–300° and 310–330°. At these two longitude regions of negative magnetic declination, large contrasting variations in both vertical and zonal drifts are observed between the two solstices. It is noticed that during the June solstice, the time of zonal drift reversal is late and the vertical drift is very small. In contrast, an earlier time of zonal drift reversal is observed to accompany a large vertical drift during the December solstice. Thus it seems that the magnetic declination can be used as an indicator for the existence of a correlation between the zonal and vertical drifts in a season at longitudes of negative magnetic declination. That is, an earlier occurrence of zonal drift reversal in local time seems to have a causal relationship with the subsequent larger vertical drift during the December solstice. Similarly, a late occurrence of zonal drift reversal results in a smaller vertical drift during the June solstice. We would expect a similar correlation existed between the two drifts at longitudes 150–170° and 190–210° where the magnetic declination is positive in the two solstices. That is, an earlier occurrence of zonal drift reversal should be observed to accompany a large vertical drift at longitudes of positive magnetic declination during the June solstice. On the contrary, a late zonal drift reversal is observed in the June solstice compared to that in the December solstice, and the vertical drift observed during the June solstice is no larger than that in the December solstice. Dissimilarities in the variations of the two drifts at longitudes of positive and negative magnetic declinations during the two solstices will play the key factor to result in the contrasting local time distributions of irregularity occurrences in the two solstices seen in Figure 1 and this is discussed in section 3.

3. Discussion

3.1. Relationship Between the Time of Zonal Drift Reversal and the Magnitude of Vertical Drift

[14] Studies of the postsunset equatorial vertical and zonal drifts indicate that a close relationship exists between the two drifts through the effects of the zonal wind and the Hall and Pedersen conductivity variations in the E and F regions [Haerendel et al., 1992; Crain et al., 1993a; Eccles, 1998a]. The two drift velocities in the low-latitude region are obtained in the following process: first, the vertical electric field (Ep) and horizontal electric field (Eϕ) are solved from the current continuity equation given by [e.g., Haerendel et al., 1992; Crain et al., 1993a]

equation image

with

equation image

and

equation image

The subscripts p and ϕ represent the components in the curvilinear dipole coordinate system, where p is directed perpendicular to equation image in the meridional plane, q is along equation image, and ϕ is the longitude. The other subscripts P and H in the conductivity σ indicate Pedersen and Hall conductivity, respectively. In general, the electric potential ψ in equation image = −∇ψ is substituted in equations (1) and (2) to solve the current continuity equation using the whole flux tube as one unit in a two-dimensional model where no field-aligned current Jq exists. The Pedersen and Hall conductivities are calculated from an ionospheric model and so is the neutral wind equation image. The vertical and zonal drifts Vp and Vϕ obtained from equation image = equation image/B2 can be shown as, by rewriting equations (1) and (2),

equation image

and

equation image

The modeled results of the postsunset ionospheric plasma drift velocities have been reported in the literature [e.g., Heelis et al., 1974; Farley et al., 1986; Haerendel et al., 1992; Haerendel and Eccles, 1992; Crain et al., 1993a, 1993b; Eccles, 1998a, 1998b; Fesen et al., 2000; Millward et al., 2001; Vichare and Richmond, 2005]. However, in almost all published reports, only the local time variation of the vertical drifts has been presented. The seasonal/local time variations of two drift velocities are not available for comparison with the ROCSAT observations during the June and December solstices. We could study the s/l distributions of the vertical and zonal drifts in local time variation by solving equations (1) and (2), but this has not been done yet. However, the local time variation between the two drifts can be examined from equations (3) and (4) with reference to previously published results [e.g., Heelis et al., 1974; Farley et al., 1986; Crain et al., 1993a]. First, it is understood from the published results that the existence of a large gradient in σP across the sunset terminator will induce a large zonal electric field equation image to result in a large postsunset PRE vertical drift. A large gradient in σP across the sunset terminator can occur at longitudes of large magnetic declination during a solstice season. Indeed, a large PRE vertical drift has been observed, as expected, at longitudes of negative magnetic declination at 280–300° and 310–330° during the December solstice. On the other hand, no such large vertical drifts are observed at longitudes of positive magnetic declination at 150–170° and 190–210° during the June solstice. Failure to achieve a high vertical drift at longitudes of positive magnetic declination during the June solstice will become clear in the discussion section of 3.3. Second, the ratio σH/σP is known to decrease rapidly from a value around unity in the E region to become negligible above 250 km, so that the zonal drift Vϕ varies approximately with the zonal wind Uϕ above the middle ionosphere as seen in equation (3) [Heelis, 2004; Eccles, 1998a]. However, when the first term (−σHEϕ/σPB) in equation (3) is larger than the second term Uϕ at the lower ionosphere, Vϕ will flow in the opposite direction of the zonal wind Uϕ. A shear in Vϕ between the low and middle ionospheres will result in a vortex flow in the evening ionosphere [e.g., Haerendel et al., 1992; Kudeki and Bhattacharyya, 1999; Eccles et al., 1999; Heelis, 2004]. The existence of this vortex flow is thought to cause wave structure perturbations and initiate the R–T instability for density irregularity occurrence in the postsunset ionosphere. Since the existence of an evening plasma vortex between the lower and middle ionospheres will only occur after the zonal plasma drift has reversed from westward to eastward in the upper ionosphere, we use the time of ROCSAT observed zonal drift reversal as the reference time for the initiation of the R–T instability. The results in Figure 4 then indicate that the triggering time for the R–T instability during the June solstice is always later than that in the December solstice, and occurs most often after 1800 LT.

[15] Furthermore, close coupling between Vϕ and Vp in equations (3) and (4) through the equations ∇ · equation image = 0 and curl equation image = 0 will lead to a result such that the peak of PRE vertical drift Vp occurs after the reversal of Vϕ in local time as has been reported by Crain et al. [1993b]. The ROCSAT data shown in Figure 3 indicates that large vertical drifts appear, indeed, after the reversal of zonal drift Vϕ at all six longitude regions. In addition, the magnitude of PRE vertical drift seen in Figure 4 further indicates that it will have a smaller value when the zonal drift reversal occurs late in the evening sector, and this is especially noticeable at longitudes of negative magnetic declination in the June solstice.

3.2. Seasonal Variation of Zonal Drift Reversal in Local Time

[16] As the zonal drift reversal at the upper ionosphere is controlled mainly by the local time variation of the zonal wind Uϕ; and the zonal wind Uϕ has not been observed to have a seasonal variation, then what can cause the solstitial variation in the time of zonal drift reversal? From equation (3), we see that the effects from the first term (−σHEϕ/σPB) and the last term (−Jp/P) can describe some seasonal variation of zonal drift reversal at the upper ionosphere. Because of a rapid decrease of the ratio σH/σP to almost nil above the altitude 250 km, only the effect of the last term (−Jp/P) is examined for comparison with the ROCSAT observation at the 600-km altitude. First, Jp takes a value from the diverted upward flow at the base of the ionosphere from the boundary input of Jϕ in the equatorial electrojet (EEJ). A finite EEJ can exist in the nighttime ionosphere after 1800 LT if a finite conductivity value exists to sustain the horizontal current density Jϕ across the sunset terminator. Therefore, observation of EEJ after 1800 LT can be used as a reference for the variation of the postsunset plasma drift at the upper ionosphere. Simulation results from the Thermosphere Ionosphere Electrodynamics General Circulation Model (TIEGCM) reported by Doumbia et al. [2007] for comparison with the ground magnetometer observations at the African sector (∼4°W) indicate that a slow decay of EEJ exists beyond 1800 LT near the dip equator during the June solstice compared to an earlier and complete disappearance before 1800 LT in the March equinox and the December solstice. We will adopt their simulation results of the seasonal variation of EEJ after 1800 LT to explain the solstitial variation of zonal drift reversals in the current report. Because of the existence of negative magnetic declination at the African sector (see Figure 5), an eastward EEJ can appear after 1800 LT due to a finite conductivity existed in the postsunset ionosphere from the asynchronous sunsets at conjugate E regions at that longitude region during the June solstice. An eastward EEJ exists after 1800 LT leads to a small upward vertical current that will yield a small positive value of the flux tube-integrated Jp/σP to move the reversal of zonal drift Vϕ to slightly after the reversal of zonal wind Uϕ as seen in equation (3). In contrast, when EEJ disappears around 1800 LT during the December solstice, the flux tube–integrated Jp/σP value will be almost nil so that the reversal of zonal drift Vϕ will occur at the same local time as the reversal of zonal wind Uϕ. Thus the zonal drift reversal during the June solstice should be observed at a later time than that in the December solstice. Indeed, the ROCSAT observations shown in Figure 4 indicate that the time of zonal drift reversal at longitudes 20–40°, near the African sector, in the June solstice is at a later local time than that in the December solstice.

Figure 5.

(top) Longitudinal variation of geomagnetic field strength at 300-km altitude dip equator and (bottom) the magnetic declination angle. The annotated shaded and hatched bars at the bottom indicate the geographic locations of the dip equator in the northern and southern hemispheres, respectively.

[17] In addition, a prolonged existence of an eastward EEJ after 1800 LT implies that the conductivity gradient is small across the sunset terminator so that only a small zonal electric field will appear at the base of the ionosphere to result in a small PRE vertical drift. Therefore, a delay in the zonal drift reversal should accompany with a small vertical drift, and an earlier reversal in the zonal drift results in a larger vertical drift. The ROCSAT observations at longitudes 20–40° seen in Figure 4, though not completely fit this picture because of large uncertainties existed in the vertical drifts in the two solstices, yet they did not show the contradiction either. A stronger support of this relationship between the zonal and vertical drifts is observed at longitudes 280–300° and 310–330° where the magnetic declination is negative. This relationship then becomes less clear at longitudes of positive magnetic declination at 150–170° and 190–210°. The cause of failing to have this simple relationship between the time of zonal drift reversal and the magnitude of the vertical drift at longitudes of positive magnetic declination is discussed in section 3.3.

[18] Furthermore, it is noted that the conductivity value in σP in the last term in equation (3) can also affect the time of zonal drift reversal. The dissimilarity in the ionospheric density distribution existing between the two solstices is well known. A large so-called annual asymmetry exists between the two solstices with a larger ionospheric density occurring in the December solstice than in the June solstice [e.g., Mendillo et al., 2005; Rishbeth and Mueller-Wodarg, 2006; Liu et al., 2007]. A larger density existed during the December solstice will result in a larger flux tube–integrated conductivity value in σP to reduce the effect of (−Jp/P) in moving the time of zonal drift reversal to occur earlier. Since the s/l variations of the flux tube integrated σP are not available at the moment, we will not evaluate their impacts on the s/l variations of zonal drift reversal in this report, even though the ROCSAT observations do indicate a higher density level at the 600-km dip equator during the December solstice than in the June solstice.

[19] Finally, it should be noted that the zonal drift can vary with altitude as has been observed by DE-2 in a report by Coley and Heelis [1989]. Thus the difference noted between the seasonal variations of zonal drift reversal observed by ROCSAT shown in Figure 2b and by JRO shown in Figure 2a could be caused by the altitudinal variation in the zonal drift. This difference could also be caused by the fact that the ROCSAT result includes all the observations taken during the moderate to high solar activity years of 2000 to 2004 in which some solar activity effect on the zonal drift reversal could be included. We have included all the ROCSAT observations made over JRO to have a better statistics in the averaged value to compare with the JRO observations. This is also consistent in our data analysis when the result in Figure 4 is used to explain the result in Figure 1 in which all the quiet time data from 1999 to 2004 were used. The impact from the uncertainty in the exact timing of zonal drift reversal in our scenario to explain the seasonal variation of irregularity occurrences in two solstices is discussed in section 3.3.

3.3. Cause of Different Behavior in the Plasma Drifts and Irregularity Occurrences Between the June and December Solstices

[20] The s/l distributions of postsunset irregularity occurrences have been explained by the ionospheric electrodynamical effect resulted from the alignment of the sunset terminator with the magnetic flux tube [e.g., Tsunoda, 1985; Su et al., 2006, 2008]. Under such circumstances, at longitudes of negative magnetic declination such as at 280–300° and 310–330° in Figure 4, we should observe a large occurrence of irregularities during the December solstice. Indeed, the irregularity occurrence rates are the largest at longitudes of negative magnetic declination during the December solstice [e.g., McClure et al., 1998; Su et al., 2006, 2008]. In this report, we propose a scenario to describe a sequence of physical processes that result in this high occurrence rate of irregularities at longitudes of negative magnetic declination during the December solstice. It goes like this: an earlier occurrence of zonal drift reversal indicates an earlier appearance of perturbation seed to initiate the R–T instability. As the instability occurs, the postsunset ionosphere that has been raised high by a large vertical drift velocity will yield a large growth rate of the instability to indicate an earlier occurrence of the irregularity. This will happen at longitudes 280–300° and 310–330° during the December solstice as seen in Figure 4. Therefore, we have a rapid increase in the irregularity occurrence rate and a large number of irregularity occurrences is resulted at longitudes of negative magnetic declination during the December solstice. However, no similar situation of such high irregularity occurrence rate was observed at longitudes of positive magnetic declination, such as at 150–170° and 190–210° during the June solstice. The observed irregularity occurrence rates during the June solstice at these longitudes were high but not as conspicuously high as that at longitudes of negative magnetic declination during the December solstice [e.g., Su et al., 2006, 2008]. Failure to achieve the highest irregularity occurrence rate at longitudes of positive magnetic declination during the June solstice can be explained by the same scenario. That is, a delay in the time of zonal drift reversal existed at these longitudes during the June solstice implies a delay to have perturbation seeds for the R–T instability initiation. The delay in initiating the instability then misses the optimal postsunset ionospheric condition to have a high instability growth rate. More importantly, the fact that not-so-high vertical drift velocities are observed at these longitudes during the June solstice further indicates that the postsunset ionosphere is not so high to have the maximum instability growth. Therefore, we fail to have the highest irregularity occurrence rate at longitudes of positive magnetic declination where the irregularity should be most likely to occur during the June solstice. Consequently, the global averaged irregularity occurrence rate increases at a slower pace and the number of irregularity occurrences is smaller during the June solstice than that in the December solstice as seen in Figure 1.

[21] What can account for the differences in the zonal and vertical drifts between the June and December solstices seen in Figure 4? We think that the equatorial geomagnetic field is the cause for the different postsunset plasma drifts at longitudes of positive and negative magnetic declinations. If the dipole axis is merely tilted with respect to the geographic axis, then the magnetic declination and the ionospheric seasonal effects at the dip equator should be mirrored between June and December solstices with a simple shift of 180° in longitude. The electrodynamical effect of the postsunset vertical drift and the resultant irregularity occurrence will be almost the same if we ignore the small difference in the solar flux input due to different Sun-Earth distances in the two solstices. However, because of the displacement of the magnetic dipole axis with respect to the geographic axis, the magnetic field strengths are different between the longitudes of positive and negative magnetic declinations. We show the longitudinal variations of the geomagnetic field strength at the 300-km altitude of the dip equator in Figure 5 (top) and the magnetic declination angle in Figure 5 (bottom). In addition, we have added the shaded and hatched regions at the bottom to indicate the locations of the dip equator in the northern and southern hemisphere, respectively. This particular annotation will show the seasonal variation of the latitudinal effect at the dip equator for the postsunset ionospheric density level that affects the flux tube integrated Pedersen conductivity. It is noted in Figure 5 that the longitudinal variation of the magnetic field shows a good contrast between the longitudes 150–170° and 190–210° where the magnetic declination is positive and the longitudes 280–300° and 310–330° where the magnetic declination is negative. The magnetic field strength at longitudes 150–170° and 190–210° is about 20% larger, on the average, than that at longitudes 280–300° and 310–330°. In addition, the magnetic declination angle at longitudes 150–170° and 190–210° is about 10° compared to the declination angle of ∼−20° existed at longitudes 310–330°. This implies that the sunset terminator will have a much better alignment with the local magnetic flux tube at longitudes 310–330° during the December solstice than with the local magnetic flux tube at longitudes 150–170° and 190–210° during the June solstice. Consequently, the longitudinal gradient of Pedersen conductivity is much larger across the sunset terminator at longitudes 310–330° during the December solstice than that at longitudes 150–170° and 190–210° during the June solstice. This will greatly affect the time of zonal drift reversal and the magnitude of the PRE vertical drift at these two longitude regions between the two solstices.

[22] Thus when we overlay Figure 4 on top of Figure 5, the effect of the magnetic field on the longitudinal variations of zonal drift reversal, the magnitude of the PRE vertical drift, and the occurrences of postsunset density irregularities will become clear. During the December solstice, the dip equator around longitude 310° has a larger seasonal density level because it is located in the southern hemisphere. The magnetic declination is negative so that the postsunset ionospheric electrodynamics has a maximum effect to result in a maximum PRE vertical drift (Eϕ/B) from a low magnetic field existed at these longitudes. An earlier reversal of zonal drift appears to initiate the R–T instability process and the large PRE vertical drift raises the ionosphere high enough to accelerate the instability growth rate to result in a maximum number of irregularity occurrences at these longitudes. In contrast, a high magnetic field strength appears halfway across the globe at the dip equator around longitude 160° where the magnetic declination positive. Thus even with the identical values of the electric fields derived in equations (1) and (2) for the two solstices, the resultant vertical drift (Eϕ/B) will be smaller at longitudes 160° during the June solstice than that at longitude 310° during the December solstice. When the time for the postsunset zonal drift reversal is further delayed during the June solstice for the instability initiation, the vertical drifts as well as the irregularity occurrences at longitude 160° will be greatly reduced in comparison with that at longitude 310° during the December solstice. A smaller number of irregularity occurrences are thus resulted during the June solstice as seen in Figure 1. The importance of the magnetic field strength on the longitudinal variation of postsunset vertical drifts in an equinox season has been reported by Vichare and Richmond [2005]. In the current report, the effect of the longitudinal variation of the magnetic field to cause different postsunset irregularity occurrences between two solstices is further demonstrated with the ROCSAT observation.

[23] As for the specific effect of the magnetic field on the time of zonal drift reversal, we could again check with the s/l variations of the term (−Jp/P) in equation (3). However, we note that although the magnetic field strength B has a ∼30% variation between the longitude regions around 160° and around 310° as seen in Figure 5, the value Jp after 1800 LT (derived from Jϕ across the sunset terminator) can have a larger variation from about nil to some finite value depending on the longitudinal gradient of Pedersen conductivity due to the s/l variations of the magnetic declination effect. Thus the ratio (−Jp/P) may not vary simply with the magnetic field strength alone but from a coupling between equations (3) and (4) that has not been studied in this report. However, the ROCSAT observation indicates that an earlier zonal drift reversal exists at longitudes 280–300° and 310–330° as seen in Figure 4 during the December solstice. Thus the value Jp after 1800 LT should have assumed an almost nil value at the longitude of negative magnetic declination as it should be during the December solstice. In contrast, the value Jp after 1800 LT should be finite at the same longitude regions during the June solstice so that (−Jp/P) is large in negative value and the time of zonal drift reversal is greatly delayed. As for the value Jp after 1800 LT at longitude 150–170° or 190–210° during the June solstice, it seems that the value is not close to nil because the magnetic declination is about 10° which did not make a good alignment with the sunset terminator that is aligned about 23° off the Earth rotation axis during the June solstice. Since the magnetic field strength is large and the plasma density is high during the June solstice at these longitudes, the term (−Jp/P) could have a similar value as that at longitudes 280–300° and 310–330°. Thus the time of zonal drift reversal at longitudes 150–170° or 190–210° is observed to be about the same as that at longitudes 280–300° and 310–330° during the June solstice. On the other hand, the relatively earlier times of zonal drift reversal during the December solstice at longitudes 150–170° and 190–210° than that in the June solstice is hard to explain. One possible explanation is that the ratio Jp/σP is similar for the two solstices at longitudes of positive magnetic declination. Such conjecture seems reasonable from the fact that the vertical drift which is closely related to the time zonal drift reversal also indicates similar values in the two solstices.

[24] Finally, we would like to mention that the uncertainty in the time of zonal drift reversal that serves as the existence of perturbation seeds to initiate the R–T instability will not alter our conclusion on the cause of the seasonal variation of irregularity occurrences seen in Figure 1. Since the existence of perturbation seeds in an instability process only serves as a necessary condition for the instability occurrence, the initial time of perturbation seed appearance is not critical as long as the seed has existed in an instability process. The postsunset ionospheric height indicated by the magnitude of the vertical drift is the key factor to determine the instability growth rate and hence the irregularity occurrence rate as well as the number of irregularity occurrences. Therefore, even if a large uncertainty exists in the observed times of zonal drift reversals to indicate that the postsunset R–T instabilities at different longitude regions in different seasons could all have been triggered around the same local time, the magnitude of the vertical drift at different longitudes can still determine a different outcome of the irregularity occurrence in different season. Under such circumstances, the longitudes of negative magnetic declination that have the largest postsunset vertical drift during the December solstice seen in Figure 4 should still have a larger number of irregularity occurrences. In contrast, the longitudes of positive magnetic declination during the June solstice that has a smaller vertical drift will have a smaller number of irregularity occurrences than that at longitudes of negative declination in the December solstice. Because the irregularity occurrences during the June solstice have missed the opportunity to have the largest occurrence number at longitudes of positive magnetic declination, the global averaged irregularity occurrence is then the smallest among the four seasons as seen in Figure 1. Since a close relationship exists between the vertical and zonal drift, we have included the averaged time of zonal drift reversal observed by ROCSAT in the scenario of a physical process to discuss the result of Figure 1.

4. Conclusions

[25] Comparing the ROCSAT observations in the zonal drift reversal, the magnitude of PRE vertical drift, and the irregularity occurrence variation in local time during the December solstice with that in the June solstice, we conclude that an earlier time of zonal drift reversal in association with a large vertical drift has resulted in a large number of postsunset equatorial density irregularity occurrences at longitude of negative magnetic declination during the December solstice. In contrast, a late reversal of zonal drift that results in a smaller postsunset vertical drift, and consequently a smaller number of irregularity occurrences is observed at longitudes of positive magnetic declination during the June solstice. This implies that the postsunset equatorial ionospheric irregularity occurrences are related to three driving factors: They are the geographic location of the dip equator to affect the ionospheric seasonal density variation, the magnetic declination angle to affect the longitudinal gradient of the ionospheric conductivity across the sunset terminator, and the strength of the geomagnetic field at the dip equator to drive the overall electrodynamics. Each individual factor to affect the postsunset ionospheric electrodynamical effect has been mentioned in many previous reports, but the three factors are put together in this report to explain the s/l variations of PRE vertical drift in relation to the zonal drift reversal, and hence to determine the seasonal variation of irregularity occurrences in Figure 1. Although we do not have a theoretical model to support our conjecture, we believe that these three factors have caused different global/seasonal/local time postsunset equatorial electrodynamics that can be used to explain the close relationship between the time of zonal drift reversal, the magnitude of vertical drift, and the occurrence frequency of the postsunset irregularities. Under such circumstances, there seems no need to add other factors such as perturbation seeds from atmospheric disturbances to initiate the R–T instability process. It is hope that the current report will stimulate further theoretical study on the s/l variations of postsunset vertical and zonal drifts in relation to the irregularity occurrences.

Acknowledgments

[26] The work was supported, in part, by NSC97-2111-M-008-013 from the National Science Council of the Republic of China and, in part, by a grant from the Asian Office of Aerospace Research and Development (AOARD) of the U.S. Air Force Office of Scientific Research (AFOSR), AOARD-07-4098. We are grateful to many NCU ROCSAT/IPEI team members for their efforts in processing the ROCSAT/IPEI data. The editorial assistance by Clia Goodwin in correcting the English writing in the first draft and the critical comments by the reviewers are also acknowledged.

[27] Amitava Bhattacharjee thanks the reviewers for their assistance in evaluating this paper.