Analysis of TEC data from the TOPEX/Poseidon mission

Authors


Abstract

[1] TOPEX/Poseidon mission has provided an extensive database of vertical TEC over the ocean since August 1992. Data from nearly 10 years of TOPEX TEC observations were analyzed to study the TEC climatology. First, TEC data were binned by season, geomagnetic activity, and solar activity to create longitudinally averaged TEC maps in magnetic latitude and local time. These maps show the annual and semiannual anomalies well known from climatological studies of NmF2 but lack the seasonal anomaly because of the longitudinally averaged binning. The equatorial anomaly is the most prominent feature in the maps, and they show strong TEC variations with solar activity but relatively weak variations with geomagnetic activity in our three Kp bins. Compared with the low solar activity conditions (F10.7 < 120), the TEC values for F10.7 ≥ 120 are much larger and the equatorial anomaly lasts longer into the night, up to midnight. During geomagnetically active periods, the TEC maps generally show a noticeable increase at low latitudes for F10.7 < 120, but this effect is barely detectable for F10.7 ≥ 120. Finally, three longitudinal bins (Indian, Pacific, and Atlantic) were added in order to see how the TEC morphology varies with longitude. The TEC measurements display strong longitudinal variations that closely follow the longitudinal variation of the magnetic declination. In the southern Pacific, where the declination is positive and large, the diurnal TEC variations significantly differ from those in the other longitude sectors, where the magnetic declination is negative in the Southern Hemisphere. Also, at noon, the phase of the longitudinal TEC variation is typically opposite to that at midnight.

1. Introduction

[2] The TOPEX/Poseidon mission, launched on 10 August 1992 [Fu et al., 1994], is a joint mission of the National Aeronautics and Space Administration (NASA) and the French space agency, Centre National d'Etudes Spatiales (CNES). In 1979, the Jet Propulsion Laboratory (JPL) of NASA started to plan a mission called the Ocean Topography Experiment (TOPEX) that would utilize a satellite equipped with an altimeter to measure the surface of the global ocean. Meanwhile, the CNES was designing an oceanographic mission called Poseidon, named for the Greek god of the sea. In order to achieve common scientific goals, the two space agencies decided to join together and design a single mission called TOPEX/Poseidon that would utilize a satellite altimeter to help them to understand how the ocean interacts with our planet [Fu et al., 1994].

[3] A challenge of the TOPEX/Poseidon mission was to improve the measurement accuracy using a state-of-the-art dual-frequency radar altimeter and three independent precision orbit determination systems. The satellite still orbits the Earth at an altitude of 1336 km (830 miles) with an inclination angle of 66° and a period of 112 min. The orbit altitude (1336 km) was selected to minimize atmospheric drag and gravity forces acting on the satellite, which makes orbit determination easier and more accurate. There are 127 orbits in each 9.916 day period (i.e., 10-day cycle: the satellite passes vertically over the same location, to within 1 km, every 10 days), and the repeat period was chosen as a compromise between spatial and temporal resolutions. The satellite orbits are close to Sun-synchronous, advancing by 2° per day. Therefore it takes about 90 days to cover all local times. The satellite covers most of the world's ocean (Figure 1) and makes measurements of the height of the ocean, using the two altimeters, one built by NASA and another built by CNES. TOPEX/Poseidon distributes the data to 38 science investigators in nine countries: USA, France, Japan, Australia, United Kingdom, Germany, Norway, South Africa, and Netherlands. The TOPEX/Poseidon mission celebrated its 10th anniversary in orbit last year (10 August 2002), and NASA decided to continue the operations through September 2003 because the satellite is still healthy and is producing high-quality data. As a follow-on, the Jason-1 mission was launched on 7 December 2001 to measure surface heights, and it will provide data of the same quality as TOPEX.

Figure 1.

TOPEX satellite ground tracks for 10 days, from day 164 to 174, in 1996 (obtained from www.jason.oceanobs.com).

[4] TOPEX is the first satellite radar altimetry mission to carry a dual-frequency radar altimeter, the NASA Radar Altimeter (NRA), operating at 13.6 GHz (Ku band) and 5.3 GHz (C band) simultaneously. Because the ionosphere has a dispersive nature, measuring at two frequencies provides both a direct estimate of the total electron content (TEC) along the ray path from the satellite to the surface of the ocean and a method for effectively removing the ionospheric delay imposed on the altimeter. The ionospheric delay is directly proportional to the electron content (Nt) along the ray path and inversely proportional to the frequency (f) squared of the radio wave

equation image

where Iono_Corr stands for the ionospheric correction, that is, the ionospheric delay. The measured electron content (Nt) is essentially equivalent to the total electron content (TEC) of the ionosphere in a column extending from the satellite to the subsatellite reflection point on the surface of the ocean. It is well known that the TOPEX TEC data have a constant bias of about 2–3 TECU [Orús et al., 2002] and we believe the bias could be up to 5 TECU. However, for the analysis in this paper, we did not consider this bias because we were only interested in the general trends that the data displayed.

[5] The measurements of TEC from the TOPEX/Poseidon mission during the last decade provide an excellent TEC database over the ocean areas, where conventional measurements are sparse. The TEC data offer a good opportunity to construct a better global ionospheric climatology and can be used for validation purposes of the current ionospheric models or assimilation techniques. The TOPEX TEC data from 1992 to 1996 had been analyzed by Codrescu et al. [1999], and later, they extended the analysis to include the data from 1992 to 1997 [Codrescu et al., 2001]. However, their extended study was focused on the comparison with DORIS TEC measurements and empirical model (IRI and Bent) results corresponding to TOPEX measurements. Their analysis of TOPEX TEC measurements was restricted to the measurements for low solar activity (F10.7 < 120), since their data set was representative of the low solar activity conditions. Currently, data from 1992 to 2001 are available. The average F10.7 of the data is 117.48 and now there is a large enough number of averaged data for relatively high solar activity (F10.7 ≥ 120). Therefore we were able to extend the analysis to relatively high solar activity conditions. In addition, we analyzed three seasonal conditions (equinox, December solstice, and June solstice) and three longitude domains (Pacific, Indian, and Atlantic oceans).

2. TEC Data and Analysis

[6] The TEC data were obtained from the NASA Physical Oceanography Distributed Active Archive center at the Jet Propulsion Laboratory (JPL PO.DAAC/NASA). The TEC measurements were taken almost every second and the data set used in our analysis has a time resolution of 18 s or about 1° of orbit (i.e., each TEC value was obtained by averaging the data for 18 s). The scatter of the 1-s TEC values about the 18-s averaged mean shows fairly constant spreads of about ±4 ∼ 5 TECU for different seasons, local times, and hemispheres. We computed the geomagnetic latitude, longitude, and magnetic local time associated with each 18-s averaged data along each orbit using “quasi-dipole coordinates” [Richmond, 1995].

[7] The database includes more than 1,000,000 18-s averaged data per full year and covers the period from August 1992 to the end of 2001. This means that the measurements cover almost a full solar cycle through the cycle 22 to 23 (Figure 2). The total number of 18-s averaged data for about 10 years that was used in our study is 9,841,119 with the average equation image = 2.154 and the average equation image = 117.474. For our initial work the data have been longitudinally averaged; that is, the measurements in all longitude sectors for a given magnetic local time contributed to the average of that particular local time bin. We therefore do not expect any longitudinal or universal time variation in this part of our analysis.

Figure 2.

Solar activity (F10.7 cm flux) during the TOPEX/Poseidon mission. The period for the TEC data used in our analysis is from August 1992 to the end of 2001.

[8] To examine the general TEC morphology, we binned the data as shown in Table 1. For the solar activity (F10.7), we binned the data with high solar activity (F10.7 ≥ 120) and low solar activity (F10.7 < 120) so that we could compare our results for F10.7 < 120 with the previous work done by Codrescu et al. [1999]. For the geomagnetic activity bins, we chose the Kp index, which is the most widely used index for magnetic activity, instead of using the integral of the hemispheric power during 36 hours preceding the measurement, which was used in Codrescu et al.'s [1999] study. The bins have three different levels in a relatively simple way; low (Kp ≤ 1.7, equation image ∼ 1.0), medium (2.0 ≤ Kp ≤ 3.3, equation image ∼ 2.5), and relatively high (Kp ≥ 3.3, equation image ∼ 4.2) geomagnetic activities.

Table 1. Binning Criteria for Geomagnetic Activity, Solar Flux, Season, and the Geomagnetic Coordinate System
Three-Hour Kp IndexF10.7 cm FluxSeason, MonthMlat and MLT
Kp ≤ 1.7F10.7 < 120Equinox (3, 4, 9, 10)1° × 1° for the general climatology
2.0 ≤ Kp ≤ 3.0F10.7 ≥ 120Jun. Sol. (5, 6, 7, 8)2° × 3.75° for the seasonal maps
Kp ≥ 3.3 Dec. Sol. (1, 2, 11, 12)(1° corresponds to 4 min in MLT.)

[9] At first, we analyzed the data only for the solar and geomagnetic activity variations with the 1° × 1° bins in geomagnetic latitude and magnetic local time (MLT), combining the data from all seasons (1° corresponds to 4 min in MLT). Then, we added seasonal bins: January, February, November, and December contributing to December solstice; March, April, September, and October to equinox; and May, June, July, and August to June solstice. However, to ensure a sufficient number of data points per bin, we increased our coordinate bins to 2° × 3.75° bins (15 min in MLT) in geomagnetic latitude and MLT. In this part of our analysis, the total number of bins was about 3 × 2 × 3 × 75 × 96 = 129,600. The number of 18-s averaged data points that contributed to a bin varied from about 2 or 3 to 80 for the 1° × 1° case and from about 10 to 300 for the 2° × 3.75° case, depending mainly on the hemispheres, F10.7, and the Kp index (see Figure 5). The hemispheric variation is due to the characteristic of the orbit of the TOPEX mission, which covers only ocean areas. Therefore there is a much better data distribution in the Southern Hemisphere than in the Northern Hemisphere. Also, the low F10.7 case has a better data distribution than the high F10.7 case because of the solar cycle variation during the mission (the declining phase of solar cycle 22 to the peak of cycle 23, see Figure 2). For the geomagnetic activities, the number of data points decrease as the activity increases.

[10] For our final analysis, we binned the TEC data into three longitudinal bins, including the Pacific, Indian, and Atlantic Ocean areas. This binning was done so that we could study the effect of magnetic declination on the TEC climatology. The additional binning gives a better data distribution than the 1° × 1° case, but a smaller number of data points than the 2° × 3.75° case (see section 6 for details).

3. General TEC Climatology

[11] In the first part of our study, we produced global, longitudinally averaged TOPEX TEC maps for low (equation image ∼ 1.0), medium (equation image ∼ 2.5), and high (equation image ∼ 4.2) geomagnetic conditions for both low (F10.7 < 120) and “relatively” high (F10.7 ≥ 120) solar activity conditions. Figure 3 shows the TEC maps for the case when all seasons are combined. Also shown in Figure 3 are the standard deviations, which indicate the variability in our bins. The longitudinally averaged TEC values are shown as a function of geomagnetic latitude and magnetic local time (MLT). Note that in our analysis we did not apply the 5-point running average that was applied by Codrescu et al. [1999]. However, we also tested the effects of the averaging procedure on these maps and virtually no differences were detected. The color scale for the TEC maps ranges from 0 to 50 TECU (1 TECU = 1016 els/m2) for F10.7 < 120 and from 0 to 100 TECU for F10.7 ≥ 120. For the standard deviation, the range is from 0 to 15 TECU for F10.7 < 120 and 0 to 30 TECU for F10.7 ≥ 120.

Figure 3.

Longitudinally-averaged TOPEX TEC maps and the standard deviation of the data for low (left), medium (center), and high (right) geomagnetic conditions for both F10.7 < 120 and F10.7 ≥ 120 (upper and lower panels, respectively). All seasons are combined to produce the TEC maps. The maps correspond to the average TEC values in each bin.

[12] Figure 3 was produced so that we could compare our analysis with the previous study by Codrescu et al. [1999]. Specifically, the combination of data from all seasons, the use of longitudinally-averaged data, the selection of 1° × 1° coordinate bins, the three Kp ranges, and the two solar activity bins were all motivated by the work of Codrescu et al. [1999]. However, the color scales in Figure 3 are not the same as in Codrescu et al. [1999] and they only considered the case of F10.7 < 120. Also, Codrescu et al. [1999] used an integral of hemispheric power to define the magnetic activity levels, whereas we used Kp bins that are roughly similar.

[13] The equatorial anomaly is the most evident feature in all of the TEC maps. During the day the dynamo electric field is eastward, and this leads to an upward plasma drift at low latitudes. The lifted plasma then diffuses downward along the geomagnetic field lines due to the gravitational force, and this results in ionization enhancements on both sides of the magnetic equator (at about ±10°). At night, the dynamo electric field is westward, which leads to a downward plasma drift and lower TEC values.

[14] With regard to the TEC maps for F10.7 < 120, Codrescu et al. [1999] found that the equatorial anomaly becomes wider in magnetic latitude and more pronounced in amplitude as the magnetic activity level increases. Our TEC maps for F10.7 < 120 also display these features. However, for F10.7 ≥ 120, these features are not that evident and the anomaly is very similar for all magnetic activity levels. Also, for F10.7 < 120, Codrescu et al. [1999] found that the anomaly crests tend to move closer together and that there appears to be a local minimum at the magnetic equator for low and medium magnetic activities but not for high magnetic activity. Our TEC maps do not really show a tendency for the anomaly crests to move closer together with increasing magnetic activity, and this is true for both the low and high F10.7 cases. Also, all of our TEC maps show a local minimum at the magnetic equator, and it does not disappear for high magnetic activity. Another interesting feature is the increase in TEC at sunrise. For F10.7 < 120, Codrescu et al. [1999] found that the rate of TEC increase in the morning is basically the same for all magnetic activity levels because solar EUV is responsible for this increase. Our TEC maps agree with this conclusion for both low and high solar activities.

[15] At auroral latitudes, Codrescu et al. [1999] found large enhancements in TOPEX TEC values for F10.7 < 120, which they attributed to an auroral contribution to TEC. These enhancements were visible in the southern hemisphere where the latitudinal coverage extends to −80° magnetic latitude but were not noticeable in the Northern Hemisphere where the coverage ends between 60° and 70° magnetic latitude. They pointed out that the enhancements are larger for low geomagnetic activity conditions than for medium and high activity conditions. In our analysis, we also found large enhancements in TEC at high latitudes, mostly in the Southern Hemisphere (see Figures 3 and 4). However, the enhancements in our TEC maps are spatially very narrow and relatively weak in the magnitude of TEC, compared with the results of Codrescu et al. [1999]. In addition, we did not see any systematic variations of the enhancements with the magnetic activity levels for either low or high solar activity. We believe that the observed enhancements in TEC values should be interpreted with caution, since they might be contaminated due to the interaction of the range measurement with ice in the Southern Hemisphere [Picot, 1998].

Figure 4.

Longitudinally averaged TOPEX TEC maps for low (left), medium (center), and high (right) geomagnetic conditions for equinox (top), December solstice (middle), and June solstice (bottom) conditions. For each season the upper and lower panels show the average TEC maps for both F10.7 < 120 and F10.7 ≥ 120, respectively.

[16] The standard deviation of the TOPEX TEC from the longitudinally averaged values is typically 30 ∼ 40% and sometimes more than 50%. The largest deviations in terms of TECU occur at the crest of the equatorial anomaly and they are larger for F10.7 ≥ 120 than for F10.7 < 120 (note different color scales). Also, with regard to the relative standard deviation, the variability reaches a maximum between about 2000 ∼ 2200 MLT (not shown in Figure 3). This is no doubt due to the variability in the dynamics of the post-sunset equatorial ionosphere.

4. Seasonal Variations

[17] There are important seasonal variations of the ionosphere, and therefore we binned the TOPEX TEC data according to season in order to study these variations. Specifically, we produced global, longitudinally averaged TEC maps for low, medium, and high geomagnetic activity for both low and high solar activities and for equinox, December solstice, and June solstice (Figure 4). Each one of the seasonal panels in Figure 4 has the same format and color scale as in Figure 3, but the standard deviations are not shown. In the TEC maps, December solstice corresponds to four months (November, December, January, February), June solstice includes May, June, July, and August, and equinox includes March, April, September, and October (see Table 1). For F10.7 < 120, Codrescu et al. [1999] also studied seasonal effects, but their binning was different than ours. They combined winter in the Northern Hemisphere with winter in the Southern Hemisphere, and summer in the Northern Hemisphere with summer in the Southern Hemisphere. Also, their spring case was a combination of TEC data from February, March, April, and the autumn months. Nevertheless, despite the difference in TEC binning, it is still useful to compare our low solar activity case to their results.

[18] For F10.7 < 120, Codrescu et al. [1999] found that the variation of TEC with magnetic local time at low latitudes was the same for all three seasons. They found that there was a minimum in TEC before sunrise, a steep increase toward a peak in the afternoon, and a slow decay at night. Both our low and high solar activity cases display these trends. Codrescu et al. [1999] also found that for F10.7 < 120 the TEC values in the equatorial anomaly are greater during equinox than in winter or summer. Our results are in general agreement with their finding, and our new results for F10.7 ≥ 120 also agree with this finding.

[19] Codrescu et al. [1999] found that for F10.7 < 120 the lowest TEC values at midlatitudes occur in winter at night (5 ∼ 10 TECU). The corresponding minimum values in summer are 10 ∼ 15 TECU, which are 30 ∼ 50% higher. Our case for F10.7 < 120 agrees with these findings. For F10.7 ≥ 120, the lowest TEC values at midlatitudes also occur in winter at night, with minimums slightly larger than the low solar activity case. The corresponding minimum values in summer (20 ∼ 25 TECU), however, are almost twice as large as the minimum values in summer for F10.7 < 120.

[20] There are several other trends and anomalies that are worth investigating. First, the TEC patterns at equinox are basically symmetric about the magnetic equator, but they are asymmetric at the solstices. This can be understood by considering the solar illumination in both hemispheres and the direction of the meridional neutral wind. The meridional wind blows from the summer to winter hemisphere at the solstices, which raises the F-layer in the summer hemisphere and lowers it in the winter hemisphere [Schunk and Nagy, 2000]. This wind effect, coupled with the asymmetry in solar ionization, account for the TEC asymmetry at the solstices.

[21] The F2-layer anomalies are interesting in that they display departures from solar-controlled behavior. The “seasonal anomaly” is that NmF2 at midlatitudes is greater in winter than in summer during the day, but the anomaly disappears at night (NmF2 is greater in summer than in winter). The “semiannual anomaly” is that at low and midlatitudes NmF2 is greater at equinox than at solstice. The “annual anomaly” describes the fact that NmF2 in December is on average greater than NmF2 in June, both during the day and at night. An alternative description is that the seasonal anomaly is greater in the Northern Hemisphere than in the Southern Hemisphere.

[22] At solstice, there is a prevailing summer-to-winter neutral circulation, upwelling in the summer hemisphere and at the equatorial latitudes, and downwelling just equatorward of the auroral oval in the winter hemisphere [Rishbeth, 1998; Rishbeth et al., 2000]. At low latitudes and lower midlatitudes, the upwelling moves the air rich in molecules to the F2-layer and decreases NmF2 from the equinoxial value, which shows the semiannual variation. At higher midlatitudes, the summer hemisphere shows a decrease in NmF2 by the same effect. However, the situation in the winter hemisphere is somewhat different because of the downwelling zone. At longitude sectors far from the magnetic poles, the downwelling occurs at relatively high latitudes where the solar zenith angle is very large and the production becomes very sensitive to the zenith angle. The zenith angle effect on NmF2 is more important than the effect of the increased O/N2 ratio due to the downwelling, and this leads to a decrease in NmF2 from the equinox (the semiannual anomaly). At longitude sectors near the magnetic poles, on the other hand, the solar zenith angle at the downwelling zone is relatively small and the production is not sensitive to the zenith angle. The result is that the increase in O/N2 ratio due to the downwelling produces larger NmF2 values at higher mid-latitudes in this longitude sectors (the seasonal anomaly). For the annual anomaly, several factors come into play. First, the solar flux change due to the different Sun-Earth distances between June solstice and December solstice will have some effect, but it is known to be very small. The asymmetrical geomagnetic field in the two hemispheres could be another factor. A further study is necessary to successfully explain the annual anomaly.

[23] One might expect to see the “anomalies” in the TEC data because the electron density near the F-region peak makes a significant contribution to TEC. During the daytime, our TEC maps (Figure 4) show the annual (larger TEC in December than in June) and semiannual (larger TEC at equinox than at solstice) anomalies at low and lower midlatitudes for both the low and high solar activity conditions. However, for F10.7 ≥ 120, the anomalies are generally stronger, and in particular, the semiannual anomaly lasts until midnight. The seasonal anomaly at higher midlatitudes, however, does not clearly appear in our TEC maps for both the low and high solar activity cases, except for the noon local time sector for F10.7 ≥ 120 (Figure 6). Part of the reason for these results is that in the topside ionosphere the annual anomaly is very large [Su et al., 1998] and the seasonal anomaly is very small [Torr and Torr, 1973]. Moreover, our longitudinally averaged binning for the TEC maps may act to mask the seasonal anomaly. The lack of a seasonal anomaly was also noted by Codrescu et al. [1999] for their case of low solar activity (F10.7 < 120).

[24] Figure 5 shows TEC (top panels) and the corresponding number of 18-s averaged TEC data points (bottom panels) versus magnetic latitude at selected magnetic local times (0300, 0900, 1500, and 2100 MLT) for low, medium, and high geomagnetic activity. Only the results for equinox are shown. The thin lines in the panels correspond to low solar activity and the thick lines to high solar activity. These curves essentially are cuts through the “equinox” panels in Figure 4. Averages from less than six 18-s averaged data points are not included in this figure.

Figure 5.

Latitudinal variation of TEC (top panels) and the corresponding number of 18-s averaged TEC data points (bottom panels) during equinox at four selected magnetic local times (0300, 0900, 1500, and 2100 MLT) for three levels of magnetic activities. The thick line in the panels represents TEC for F10.7 ≥ 120 and the thin line represents TEC for F10.7 < 120. Averages from less than six 18-s averaged data points are not included.

[25] Our choice of MLTs was motivated by the work of Codrescu et al. [1999], who selected these local times in their study for F10.7 < 120. However, in their figures all seasons were averaged together, whereas only equinox results are shown in Figure 5. Nevertheless, it is useful to compare results. First, for all three magnetic activity levels and at all four times (0300, 0900, 1500, and 2100 MLT), Codrescu et al. [1999] found distinct TEC enhancements in association with the southern auroral oval, but not for the northern oval because of the lack of data. However, our equinox results in Figure 5 for both the low and high solar activity cases do not show such distinct enhancements. In general, both the shape of the curves and the TEC magnitudes that we get for F10.7 < 120 and equinox conditions are consistent with those obtained by Codrescu et al. [1999] when all seasons were averaged together.

[26] An examination of the panels for TEC in Figure 5 indicates that the TEC curves are very similar for all three magnetic activity levels at a given MLT, even though the low solar activity cases show the tendency of slight increase in TEC as the activity levels increase. Therefore before comparing the low and high solar activity cases, it is useful to select one magnetic activity level and add more magnetic local times (MLTs). Figure 6 shows TEC curves versus magnetic latitude for medium Kp, eight MLTs (0300, 0600, 0900, 1200, 1500, 1800, 2100, and 2400 MLT), and three seasons (equinox, December solstice, and June solstice). Similar to Figure 5, averages from fewer than six 18-s averaged data points are excluded. Several interesting features can be identified. First, TEC for F10.7 ≥ 120 is larger than TEC for F10.7 < 120 at all times and for all three seasons. However, at times (0300, 0600, and 2400 MLT), it is only slightly larger (particularly in winter hemispheres), while at other times it is much larger. Also, the equatorial anomaly peaks, particularly in the evening, are more distinct for F10.7 ≥ 120 than for F10.7 < 120. Finally, a comparison of the winter hemispheres at June solstice with the winter hemispheres at December solstice and a comparison of the summer hemispheres at June solstice with the summer hemispheres at December solstice indicate that they are similar, except during the daytime (1200 and 1500 MLT). At these times, the low-latitude TEC values in December solstice are larger than those in June solstice, which shows the annual anomaly (also see Figure 4). The same result occurred for low and high Kp (not shown in Figure 6). This is one of the reasons why we separated seasons and hemispheres.

Figure 6.

Local time variations (every 3 hours in MLT) of the TEC patterns for June and December solstice and equinox. Each panel shows longitudinally averaged TEC curves for low and high solar flux conditions but only for the medium geomagnetic activity condition. The thick line represents TEC for F10.7 ≥ 120 and the thin line represents TEC for F10.7 < 120. Averages from less than six 18-s averaged data points are not included.

5. Geomagnetic Activity Effects

[27] Although the average Kp of our high geomagnetic activity bin (equation image ∼ 4.2) is too small to represent storm periods, it is useful to determine whether or not the TEC variation with Kp is consistent with our current knowledge of ionospheric behavior during storms. The general behavior of the F-region during geomagnetic storms can be described as follows. Negative phases (decreases in Ne) are always observed at high latitudes. At midlatitudes, both positive (increases in Ne) and negative phases can occur, but the latter is more common. At low latitudes, positive phases tend to dominate at the times when storm effects are detectable. Although the storm morphology can be complicated, it can be understood by considering changes in the neutral composition (the O/N2 ratio), temperature, and winds as well as the dynamo electric fields at low latitudes [cf. Fuller-Rowell et al., 1996, and references therein].

[28] Figures 3, 4, and 5 show the effect on TEC due to changes in geomagnetic activity, although most of the effects that we will discuss are barely detectable. For F10.7 < 120, Codrescu et al. [1999] found that TEC increases occur at low latitudes as magnetic activity increases. They also found that the increase in peak TEC values, in the high geomagnetic activity bin compared with the low activity bin, increases with MLT from about 20% at 0300 MLT to close to 50% at 2100 MLT. We primarily see TEC increases at low latitudes as magnetic activity increases for our low (F10.7 < 120) solar activity case, but for F10.7 ≥ 120 the TEC values at low latitudes are similar for all magnetic activity levels. Also, the variation in peak TEC values with MLT is not clear in our results. At midlatitudes, Codrescu et al. [1999] primarily found negative TEC effects as magnetic activity increased. We also get negative TEC effects with increasing Kp for both our low and high solar activity cases but barely detectable.

[29] With regard to the seasonal responses to magnetic disturbances, measurements indicate that summer displays both positive and negative phases, but the phase tends to be positive in winter [Danilov and Lastovicka, 2001]. Negative phases at midlatitudes are evident at several local time sectors for high solar activity: 0900 MLT for June and December solstices (not shown) and 1500 MLT for equinox (Figure 5). However, in both local time sectors, the seasonal preference described by Danilov and Lastovicka [2001] is barely detectable.

[30] The effects of geomagnetic storms on the ionosphere are known to be significant (driven by prompt penetration electric fields, polarization jets, etc.), but the effects on TEC seem to be very small in our analysis. This is possibly due to our relatively simple binning procedure we used for the geomagnetic activity. Clearly, a more detailed and systematic study of the storm time variations of TEC will be necessary to elucidate such effects.

6. Longitudinal Variations

[31] Up to this point, we only considered longitudinally averaged TEC data because our analysis was focused on the diurnal variation of TEC for different solar, seasonal, and magnetic activity conditions. Now, it is instructive to consider longitudinal variations. For this analysis, we selected three longitudinal bins, including the Pacific (150 ∼ 280°E), Atlantic (280 ∼ 30°E), and Indian (30 ∼ 150°E) sectors. These sectors were selected because of the different characteristics of the magnetic declination (Figure 7). The declination in the Pacific sector is positive and relatively large, up to 50° at midlatitudes in the Southern Hemisphere. The Atlantic sector has a relatively small negative declination in both hemispheres. In the Indian sector, the declination is negative and very large, up to −60° in the Southern Hemisphere, but it is positive and small in the Northern Hemisphere.

Figure 7.

Three different longitude sectors (Pacific: 150 ∼ 280°E, Atlantic: 280 ∼ 30°E, and Indian: 30 ∼ 150°E in geographic longitude) used in our analysis. The figure shows the location of the magnetic equator and the declination angle.

[32] Because the TEC data were divided among three additional bins, the geomagnetic local time bin was increased to 15° (1 hour MLT) with the same 2° magnetic latitude bin in order to guarantee that we had enough data in each bin. Furthermore, we used three bins for F10.7 solar flux instead of two bins, since the previous “high” solar activity bin (F10.7 ≥ 120) does not represent real high solar activity conditions. The new bins correspond to low (F10.7 < 120), medium (120 ≤ F10.7 < 150), and high (F10.7 ≥ 150) solar activities. Typically, the number of 18-s averaged data points in a bin reached up to 500. Figure 8 shows TEC maps in the three longitude sectors for three seasons, both low and high solar activities, and geomagnetically moderate conditions (combined low and medium activities, equation image ∼ 1.7). We excluded high Kp and medium F10.7 bins for this analysis. The panels in Figure 8 show TEC versus geomagnetic latitude and MLT. The color scales for TEC range from 0 to 50 TECU for F10.7 < 120 and from 0 to 100 TECU for F10.7 ≥ 150, although the maximum TEC value in the maps reaches up to 120 TECU for the high solar activity condition.

Figure 8.

TEC maps in three longitude sectors (three columns) for three seasons, both low and high solar activities, and geomagnetically moderate conditions (equation image ∼ 1.7). Each panel shows TEC versus geomagnetic latitude and MLT. The color scale for TEC ranges from 0 to 50 TECU for low solar flux and from 0 to 100 TECU for high solar flux.

[33] Besides the obviously high TEC values for high F10.7, the most striking feature in Figure 8 is the relatively strong longitudinal dependence of TEC. Note that the longitudinal variation of TEC follows the longitudinal variation of the magnetic declination in Figure 7. Specifically, there is a considerable TEC variation in the Southern Hemisphere in all three longitudinal sectors, but relatively small TEC variation in the Northern Hemisphere. Note that in the southern Pacific sector, where the declination is positive and large, the TEC variation is significantly different than the other two longitudinal sectors, where the declination is negative in the Southern Hemisphere. In the Northern Hemisphere, however, the longitudinal variation of the declination is very small. Furthermore, the TOPEX TEC data in the Northern Hemisphere are spatially very limited because of the huge land mass structure. Consequently, our TEC maps show relatively weak TEC variations with respect to the longitude in the Northern Hemisphere.

[34] The declination effect on TEC can be clearly seen in Figure 9, where we show a scatter plot of TEC versus longitude at −45° magnetic latitude. Scatter plots are shown for noon (top panels) and midnight (bottom panels) for both low (left panels) and high (right panels) solar activities. Three sets of panels are shown, corresponding to equinox (Figure 9a), December solstice (Figure 9b), and June solstice (Figure 9c). Also shown in these panels are the median TEC values versus longitude and the spread in TEC that corresponds to the lower and upper quartiles. The median TEC values were calculated every 10° in longitude with a 20° wide bin, and median values calculated with less than 10 data points were not included. For these panels, as we mentioned earlier, the high Kp bins were excluded. Finally, note that the TEC scale is different for the low and high solar activity cases.

Figure 9.

Scatter plots for TEC and the declination angle versus geographic longitude at −45° magnetic latitude. Scatter plots are shown for noon (top) and midnight (bottom) for both low (left) and high (right) solar activity. The median values of TEC and corresponding upper and lower quartiles (as an error bar) are shown for equinox (a), December solstice (b), and June solstice (c). All cases are for the moderate geomagnetic activity (equation image ∼ 1.7). Note that the high solar activity is now defined to be F10.7 > 150. Median values from less than 10 averaged data points are not included.

Figure 9.

(continued)

Figure 9.

(continued)

[35] In general, Figures 9a9c clearly show a strong correlation between TEC and the magnetic declination. At midnight, TEC closely follows the declination in December solstice for both low and high solar activities and in equinox for F10.7 ≥ 120. Even in June solstice for both solar activities and in equinox for F10.7 < 120, it is possible to see the TEC correlation with declination, although it is weak. On the other hand, at noon the longitudinal variation of TEC is opposite to that at midnight. The opposite variation of TEC with declination at noon and midnight is due to the different directions of the zonal neutral wind at these local times. At noon, the average zonal wind is westward in the Southern Hemisphere. This westward zonal wind moves the ionosphere upwards when there is a negative declination, which acts to increase TEC and pushes the ionosphere downwards when there is a positive declination, which decreases TEC. On the other hand, at midnight, the average zonal wind is eastward in the Southern Hemisphere, which causes exactly the opposite effects on TEC to the cases at noon; decrease in TEC with a negative declination and increase in TEC with a positive declination (Figure 9).

[36] Another interesting feature shown in Figures 9a9c is that at around 250°E longitude, where the declination is the largest, the TEC values at midnight are larger than those at noon in December solstice for both low and high solar activities and in equinox for F10.7 > 150. This surprising feature can also be found in the southern Pacific sector at equinox and December solstice in Figure 8. The southern Pacific sector, where the declination is large and positive, shows remarkably large TEC values in the evening through the midnight compared with the other two sectors. In this longitude sector, the large electron density in the afternoon seems to be maintained through the nighttime by the effects of the zonal wind combined with the very large and positive declination.

[37] There is a considerable change in the correlation between the longitudinal variation of TEC and declination from daytime to nighttime. To see how the longitudinal variations change with local time, Figures 10a10c show the median values of the TEC variation every 3 hours in MLT for the low and high F10.7 solar flux cases. It appears that the phase of the TEC variations moves toward the east with increasing local time for all three seasonal cases but particularly for the December solstice condition. Also, in all three seasonal cases, particularly June solstice, the TEC scatter at midnight is smaller than that during the day. This is due to the fact that Figures 10a10c show TEC values and the lower TEC at night produce a smaller TEC scatter with regard to TECU. However, the TEC scatter during the daytime and nighttime are similar from the percentage standpoint.

Figure 10.

Local time variations (every 3 hours in MLT) of the median value of TEC at −45° MLAT with the corresponding error-bar for the lower and upper quartiles of the data for equinox (a), December solstice (b), and June solstice (c). The left and right panels represent the plots for low and high solar activities, respectively. All cases are for moderate geomagnetic activity (equation image ∼ 1.7). Median values from less than 10 averaged data points are not included.

Figure 10.

(continued)

Figure 10.

(continued)

7. Conclusion

[38] The TOPEX/Poseidon mission has provided an excellent database of vertical TEC over the oceans, where conventional measurements are sparse. The data cover the period from the middle of 1992 to the end of 2001, and therefore the measurements span almost a full solar cycle. We analyzed these data in order to study the TEC climatology that is inherent in the database. First, 18-s averages were taken of the 1-s data, which left more than one million TEC values per year. The total number of averaged data for about ten years that was used in our study is 9,841,119. Next, these nearly ten million TEC values were binned according to solar activity (F10.7 < 120 and F10.7 ≥ 120), magnetic activity (Kp ≤ 1.7, 2.0 ≤ Kp ≤ 3.3, Kp ≥ 3.3), season (equinox, June and December solstices), and magnetic latitude and MLT (1° × 1° for the general climatology and 2° × 3.75° for the seasonal TEC maps). This binning resulted in longitudinally averaged TEC values. Finally, the binning was further divided in order to study longitudinal effects, and bins were created to cover the Pacific, Indian, and Atlantic Ocean areas.

[39] In a previously published study, Codrescu et al. [1999] examined the TEC climatology for the TOPEX data covering the period from 1992 to 1996, which limited their study to the low solar activity case. In their seasonal analysis, they combined TEC data from both the Northern and Southern Hemispheres to create their summer and winter bins, whereas we used equinox, December solstice, and June solstice bins. They used low, medium, and high magnetic activity bins, and these bins are similar to what we adopted. However, Codrescu et al. [1999] only considered longitudinally averaged data, whereas we also studied the longitudinal variation of TEC and how it relates to the magnetic declination.

[40] In what follows, we provide a fairly complete summary of our findings, and when appropriate, we compare our results to those obtained by Codrescu et al. [1999]. This detailed summary should not only be valuable for summarizing the global TEC climatology, but the features discussed can also be used to validate ionospheric (climatology, numerical, and data assimilation) models over the ocean areas.

7.1. General TEC Climatology

[41] The TEC value for F10.7 ≥ 120 is much larger than the one for F10.7 < 120 at all local times and for all three seasons. For F10.7 ≥ 120, the equatorial anomaly is more distinct, especially during the evening (Figures 3, 4, and 6).

[42] Codrescu et al. [1999] found that the equatorial anomaly becomes wider in magnetic latitude and more pronounced in amplitude as the magnetic activity level increases. Our TEC maps for F10.7 < 120 also display these features. However, for F10.7 ≥ 120, these features are not evident and the anomaly is very similar for all magnetic activity levels (Figure 3).

[43] Codrescu et al. [1999] found that the anomaly crests tend to move closer together with the magnetic activity levels and that there appears to be a local minimum at the magnetic equator for low and medium magnetic activities but not for high magnetic activity. Our TEC maps do not show these tendencies for both the low and high F10.7 cases and show a local minimum for all magnetic activity levels (Figure 3).

[44] Codrescu et al. [1999] found that the rate of TEC increase in the morning is basically the same for all magnetic activity levels and we agree with this result for both our low and high solar activity cases.

[45] At auroral latitudes, Codrescu et al. [1999] found that the auroral contribution to TEC is visible in the Southern Hemisphere but is not noticeable in the Northern Hemisphere. We also found large enhancements in TEC at high latitudes mostly in the Southern Hemisphere. However, the enhancements are spatially very narrow and relatively weak in the magnitude of TEC, compared with their results (Figures 3 and 4).

[46] They pointed out that the enhancements are larger for low geomagnetic activity than for medium and high activities. However, we did not see any systematic variations of the enhancements with the geomagnetic activity levels for either low or high solar activity (Figures 3 and 4).

[47] The standard deviation is typically 30 ∼ 40% and sometimes more than 50%. The largest deviation occur at the crest of the equatorial anomaly and they are larger for F10.7 ≥ 120 than for F10.7 < 120. Also, with regard to the relative standard deviation, the variability reaches a maximum between about 2000 ∼ 2200 MLT.

7.2. Seasonal Variations

[48] Codrescu et al. [1999] found that the variation of TEC with magnetic local time at low latitudes was the same for all three seasons and we agree with this result for both the low and high solar activity cases (Figures 4 and 6).

[49] Codrescu et al. [1999] found that the lowest TEC values at midlatitudes occur in winter at night (5 ∼ 10 TECU), the corresponding minimum values in summer are 10 ∼ 15 TECU. Our case for F10.7 < 120 agrees with these findings. For F10.7 ≥ 120, the lowest TEC values at midlatitudes also occur in winter at night, with minimums slightly larger than the low solar activity case. The corresponding minimum values in summer (20 ∼ 25 TECU), however, are almost twice as large as the minimum values in summer for F10.7 < 120 (Figures 4 and 6).

[50] The TEC patterns at equinox are basically symmetric about the magnetic equator, but they are asymmetric at the solstices (Figure 4).

[51] During the daytime, our TEC maps show the annual and semiannual anomalies at low and lower midlatitudes for both low and high solar activities. However, for F10.7 ≥ 120, the anomalies are generally stronger and in particular, the semiannual anomaly lasts until midnight. However, the seasonal anomaly does not clearly appear in our TEC maps for both low and high solar activities except for the noon local time sector for F10.7 ≥ 120 (Figure 4).

[52] In the hemispheric comparison in the June and December solstices, the low-latitude TEC values in December solstice are larger than those in June solstice during the daytime (the annual anomaly), but at night, they are similar (Figure 6).

7.3. Geomagnetic Activity Effects

[53] Codrescu et al. [1999] found that TEC increases occur at low latitudes as magnetic activity increases and the increase in peak TEC values, in the high Kp bin compared with the low Kp bin, increases with MLT from about 20% at 0300 MLT to close to 50% at 2100 MLT. We primarily see TEC increases at low latitudes as magnetic activity increases for F10.7 < 120, but for F10.7 ≥ 120 the low-latitude TEC values are similar for all magnetic activity levels. Also, the variation in peak TEC values with MLT is not clear in our results (Figures 3 and 4).

[54] Codrescu et al. [1999] found the negative TEC effects at midlatitude as magnetic activity increased. We also get negative TEC effects with increasing Kp for both low and high solar activities but barely detectable.

[55] With regard to the seasonal responses to magnetic disturbances, negative phases at midlatitudes are evident at several local time sectors for F10.7 ≥ 120; 0900 MLT for June and December solstices and 1500 MLT for equinox. However, the seasonal preference described by Danilov and Lastovicka [2001] is barely detectable.

[56] The effects of geomagnetic storms on the ionosphere are known to be significant, but the effects on TEC seem to be very small in our analysis, possibly due to our relatively simple binning procedure for the geomagnetic activity.

7.4. Longitudinal Variations

[57] Besides the obviously high TEC values for high F10.7, the most striking feature is the relatively strong longitudinal dependence of TEC, which is related to the longitudinal variation of the magnetic declination. Specifically, in the southern Pacific sector, where the declination is positive and large, the TEC variation is significantly different than the other two longitudinal sectors, where the declination is negative in the southern hemisphere (Figure 8).

[58] In the Northern Hemisphere, the longitudinal variation of the declination is very small and the TOPEX TEC data are spatially very limited because of the huge land mass structure. Consequently, our TEC maps show relatively weak TEC variations with respect to the longitude in the Northern Hemisphere.

[59] At midnight, the longitudinal variation of TEC closely follows the longitudinal variation of declination for most of the seasonal cases. At noon, on the other hand, the longitudinal variation of TEC has an opposite phase to the one at midnight (Figures 9a9c).

[60] At around 250°E longitude, where the declination is the largest, the TEC value at midnight is even larger than the noontime value in December solstice for both low and high solar activities and in equinox for F10.7 > 150 (Figures 9a9c). The large electron density in the afternoon seems to be maintained through the nighttime by the effects of the zonal wind combined with the declination in this longitude sector.

[61] The phase of the TEC variation moves toward the east with increasing local time, for all three seasonal cases but is particularly evident for the December solstice condition (Figures 10a10c).

Acknowledgments

[62] This research was supported by the DoD Multidisciplinary University Research Initiative (MURI) via grant N00014-99-1-0712 and by NSF grant ATM-0000171 to Utah State University.

[63] Arthur Richmond thanks Dieter Bilitza and another reviewer for their assistance in evaluating this paper.

Ancillary