IHFACs: reproducing science-grade magnetometer observations
The platform magnetometer data used in this study can reproduce low-/mid-latitude IHFAC climatology reported previously by science-grade magnetometers. Specifically, we have demonstrated in “Results” section that platform magnetometer data of CryoSat-2 and GRACE-FO successfully reproduce IHFAC climatology as reported previously. Statistical distributions of the vertical currents at CryoSat-2 and GRACE-FO altitudes generally agree with previous Swarm and CHAMP studies. The agreement at different altitudes is as expected because IHFACs are, by definition, a field-aligned (then nearly altitude-independent) system from one hemisphere to the other. As an exercise, we have also plotted the IHFACs in the Apex latitude vs. MLT space, where the Apex latitude is defined by a field-aligned mapping of data points down to 110 km irrespective of the actual observation altitudes (Richmond 1995). The results for CryoSat-2 and GRACE-FO are shown in Figs. 6 and 7, respectively. We can see that Apex latitudes of peak IHFACs at different satellite data (at different altitudes) generally agree with each other, which supports that the detected signals are due to IHFACs (i.e., a current system aligned with the background magnetic field). GRACE-FO results in Fig. 7 exhibit less clear distributions than in Fig. 6, possibly because of the less amount of data (only about 1.5 years) and higher noise level, as we will discuss later in this section.
Encouraged by the promising results from CryoSat-2 (Fig. 2) and its long-term (> 8 years, much longer than the < 2 years of GRACE-FO) data accumulation, we have further subdivided CryoSat-2 data according to magnetic longitude (MLON) so that the data can be binned in the MLAT-MLT-MLON-season space. The MLON bin size is 5°. Then, we reassemble the vertical current data into separate global maps for three seasons and three MLT sectors: dawn (06–09 MLT), noon (10–14 MLT), and dusk (15–21 MLT). After hemispheric decomposition as described in “Results” section, hemispherically anti-symmetric components (i.e., IHFAC) are presented in the nine subpanels in Fig. 8. Note also that a 3-by-3 median filter is applied to all the panels to enhance visual clarity. Each row corresponds to a season: combined equinoxes, June solstice, and December solstice from top to bottom, respectively. Each column from left to right represents an MLT sector: dawn, noon, and dusk in that order.
Overall, Fig. 8 exhibits well-organized patterns, most of which are in reasonably good agreement with previous Swarm observations (Park et al. 2020, Figure 1) despite the different precession rates (i.e., different degrees of season-MLT aliasing) of Swarm and CryoSat-2: see Fig. 1a. As for dependence of IHFACs on longitudes, Park et al. (2020) emphasized a few notable features in the Swarm global maps near local noon, which are relevant to the middle column in our Fig. 8 highlighted by a red rectangle. During combined equinoxes and June solstice, noontime low-latitude IHFACs in Swarm data showed wavenumber-4 or -5 patterns along the zonal direction, which is also manifest in our Fig. 8e and partly for the equinoctial data in Fig. 8d: see the bead-like structures around the equator annotated by vertical arrows. Also, Swarm encountered a clear ‘C’-shaped pattern in the noontime maps, which traversed from the North Atlantic to South America during combined equinoxes and June solstice. CryoSat-2 data also exhibit this ‘C-shaped’ pattern near the white dashed curves in Fig. 8d, e. The C-shape in the SH is nearly collocated with the South Atlantic Anomaly, which is annotated with solid white curves (magnetic field magnitude of 23,000 nT and 26,000 nT). While some previous studies reported peculiar behavior of IHFACs around the SAA (e.g., Park et al. 2011; Lühr et al. 2019), they basically focused on low-latitude (|MLAT|< 35°) IHFACs. Our Fig. 8 shows the first evidence that not only low-latitude (|MLAT|< 35°) IHFACs but also mid-latitude ones (|MLAT|> 35°) reflect the shape of the SAA for noontime during combined equinoxes and June solstice. This C-shape may be partly explained in terms of enhanced conductance in the SAA and concomitant summer-like Sq currents, as discussed extensively in Lühr et al. (2019). For example, low-latitude IHFACs during noontime June solstice generally flow from the winter to the summer hemisphere (i.e., from the SH to the NH). On the other hand, IHFACs around the SAA flow from the NH to the SH, possibly due to the weak background magnetic field and locally enhanced ionospheric conductivities, which result in a locally persistent summer-like condition in the SH. As equinoctial IHFAC patterns largely follow those of June solstice, we can also expect that equinoctial noontime IHFACs have a C-shape near the SAA. Note that similar, albeit less conspicuous, signatures can be seen for December noon IHFACs (Fig. 8f). However, the above-mentioned theory (‘locally summer-like Sq’) cannot explain why such C-shapes are hardly identifiable at other MLTs. Near dawn and dusk, the NH and SH E-regions may experience significantly different insolation depending on seasons and longitudes. This can add complexity to the MLON-MLAT distribution of IHFACs at dawn and dusk. We also speculate that IHFAC distributions can be controlled by an involved interplay between the SAA-induced high conductance and atmospheric tidal effects in the wind, but a dedicated future study is necessary to verify it.
During December solstice, noontime low-latitude IHFACs in Park et al. (2020, Figure 1f) exhibit wavenumber-1 structure, with the polarity changing above the Atlantic and Western Pacific Oceans. Figure 8f also reproduces this feature, as highlighted by horizontal bidirectional arrows. December dusk-side IHFACs in Fig. 8i have strongest southward currents in the mid-latitude Pacific region, which also agrees with Swarm observations (Park et al. 2020, Figure 1i). The consistency between CryoSat-2 (Figs. 2 and 8) and Swarm (Park et al. 2020) confirms the ability of platform magnetometers, after proper post-processing, to investigate low-/mid-latitude ionospheric currents.
Figure 9 is the same as Fig. 8, but for GRACE-FO1. The locations of the horizontal/vertical arrows and C-shaped curves for annotation are the same as in Fig. 8 for the purpose of a direct comparison. Although the data are noisier than in Fig. 8 and containing data gaps (white area), we can still identify the salient features we highlighted in Fig. 8: see the regions inside the red rectangle (near-noon MLT). Wavenumber-4 or -5 structures in the zonal direction can be seen in the noontime low-latitude currents during combined equinoxes and June solstice (panels d and e) as well as C-shaped structures passing South America and the Atlantic Ocean. On the other hand, the wavenumber-1 structure becomes manifest for noontime December solstice at low latitudes (panel f). All these features agree with those in Fig. 8 and previous reports such as Lühr et al. (2019) and Park et al. (2020).
F-region dynamo currents
The platform magnetometer data could only partially identify the F-region dynamo currents, at least in its traditional sense as reported by previous studies (e.g., Lühr and Maus 2006). In Figs. 4, 5 we can see downward equatorial currents at 12–15 MLT during combined equinoxes and December solstice, as reported in previous studies. However, in other MLT sectors we fail to extract conventionally known signatures of F-region dynamo currents, such as downward currents before noon and clear upward currents near dusk. According to previous studies, the F-region dynamo currents are generally weaker than IHFACs (e.g., Lühr et al. 2015, Figure 4), which can be one reason why the signatures cannot pop out clearly in Figs. 4, 5. Also, the dynamo currents depend significantly on solar activity and altitude: they decrease with increasing altitudes in the F-region (Maute and Richmond 2016, Figure 9) and with decreasing solar activity (Maute and Richmond 2016, Figure 7). Considering higher altitudes of CryoSat-2 (717 km) and GRACE-FO (490 km) than those of CHAMP (< 450 km) and Swarm-A/C (< 470 km), it would be natural for the former satellites to encounter weaker currents (possibly below the detection limit of platform magnetometers) of the F-region dynamo than the latter. Also, both CryoSat-2 and GRACE-FO operate within Solar Cycle 24, which is known to be weaker than previous cycles: the two satellites cannot record as strong signatures of the F-region dynamo currents as CHAMP did during the last solar cycle (e.g., Lühr and Maus 2006). Especially, GRACE-FO has been operating right around the current solar minimum. We expect that further data accumulation in the future may reveal the dynamo currents more clearly because the next solar maximum is coming in a few years, and satellite altitudes will be reduced by continuous atmospheric drag.
Relationship with currents driven by gravity and pressure gradients
The vertical currents analyzed in this study are deemed (1) IHFACs at off-equatorial latitudes and (2) F-region dynamo currents near the equator. While the low-/mid-latitude F-region ionosphere also hosts currents driven by gravity and plasma pressure gradient (e.g., Alken et al. 2016; Maute and Richmond 2017), we do not expect that they have strong effects on our analyses. First, the gravity and pressure gradient currents are mainly in the zonal direction while we focus on vertical currents. Second, these two zonal currents tend to cancel above the F-region peak altitude (e.g., Maute and Richmond 2017, Abstract): that is, where GRACE-FO and CryoSat-2 are located. Third, the periods examined here are mainly during low solar flux, when the ionosphere is in general more tenuous and gravity and pressure gradient currents are weaker than during high solar flux. Fourth, vertical field-aligned currents feeding the two zonal currents are, according to Maute and Richmond (2017 Figure 1), hemispherically symmetric at off-equatorial latitudes and nearly zero at the equator. These current directions are different from those of IHFACs (hemispherically anti-symmetric off the equator) and F-region dynamo currents (nonzero at the equator) addressed in this study.
Estimating noise levels of platform magnetometer data
It would give useful technical information to check the overall fluctuation levels of the vertical current density from platform magnetometer data. Figure 10 has the same structure as Fig. 2, but the color represents absolute values of the ‘adjacent difference’ (or first-order time derivative) of unfiltered vertical current density. We use ‘unfiltered’ data to highlight fluctuation levels in the raw data. Also, instead of the combined data out of all the three FGMs onboard CryoSat-2, we use the single FGM1 in this section, which is representative of all the three magnetometers (Olsen et al. 2020). We first calculate the adjacent difference in the time series of vertical current density: j(t + dt) − j(t), where j is unfiltered vertical current density, t is time, and dt is temporal resolution of the data (4 s for CryoSat-2). Then, its absolute magnitude (|j(t + dt)-j(t)|) is bin-averaged as a function of MLAT, MLT, and season. The results can give an idea as to how intense the data fluctuations (presumably due to noise of platform magnetometers) are. In Fig. 10, the overall fluctuation levels of CryoSat-2 data are confined to a narrow range between 150 nA/m2 and 180 nA/m2. The fluctuation levels exhibit little dependence on the season, are slightly stronger at |MLAT|< 35° than at |MLAT|> 35°, and have shallow minima near 00 and 12 MLT. These features are different from distributions of geophysical current fluctuations, as reported by earlier studies. According to Aoyama et al. (2017, Figures 1–2) and Yin et al. (2019, Figure 9–10), both of which were based on science-grade magnetometer data onboard Swarm, geophysical magnetic fluctuations (and concomitant current fluctuations) significantly depend on season and MLT. Hence, it is reasonable to conclude that the fluctuations in Fig. 10 mostly originate from instrumental noise and artificial disturbances. The adjacent difference (150–180 nA/m2) much larger than average IHFAC magnitudes (a few nA/m2) suggests that long-term accumulation of the CryoSat-2 magnetic data (e.g., > 8 years as in this study) is necessary for getting well-organized IHFAC distributions as shown in Fig. 2. A large amount of data can reduce the standard error of the mean, which decreases with the square root of the bin population.
At the moment, we do not have a good explanation for the low fluctuation levels at mid-latitudes (|MLAT|> 35°) and near the noon–midnight meridian. We speculate that oblique incidence of sunlight on the left or right side of the satellite may disturb the measurement, but further analyses in the future are warranted to verify this conjecture.
Figure 11 is the same as Fig. 10, but for GRACE-FO1. The fluctuation levels are generally stronger than those of CryoSat-2. The stronger fluctuations cannot be attributed solely to the difference in sampling rates (4 s for Cryosat-2 and 1 s for GRACE-FO) because both are spot reading values (that is, neither has been time-averaged to get a smoother signal). The distribution of fluctuation levels in the MLAT-MLT space is complicated and does not conform to previous studies such as Nakanishi et al. (2014, Figure 5) and Aoyama et al. (2017). For example, Aoyama et al. (2017, Figures 2 and 4) report a clear local minimum of fluctuation intensity near the equator and no preference for the summer hemisphere during December solstice. Neither of the features can be seen in our Fig. 11, which suggests that Fig. 11 is also significantly affected by instrument noise and artificial disturbances from the satellite body.
Figure 12 compares fluctuation levels of vertical current density between (a) Swarm, (b) CryoSat-2 (FGM 1 data), and (c) GRACE-FO for an example day on 01 June 2018. For CryoSat-2 and GRACE-FO, black and red dots represent unfiltered and filtered data (with ~ 20-s low-pass filters; see “Satellites, Instruments, and Data Processing Methods” section), respectively. As for Swarm, we just present unfiltered data. Figure 12 shows that the unfiltered platform magnetometer data (black dots: CryoSat-2 and GRACE-FO) exhibit stronger scatter than the science-grade magnetometer data onboard Swarm. However, even Swarm data exhibit significant scatter, which we assume geophysical, in comparison to the averaged IHFAC strength of several nA/m2. Platform magnetometer data after filtering (red dots) have significantly reduced fluctuation levels, which become nearly comparable to that of Swarm data. Though the fluctuation levels are still larger than several tens of nA/m2 in both the Swarm and filtered platform magnetometer data, long-term accumulation of those data can reproduce the climatology of IHFACs (of the order of one nA/m2), as we have demonstrated throughout this paper. We also checked the standard error of the mean for Figs. 2, 3, and found that the errors are generally smaller than the magnitude of IHFACs (figures not shown). Note that such low-amplitude currents in platform magnetic field data may depend on the chosen stability of calibration parameters. However, it is an encouraging result that well-known geophysical signals such as IHFACs clearly emerge after data stacking.