Traits of sub-kilometre F-region irregularities as seen with the Swarm satellites

Abstract. During the night, in the F-region, equatorial ionospheric irregularities manifest as plasma depletions observed by satellites, and they may cause radio signals to fluctuate. In this study, the distribution characteristics of ionospheric F-region irregularities in the low latitudes were investigated using 16 Hz electron density observations made by a faceplate which is a component of the electric field instrument (EFI) onboard Swarm satellites of the European Space Agency (ESA). The study covers the period from October 2014 to October 2018 when the 16 Hz electron density data were available. For comparison, both the absolute (dNe) and relative (dNe∕Ne) density perturbations were used to quantify the level of ionospheric irregularities. The two methods generally reproduced the local-time (LT), seasonal and longitudinal distribution of equatorial ionospheric irregularities as shown in earlier studies, demonstrating the ability of Swarm 16 Hz electron density data. A difference between the two methods was observed based on the latitudinal distribution of ionospheric irregularities where (dNe) showed a symmetrical distribution about the magnetic equator, while dNe∕Ne showed a magnetic-equator-centred Gaussian distribution. High values of dNe and dNe∕Ne were observed in spatial bins with steep gradients of electron density from a longitudinal and seasonal perspective. The response of ionospheric irregularities to geomagnetic and solar activities was also investigated using Kp index and solar radio flux index (F10.7), respectively. The reliance of dNe∕Ne on solar and magnetic activity showed little distinction in the correlation between equatorial and off-equatorial latitudes, whereas dNe showed significant differences. With regard to seasonal and longitudinal distribution, high dNe and dNe∕Ne values were often found during quiet magnetic periods compared to magnetically disturbed periods. The dNe increased approximately linearly from low to moderate solar activity. Using the high-resolution faceplate data, we were able to identify ionospheric irregularities on the scale of only a few hundred of metres.



Data and Methods
The Swarm mission is made up of three same satellites (Swarm A, B, and C) with an orbital speed of around 7.5 km s −1 in polar orbits. Each satellite is equipped with an Electric Field Instrument (EFI) in addition to other payloads. The EFI consists of LPs, and Thermal Ion Imagers (TII) (Knudsen et al., 2017). The LPs measure simultaneously the electron density (N e ), electron temperature (T e ), and spacecraft potential at a frequency of 2 Hz along the satellites' track. The Swarm satellites also 5 measure N e at a frequency of 16 Hz with a faceplate only when the TII is inactive. The Plasma density is derived from the faceplate current assuming that it is carried by ions hitting the faceplate due to the orbital motion of the spacecraft (Buchert, 2016). By October 2018, Swarm A and C were orbiting next to each other in polar orbits (inclination angle of 87.35 • ) at about 1.5 • longitudinal spacing (Kil et al., 2019) and 442 km altitude above sea level over the low latitude region, and Swarm B was orbiting at an altitude of about 506 km (inclination angle of 87.75 • ). In a day, the swarm satellites complete about 16 10 orbits with an average orbital period of about 91.5 min. Swarm satellites regress in longitude around 22.5°between orbital ascending nodes. Swarm A and C need about 133 days to complete all 24 hours of local time and Swarm B needs about 141 days (Xiong et al., 2016b). Data sets measured by Swarm can be downloaded from http://earth.esa.int/swarm.
The investigations done in this study are based on the 16 Hz N e faceplate data collected for the period of October 2014 to October 2018. 15 The identification criteria adopted for quantifying ionospheric irregularities have been a matter of concern. Some earlier studies (e.g, Kil and Heelis, 1998;McClure et al., 1998;Burke et al., 2003;Su et al., 2006;Kil et al., 2009;Dao et al., 2011, etc) used relative plasma density disturbance to identify ionospheric irregularities while others (e.g, Lühr et al., 2014;Buchert et al., 2015;Xiong et al., 2016b, etc) took absolute density disturbance. However, Huang et al. (2014) used the 512 Hz Communication / Navigation Outage Forecasting System (C / NOFS) satellite's measurements of ion density and found that 20 when the relative and absolute density disturbances are used independently, the likelihood of irregularities occurring and their variation with local time differ. The C / NOFS satellite was in a low tilt orbit, so the bubbles were sampled zonally. Important differences basing on latitudinal distribution of ionospheric irregularities using different criteria could not be addressed by Huang et al. (2014). The polar-orbiting Swarm satellites sample bubbles in a meridional direction and give an opportunity to check the difference in their latitudinal distribution using different identification criteria. 25 For comparison purposes, two methods were adopted for the polar-orbiting Swarm satellites to quantify the level of electron density irregularities. In the first method, the 16 Hz N e measurements were passed through a 2-s (32 data points) running mean filter corresponding to a wavelength of about 15 km. From the original observations, the filtered data were subtracted to obtain the residual dN e = N e − N e ; where N e is the mean of N e at a 2-s interval. The standard deviation of the residuals which represents the density perturbation, ∆N e was then calculated for every 32 data points. Basu et al. (1976) found that, 30 on a global scale, ∆N e = 1 × 10 −10 m −3 represents the percentage occurrence of 140 MHz scintillations. Xiong et al. (2010) used absolute density disturbance thresholds of 5 × 10 10 m −3 and 3 × 10 10 m −3 respectively to identify density irregularity structures on CHAMP and GRACE observations. Wan et al. (2018) adopted absolute density perturbation > 5 × 10 10 m −3 to identify ionospheric irregularities from Swarm. Basing on the method used in the current study, only batches with ∆N e greater than 0.25 × 10 10 m −3 were considered to be significantly irregular and selected for extra processing and analysis.
In the second method, ∆N e was divided by N e to obtain the relative perturbation, ∆N e /N e . There is no specific threshold definition to be used when ∆N e /N e identifies irregularities (Huang et al., 2014;Wan et al., 2018). Kil and Heelis (1998) determined the likelihood of occurrence of relative disturbance > 1%(0.01) and 5%(0.05) from Atmospheric Explorer-E (AE -E) satellite data. AE -E data was also used by McClure et al. (1998), but relative disturbances > 0.5%(0.005) were used 5 to identify irregularities. To identify the occurrence of ROCSAT-1 irregularities, Su et al. (2006) and Kil et al. (2009) used a threshold of 0.3%(0.003) for the relative disturbance. Huang et al. (2014) used high-resolution ion density measurements from C / NOFS satellite and took relative perturbation > 1%(0.01). Wan et al. (2018) considered ∆N e /N e values larger than 20%. In the current study, only batches with ∆N e /N e > 0.01 were considered to be significantly irregular and used for further analysis basing on the methods adopted. The results are presented and discussed in the following section.

Results and Discussions
The high-resolution Swarm faceplate N e data were used to characterize ionospheric irregularities using procedures described in Sect. 2. Figure 1 shows examples of N e results for arbitrary passes of Swarm A and C on 2014-10-06 and 2015-07-03, respectively from the LP and faceplate to highlight the capability of the 16 Hz N e data for observations of irregularity density structures. In Fig. 1 Hz data as shown in the zoomed-in sections in Fig. 1. One of the drawbacks associated with the 16 Hz N e data, as mentioned earlier, is that it is only recorded when the TII is inactive. Therefore, to check data availability, plasma density structures, these depletions are ionospheric irregularities. The RTI is the most known mechanism that causes irregularities in low latitudes (Kelley, 2009;Kintner et al., 2007). The lower ionospheric layer declines rapidly during the night compared to the top layer. This creates a sharp vertical gradient of plasma density directed upwards, contrary to the gravitational force's direction of action. For such unstable arrangement, irregularities in the F -region at the bottom intensify and drift up, creating more complex plasma structures that extend to higher altitudes along magnetic field lines (Woodman and 15 La Hoz, 1976;Abdu, 2005;Kelley, 2009). In general, ionospheric irregularities are more intense at the Equatorial Ionization Anomaly (EIA) belts (±15 • QLat) than at the geomagnetic equator as observed in Fig. 2. However, from Fig. 2, the event presented for Swarm A and C on 2015-03-07 shows high values of ∆N e /N e even at the magnetic equator. Huang et al. (2014) also observed that the relative and absolute perturbations were both able to capture fluctuations in ion density measurements made along C / NOFS tracks during 2008 − 2012 in the zonal direction. However, it was not possible to see a more detailed 20 latitude distribution using C / NOFS satellite because it covered a small latitude range of about ±13 • due to its low inclination angle of about 13 • . The local time distribution characteristics of ionospheric irregularities were also determined and the results are presented and discussed in the following subsection.

Local Time Distribution of Ionospheric Irregularities
It is known from many studies (e.g, Kil and Heelis, 1998;Burke et al., 2004;Su et al., 2006;Stolle et al., 2006;Dao et al., 25 2011;Huang et al., 2014;Xiong et al., 2016b;Wan et al., 2018, etc) that ionospheric irregularities in the low latitudes occur   than 25 • over each independent station were considered to reduce the multipath effects. ROT I values > 0.5 TECU/min 15 (1 TECU = 10 16 el/m 2 ) were classified as irregularities/scintillations (Ma and Maruyama, 2006). Figure 5 presents the percentage occurrence of ROT I > 0.5 TECU/Min in 1-hour local time bins for the different IGS stations and seasons. The trend followed by local time distribution of ROT I seems to closely agree with that of ∆N e and ∆N e /N e in the equinoxes and December Solstice. As expected the percentage occurrence of ionospheric irregularities is higher mainly for the IGS stations in the African longitude even in June Solstice (Yizengaw et al., 2014). However, the enhanced post-midnight irregularities seen

Seasonal and Longitudinal Distribution of Ionospheric Irregularities
The Swarm mission's 16 Hz N e data collected over the 5-year period (2014 − 2018) has a credible global spatial and temporal coverage that is sufficiently good for examining the seasonal and longitudinal distribution of ionospheric irregularities in low longitude. The occurrence rate of ionospheric irregularities does not always correspond to the highest amplitude of irregularity structures from the results presented by Wan et al. (2018). Therefore, here we concentrate on the magnitude of ionospheric irregularities other than the rate of occurrence. Zakharenkova et al. (2016) compared Swarm A and B 1-s N e data and revealed satellite-to-satellite differences related to altitude, longitude, and local time. Here, we also show the results for all the three satellites separately. Figure 6 shows the seasonal and longitudinal distribution of ∆N e during the period of study in geographic  The first noticeable feature in Fig. 6 and Fig. 7 is that almost all the irregularities occur within the EIA belts between about ±15 • −±20 • magnetic latitudes. However, Fig. 6 shows that absolute variations of ∆N e are observed with a gap of low values at the magnetic equator, while in Fig. 7 maximum values of ∆N e /N e extend from the northern crest to the southern crest, 10 including the magnetic equator. A clear picture of the density variations across the magnetic equator is seen in a scatter plot of the irregularities as a function of latitude as shown in Fig. 8. Some earlier studies (e.g, Burke et al., 2004;Su et al., 2006, etc) observed a normal-like distribution that peaks at the quasi -dipole equator and gradually decreases towards higher latitudes,    Fig. 6 and Fig. 7 is consistent with earlier studies irrespective of the criteria adopted (e.g., Su et al., 2006;Huang et al., 2001;Burke et al., 2004;Park et al., 2005;Huang et al., 2014;Zakharenkova et al., 5 2016;Wan et al., 2018). RTI is known to intensify after sunset, causing severe irregularities when the day-night terminator is aligned with the plane of the magnetic field that occurs in the equinox (Tsunoda, 1985;Burke et al., 2004;Gentile et al., 2006;Yizengaw and Groves, 2018).
One of the challenges has been explaining the mechanism governing the longitudinal distribution of irregularities. Tsunoda (1985) proposed a model based on the magnetic declination to explain the distribution of ionospheric irregularities. However, 10 this model could not explain the high occurrence of irregularities in June solstice over the African longitude. The longitudinal distribution of irregularities has also been attributed to gravity waves originating from the thermosphere (Yizengaw and Groves, 2018, and references therein). Yizengaw and Groves (2018) also added that the intertropical convergence zone (ITCZ) position, which are sources of gravity waves, may explain the longitudinal irregularity dependence observed. Kil et al. (2004) suggested that the longitudinal distribution at EIA latitudes of absolute electron density affects the occurrence of irregularities. Using 15 DMSP data, Huang et al. (2001), Huang et al. (2002), andBurke et al. (2004) showed that the pattern of precipitation of the inner radiation belt's energetic particles explains the pattern of irregularities. Among other parameters, the growth rate of equatorial ionospheric irregularities is controlled by the electron density gradient. Ionospheric irregularities in the equatorial and low latitudes can cascade upwards and along the magnetic field lines to the EIA belts characterized by high background N e and steep gradients in density (Muella et al., 2010). From both local time and longitudinal perspectives, Wan et al. (2018) confirmed that the depletion amplitudes of irregularities are closely linked to the background electron density intensity. Xiong et al. (2016a) concluded that GPS signal reception may be interfered by 5 small-scale plasma density structures with large-density gradients in zonal and meridional directions. Here, we attempt to compare the seasonal and longitudinal distribution of electron density gradient in the meridional direction along the tracks of the Swarm satellites with the magnitudes presented in Fig. 6 and Fig. 7  depletion was divided by the corresponding latitudinal distance in degrees. Figure 9 presents the N e gradient, ∇N e classified in different seasons for Swarm A, C, and B independently. Seasonal and longitudinal distribution of ∇N e generally shows the 10 same pattern as that of ∆N e and ∆N e /N e . In regions with steep gradients in N e , the highest values of ∆N e and ∆N e /N e are often found. Therefore, the amplitudes of ionospheric irregularities closely depend on background electron density (Wan et al., 2018) and steep N e gradient globally, as expected.  Sobral et al., 2002;Huang et al., 2002;Gentile et al., 2006;Stolle et al., 2006;Li et al., 2009;Basu et al., 2010;Sun et al., 2012;Carter et al., 2013;Huang et al., 2014). By using different criteria, Huang et al. (2014) determined the solar activity dependence of the occurrence of irregularities. In addition 10 to solar activity, we also used different criteria to check the effects of magnetic variability on the distribution characteristics of irregularities in low latitudes.
Scatter plots of (a) ∆N e and (b) ∆N e /N e as functions of F10.7 are shown in Fig. 11 for Swarm A, B, and C, independently.
Each panel of Fig. 11 contains a linear fit and the correlation coefficient R. In general, both ∆N e and ∆N e /N e show weak positive correlation with F10.7. However, a higher correlation was obtained between ∆N e and F10.7 (maximum of 0.37 for 15 Swarm A) compared to ∆N e /N e and F10.7. For Swarm A alone, Fig. 12 shows the solar variation effect on seasonal and longitudinal distribution of ionospheric irregularities. The results are divided into two major columns (distribution with respect to ∆N e to the left and ∆N e /N e to the right). In each major column, there are two sub-columns, one for low solar activity (F 10.7 < 140) and the other for moderate solar activity (140 F 10.7 < 180). It is important to point out that a reduced number of days were used to generate the climatology maps when 140 F10.7 < 180 compared to when F10.7 < 140. In Fig. 12 Circulation Model (MTIEGCM), Vichare and Richmond (2005) showed that upward evening drift increases at a similar rate in all longitude sectors with solar activity. Therefore, the high occurrence of irregularities during moderate or high solar activity period may be because of the atmospheric driver for the zonal electric field which intensifies during moderate/high solar activity, causing an increase in the RTI growth rate. pre-midnight plasma data, Huang et al. (2001) found that the rate of occurrence of irregularity and geomagnetic activity were negatively correlated. We also examined the geomagnetic effect on the seasonal and longitudinal distribution of irregularities as presented in Fig. 14 for Swarm A. The results are divided into two major columns (distribution with respect to ∆N e to the left and ∆N e /N e to the right). In each major column, there are two sub-segments, one for calm geomagnetic occasions (Kp 10 < 3) and the other for geomagnetically disturbed periods (Kp 3). From Fig. 14, high values of both ∆N e and ∆N e /N e are frequently observed when Kp < 3. Geomagnetic activity affects irregularity occurrence in the low latitudes in two noteworthy ways i.e., by the brief entrance of auroral electric fields (Fejer, 1991;Kikuchi et al., 1996) and by the unsettling influence of dynamo effects (Blanc and Richmond, 1980). The second mechanism produces disturbance electric fields which last for a long time. The disturbance electric fields are westward after sunset (Blanc and Richmond, 1980;Huang et al., 2005;Abdu, 2012).  It is important to note that the well-known trend in the longitudinal distribution of ionospheric irregularities for some seasons may not be clearly observed in Fig. 12 and Fig. 14 because of limited data after categorizing with respect to Kp or F 10.7.

Conclusions
In this study, we have used Swarm N e data measured by the faceplate at a frequency of 16 Hz to examine the distribution characteristics of ionospheric irregularities in the equatorial and low latitude ionosphere from 2014−2018 when the 16 Hz data 5 was available. Two methods (absolute and relative perturbation) were used to quantify the level of ionospheric irregularities.
Both methods were able to capture fluctuations in electron density along single satellite passes. Basing on the large number of Swarm low latitude crossings for the years 2014 − 2018, the local time, seasonal and longitudinal distribution of ionospheric irregularities in the low latitudes were examined. We demonstrated the importance of steep density gradients for the generation and distribution of ionospheric irregularities in the low latitudes. We also checked the effects of geomagnetic and solar activity Also, symmetry about the magnetic equator is observed with ∇N e . Therefore, in addition to the background electron  density presented by Wan et al. (2018), the longitudinal distribution of ionospheric irregularities also depends on steep electron density gradients as expected.
(iv) The occurrence of ionospheric irregularities quantified using ∆N e shows a weak positive correlation with F 10.7 and the correlation is even lower with ∆N e /N e . The seasonal and longitudinal distribution of the ionospheric irregularities shows slightly different trends between ∆N e and ∆N e /N e . On the other hand, the distribution of ionospheric irregularities is 5 still lower during the geomagnetically disturbed period than in quiet times.
Despite the obvious limitations of using polar-orbiting satellites to monitor equatorial electrodynamics, Swarm has provided credible distribution characteristics of ionospheric irregularities in the low latitude region with data accumulated in five years (2014 − 2018). In general, the initial observations of the distribution characteristics of ionospheric irregularities using the 16 Hz N e data are in good agreement with earlier works that have addressed similar concepts. This has demonstrated the ability 10 of Swarm faceplate N e data for ionospheric studies. Therefore, the 16 Hz faceplate data is a useful measurement that can be adopted in order to understand ionospheric irregularities.