Characteristics of GPS TEC variations in the polar cap ionosphere
Abstract
This paper presents statistical characteristics (occurrence rate, amplitude, and frequency) of low-frequency (<100 mHz) variations in total electron content (TEC) observed in the polar cap ionosphere. TEC variations were primarily associated with mesoscale (tens to hundreds of kilometers) ionization structures and were observed by five Global Positioning System (GPS) receivers over a 6 year period (2009–2014). The altitude of ionization structures was estimated by using colocated ionosonde radars. High data rate receivers combined with broad spatial coverage of multisatellite TEC measurements provided high-resolution magnetic local time/latitude maps of TEC variation characteristics, which were examined as a function of solar cycle and season. These high-resolution maps improve upon the current observational picture of mesoscale structuring in the polar cap and provide accurate links to particular magnetospheric source regions. Occurrence of TEC variations was consistently highest in dayside regions mapping to low latitude and plasma mantle boundary layers, while largest-amplitude TEC variations were observed in dayside regions close to the polar cusp, and lower latitudes around midnight. Occurrence and amplitude of TEC variations increased significantly during the ascending phase of the solar cycle, independent of solar wind conditions, while seasonal statistics showed highest dayside occurrence and amplitude in winter months, lowest in summer, and highest nightside occurrence and amplitude around equinox. A surprising result in the frequency distributions of TEC variations was discrete frequencies of about 2 and 4 mHz, which appeared to originate from regions corresponding to the plasma mantle, immediately poleward of the polar cusp.
Key Points
- GPS TEC measurements provide a detailed statistical picture of mesoscale ionospheric structuring in the polar cap
- Occurrence rate and amplitude of TEC variations depend on magnetic coordinates, solar cycle, and season
- TEC variations at preferential frequencies were observed in regions poleward of the dayside cusp
1 Introduction
The ionosphere in the polar cap region is structured on a wide range of temporal and spatial scales due to direct exposure to the highly dynamic solar wind and magnetospheric boundary layers. Due to an incomplete observational picture and low understanding of the generation and morphology of polar cap ionization structures, our current ability to forecast irregularities in the polar cap ionosphere is poor [e.g., Baker et al., 2008]. In addition, ionospheric models such as the International Reference Ionosphere (IRI) and Klobuchar Ionospheric Model are unreliable and inaccurate in polar cap regions [e.g., Themens et al., 2014]. Reliable operation of Global Navigation Satellite Systems (GNSS) and high-frequency (HF) communication links in the polar cap requires an improved observational picture of the polar cap ionosphere and empirical models of ionospheric structure suitable for high latitudes. The intent of this study is to improve upon the currently incomplete observational picture of mesoscale (tens to hundreds of kilometers) structuring of the polar cap ionosphere and to investigate the source and generation mechanisms of ionization structures that manifest as variations in TEC.
Several coupling processes are known to significantly impact the polar cap ionosphere including magnetic reconnection between geomagnetic and interplanetary field lines [Lester et al., 2006], viscous interaction of the solar wind and magnetosphere boundary [Axford and Hines, 1961], electrodynamic coupling of the ionosphere and outer magnetosphere via magnetic field-aligned currents [Iijima et al., 1978], and direct precipitation of solar wind plasma into the ionosphere [Newell et al., 2009]. How coupling processes materialize and evolve is not fully understood. Electric fields, currents, and plasma drifts are generated in the polar cap ionosphere due to solar wind, magnetosphere, and ionosphere (SW-M-I) interactions, which are closely linked to instabilities in ionospheric plasma occurring on a wide range of time and spatial scales. Variations in plasma density, turbulence in plasma flows, plasma convection and heating, auroral brightening, and neutral wind disturbances are all ubiquitous features of the polar cap ionosphere.
Solar wind plasma has direct access to the magnetosphere via the polar cusps, points of diverging magnetic field on the dayside magnetosphere. Polar ionosphere regions magnetically mapping to the cusp usually span 1.5–2.5 h local time across noon and 1.0°–1.5° latitude, with the exact location and span of the cusp depending heavily on interplanetary magnetic field (IMF) orientation [Newell et al., 2004]. Cusp particle populations have compositions and energies comparable to solar wind particles, which are characterized by relatively low energy (tens of eV) electrons, high kinetic energies compared to thermal energies, and a high density of ions [Newell and Meng, 1992; Woch and Lundin, 1992]. Observations have indicated that precipitation of accelerated ions near the equatorward edge of the cusp can be energetic (>1 keV) enough to ionize the E region ionosphere [e.g., Galand and Richmond, 2001]. Magnetospheric boundary layers in the immediate vicinity of the cusp can be described as interfaces or transition layers between shocked solar wind plasma of the magnetosheath and the magnetopause. The low-latitude boundary layer (LLBL) and mantle boundary layers are located equatorward and poleward of the cusp, respectively, and have magnetic footprints in the polar ionosphere extending across noon into the afternoon and morning sectors between 06:30 and 17:00 magnetic local time (MLT) [Newell et al., 2004]. The LLBL extends to higher latitudes in the dawn-dusk sectors flanking the cusp. The plasma mantle typically consists of lower energy particles decelerated in the magnetosheath, and a significantly lower density of particles compared to the cusp, while particles of LLBL origin are slightly energized compared to those of the cusp [e.g., Newell and Meng, 1994]. High-energy particle precipitation is a known feature of ionospheric regions mapping to the mantle and LLBL, mainly due to disruptions of magnetopause currents and subsequent acceleration of energetic particles via field-aligned currents into the polar cap [Newell et al., 1991; Echim et al., 2008]. A high occurrence of discrete auroral forms associated with this energetic precipitation is a well-known feature of morning-afternoon ionospheric regions mapping to the dawn-dusk flanks of the LLBL [e.g., Lorentzen and Moen, 2000].
Dayside extensions of the boundary plasma sheet (BPS) and central plasma sheet (CPS) also potentially influence lower dayside polar cap latitudes covered by GPS receivers, according to maps of Newell et al. [2004]. Average energies of precipitating BPS and CPS electrons are typically greater than 1 keV, which more resemble auroral precipitation compared to energies on the order of tens to hundreds of eV characteristic of boundary layers impacting higher latitudes.
High-energy substorm precipitation originating from the magnetospheric plasma sheet can also impact polar cap latitudes since the auroral oval expands both equatorward and poleward during substorms. The breadth of the expansion varies from event to event, with precipitation and aurora associated with larger substorms typically reaching farther south and deeper into the polar cap than less intense events. The nightside oval commonly reaches 75° magnetic latitude (MLat), with observations of substorm phenomena above 80° MLat possible but rare. Watson et al. [2011] showed nighttime TEC variations of up to 6 TECU (1 TECU = 1016 electrons/m2) associated with high-energy (tens to hundreds of keV electrons) substorm precipitation in the polar cap, at magnetic latitudes up to ~83° MLat. Sustained TEC variations with discrete frequencies in the range of ~8–33 mHz were also observed, possibly associated with ULF wave modulation of precipitation into the polar cap [e.g., Weatherwax et al., 1997].
Latitudes deep in the polar cap, beyond regions mapping directly to boundary layers and the plasma sheet, are characterized by a steady, low-intensity, largely homogeneous precipitation of low-energy ions and electrons (tens to hundreds of eV) called the polar rain. More intense, high-energy precipitation is occasionally observed in localized regions within the polar cap, typically associated with localized intensification of the polar rain [Newell et al., 2009]. These high-intensity events often produce visual aurora in the lower F and E layers.
Mesoscale (tens to hundreds of kilometers) structures in the polar cap consist mainly of polar patches and polar cap aurora, which are ionospheric ionization structures that vary in geometry and evolution, and have a broad range of generation mechanisms and source regions. Smaller-scale (microscale—less than a few kilometers) structures are known for the rapid phase and amplitude fluctuations (scintillations) they produce in transionospheric radio signals and are beyond the scope and time resolution of data used in this study. Larger-scale structures (macroscale—e.g., tongues of ionization), comparable to the size of the polar cap, are often extensions of midlatitude storm-enhanced densities and also occur on time and spatial scales outside the interest of this study.
Polar cap patches are F region density enhancements/depletions (>100 km) most often observed during periods of southward IMF (negative IMF Bz). The majority of observations indicate that patches form around noon near the polar cusp and subsequently follow the prevailing plasma convection through the polar cap [e.g., Buchau et al., 1983; Weber et al., 1984]. Past studies have indicated that patches originate from higher density plasma (e.g., more sunlit plasma) at lower latitudes, which is transported into the polar cap in the form of a “tongue of ionization”. This enhanced density plasma can then be “chopped up” by variations in the IMF and open-closed field line boundary as it is drawn into the polar cap [e.g., Anderson et al., 1988], forming patches. Other suggestions of patch origin include external plasma sources such as ionization due to precipitation from the cusp region [Walker et al., 1999], density depletion in the cusp region due to enhanced plasma recombination resulting from ion heating [Tsunoda, 1988], or from auroral region precipitation of the dawnside convection cell, where ionospheric plasma is exposed to precipitation for long durations due to the corotation effect [MacDougall and Jayachandran, 2007]. Several studies have reported GPS TEC observations of polar cap patches [e.g., Burston et al., 2010; Krankowski et al., 2006], with reported TEC enhancements of up to 10–15 TECU. Magnitude, spatial scales, and temporal scales of electron density variations associated with patches can be quite random, typically resulting in lower frequency (<2 mHz) variations in TEC.
Auroral forms observed in the polar cap are often interpreted as signatures or “footprints” of electrodynamic processes in the magnetospheric boundary layers or plasma sheet. Structure and dynamics of auroral forms largely depend on their source region [e.g. Samara and Michell, 2013]. Poleward moving auroral forms (PMAFs) are midday auroral intensifications that either originate in the auroral region and are convected into the polar cap, or are a result of solar wind plasma entering the magnetosphere via boundary layers. PMAFs are thought to be a consequence of periodic intensifications of solar wind plasma entry associated with periodic enhancements of reconnection rates equatorward of the cusp [e.g., Fasel, 1995]. Observations of periodic PMAF occurrence have shown 1–4 min (4.2–16.7 mHz) periodicity due to modulation of the reconnection [Thorolfsson et al., 2000; Taguchi et al., 2012]. Jayachandran et al. [2012] introduced observations of quasiperiodic variations in GPS TEC associated with poleward moving Sun-aligned arcs (PMSAAs). In that study, concurrent all-sky images showed multiple morning side features originating adjacent to the auroral oval and propagating poleward, while multiple GPS satellites from two receivers showed TEC variations of 1–4 TECU in magnitude with discrete frequency components in the range of 1.6–22.8 mHz. These frequencies were associated with poleward motion of the arcs as well as smaller-scale structuring within the arcs. Jayachandran et al. [2009a] presented GPS TEC observations of a polar cap arc, which resulted in TEC enhancements of 0.3–2.0 TECU associated with E region ionization.
Periodic variations in riometer absorption and optical observations of pulsating aurora in the polar regions have been attributed to modulation of particle precipitation by magnetospheric ULF waves. Weatherwax et al. [1997] observed significant variations in riometer absorption associated with high-energy substorm precipitation in the Antarctic polar cap. Variation frequencies were of 1–4 mHz and most likely a result of modulation of the precipitation by Pc5 Alfven waves. Similar frequencies were simultaneously observed in ground magnetometer data. Since observations from multiple stations showed discrete frequencies of oscillation which occurred in a latitudinally narrow region (~1°), the authors speculated this event resulted from compressional mode excitation of field line resonance on closed field lines threading the plasma sheet.
In this study, high-resolution (1 Hz) GPS TEC measurements of the Canadian High Arctic Ionospheric Network (CHAIN) [Jayachandran et al., 2009b] were used to statistically examine the characteristics of TEC variations in the polar cap ionosphere. TEC measurements provide high temporal and spatial resolutions due to multisatellite measurements available to each receiver on the ground. Past studies have shown that GPS TEC is an effective tool for studying small-scale structuring within polar cap irregularities [Jayachandran et al., 2012] and tracking the horizontal motion of irregularities [Watson et al., 2011; Jayachandran et al., 2011].
Aside from better understanding of the physics, results from this work will contribute to ongoing improvements of satellite- and ground-based communication capabilities at high latitudes. Ionospheric irregularities result in significant GNSS positioning errors [Lanzerotti, 2001], as well as enhancements or severe limitations of high-frequency (HF) radio communication capabilities [Zaalov et al., 2003]. Small-scale (microscale) irregularities and sharp density gradients associated with polar patches and other ionization structures are known to produce scintillations, cycle slips, and loss of lock in GNSS signals traversing the high-latitude ionosphere [e.g., Weber et al., 1984]. Since the source and mechanism of polar cap ionosphere irregularities are not well understood, the ability to forecast and mitigate effects on navigation and communication signals is limited. Due to increasing infrastructure, air and marine traffic, and exploration of natural resources in the Arctic, both Government and industry have significant interest in improving navigation and communication capabilities in this region. Statistical results of this study can potentially improve capabilities to forecast occurrence of particular plasma irregularities based on solar wind conditions and local time/latitude, thus increasing capabilities to predict and correct for subsequent effects on ionospheric radio signals. Linking TEC variations to particular generation mechanisms and sources regions will also improve capabilities to resolve and model effects on communication signals and navigation links.
Statistical results have also shown that occurrence rate and amplitude of TEC variations have a strong dependence on a number of solar wind parameters, including the BX, BY, and BZ components of the IMF, as well as solar wind speed. The solar wind dependence of TEC variations is presented in a separate manuscript titled “GPS TEC variations in the polar cap ionosphere: Solar wind and IMF dependence,” which follows this manuscript.
2 Data and Method of Analysis
Figure 1 shows geomagnetic coordinates of the five colocated CHAIN GPS receivers and CADI ionosondes used in this study. CHAIN GPS receivers are GPS Ionospheric Scintillation and TEC Monitors (GISTMs) model GSV4000B [Van Dierendonck and Arbesser-Rastburg, 2004]. A GISTM consists of a NovAtel OEM4 dual-frequency receiver with special firmware specifically configured to measure amplitude and phase scintillation derived from the L1 frequency GPS signals and ionospheric TEC derived from the L1 and L2 frequency GPS signals. This receiver is capable of tracking and reporting scintillation and TEC measurements from up to 10 GPS satellites in view. Phase and amplitude data are sampled and logged, either in raw form or detrended, at a rate of 50 Hz. Receivers are currently fed by a NovAtel GPS-702 antenna. Temporal resolution for TEC data in this study is 1 s. The spatial resolution of TEC measurements varies with the speed of the signal path through the ionosphere, which, in turn, varies with altitude of ionization and satellite elevation. For 1 s TEC measurements and a satellite elevation cutoff of 25°, the spatial resolution is between 18 and 350 m.
There were large gaps in data unavailability in Hall Beach (February–June 2013) and Resolute (May–August 2014). Taking these large gaps into account, in addition to shorter periods of data outages and poor data quality, total combined data availability from all five receivers during the 6 year period of interest was 92%.
TEC is electron volume density integrated along the GPS satellite to receiver raypath, with 1 TEC unit (TECU) defined as 1016 electrons per square meter. Satellite motion produces a long-period trend in the TEC since the time a transmitted GPS signal spends in the ionosphere increases with satellite elevation. This trend, in addition to lower frequency and diurnal TEC variations, is removed from the TEC by using a third-order Butterworth high-pass frequency filter at 0.37 mHz (45 min). GPS satellites are identified by their pseudo random noise (PRN) number, which refers to each satellite's unique PRN code.
(1)
(2)All available GPS TEC data for 1 January 2009 to 31 December 2014 from five CHAIN GPS receivers was collected and analyzed. This study presents statistical analysis for TEC variations with peak-to-peak amplitudes greater than 1 TECU. Total vertical TEC in the polar cap is typically in the range of 1–30 TECU [Themens et al., 2013]. This study is limited to TEC measurements collected from satellites at elevation angles greater than 25°, and TEC constituting at least 1 h of continuous measurements at greater than 25° satellite elevation, with no cycle slips. Cycle slips occur when sharp ionospheric density gradients result in loss of lock between the GPS satellite and ground receiver. To collect frequency statistics of TEC variations, dynamic power spectra of TEC were calculated using the S transform [Stockwell et al., 1996; Mansinha et al., 1997]. TEC variations at frequencies less than 100 mHz (10 s) are considered in this study. Detrending of the TEC also limits this study to TEC variations greater than 0.37 mHz. This 0.37–100 mHz frequency range corresponds primarily to mesoscale (tens to hundreds of kilometers) structures in the ionosphere.
The calculated TEC is often called slant TEC since GPS signals travel on a slanted path with respect to zenith of the ground receiver (unless the satellite is directly overhead the receiver). As a result, the estimated local time and latitude for a particular TEC variation depends on the ionospheric altitude from which the disturbance is estimated to originate from. This is illustrated in Figure 2, which shows magnetic local times and latitudes of GPS satellite-to-receiver raypaths at points where they intersect the ionosphere at 110 km (left) in the E region and 270 km (right) in the F region. Altitude adjusted, corrected geomagnetic (AACGM) coordinates are used for magnetic latitude and local time. Each arrow in the figure represents an individual satellite-to-receiver raypath and shows magnetic coordinates for a particular raypath over time as the corresponding GPS satellite moves overhead. The point at which a satellite-to-receiver raypath crosses the ionosphere at a certain altitude is called the ionospheric pierce point (IPP) for that raypath. For 110 km and 270 km, 1 day (24 h) of IPP trajectories are plotted for all five receivers (color coded) and for raypaths from all visible satellites above 25° elevation. Most satellites spend about 1–4 h above 25° elevation for receivers in the polar cap. Due to the slanted path of the satellite signal with respect to receiver zenith, IPPs at lower altitudes in the ionosphere are more concentrated around receiver latitude than IPPs for higher altitudes. This is clearly seen in Figure 2, where IPP trajectories at 270 km altitude provide more coverage in latitude, while IPPs at 110 km are concentrated around the latitude of the respective receiver.
In this study, CADI ionosondes were used to estimate both the altitude in the ionosphere from which variations in TEC originate, and the horizontal drift velocity of the ionospheric plasma. CHAIN CADIs employ vertical sounding at HF frequencies in the range of 0.1–22.5 MHz to obtain a bottomside profile of the ionosphere. CADI fixed frequency (group range) measurements are available at 30 s resolution. CADIs also measure the bulk drift velocity at 30 s resolution using a multireceiver interferometry technique. For purposes of estimating magnetic local time and latitude of TEC variations, ionospheric altitudes of TEC variations were estimated based on CADI ionosonde group range measurements. For each TEC variation, the lowest altitude at which at least 10 dB of reflected power was observed in the group range was taken as the altitude contributing to the variation in TEC.
For presenting results in this paper, TEC variations have been divided into two categories: variations in TEC entirely due to ionization in the F region and variations in TEC which involve ionization of the E region. There is no sharp boundary between E and F layers, and thus, the cutoff altitude for classifying E and F region structures is somewhat arbitrary. For CADI observations of ionization structures at 120 km and below, TEC variations were classified as “E region variations,” while for CADI observations of ionization at 150 km and above, TEC variations were classified as “F region variations.” Ionization structures observed between 120 km and 150 km altitude were included in both E region and F region statistics. Note that observation of E region ionization in the CADI group range does not discount the possibility of simultaneous F region structuring that also results in variations in TEC. Thus, while TEC variations observed simultaneously with E region ionization are classified as E region variations, ionization of the F region may also contribute to these E region variations in TEC.
Figure 3 shows an example of TEC variations due to entirely F region ionization, as observed by the GPS receiver and CADI located in Pond Inlet. Figure 3a shows the CADI group range at 5 mHz, which is the altitude and power of the reflected 5 mHz signal over time. Structuring in the F region ionosphere is indicated by the varying height of the reflected signal, which varies from 250 to 350 km over the entire 16:00–19:00 UTC time interval. Multiple GPS satellites linked to the Pond Inlet receiver observed significant TEC variations due to F region disturbances over this time interval. Figure 3b shows one example of TEC from PRN 18, where TEC variations with peak-to-peak amplitude of 1–5 TECU where observed from 16:00 to 19:00 UTC due to the structuring in the F region. Figure 3c shows the dynamic power spectrum of TEC from PRN 18, which revealed significant frequency components in the range of 0–30 mHz. Universal time coordinated (UTC), magnetic local time on the Pond Inlet CADI, and magnetic local time and latitude of the 270 km IPP of PRN 18 are indicated on the time axis.
An example of TEC variations involving E region ionization is shown in Figure 4. Figure 4a shows CADI 4 MHz group range over a 3 h interval of disturbed ionosphere above Cambridge Bay. Starting around 13:20 UTC and observed until 16:00 UTC, significant enhanced ionization occurred in the E region ionosphere, as indicated by vertical streaks at E region altitudes as low as 110 km. Coinciding with the appearance of enhanced E region ionization was the onset of significant TEC variations observed by multiple GPS satellite signals linked to the Cambridge Bay receiver. Measurements of PRN 7 are shown in Figure 4b, where variations of 1–4 TECU were observed during enhanced E region ionization. The TEC dynamic power spectrum in Figure 4c shows discrete spectral components in the range of 1–20 mHz from 13:20 to 16:00 UTC.
There are several factors at play which determine the observed frequency of TEC variations. These factors include the motion, geometry, and lifetime of ionization structures, as well as the horizontal velocity of GPS raypaths due to satellite motion. Aside from the motion of GPS raypaths, the degree to which each of these factors will impact TEC variations depends largely on the source and generation mechanism of the atmospheric ionization, and how these ionization structures evolve. The motion of GPS raypaths through the ionosphere can potentially result in unwanted “Doppler shifts” in TEC variations, which result from the relative motion of ionospheric structures and GPS raypaths observing these structures. Since the purpose of this study is to investigate characteristics of ionospheric ionization structures, TEC variation frequencies should ideally reflect the motion, geometry, and lifetime of ionization structures. An impact on observed frequencies due to GPS satellite motion is undesirable.
(3)Figure 5 shows the total number of TEC measurements made by GPS satellites during the 6 year study, at a cadence of 1 s. “Number of observations” is represented by color in the color contour plots and is plotted over 00:00–24:00 magnetic local time (MLT) and 74°–90° magnetic latitude (Mlat). Number of observations was collected in bins of 00:15 MLT and 0.25° MLat. Figure 5 (left) shows F region observations at 270 km, and Figure 5 (right) shows E region observations at 120 km. With the exception of latitudes greater than 89.0° MLat and less than 75° MLat, there were more than 106 observations of the F region throughout the polar cap. Latitudes where the total number of observations was less than 105 are not included in this study. For F region observations, this includes geomagnetic latitudes greater than 89°. For E region observations, this includes latitudes less than 75.5°, between 80.0° and 80.5°, and between 85.0° and 86.5°.
3 Results
(4)
(5)3.1 Yearly Occurrence Rate and Amplitude
Yearly occurrence rate and amplitude statistics of TEC variations are shown in Figure 6. For comparison of occurrence and amplitude to solar activity level, the mean annual sunspot number for 2009 to 2014 (source: WDC-SILSO, Royal observatory of Belgium, Brussels; http://www.sidc.be/silso/datafiles) is plotted in Figure 6a. There were almost no sunspots observed during solar minimum in 2009, while a modest increase in the number of sunspots from ~25 in 2010 to ~113 in 2014 reflects the weak 24th solar cycle.
Figure 6b plots the yearly occurrence rate of TEC variations for events arising from F region ionization (top row) and events involving E region ionization (bottom row). Occurrence rate is represented by color, where, for example, an occurrence of 0.4 indicates that TEC variations of at least 1 TECU were observed 40% of the time. Latitudes at which insufficient observations were made for statistical purposes are colored white. Black asterisks mark local maxima in polar plots.
As shown in Figure 6b, average annual occurrence rate was consistently highest in localized regions on the dayside for years 2009–2014. Dayside F region occurrence rates peaked at 0.35–0.45 for all years. For 2009–2011, two distinct occurrence maxima in the prenoon (09:30–10:00 MLT, 80.0° MLat) and postnoon (13:00–14:00 MLT, 78.5° MLat) sectors were observed. For 2012–2014, a third peak in occurrence rate was observed in the morning around 08:00–09:00 MLT and 77.0–78.0° MLat.
Dayside occurrence of E region TEC variations peaked at 0.28–0.35 for all years. For 2009–2013, E region occurrence was highest in the morning-afternoon sectors at 77°–78° MLat, with highest occurrence observed in the morning at 81°–82° MLat in the year 2014.
Figure 6b also shows a year-by-year increase in occurrence rate throughout the polar cap. Significant nightside TEC variations were rare in 2009, with occurrence rates below 0.02 in some regions, while in 2014, occurrence of F and E region TEC variations was greater than 0.15 in most polar cap regions. Occurrence rate was consistently lowest near the magnetic pole, with E region occurrence below 0.1 at 87°–90° MLat for all years. The systematic increase in yearly occurrence rate is evident in Figure 6c, which plots the occurrence rate of F (blue) and E (green) region TEC variations integrated across all local times and latitudes greater than 74° MLat.
The annual mean peak-to-peak amplitude of F region (top row) and E region (bottom row) TEC variations is plotted in Figure 6d for 2009–2014. On average, the amplitude of TEC variations in both E and F regions clearly increased from year to year with solar activity, with highest-amplitude variations observed on the dayside for all 6 years. In 2009, peak average amplitudes of 2.2–3.0 TECU were observed in the morning-afternoon sectors, at lower latitudes relative to peak occurrence rates in Figure 6b. From 2009 to 2014, average amplitudes of F region TEC variations peaked around noon and increased from 2.9 to 5.2 TECU, while amplitudes of E region variations peaked slightly postnoon and increased from 3.6 to 4.8 TECU. A peak in average amplitude was also observed at lowest latitudes close to midnight (slightly premidnight) for most years. This midnight amplitude peak was most evident for TEC variations involving E region ionization and increased from 2.4 to 3.1 TECU from 2009 to 2014. Possibly related to the premidnight peak was an early morning minimum in F region amplitude at lowest latitudes between 02:00 and 05:00 MLT.
(6)
As shown in Figure 7a, the probability of observing significant TEC variations on the nightside and deep in the polar cap is higher closer to solar maximum, regardless of solar wind conditions. On the other hand, occurrence rate of TEC variations on the dayside was consistently high (>0.4) for all years, during periods where solar wind-magnetosphere coupling rates were also large. Figure 8b also shows that average amplitudes of TEC variations throughout the polar cap are significantly larger closer to solar maximum. The stage of the solar cycle will therefore be an essential consideration for any empirical model of TEC variations or mesoscale structuring in the polar cap ionosphere.
3.2 Seasonal Occurrence Rate and Amplitude
Figure 8 shows the seasonal occurrence (a), integrated occurrence (b), and average amplitude (c) of TEC variations for all 6 years. Seasons are classified such that summer and winter are centered on summer and winter solstice, while fall and spring are centered on autumnal and vernal equinox, respectively. Each season is 3 months in duration.
Figure 8a shows significant variability in occurrence rates with season, with highest occurrence consistently observed in morning-afternoon sectors around 77°–78° MLat and at 80°–82° MLat in the morning sector. Highest dayside occurrence rates were observed in the winter months, where average occurrence of F region variations reached as high as 0.48 in localized dayside regions and E region occurrence peaked at 0.39 in the morning sector. Dayside occurrence of F region variations in spring and fall months were comparable, with the exception of notably lower occurrence rates during spring months at 76°–78° MLat in the morning sector. Dayside E region occurrence was slightly higher in spring compared to fall. While F region occurrence was lowest in summer months in all dayside regions, peaking at 0.32–0.34 in the morning-afternoon sectors, occurrence of dayside E region variations at 81°–82° MLat was highest in summer. Dayside E region occurrence below 79° MLat was comparable for all seasons in the afternoon sector and lowest in the summer in the morning sector.
Despite the high winter occurrence of F region TEC variations on the dayside, F region occurrence rates on the nightside and close to the magnetic pole were lowest in the winter months. Highest nightside occurrence was in the spring, greater than 0.20 for latitudes below 78.0° MLat, while nightside occurrence during fall and summer ranged from 0.15 to 0.20. F region occurrence below 78.0° MLat was lowest at local times of 21:00–22:30 MLT for all four seasons, while occurrence for higher latitudes approaching the magnetic pole ranged from 0.05 to 0.15 for all four seasons and was highest in spring.
Unlike TEC variations due to F region ionization, nightside occurrence of E region variations was highest in winter and fall, up to 0.17 at latitudes below 79.0° MLat. Nightside E region occurrence was slightly lower during summer and spring months. E region occurrence at higher latitudes approaching the magnetic pole was consistently low (<0.10) for all seasons, with significant E region ionization rarely observed in the fall months near the magnetic pole (<0.02 occurrence rate).
As shown in Figure 8b, occurrence integrated over all local times and latitudes did not vary significantly with season. Integrated F region occurrence was highest in spring and lowest in summer, while integrated E region occurrence was highest in summer and spring and lowest in winter and fall.
As shown in Figure 8c, largest-amplitude TEC variations were, on average, observed in the winter months, while lowest-amplitude variations were observed in the summer. Amplitudes of F region variations peaked around noon at 5.4 TECU in the winter and slightly postnoon at 3.2 TECU in the summer, while E region amplitudes peaked postnoon at 4.3 TECU in winter and 2.9 TECU in summer. Notable in winter months were large-amplitude F regions variations greater than 4.0 TECU extending to the magnetic pole, while average F region amplitudes near the magnetic pole in summer and spring were less than 2.5 TECU. An equinox asymmetry was evident in the average amplitudes of both F and E region TEC variations, with significantly higher amplitude TEC variations observed in fall months compared to spring throughout the polar cap. Lowest-amplitude F region variations were observed in the early morning sector (02:00–05:00 MLT) for all seasons. The midnight peaks in E region amplitudes also observed in Figure 6d arose primarily from winter and fall months. Below 78.0° MLat, average fall and winter amplitudes peaked at 3.0–3.3 TECU around midnight, while average nightside amplitudes in summer and spring were less than 2.3 TECU.
Seasonal occurrence rates and amplitudes shown in Figure 8 are six year averages (2009–2014). Seasonal patterns in each individual year were similar to patterns shown in Figure 8, although average occurrence rates and amplitudes varied from year to year (Figure 6). Seasonal occurrence and amplitude for 2009 would therefore be smaller compared to Figure 8 averages, while seasonal statistics in 2014 would show larger occurrence rates and amplitudes. An exception to this trend is the lower latitude peaks in occurrence rate observed in the morning and afternoon for F region variations, which were largest in 2012.
3.3 Frequency Statistics
Figure 9 shows histograms for TEC variation frequency for all F region (a) and E region (b) TEC variations observed from 2009 to 2014. Plots show the number of observed events as a function of frequency from 0 to 20 mHz, using frequency bins of 0.125 mHz. As indicated in each plot, 1,954,186 events involving F region ionization were observed, while 1,188,969 events involving ionization of the E region were observed. Not shown are occurrence distributions for frequencies greater than 20 mHz. Occurrence decreased significantly with increasing frequency, with 121,369 F region events and 101,144 E region events observed at frequencies from 20 to 100 mHz. An exponential fit was applied to each distribution (red dashed line) using a nonlinear least squares method, with the corresponding exponential equation also shown in each panel. Note the cutoff frequency of 0.37 mHz due to detrending of the TEC. Distributions for TEC variations due to both F and E region ionization peaked at 0.9–1.0 mHz, with generally decreasing number of observed events with increasing frequency. Distinct peaks were also evident at 1.9–2.0 mHz and 3.9–4.0 mHz in the distribution for F region variations and at 2.0–2.1 mHz and 3.9–4.0 mHz in the distribution for E region variations. In the exponential fits, a less negative exponent indicates a greater proportion of TEC variations observed at higher frequencies. Exponential fits for F and E region TEC variations resulted in exponents of −0.43 and −0.34, respectively, indicating that a greater proportion of higher frequencies were observed for TEC variations involving E region ionization.
In Figure 10, frequency distributions are plotted based on magnetic footprints of magnetospheric and solar wind source regions. Local times and latitudes of magnetospheric source regions were determined from maps of particle precipitation regions of Newell et al. [2004], which depend on the orientation of the IMF in the Y-Z GSM plane. As detailed in Newell et al. [2004], these precipitation maps were based on statistical averages of DMSP spacecraft particle observations over an 11 year period. Figure 11 shows average magnetic local times and latitudes of regions mapping to the polar cusp, low-latitude boundary layer (LLBL), plasma mantle boundary layer, and magnetospheric boundary plasma sheet (BPS). Nightside local times and regions poleward of dayside boundary layers correspond to the open polar cap, which maps directly to the solar wind along “open” magnetic field lines.
Figure 10 shows frequency distributions for TEC variations in (a) the open polar cap on the dayside (06:00–18:00 MLT), (b) the polar cap on the nightside (<06:00 MLT, >18:00 MLT), (c) locations mapping to the mantle boundary layer of the magnetosphere (dayside and poleward of the cusp), (d) locations mapping to the low-latitude boundary layer (LLBL) of the magnetosphere (dayside, equatorward of, and flanking the cusp), and (e) locations at the magnetic footprint of the polar cusp. Total number of observed events and exponential fits (red dashed lines) are also shown in each panel. Distributions for all source regions peaked around 0.9–1.0 mHz, with a generally decreasing number of observed events with increasing frequency. Discrete peaks around 2.0–2.1 mHz and 3.9–4.0 mHz were most prominent in the distribution for latitudes and local times corresponding to the plasma mantle, with the number of events observed at 2.0–2.1 mHz comparable to the 0.9–1.0 mHz peak. Distinct ~2 and ~4 mHz peaks were also evident in the distribution for dayside polar cap regions mapping to the solar wind, while less pronounced peaks at these frequencies were observed for nightside polar cap regions connected directly to the solar wind. Distinct peaks at ~2 and ~4 mHz were less obvious in the distribution for regions mapping to the LLBL. A distinct peak at 2.0–2.1 mHz was also observed in the distribution for regions mapping to the polar cusp; however, this distribution was less clear due to relatively low number of events observed (18,723).
The exponential fits indicate that a greater proportion of higher-frequency TEC variations were observed in ionospheric regions mapping to the plasma mantle, LLBL, and polar cusp, compared to open polar cap regions mapping directly to the solar wind. Higher frequencies were least likely to be observed in nightside regions of the polar cap. For mantle, LLBL, and cusp regions, ~22% of observed TEC variations were less than 2.0 mHz, while 39% and 44% of TEC variations observed in the dayside and nightside open polar cap, respectively, were less than 2.0 mHz. For mantle, LLBL, and cusp regions, 44–48% of observed TEC variations were less than 4.0 mHz, 74–78% were less than 10.0 mHz, and 91–94% were less than 20.0 mHz. For dayside open field lines, 68% of TEC variations were less than 4.0 mHz, 92% were less than 10.0 mHz, and 98% were less than 20.0 mHz. For nightside open field lines, 73% of TEC variations were less than 4.0 mHz, 93% were less than 10.0 mHz, and 98% were less than 20.0 mHz.
Exponential fits for Mantle, LLBL, and cusp regions resulted in exponents ranging from −0.20 to −0.22, while exponential fits for dayside and nightside open polar cap field lines resulted in exponents of −0.41 and −0.52, respectively. Figure 12 shows the distribution in magnetic latitude and local time for exponents of exponential fits. For this figure, exponential fits were applied to frequency distributions in bins of 00:30 MLT, 0.5° MLat. Exponents of up to −0.15 in the range of 08:00–15:00 MLT and 77.0–81.0° MLat indicate that higher-frequency TEC variations were more often observed across the dayside. As presented in the previous section, occurrence of both F and E region TEC variations tended to be highest in these regions. Figure 10 also indicates that higher-frequency TEC variations occurred more often in the morning sector, where exponents < −0.24 spanned a broader range of latitudes (74.0°–83.0° MLat) than similar exponents of afternoon distributions. Higher-frequency TEC variations were least likely to be observed close to the magnetic pole at latitudes >87.0° MLat, and on the nightside (after 21:00 MLT and before 03:00 MLT) at latitudes >78.0° MLat.
4 Discussion
4.1 Occurrence Rate of TEC Variations
Low-frequency (<100 mHz) TEC variations greater than 1 TECU result primarily from ionization due to energetic particle precipitation or convection (drift) of ionization structures such as polar cap patches and PMAFs. Occurrence rate of TEC variations associated with both F and E region ionization was consistently highest on the dayside and peaked in three localized regions in the morning-afternoon sectors. At around 76.0°–80.0° MLat, highest occurrence was observed in the morning and afternoon sectors at 07:00–10:00 MLT and 13:00–15:00 MLT, which flanked a region of lower occurrence across noon. According to energetic particle precipitation maps of Newell et al. [2004], these high-occurrence regions magnetically map to the postnoon and prenoon regions of the LLBL, flanking the polar cusp around noon. The third occurrence maximum was in the morning sector at 80.0°–82.0° MLat and 09:00–11:00 MLT, which magnetically maps to the morning sector plasma mantle according to Newell et al. [2004]. For F region variations, high occurrence at 80.0°–82.0° MLat was also observed across noon and into the afternoon sector, corresponding to noon and postnoon regions of the plasma mantle. In addition, Newell et al. [2009] showed similar prenoon and postnoon maxima in occurrence and number flux of energetic electron precipitation (mainly 80 eV to tens of keV), at latitudes ranging from 76.0° to 82.0° MLat depending on solar wind conditions. These results suggest that the localized high occurrence of dayside TEC variations may be a result of energetic electron precipitation of LLBL and mantle origin. Peak dayside occurrence rate in Figures 6b and 7a increased only slightly from year to year, with small shifts in the local time and latitude of peak occurrence. These shifts may reflect changing solar wind conditions associated with the ascending phase of the solar cycle and thus change in dayside location of magnetic reconnection and magnetospheric topology. The dependence of TEC variations on solar wind and IMF parameters is explored in a paper following this study.
Although there was a consistently high occurrence of TEC variations at localized dayside regions for all solar activity levels, average yearly occurrence rates of F and E region TEC variations in most other polar cap regions increased year by year with the ascending phase of the solar cycle (Figures 6b, 6c, and 7a). TEC variations rarely occurred on the nightside and near the magnetic pole in the year 2009 (<5% occurrence), even when solar wind-magnetosphere coupling rate were high. In 2014, TEC variations were observed more than 15% of the time in most polar cap regions, and more than 25% of the time when solar wind-magnetosphere coupling was high. This suggests less structuring, or at least less intense structuring (<1 TECU) of the polar cap ionosphere in years when solar activity is low, and that the phase of the solar cycle is an important consideration in any attempt to predict or model irregularities in the polar ionosphere. Occurrence of polar cap patches is known to increase with solar activity [Dandekar, 2002] and likely contributed to the increased occurrence of F region TEC variations throughout the polar cap. Increased occurrence of E region ionization indicates increased occurrence of energetic particle precipitation throughout the polar cap with increased solar activity. Nightside occurrence of F and E region variations decreased with increasing latitude, which may be indicative of auroral activity expanding poleward from lower latitudes into the nightside polar cap. Increased substorm activity near solar maximum is associated with more frequent poleward expansion of auroral activity, and expansion of auroral activity deeper into the polar cap. Significant TEC variations due to high-energy particle precipitation (>10 keV electrons) in the nightside polar cap has previously been reported by Watson et al. [2011].
Largely due to seasonal variation in magnetic dipole tilt and ionospheric conductivity, past studies have shown a seasonal favorability for certain generation mechanisms of ionization structures in the polar cap. For example, Newell et al. [1996] showed that discrete aurora resulting from field-aligned accelerated electrons occur more often in the winter ionosphere and less in the summer, mainly due to the efficiency of the ionospheric feedback mechanism when ionospheric conductivity is low. In general, dayside occurrence of F region TEC variations was highest in winter, where occurrence greater than 0.4 was observed over a broad range of local times and latitudes (Figure 8a). Summer occurrence on the dayside was lowest, peaking around 0.35 at localized regions in the morning and afternoon. Seasonal dependence was greatest in the morning sector at 76.0°–80.0° MLat, with highest F and E region occurrence in the winter (0.48 and 0.39, respectively) and lowest in the summer (0.23 and 0.22, respectively). This seasonal trend indicates that a mechanism which produces high-energy particle precipitation and operates favorably in winter, such as the ionospheric feedback mechanism, contributed to TEC variations in the morning sector. On the other hand, Frey et al. [2004b] discussed observations of dayside aurora generated by a specific mechanism favorable to the highly conductive, summer polar ionosphere, with occurrence highest in the postnoon region. Although afternoon occurrence was still highest in winter, this so-called high-latitude dayside aurora (HiLDA) may contribute to the relatively high occurrence (compared to morning) of F and E region TEC variations observed at 76.0°–80.0° MLat in the summer. High summer and low winter occurrence of E region TEC variations around 80.0°–82.0° MLat in the morning sector also indicates that precipitating particles with sufficient energy to ionize the E layer (>5 keV electrons, 20 keV protons [Rees, 1963, 1982]) in this region occurred most often in the summer.
In contrast to dayside occurrence, occurrence of F region variations near the polar cap and on the nightside was highest closer to equinox (fall and spring) and lowest in the winter. The well-known semiannual variation [Russell and McPherron, 1973; Weigel, 2007] is an average increase in overall geomagnetic activity, including auroral activity around the spring and fall equinox, which may reflect the increased occurrence in nightside regions of the polar cap around equinox. Several studies have reported that winter is the favorable season for polar patch formation, largely due to increased recombination rates of ionospheric plasma in the summer and decreased rates of plasma entry into the polar cap from lower latitudes [e.g., Dandekar, 2002; Wood and Pryse, 2010]. Some reports have indicated that patches still occur frequently in summer months but are significantly less intense due to the higher background plasma density of the sunlit polar cap [e.g., Sojka et al., 1994]. Seasonal occurrence for F region TEC variations gives no indication of a winter preference for occurrence of polar patches, and it is difficult to resolve whether high nightside spring and fall occurrence is due to polar patches or other precipitation sources. The consistent low occurrence around the magnetic pole for seasonal F region TEC variations indicates that auroral precipitation contributes to the high equinox occurrence rate on the nightside, since polar patches are observed throughout the polar cap, including near the magnetic pole.
Seasonal occurrence of TEC variations around 80° MLat were consistent with results of Krankowski et al. [2006], which examined the occurrence of TEC variations observed by the Casey (CAS1) (−80.66°, 159.10° geomagnetic coordinates) GPS receiver during the year 2001 (solar maximum). For this latitude, the authors reported maximum occurrence across the dayside and low occurrence across the nightside in the winter, occurrence at all local times around equinox, and low occurrence during summer. Krankowski et al. [2006] reported a slightly different local time dependence across the dayside in the winter, with highest occurrence between 10:00 and 18:00 MLT. Results in Figure 7a indicated a morning affinity for TEC variation occurrence around 80° MLat, with highest occurrence between 07:00 and 15:00 MLT. Note that Krankowski et al. [2006] attributed all observed TEC variations to “polar patch activity.”
The statistical occurrence rate of phase scintillation due to microscale (hundreds of meters to ~10 km) structuring in the high-latitude ionosphere was examined by Prikryl et al. [2015], for the years 2008–2013. The solar cycle dependence of scintillation activity observed during periods of moderate-high geomagnetic activity (Kp > 2) reported by Prikryl et al. [2015], was similar to the solar cycle dependence of TEC variations shown in Figure 7a. Occurrence of both microscale and mesoscale structures was highest on the dayside for all years but increased throughout the polar cap closer to solar maximum. Seasonal occurrence of phase scintillations in Prikryl et al. [2015] was also similar to seasonal patterns in Figure 8a, with peak dayside occurrence of scintillations in the winter, lowest dayside occurrence in the summer, and largest nightside occurrence around spring-fall equinox. These comparisons indicate that smaller-scale, scintillation producing structures are possibly related to mesoscale structures in the polar cap. Past studies have shown scintillation activity due to smaller-scale structuring within larger-scale auroral arcs [e.g., Smith et al., 2008] and polar patches [Krankowski et al., 2006].
Depletions and enhancements in ionosphere plasma density, as well as horizontal density gradients associated with ionization structures, can have severe impacts on GNSS positioning systems. A 1 TECU change in the plasma density along the satellite to receiver raypath corresponds to a range delay of about 16 cm, which is problematic for single-frequency GPS users that rely on ionospheric models that do not account for mesoscale ionization structures in the polar cap. More problematic is the density gradients and smaller-scale structuring often embedded within mesoscale structures, which can result in signal loss-of-lock and loss of positioning capabilities. In addition, the high occurrence of E region structuring suggests that communication and navigation applications will frequently encounter operational issues when operating in the polar cap, such as absorption of HF radio broadcasts. A 1 TECU increase in vertical TEC due to E/D region ionization can result in up to 2.5 dB absorption of a 30 MHz signal in the ionosphere [Watson et al., 2011]. Note that years 2009–2014 were part of a relatively weak solar cycle, and thus, future, stronger solar cycles may result in further increases in polar cap ionospheric variability.
4.2 Average Amplitude of TEC Variations
Largest-amplitude E and F region TEC variations were, on average, observed at 11:00–13:00 MLT and 74.0–78.0° MLat, which covers the local time gap between morning-afternoon peaks in occurrence rate observed at similar latitudes. As seen in Figure 6d, the average amplitudes of TEC variations in 2009 were relatively low, with prenoon, postnoon, and midnight maxima of 1.9–2.7 TECU for both F and E region variations. Average amplitudes for afternoon and nightside events involving E region ionization were higher than those involving only F region ionization in 2009, indicating that the most intense events in these regions involved high-energy particle precipitation. Highest average amplitudes in 2009 were observed equatorward of occurrence maxima observed in Figure 6d, indicating that the most intense particle precipitation originated equatorward of regions with the highest occurrence of particle precipitation. High-intensity precipitation producing the largest-amplitude TEC variations in 2009 possibly originated from the dayside extension of the plasma sheet or the LLBL [Newell et al., 2009]. Average amplitudes around noon in 2009 were significantly lower than in subsequent years.
For 2010–2014, highest-amplitude F region variations were observed around noon, with a significant increase in average amplitude across the dayside with increasing solar activity. Occurrence statistics indicated that the largest-amplitude F region variations below 78.0° MLat and around noon occurred relatively infrequently, with a 0.05–0.25 occurrence rate. High noon amplitudes indicate that the most intense TEC variations occurred in regions corresponding to the polar cusp, or the LLBL equatorward of the cusp (according to precipitation maps of Newell et al. [2004]). Large-amplitude TEC variations may have resulted from intense cusp/LLBL particle precipitation associated with high solar output, or possibly increased occurrence of high-density polar cap patches originating from the cusp region. Large E region amplitudes for high solar activity levels indicates that high-intensity, high-energy particle precipitation was responsible for at least some of the large-amplitude variations observed in this region. E region ionization near the equatorward edge of the cusp due to precipitation of accelerated ions (2–5 keV) was observed by Nilsson et al. [1998].
Polar cap patches, which can result in large-amplitude TEC variations of up to 10–15 TECU, would seemingly make a significant contribution to the average amplitudes of F region variations, especially since the occurrence rate and intensity of polar patches increases with solar activity [Buchau et al., 1983; Dandekar, 2002]. It is unclear whether the lifetime of polar patches or in other words the dissipation of polar patches as they convect across the polar cap could impact the decreasing average amplitude of TEC variations away from noon on the dayside. Previous reports have indicated that polar patches can have long enough lifetimes to completely traverse the entire polar cap and enter the nightside auroral region [e.g., Pedersen et al., 2000].
Amplitudes of TEC variations also reflect the relative intensity of polar patches compared to the background plasma density of the polar cap ionosphere. Low summer amplitudes of F region variations throughout the polar cap may reflect previous reports of low polar patch density relative to the background ionosphere during summer months, mainly a result of persistent solar photoionization (higher background density) and higher recombination rates of polar patch ions [e.g., Wood and Pryse, 2010]. Similarly, high winter amplitudes in F region variations throughout the polar cap, including near the magnetic pole, is consistent with high patch-to-background density ratio typical of the dark polar cap ionosphere. F region amplitudes in the fall were also relatively high, while spring amplitudes were slightly smaller than fall amplitudes but larger than summer amplitudes.
While occurrence rates of TEC variations in fall and spring seasons were somewhat similar (Figure 8a), average amplitudes of TEC variations were significantly larger in fall compared to spring (Figure 8c). While the semiannual variation in geomagnetic activity describes an average increase in geomagnetic activity at spring and fall equinox, statistics of Newell et al. [2013] showed that solar wind-magnetosphere coupling rates have a significant seasonal dependence resulting from a combination of the Russell-McPherron effect [Russell and McPherron, 1973] and the elliptical orbit of the Earth. The Russell-McPherron effect describes, on average, a larger magnitude BZ (north-south) IMF component around spring and fall equinoxes due to the Earth's magnetic dipole tilt relative to the Earth-Sun line, while the elliptical orbit of Earth results in a larger IMF magnitude when Earth is closer to the Sun. These combined effects result in highest solar wind-magnetosphere coupling rates in November and lowest in summer (June–July), which is consistent with the large TEC amplitudes observed in winter and fall, and small amplitudes observed in summer.
Except for the year 2012, amplitudes of E region TEC variations peaked around midnight below 78.0° MLat. Large-amplitude TEC variations associated with E region ionization on the nightside are most likely due to poleward expansion of intense auroral precipitation associated with substorms. Low amplitudes around midnight in 2012 may be due to increased occurrence of less intense precipitation events for that year. Midnight amplitudes of E region variations were also highest in winter and fall months, consistent with the statistically low summer occurrence of substorms [Newell et al., 2013].
Lowest F region amplitudes were typically observed below 78.0° MLat in the early morning around 02:00–06:00 MLT. Low morning amplitudes may also be related to a statistical morning-afternoon asymmetry in number flux of monoenergetic electron precipitation [Newell et al., 2009], or possibly the premidnight proclivity for substorm onsets [Frey et al., 2004a]. A notable feature in the average amplitude of F region TEC variations was an extension of high average amplitudes well into the afternoon around 78.0°–82.0°, which most obvious in the amplitude distributions for years 2010–2014 in Figure 6d. One possible explanation is a statistically high number flux of precipitating electrons with monoenergetic energy spectra around these latitudes in the post noon sector, as reported by Newell et al. [2009].
4.3 Spectral Frequency of TEC Variations
Frequencies of TEC variations reflect the geometry, motion, and lifetime of ionization structures in the ionosphere. At this point, we cannot resolve the relative contribution of each characteristic, and thus, the spectral content of TEC variations reflects a combination of these characteristics. The most interesting features in the distribution of TEC frequencies (Figures 9 and 10) are the distinct peaks observed around 2.0 mHz and 4.0 mHz. A critical issue regarding these peaks is whether they were a product of the data collection/analysis (e.g., detrending of TEC and motion of GPS satellites) or resulted from a real physical process involved in the structuring of the polar cap ionosphere. In the data collection and analysis for this study, only events involving at least 1 h of continuous TEC measurements with no cycle slips were included, and it was verified that side effects of the detrending process (e.g., filtering edge effects) did not influence statistical results. To investigate the possibility of a real physical process-producing TEC variations at preferential frequencies, frequency distributions for varying solar wind conditions (not shown) and for variations involving F and E region ionization (Figure 8) were examined. These investigations revealed no significant dependence of occurrence of frequencies around 2.0 and 4.0 mHz on ionospheric altitude or solar wind conditions. As shown in Figure 9, discrete 2.0 and 4.0 mHz frequencies had a high occurrence in ionospheric regions mapping to the mantle boundary layer and, to a lesser extent, regions mapping to the open polar cap on the dayside (>06:00 MLT and <18:00 MLT). Distinct, but significantly less prominent peaks around these frequencies were also observed in the distribution for regions mapping to the nightside polar cap. The frequency distribution for the LLBL, on the other hand, showed very little modulation of TEC variations at 2.0 and 4.0 mHz, indicating that these preferential frequencies occurred in regions mainly poleward of the LLBL. If these preferential frequencies were to arise from the data collection and analysis process, one would expect to consistently observe these distinct frequencies independent of local time and latitude. These distributions indicate that some physical process associated the mantle or open polar cap modulates the structure or motion of the ionosphere in such a way that results in TEC variations around 2.0 and 4.0 mHz. Less prominent peaks for open polar cap regions indicate that this modulation process, or structures resulting from this process, dissipated at latitudes poleward of the plasma mantle and in the nightside polar cap.
As observed in Figure 12, higher-frequency TEC variations were more often observed on the dayside polar cap, in regions mapping to magnetospheric boundary layers. Frequency distributions with the greatest proportion of higher-frequency variations (exponents < −0.24) were observed in regions corresponding to the mantle, LLBL, and dayside extension of the plasma sheet boundary layer [Newell et al., 2004], and mainly in regions with a statistically high occurrence of discrete aurora and number flux of high-energy precipitating electrons with broadband energy spectra [Newell et al., 2009]. A high occurrence of higher-frequency TEC variations indicates a higher occurrence of small-scale structuring, higher dayside convection speeds, and, in general, a more dynamic and turbulent ionosphere. Smaller-scale structuring is possibly, in part, due to spatially and temporally inhomogeneous particle precipitation flux characteristic of the magnetospheric boundary layers, as well as the generally more turbulent environment expected from direct solar wind-magnetosphere interaction on the dayside. Higher-frequency TEC variations will also typically indicate stronger horizontal gradients in ionospheric density, increasing the potential for GNSS signal issues such as cycle slips and loss of lock.
5 Summary and Conclusions
Five high data rate CHAIN GPS receivers in the polar cap were used to statistically examine the characteristics of TEC variations greater than 1 TECU arising from mesoscale (tens to hundreds of kilometers) ionization structures, over a period of 6 years. Occurrence rate of TEC variations associated with both F region and E region ionization was highest on the dayside (0.2–0.5) and peaked in localized morning and afternoon regions mapping to the LLBL and plasma mantle. Occurrence of TEC variations observed deep within the polar cap and on the nightside had a strong dependence on solar cycle, with occurrence rates in these regions below 0.05 close to solar minimum (2009) and above 0.2 close to solar maximum (2015). This increase in occurrence was independent of solar wind conditions and may result from increased occurrence and intensity of polar cap patches (F region), as well as more frequent expansion of the auroral region deeper into the polar cap during periods of high solar activity. Dayside occurrence of TEC variations was generally highest in winter (up to 0.4), suggesting a precipitation mechanism that operates favorably under conditions of low ionospheric conductivity. Occurrence of TEC variations due to E region ionization in the afternoon and at higher morning latitudes (>80° MLat) peaked in the spring and summer, suggesting a spring-summer proclivity for higher-energy precipitation in these regions. Seasonal occurrence of F region variations away from the dayside was highest around equinox (0.1–0.2), suggesting a contribution from the well-known semiannual variation in geomagnetic activity. High winter and fall occurrence of nightside E region TEC variations at lower latitudes (0.1–0.2) indicated a contribution from intense substorm precipitation, while E region ionization at higher latitudes near the magnetic pole was rarely observed (<0.05 occurrence rate).
Largest-amplitude F region TEC variations were generally observed around noon on the dayside, in regions mapping to the LLBL and polar cusp, while average E region amplitudes peaked in the postnoon sector. Average amplitudes of both F and E region TEC variations also had strong solar cycle dependence, with significantly larger variations observed near solar maximum compared to solar minimum, independent of solar wind conditions. This may be a result of increased intensity of precipitation from magnetospheric boundary regions and the plasma sheet, increased density of polar cap patches, and auroral expansion into polar cap regions near solar maximum. Average seasonal amplitudes of F region variations were highest in winter and lowest in summer throughout the polar cap, consistent with a high ratio of patch-to-background density in winter. Relatively large F and E region TEC variations observed in winter and fall was possibly due to the November peak in solar wind-magnetosphere coupling that has been reported in previous studies.
In general, spectral frequency (0.37–100 mHz) of TEC variations followed an exponential distribution, dominated by frequencies of around 1.0 mHz. The exact shape of this distribution depended heavily on local time and latitude, with the greatest proportion of higher-frequency variations observed on the dayside. An unexpected result in the frequency distributions was distinct peaks around 2.0 mHz and 4.0 mHz. These preferential frequencies arose primarily from TEC variations observed in regions mapping to the mantle boundary layer, and to a lesser extent the open polar cap. The strong dependence of these ~2 and ~4 mHz preferential frequencies on local time and latitude indicates that they were a result of a real physical process that modulated the ionospheric structure and produced TEC variations at these particular frequencies.
These results improve upon the current observational picture of the climatological behavior of mesoscale structuring in the polar cap ionosphere. As demonstrated, GPS TEC measurements from multiple polar cap receivers allows for high-resolution analysis of statistical results in local time and latitude, which paves the way for more accurate and detailed empirical models of ionospheric variability based on TEC. Reliable observations and models of ionospheric irregularities are currently lacking in polar cap regions and are essential for reliable operation of GNSS navigation systems and HF radio communication in the polar regions. A second manuscript, following this manuscript, examines occurrence rate and amplitude of TEC variations as a function of solar wind-magnetosphere coupling rate and orientation of the interplanetary magnetic field (IMF). Empirical functions for characteristics of TEC variations, based on magnetic local time/latitude, solar activity level, season, and solar wind condition, will also be presented in a future manuscript.
Acknowledgments
Infrastructure funding for CHAIN was provided by the Canadian Foundation for Innovation and the New Brunswick Innovation Foundation. CHAIN operation is conducted in collaboration with the Canadian Space Agency. Science funding is provided by the Natural Sciences and Engineering Research Council of Canada. Computing resources were provided by ACENET (https://www.ace-net.ca/wiki/ACENET). CHAIN data are available online through chain.physics.unb.ca. Solar wind data were obtained from the OMNI database on CDAWeb (http://cdaweb.gsfc.nasa.gov/).
