On the Annual Asymmetry of High‐Latitude Sporadic F

The polar F region ionosphere frequently exhibits sporadic variability (e.g., Meek, 1949, https://doi.org/10.1029/JZ054i004p00339; Hill, 1963, https://doi.org/10.1175/1520‐0469(1963)020<0492:SEOLII>2.0.CO;2). Recent satellite data analysis (Noja et al., 2013, https://doi.org/10.1002/rds.20033; Chartier et al., 2018, https://doi.org/10.1002/2017JA024811) showed that the high‐latitude F region ionosphere exhibits sporadic enhancements more frequently in January than in July in both the northern and southern hemispheres. The same pattern has been seen in statistics of the degradation and total loss of GPS service onboard low‐Earth orbit satellites (Xiong et al. 2018, https://doi.org/10.5194/angeo‐36‐679‐2018). Here, we confirm the existence of this annual pattern using ground GPS‐based images of TEC from the MIDAS algorithm. Images covering January and July 2014 confirm that the high‐latitude (>70 MLAT) F region exhibits a substantially larger range of values in January than in July in both the northern and southern hemispheres. The range of TEC values observed in the polar caps is 38–57 TECU (north‐south) in January versus 25–37 TECU in July. First‐principle modeling using SAMI3 reproduces this pattern, and indicates that it is caused by an asymmetry in plasma levels (30% higher in January than in July across both polar caps), as well as 17% longer O+ plasma lifetimes in northern hemisphere winter, compared to southern hemisphere winter.


High-Latitude Ionospheric Variability
Sporadic enhancements of F region ionization are frequently observed at high latitudes (e.g., Hill, 1963;Meek, 1949). Over the course of a few minutes, the bulk plasma density increases by up to an order of magnitude, typically accompanied by horizontal drift speeds in excess of 1 km/s and intense density structuring. This type of observation has been given several different names, including "patches" and "tongues of ionization." It is now generally accepted that at least the large F region enhancements are made up of photoionized plasma convected into the polar caps from the subauroral dayside (e.g., Carlson, 2012). Recently, there has been controversy over the annual occurrence rate of this phenomenon, with Noja et al. (2013) first reporting that sporadic F region enhancements are seen more frequently in January than July in both hemispheres. Noja et al.'s (2013) conclusion is opposite to longstanding expectations of what happens in the southern hemisphere, where it was believed that there would be more enhancements in July (which is local winter there).
The prevailing argument over the last several decades has been that the high-latitude F region should appear less variable in local summer because the background ionization is much higher then and so the contrast is reduced (e.g., Foster, 1989 ;Sojka et al., 1994). Notable observations that appear to support that theory include Coley and Heelis (1998) and Spicher et al. (2017), who counted more F region density ©2019. The Authors. This is an open access article under the terms of the Creative Commons Attribution License, which permits use, distribution and reproduction in any medium, provided the original work is properly cited. spikes in local winter in each hemisphere. Chartier et al. (2018) explained that the apparent contradiction between Noja et al. (2013) and the others is the result of different counting metrics: Noja et al. (2013) tested for a proxy of ΔN (topside TEC enhancements >4 TECU), whereas the typical approach has been to look for changes in ΔN/N, expressed perhaps most famously in Crowley's (1996) test specifying that a patch must be greater than twice the background density. This choice of a different metric explains why Noja et al. (2013) got a different result than the others, but it does not explain why ΔN should have an annual variability with a marked minimum in July. Additionally, it should be noted that ΔN appears to be the more important quantity from a technological perspective: Xiong et al. (2018) showed that high-latitude GPS signal loss from low-Earth orbit (LEO) has the same annual variability as was seen by Noja et al. (2013) and Chartier et al. (2018).
Given these unexpected observations and the link to a major form of space weather, it is important to find the cause of this annual F region variability. It has been noted (e.g., by Xiong et al., 2018) that the pattern matches the well-known global ionospheric annual asymmetry, first reported by Berkner et al. (1936) and Berkner and Wells (1938). The ionospheric annual asymmetry constitutes a 20-40% increase in the global F layer plasma density in December/January compared to that in June/July (e.g., Mendillo et al., 2005 ;Rishbeth & Müller-Wodarg, 2006 ;Zeng et al., 2008). Another important factor to consider is the lifetime of plasma in the polar caps. Wood and Pryse (2010) have put forward the theory that reduced plasma lifetimes in the northern polar cap during summer explain the relative lack of ionospheric variability seen there. Their parameter estimation indicates that this effect is caused primarily by variation in the chemical composition of the atmosphere, which, in summer, both reduced the plasma density drawn into the polar cap and enhanced plasma loss by recombination.

Method
The goal of this investigation is to determine the annual pattern of sporadic high-latitude F region enhancements at large horizontal scales (hundreds of kilometers and larger), and to understand its causes. Groundbased GPS tomography provides the best observing platform due to its broad, continuous spatial coverage and widespread year-round availability. The ionospheric F layer is usually the densest and always the thickest layer, so it provides the dominant contribution to GPS TEC. Ground GPS also provides a contrast to the LEO satellite-based approaches that have been employed previously. Results from the SAMI3 model are used to understand whether these observations can be explained by current physical understanding (as constituted in the model).
Tomographic reconstructions of ionospheric electron density are produced using the University of Bath's Multi-Instrument Data Analysis Software (MIDAS; Mitchell & Spencer, 2003;Spencer & Mitchell, 2007). The algorithm applies a constrained nonlinear approach to observations taken by a global network of dual-frequency GPS receivers. Electron density images are integrated vertically to obtain ionospheric Total Electron Content (TEC), whose dominant contribution comes from the F region electron density. In the present configuration, the horizontal grid extent is~300 km (northern hemisphere) and~500 km (southern hemisphere due to reduced data coverage there), with the grid covering all latitudes >45°. Data coverage is good in both polar caps, although there are large gaps in the oceans in the southern hemisphere. Poorly resolved regions of the images are removed from the analysis using the adaptive resolution mapping technique described in .
The SAMI3 model (Huba et al., 2000;Huba et al., 2008) is used to assess whether our observations are consistent with current physical understanding of the ionosphere. SAMI3 solves for the dynamic plasma and chemical evolution of seven ion species (H + , He + , N 2 + , O + , N + , NO + , and O 2 + ) on a field-aligned magnetic apex coordinate grid extending up to 87°MLAT. Photoionization is calculated using solar flux from the Flare Irradiance Spectral Model by Chamberlin et al. (2008), which is driven by Solar Dynamics Observatory Extreme Ultraviolet Variability Experiment data. SAMI3 contains a self-consistent electric potential solver that is seamlessly combined with an imposed high-latitude potential from Weimer's (2005) model (driven by solar wind parameters observed by the Advanced Composition Explorer). The Hardy model (Hardy et al., 1985(Hardy et al., , 1989 provides auroral electron and ion precipitation estimates based on the K P index. The neutral atmosphere is specified by the Horizontal Wind Model 2014 by Drob et al. (2015) and the Naval Research Laboratory's Mass Spectrometer Incoherent Scatter Model 2000 of neutral atmospheric densities by Picone et al. (2002). These drivers represent statistically average estimates of conditions based on historical data sets and globally defined input drivers. Therefore, SAMI3 is not expected to provide accurate instantaneous predictions, but can provide insights into climatological behavior.
The year 2014 is chosen for analysis because it is the most recent year of high solar activity, when ionospheric variability is known to be larger. Output is produced for 1-28 January and July to cover the most pronounced part of the annual asymmetry identified in Chartier et al. (2018). These two periods had a similar range of insolation, as approximated by the F 10.7 index, as can be seen from Figure 1. Likewise geomagnetic activity is similar, with both periods exhibiting low to moderate levels of activity. To provide a geophysical reference point, the apex magnetic coordinate system (Richmond, 1995), referenced at 300-km altitude, is used to identify 60°and 70°MLAT bounds within the images. The zone poleward of 70°MLAT is considered to be the polar cap for this investigation.
The first step is to test Noja et al.'s (2013) conclusions regarding the annual variability of ΔN. This is achieved by determining the median, range, and standard deviation of TEC values from the MIDAS observations and SAMI3 model in both polar caps (up to 87°MLAT for direct comparison between model and observations) in January and July 2014. The median TEC serves as a proxy for average F region density, N, while the range captures the extremes of polar F region variability, referred to hereafter as ΔN. The range is calculated as maximum-minimum of TEC values observed or modeled in each polar cap over each month. We emphasize that range is not necessarily representative of the variability of the whole population of polar TEC values, which is indicated by the standard deviation. Rather, the range provides information on sporadic F region enhancements, which are the target of this investigation.
The second step is to calculate the plasma lifetime in the model. The lifetime is defined as the average time an ion exists before undergoing charge exchange or recombining. Since this investigation focuses on the F region, where O + dominates, the O + lifetime is used. At a point location, the O + lifetime, τ, is simply the inverse of its loss rate Ψ (which occurs through charge exchange or recombination). For diagnostic purposes, the O + lifetime is integrated from 200-400-km altitude, weighted according to the O + density at each point, and averaged across the polar cap. This calculation is shown in equation (1).
The equations used to calculate the O + loss rate Ψ can be found in section 6 of Huba et al. (2000). The loss rate depends on the number density of molecular neutral species (O 2 and N 2 ) available for charge exchange, as well as a weak dependence on ion temperature. Figure 2 shows an example of high-latitude TEC from MIDAS and SAMI3 in both hemispheres at 19 UT on 23 January 2014. In the northern hemisphere, enhanced TEC enters the polar cap as a tongue of ionization from the Canadian sector, which is just postnoon in this image. In the southern hemisphere, a similar tongue of ionization forms from a somewhat later UT sector (note that the Earth's rotation appears counterclockwise in the north but clockwise in the south). Agreement between model and

10.1029/2019SW002305
Space Weather observations appears to be good in both hemispheres, given the limitations of data coverage and of model drivers.

Space Weather
To quantify TEC variability in the polar caps, we perform a statistical assessment of all the MIDAS and SAMI3 TEC values between 70 and 87 MLAT for 1-28 January and July 2014. Means and ranges for each month are provided in Table 1.
The model is consistent with the data in most regards, showing similar median values (within 1 TECU in all cases) and the same trends across all three statistical measures, although the data show a much larger range of values.  The results in Table 1 show that the southern polar cap is far more variable than the north, both in terms of seasonal swings and within each month. By far the highest median, standard deviation, and range are observed in southern hemisphere summer (22%, 41%, and 51% larger than the next-highest values observed). Related to this, the well-known global annual asymmetry in ionospheric plasma levels is also present in this polar cap data: the combined January medians are 30% higher than in July in the data (27% higher in the model). The annual asymmetry reinforces local seasonal swings in the south and counteracts them in the north.
As is to be expected, the medians are always higher in local summer than in local winter, and the standard deviations follow this seasonal pattern. However, surprisingly, the ranges are always larger in January than in July. The northern polar cap has a larger range of TEC values in winter despite the fact that the median and standard deviation are smaller than in summer. Figure 4 shows a daily breakdown of the data from which these statistics were drawn.
These results confirm that ΔN is larger in January than in July in both polar caps, and that the SAMI3 model reproduces this variability. Considering that N is higher in local summer, it might be expected that ΔN should also be higher then. From that perspective, the northern hemisphere is actually the anomalous one. Wood and Pryse (2010) found that ΔN is suppressed in northern hemisphere summer primarily because of decreased plasma lifetimes found there. To quantify this effect in our SAMI3 runs, we calculated the average O + plasma lifetimes between 200-400 km and 70-87 MLAT. An example of this quantity is shown in Figure 5.
Inside the polar caps, O + lifetimes are longest in northern winter and shorter in local summer in both hemispheres. Generally, O + lifetimes are anticorrelated with neutral temperature (e.g., shorter lifetimes are seen around noon and in local summer) due to the upwelling of molecular neutral species that occurs when temperatures are higher. Monthly averaged O + lifetimes are shown in Table 2.
Winter O + lifetimes are 17% longer in the northern hemisphere, whereas summer lifetimes are 5% longer in the northern  hemisphere. This hemispheric asymmetry is a contributing factor in the different seasonal patterns seen in both hemispheres. The density of O 2 and N 2 in the F region is the most important variable in determining these lifetimes.

Discussion
GPS-derived images of the high-latitude ionosphere in both hemispheres confirm the pattern first observed by Noja et al. (2013): sporadic F region variability (ΔN) is larger in January than in July in both hemispheres. These observations provide the first ground-based confirmation of that surprising pattern, refuting any suggestion that it is merely the result of topside scale height variations or an accident of LEO satellite observational coverage. This pattern has now been observed using three types of instrument (LEO upward GPS, Langmuir Probe, ground-based GPS tomography) and over 10 years (2001-2004 and 2013-2018). For several decades, the prevailing theory has been that the high-latitude F region appears less variable in local summer than in winter due to reduced contrast against higher ambient levels of ionization. While ambient ionization (N) is higher in local summer (and therefore ΔN/N is smaller in summer than in winter), the contrast theory is clearly incomplete in that it does not explain the observed variation of ΔN.
Here we show that the pattern of ΔN being larger in January than in July in both hemispheres is successfully modeled by SAMI3 driven by climatological inputs (FISM photoionization, Weimer convection, HWM neutral winds, MSIS neutral atmosphere). The ultimate driver of this behavior appears to be a combination of several factors. A priori, we might expect ΔN to be larger when N is larger. From that perspective, the southern hemisphere behavior is as expected (ΔN larger in local summer) and the northern hemisphere is anomalous in that ΔN is larger in local winter. We have identified two factors that contribute to this anomalous behavior: 1. The ionospheric annual asymmetry (~30% more plasma present in January than in July) extends to the polar caps. Averaged across both polar caps, 30% higher TEC is observed in January as compared to July. This effect counteracts seasonal variations in the north and reinforces them in the south. 2. Modeling indicates that O + lifetimes are especially long in northern winter, around 17% higher than in southern winter. Therefore, sporadic enhancements of the polar F region last longer in northern winter than elsewhere.
Given that sporadic F region enhancements are made up of photoionized plasma convected into the polar caps from the subauroral dayside (e.g., Carlson, 2012), the well-known global annual asymmetry (e.g., Mendillo et al., 2005 ;Rishbeth & Müller-Wodarg, 2006) also plays an important role in driving the observed pattern of variability, as it controls plasma levels in the supply region. These factors combine in SAMI3 to reproduce the pattern of variability observed here and by Noja et al. (2013) and Chartier et al. (2018), with ΔN (as characterized by the range of TEC values present in the polar cap) larger in January than in July in both hemispheres.
By far the highest densities and most variable conditions were observed in southern summer-the mean, standard deviation, and range of TEC values there were 22%, 41%, and 51% larger than the next-highest values observed, respectively. This is in part because the annual asymmetry reinforces seasonal variations in the southern hemisphere, but also because the southern magnetic pole is displaced further from the geographic pole than is the case in the north (15°versus 7°in 300-km apex coordinates). This larger offset of the south magnetic pole allows denser plasma (from lower geographic latitudes) to enter the polar convection pattern in the southern hemisphere.

Conclusions
Ground GPS-based TEC observations from 2014 indicate that the ionosphere poleward of 70 MLAT has a larger range of values in January than in July in both hemispheres. First-principle modeling reproduces the observed variability using standard climatological drivers. The pattern of variability in the southern hemisphere directly contradicts the longstanding theory that the high-latitude ionosphere is more variable in local winter. From the alternative perspective that ΔN should vary proportionally to N, the northern hemisphere is actually the anomalous one. Our model analysis indicates that the pattern observed arises because of an asymmetry in winter plasma lifetimes (17% longer in northern winter than in southern winter) and 30% higher electron densities in January as compared with July. Figure A1. Distributions and statistical characteristics of TEC values between 70-87 MLAT in January and July 2014. MIDAS GPS-derived data on the left (a-d) and SAMI3 model on the right (e-h). The model shows the same trends as the data, although the range of values is larger in the data. Means, medians and standard deviations are larger in local summer, whereas ranges are large in January in both hemispheres due to long tails of large values.