Characterization of high-latitude ionospheric scintillation of GPS signals



[1] This paper presents statistical analysis of arctic auroral oval ionospheric scintillation events during the current solar maximum based on high-rate Global Positioning System data collected in Gakona, Alaska (62.39°N, 145.15°W) from August 2010 to March 2013. The objective is to gain a better understanding of the climatology and morphology of ionospheric scintillation in high-latitude regions. A scintillation event filter, multipath identification procedures, and other processes are applied to exclude nonscintillation related signal intensity and phase fluctuation and to extract scintillation events with S4 index above 0.12 and phase sigma above 6° from over 657 days of data. A total of over 5800 scintillation events were identified; most of them show phase fluctuations, only 10% of the phase fluctuations are accompanied by weak amplitude scintillation. Based on the occurrence time, signal direction of arrival, intensity, and duration of these scintillation events and the solar and geomagnetic activities associated with these events, diurnal, seasonal, spatial, and solar activity dependencies are derived and presented in the paper.

1 Introduction

[2] Ionospheric scintillation refers to the random amplitude and phase fluctuations of radio signals after propagating through plasma irregularities [Yeh and Liu, 1982]. These irregularities occur more frequently in high-latitude and equatorial regions, especially during solar maxima [Basu et al., 2002]. Occurrence of scintillation is difficult to predict and model because of the temporal and spatial variability of the ionosphere and solar activities that are driving forces of space weather phenomena [Kintner et al., 2007]. Ionospheric scintillation affects all space-based communication, surveillance, broadcasting, and navigation systems. Its impact on Global Navigation Satellite System (GNSS) receivers is of particular importance as the effect may range from degradation of positioning, velocity, and timing (PVT) accuracy to receiver loss-of-lock. With the increasing demand on GNSS applications, understanding of characteristics and effects of ionospheric scintillation on GNSS signals and receivers has gained worldwide attention from both scientific research and engineering application fields.

[3] Monitoring and modeling of ionospheric scintillation has been an active research area for over half of a century. Two general approaches in the area are physics-based modeling of radio wave propagation through ionospheric irregularities [Booker et al., 1950; Ratcliffe, 1956; Briggs, 1975; Rino, 1979a, 1979b; Secan et al., 1995, 1997; Knight, 2000; Dyrud et al., 2006; Wernik et al., 2007; Beniguel et al., 2009; Carrano et al., 2012; De Oliveira Moraes et al., 2012; Ghafoori, 2012] and data-driven statistical analysis of the climatology and morphology of scintillation from ground-based and satellite-based measurements [Aarons, 1982; Pi et al., 1997; Pullen et al., 1998; Basu et al., 2002; Kintner et al., 2004; Smith et al., 2008; Humphreys et al., 2010; Morrison, 2010; Prikryl et al., 2011; Liu et al., 2013]. While the physical model-based approach has provided an important understanding of causes and effects of scintillation phenomena, data-driven statistical studies of scintillating radio signal behavior have helped to paint a more comprehensive picture of the occurrence and extent of scintillation activities on a global scale.

[4] Much of the previous research regarding the scintillation climatology has been conducted at the equatorial regions where the strongest levels of scintillation can be observed. Results indicate that most low-latitude ionospheric scintillation and irregularities are observed between the postsunset and premidnight period [Basu et al., 2002; Cervera and Thomas, 2006; Su et al., 2006; Li et al., 2007]. Typically, both amplitude and phase scintillation can be detected in the same event. A seasonal dependency has also been discovered as more scintillation events occur around the equinoxes (spring and fall) [Aarons, 1982; Beniguel et al., 2009].

[5] Contrary to scintillation at low latitude, scintillation observed at high-latitude regions is generally moderate and phase scintillation is more frequent and intense than amplitude scintillation [Aarons, 1997; Skone et al., 2008; Azeem et al., 2013]. It should be noted here that the term “phase scintillation” used in this paper refers to the phase fluctuations with spectral content above the phase detrending filter cutoff frequency of 0.1 Hz. Inside the auroral oval, scintillation is usually a nighttime phenomenon while polar cap scintillation exists at all local times [Rino et al., 1983]. In recent years, many research findings have been reported based on data collected in Northern Europe and auroral/polar areas in both the hemispheres [Kersley et al., 1988, 1995; Gola et al., 1992; Nichols et al., 1999; Mitchell et al., 2005; Aquino et al., 2005; De Franceschi et al., 2008; Meggs et al., 2008; Li et al., 2010; Alfonsi et al., 2011; Garner et al., 2011; Deshpande et al., 2012; Kinrade et al., 2012]. Cumulative frequency of occurrence of phase and amplitude scintillation and distributions of total electron content (TEC) changes obtained from stations in Europe were given by Rodrigues et al. [2004]. Their results indicate a correlation between scintillation and magnetic field activity index Kp. A strong influence of interplanetary magnetic field (IMF) on ionospheric irregularity formation and movement was reported by De Franceschi et al. [2006]. High-latitude scintillation characteristics at solar minimum were presented by Li et al. [2010], Ngwira et al. [2010], and Prikryl et al. [2010] based on data obtained from stations in Arctic and Antarctic regions. The seasonal pattern of occurrence of scintillation showed a winter maximum while the diurnal pattern was location dependent. A more comprehensive analysis on bipolar scintillation climatology at solar minimum was recently published by Alfonsi et al. [2011] in which the influences of location, season, and geomagnetic field were provided in detail.

[6] The objective of this paper is to present comprehensive statistical analysis of scintillation events based on relatively continuous measurements obtained while on the upswing of the current solar maximum. Since 2009, an event driven autonomous ionospheric scintillation monitoring and data collection system was established by the author's research group in Gakona, Alaska (geographic: 62.39°N, 145.15°W; geomagnetic: 63.54°N, 92.22°W) [Pelgrum et al., 2011; Vikram et al., 2011; Taylor et al., 2012]. The system records multiconstellation GNSS data using an array of commercial ionospheric scintillation monitoring (ISM) receivers and radio frequency front ends. This paper presents the results obtained by a GSV4004B ISM receiver at a sampling rate of 50 Hz on GPS L1(1575.42 MHz) power and phase measurements [Van Dierendonck et al., 1993, 1996]. A set of thresholds have been used to filter scintillation events from the raw data based on the values of the two commonly used scintillation indices: the S4 index, which is the detrended, normalized signal intensity standard deviation, and the phase scintillation indicator σφ , which is the detrended carrier phase standard deviation. We set the thresholds for S4 to be 0.12 and for σφ to be 6° in this study. These values are selected based on evaluations of background signal intensity and phase fluctuations at the data collection site by the same receiver [Taylor et al., 2012]. In an earlier publication, a set of higher thresholds of 0.15 and 15° were used for S4 and σφ, respectively [Jiao et al., 2013]. The higher thresholds were chosen to reflect events that can cause a considerable impact on GNSS measurement accuracy and reliability, while the lower thresholds used in this paper are selected for ionosphere irregularity studies. About 5800 scintillation events have been extracted after applying the lower thresholds. Based on these events, several statistical distributions of high-latitude ionospheric scintillation have been obtained including S4 and σφ magnitude distributions, scintillation event duration distributions, the diurnal, seasonal, and spatial dependency of scintillation occurrence, and correlation between scintillation and local geomagnetic field fluctuations. These results clearly show that the general trend in both scintillation event magnitude and frequency is largely affected by solar activities as we progress into a relatively mild solar maximum. A clear seasonal pattern has also been observed as more scintillation events were found to occur in spring and fall. Furthermore, strong correlations between scintillation event occurrence frequency and magnitude with deviations of local magnetic field measurements obtained from local magnetometers were also observed.

[7] The remainder of this paper is organized as follows: the data collection system setup and methods used in the analysis are described in section 2. Section 3 focuses on discussions of the results, and conclusions are summarized in section 4.

2 System Setup and Methodology

[8] Since 2009, a GNSS receiver array has been established at Gakona, Alaska which is located within the northern auroral oval. Figures 1 and 2 show the aerial view and diagram of the receiver array configuration since a system update in August 2012. A software radio front-end, built by the Ohio University Avionics Engineering Center, was used to collect narrowband GPS L1 and L2 (1227.60 MHz) intermediate frequency (IF) data [Gunawardena et al., 2008]. Two Universal Software Radio Peripheral version 2 radios by Ettus Research were configured to collect wideband GPS L5 (1176.45 MHz) and GLONASS L1 and L2 IF data [Peng and Morton, 2010]. Two commercial NovAtel (model OEMV3) receivers provided additional measurements for plasma drift calculations [Wang et al., 2012]. A GSV4004B ISM receiver by GPS Silicon Valley [Van Dierendonck et al., 1993, 1996; Van Dierendonck and Arbesser-Rastburg, 2004] was used to record scintillation data and generate outputs for a scintillation event trigger [Taylor et al., 2012]. Only data recorded by the GSV4004B ISM receiver will be presented in this paper. This receiver, modified from EuroPak-3 M, can track up to 10 GPS satellites at the L1 and L2 frequencies. It generates 50 Hz phase and amplitude measurements on L1, while on L2 only 1 Hz phase, pseudorange and carrier to noise ratio (C/N0) measurements are obtained from the semicodeless tracking of the P code [Van Dierendonck et al., 1993, 1996]. Before August 2012, the receiver was connected to antenna 1 which was approximately 4 km north of its current position [Pelgrum et al., 2011]. The main impact of relocating the receiver is the change in multipath environment. Most of the multipath effects are removed from the scintillation measurements largely by the thresholds discussed later.

Figure 1.

Aerial view of the antenna array at Gakona, Alaska since August 2012.

Figure 2.

Experimental setup at Gakona, Alaska since August 2012.

[9] Using the signal intensity and carrier phase generated by the GSV4004B ISM, we compute S4 and σφ according to equations (1) and (2) below. S4 describes amplitude scintillation which is the standard deviation of the received signal power normalized to average signal power [Briggs and Parkin, 1963]. σφ is defined as the standard deviation of the detrended carrier phase [Yeh and Liu, 1982].

display math(1)
display math(2)

where <> represents the average value over the interval of interest which is traditionally taken as a 60 s period. I is the detrended signal intensity and δφ is the detrended carrier phase. Sixth order low-pass and high-pass Butterworth filters with cutoff frequency of 0.1Hz have been used to detrend L1 signal intensity and carrier phase measurements, respectively [Van Dierendonck et al., 1993]. It should be noted here that the detrending filters may cause “phase without amplitude” scintillation phenomena frequently observed in high-latitude regions [Forte and Radicella, 2002; Beach, 2006; Mushini et al., 2012]. The integration interval used to compute the scintillation indices is 10 s, so that every 500 sampling points are averaged to generate one S4 or σφ value. This interval was chosen to most effectively highlight scintillation features based on evaluations of several different time intervals between 10 s and 60 s.

[10] In order to extract scintillation events from the raw data collected by the receiver, criteria were established to define a scintillation event. A scintillation event occurs at a time interval during which either S4 or σφ surpasses their corresponding preset threshold values. The thresholds are selected based on studies conducted by Taylor et al. 2012 using data collected by the same receiver at the same location. Baseline background S4 and σφ distributions including multipath effects were established. The results indicate that receiver background noise and environmental multipath effect induced S4 or σφ values do not exceed 0.15 and 6°, respectively for GPS L1 CA signal for the receiver at this location. If higher values of thresholds were adopted, some weak scintillation events would be missed. For navigation applications, such weak scintillation does not have a major impact on receiver operation and higher threshold values can be used to allow more focused studies of high impact scintillation event extraction and analysis. A set of relatively high threshold values of S4 = 0.15 and σφ  = 15° (or 0.26 rad) were applied by Jiao et al. 2013 for this purpose.

[11] In this paper, a set of relatively low threshold values at S4 = 0.12 and σφ  = 6° (or 0.1 rad) are used to detect scintillation events. Both S4 and σφ are calculated using measurements on L1 only, and either index passing the threshold makes a potential event. The conservative low thresholds were chosen to not exclude weak scintillation events which are indicators of ionospheric irregularity occurrence and plasma drift.

[12] The following is a summary of the scintillation event criteria:

  1. [13] Only events with elevation angles above 30° are considered in order to minimize potential multipath effects.

  2. [14] Scintillation index values have to remain above the threshold value for a minimum of 30 s to qualify as a scintillation event. This minimum time window is selected based on heuristic analysis of the data to exclude most nonscintillation related signal intensity and phase disturbances.

  3. [15] An event detected within 5 min of the end of another event trigger is not counted as an additional scintillation event. Careful visual inspection showed that if two events are separated by about 5 min or less, the interval in between the events is not completely “quiet”. Instead, there are weak scintillation fluctuations below the threshold values. By combining two consecutive events that are separated by less than 5 min as one single event, we lose one event count, but we increase the event duration. Therefore, the scintillation effects are still conserved to some extent.

  4. [16] If multiple PRNs are experiencing scintillation simultaneously, each PRN's scintillation is treated as a separate event. The reason for this criterion is that irregularities are local phenomena and different satellite signals are most likely propagating through different irregularities. However, it is also possible that two or more satellite signals may pass through the same irregularity, especially when the satellite paths are near each other. In this study, the latter possibility is ignored to simplify analysis process. As a result, the number of scintillation events may be an overestimation of the number of irregularities.

  5. [17] The starting time and ending time of a scintillation event are the times S4 and σφ exceed and drop below their corresponding thresholds, respectively, while satisfying conditions 1, 2, and 3. Scintillation event duration is the difference between the ending time and the starting time.

  6. [18] Carrier cycle slips were detected and repaired during the scintillation event detection process. All cycle slip and loss-of-lock events are extracted and analyzed to determine whether they are the results of real scintillation or due to other factors such as interference, signal blockage, multipath, or anomalies of satellite transmission.

[19] Clearly, not all the events that satisfy the above criteria are scintillation events. Two major types of nonscintillation features that may interfere with the detection process are multipath and cycle slips/loss-of-lock caused by nonscintillation factors. These features must be excluded in order to make an accurate characterization of the statistical properties of scintillation.

[20] According to [Taylor et al., 2012], the threshold values and the elevation masking angle discussed above can exclude most multipath events, although a few strong multipath events may still be included in the potential scintillation event list. For these strong multipath events, we can identify them based on GPS satellite orbit repeatability as illustrated in Axelrad et al. [2005]. For stationary GPS receivers, a multipath reflection will repeat itself approximately 4 min ahead of its appearance on the previous day, although the shape and duration of repeating multipath features can vary slightly. During the research, it has been found that some repetitive multipath lasted more than 3 weeks. By comparing events on consecutive days, most multipath in the original detection list can be ruled out. In addition, variations in the code and carrier divergence, which is the measured difference between group delay and carrier phase advance, can also be utilized to verify the existence of multipath [Van Dierendonck et al., 1993, 1996].

[21] Cycle slip and loss-of-lock detection and correction were performed based on the method in Liu [2011] in which combinations of Melbourne-Wübbena Wide-Lane and TEC rate equations are solved to determine the number of cycle slips. During the data processing, it was found that in some cases raw carrier phase measurements had large jumps on L1 (e.g., 105 cycles) while corresponding changes on L2 P(Y) signals were much smaller (usually no more than 50 cycles). This phenomenon is abnormal since L2 P(Y) tracking is more susceptible to scintillation than that of L1 CA signal. More than 95% of this type of anomaly occurred with PRN 21 and a few with PRN 4 and other satellites in the data. These events are excluded from the scintillation event list and as they are more likely due to satellite signal anomalies rather than ionospheric irregularities.

[22] It is important to note that the above criteria and process contain some degrees of arbitration in defining scintillation events. These artificially imposed rules are necessary to conduct statistical analysis of scintillation.

3 Results and Discussion

[23] The data analyzed in this paper was collected from August 2010 to March 2013. Occasional system outages due to hardware, software, or environmental issues led to missing data during this time period. Table 1 summarizes the number of effective days during which data were recorded in each month followed by their corresponding percentage of days. Fractional days are combined to form a whole number of days. For example, there are 248.6 effective days of data for 2011, dividing this number by 365 which is the total number of days in that year, yielding a percentage of 68.1%.

Table 1. Number and Percentage of Days of Effective Data Analyzed
MonthDay #PercentDay #PercentDay #PercentDay #Percent
JanuaryNot available311007.825.221.870.3
April28.996.300Under processing
Grant total in 2010, 2011, 2012, and 2013657.267.5

[24] In the time span of 32 months (2.67 years) during the data collection experiment, a total of 657 equivalent days of data have been obtained. Approximately 6000 events were extracted according to the lower S4 and σφ thresholds stated above, while about 3000 events were extracted based on the higher thresholds. Only results based on the lower thresholds will be discussed in this paper.

3.1 S4 and σφ Distributions

[25] Figures 3a and 3b are distributions of the maximum S4 and σφ during scintillation events extracted from the data. Figure 3a takes into consideration all amplitude scintillation events, while Figure 3b contains all phase scintillation events. A scintillation event having both amplitude and phase scintillation is included in both figures. The first few bins in the plots make up a large portion, indicating that most scintillation events recorded are relatively weak. Cases of strong amplitude scintillation with the S4 > 0.5 were rare with accumulative probability of slightly more than 1%. On the contrary, strong phase scintillation with the σφ  > 30° (or 0.5 rad) was much more frequent with an accumulative probability of about 12.4%.

Figure 3.

(a) Maximum S4 and (b) maximum σφ distributions.

[26] We also examined how the amplitudes of S4 and σφ changed over the 32 month time period. Figure 4 shows the average maximum S4 and σφ values of all the events observed in each season. The seasons are defined as: spring (SP)—March to May; summer (SU)—June to August; fall (FA)—September to November; winter (WI)—December to February of the following year. From the amplitude scintillation plot, it can be observed that the S4 index on average did not vary much during the time period and in general the values stayed low. However, phase scintillation displayed a seasonal pattern showing that spring and fall have more intense activities. Moreover, a slow upward trend can also be observed in each season (except in winter 2012) which corresponds to an increase in solar activity over the 32 months. Finally, in order to show the correlation between the solar activities and scintillation events, Figure 4 also includes the monthly smoothed sunspot number which is averaged over each season, as an indicator of solar activities. Note that the monthly smoothed sunspot numbers on and after February 2013 are an estimate from the Solar Influences Data Analysis Center.

Figure 4.

Seasonal average maximum S4 values for amplitude scintillation events and average maximum σφ values for phase scintillation events. Solid dots are average monthly smoothed sunspot numbers. FA – Fall, WI – Winter, SP – Spring, SU – Summer.

[27] As was mentioned before, all the results shown in this section are based on measurements on GPS L1 CA signal. The L2 P(Y) tracking of the receiver is based on semicodeless tracking. Thus, the measurements on L2 band are noisy and the tracking loop is less robust during scintillation. As a result, the scintillation indicators on L2 were not considered. Nevertheless, some interesting quantitative relationship between the receiver L1 CA and L2 P(Y) tracking during scintillation can be observed. Figure 5 is a statistical distribution showing the probability of cycle slips or loss-of-lock on L2 when L1 maintains tracking during scintillation events. This probability is obtained by dividing the number of cycle slips or loss-of-lock on L2 by the total number of events within a certain range of maxσφ value on L1. The horizontal axis in Figure 5 shows the L1 σφ values corresponding to when the L2 P(Y) signal lost lock. Only phase scintillation has been considered since amplitude scintillation was generally too small to cause problems to the carrier tracking loop in the data. This distribution is as expected as the probability of loss-of-lock or cycle slips on L2 increases as the L1 σφ value increases. When the L1 σφ value is more than 60°, it was almost definitive that the receiver would have trouble tracking the L2 P(Y) signal. On the other hand, even when the σφ was relatively low (around 15°), the L2 P(Y) signal could still lose lock or experience cycle slips. A 3° exponential polynomial fitting curve was generated based on the scatterplot data points to fit the trend of the distribution (the solid curve).

Figure 5.

Probability of loss-of-lock/cycle slip on L2P(Y) during phase scintillation observed on L1 CA.

3.2 Event Duration

[28] The duration of a scintillation event is an important indicator of background ionosphere plasma distributions and dynamics. Longer scintillation events also have more negative impact on GNSS receiver's performance. Figure 6a shows the overall event duration distribution regardless of event type (amplitude or phase) with a mean of about 9 min and standard deviation around 12.5 min. Note that in rare cases some events lasted more than an hour. Figures 6b and 6c compare amplitude scintillation event duration and number of events with those of phase scintillation. The average durations are about 3.7 and 9.7 min for amplitude and phase scintillation, respectively. The latter is nearly three times the former. Additionally, phase scintillation has been observed nearly an order of magnitude more often than the amplitude scintillation events. Considering the intensity, duration, and occurrence frequency, it is clear that of phase scintillation effects dominate at high latitudes and at GPS frequencies when one uses 0.1 Hz cutoff frequency for the detrending filter.

Figure 6.

Duration distributions for (a) all scintillation events, (b) amplitude scintillation, and (c) phase scintillation.

[29] Similarly, the average duration of events occurring in each season is illustrated in Figure 7. For both amplitude and phase scintillation, events in spring and fall are generally longer than those in summer and winter. The influence of solar activity, again, can be observed in the trend of each season.

Figure 7.

Seasonal average event duration histograms of amplitude, phase, and combined scintillation.

3.3 Event Temporal Density and Diurnal Dependency

[30] Scintillation event temporal density is defined as the number of events recorded during a certain period of time, which helps to identify the most likely period within a day when scintillation events may occur. Based on all the events extracted using the lower thresholds from the entire available data set, overall event occurrence probability distributions have been established in Figures 8a–8c as a function of Coordinated Universal Time (UTC), hours after sunset, and magnetic local time (MLT), respectively. A clear peak appeared at around UTC noon, and most events were concentrated between 7:00 and 17:00 h UTC which is between 10:00 P.M. and 8:00 A.M. local time or 9:00 P.M. and 7:00 A.M. during daylight saving time (LT = UTC − 9). The same data were converted to hours after local sunset and magnetic local time as are shown in Figures 8b and 8c. Clearly, most scintillation events in Alaska were observed during either geographic or magnetic local nighttime. This result is similar to observations acquired in low-latitude regions where scintillation is mostly detected starting 1.5–2 h postsunset [Basu et al., 2002; Cervera and Thomas, 2006].

Figure 8.

Overall hourly event occurrence probability versus (a) UTC, (b) hours postsunset, and (c) magnetic local time.

[31] Similarly, seasonal variation in average number of events per day is shown in Figure 9, in which an evident seasonal pattern can be observed. A stack histogram is presented so that the effective average event number per day in each season can be obtained by summing up the segments. The bottom, top, and middle segments represent phase only, amplitude only, and joint amplitude and phase scintillation. In general, summer and winter experienced fewer amplitude and phase scintillation events per day than spring and fall.

Figure 9.

Seasonal average event number per day for phase scintillation events, concurrent amplitude and phase scintillation events, and amplitude scintillation events. Solid dots represent average monthly smoothed sunspot numbers.

[32] The overall ascending trend in Figure 9 stopped at spring 2012, and descends significantly after that. The cause of this transition is correlated with the dip in solar activity, which can be observed on the monthly smoothed sunspot numbers represented in solid dots. The results show that scintillation occurrence frequency is mainly influenced by solar activity and is also modulated by the season.

3.4 Event Spatial Distribution

[33] GPS satellite path over a 24 h period and amplitude, phase, and combined amplitude and phase scintillation event spatial distribution plots are given in Figure 10. The plots are presented in a sky view form, in which the center is the receiver local zenith and circles around it are equal-elevation contours. The elevation masking angle is set to 30° to minimize multipath effects. Each event was marked on the plot using the average elevation and azimuth after being weighted by the amplitudes of the two scintillation indices. Therefore, a brighter pixel corresponds to larger number and more intense events. In each plot, the brightest region is near the center at elevation from 75° to 80° and azimuth from 180° to 210°. This is around the local magnetic zenith which corresponds to the direction at which the magnetic field lines enter the Earth at northern high latitudes. At Gakona, Alaska, the local magnetic zenith is at 76° elevation and 202° azimuth [Pedersen et al., 2003; Pelgrum et al., 2011]. This direction is known to be a scintillation-active spot with little multipath contamination because of its relatively high elevation. Amplitude scintillation in Figure 10c is especially concentrated around this local magnetic zenith, whereas phase scintillation appeared to also frequently occur elsewhere other than the zenith (Figure 10d). Two regions near the edge in the plots are also brighter. Apart from scintillation, they could also be contributed by denser satellite paths in the area and multipath that overlay on scintillation-induced fluctuations.

Figure 10.

(a) The 24 h satellite path; scintillation spatial occurrence for (b) all scintillation events, (c) amplitude scintillation, and (d) phase scintillation (grid size = 120 × 120).

3.5 Correlation With Geomagnetic Activities

[34] It is known that high latitude ionospheric scintillation is many a times driven by the interplanetary magnetic field and magnetosphere processes and is correlated with solar flares and coronal mass ejections [Basu et al., 2002]. Previous studies have discussed the dependence of high-latitude scintillation occurrence on geomagnetic activity using the interplanetary magnetic field (IMF) data [Li et al., 2010; Alfonsi et al., 2011] and the global magnetic index Kp [Beniguel et al., 2009]. In this paper, we present a quantitative correlation of scintillation with local geomagnetic field disturbances and ionospheric scintillation based on local fluxgate magnetometer measurements.

[35] Figures 11a and 11b show the correlation between scintillation indices and the magnitude of local geomagnetic field deviations based on data collected during 3 days with strong scintillation events— 1 March 2011, 9 July 2012, and 15 July 2012. The average phase scintillation index was obtained over a 1 h interval for all satellite paths within the vicinity of the local magnetic zenith (elevation: 78° ± 5°, azimuth: 195° ± 20°). The magnitude of the local geomagnetic field deviations, depicted on the y axis, is calculated from the root square of the standard deviations of the three components of the local magnetic field over a 1 h period. A linear correlation can be observed in both Figures 11a and 11b. The correlation coefficients between the geomagnetic field deviation magnitude and σφ and S4 are 0.87 and 0.73, respectively, clearly indicating a stronger correlation between phase scintillation and geomagnetic field disturbances. Similar correlation values are obtained between deviations of D, H, and Z components of the geomagnetic field disturbances and the phase and amplitude scintillation indices [Jiao, 2013].

Figure 11.

Correlation between geomagnetic field deviation and average S4 and σφ for satellites near magnetic zenith (integrate interval = 60 min). Data collected on 1 March 2011, 9 July 2012, and 15 July 2012.

[36] A more striking way to demonstrate the relationship between geomagnetic field disturbance and the level of phase scintillation is shown in Figure 11. Instead of correlating scintillation index values with the geomagnetic field disturbance values measured at the same time, we compared maximum σφ values of phase scintillation events with their corresponding local magnetometer measured peak-to-peak field disturbance magnitude. Figure 12 is generated based on 665 phase scintillation events extracted from 11 days of strong phase scintillation. The dates include 5 days from 2011 (1 March, 10 March, 2 May, 28 May, and 1 November) and 6 days in 2012 (15 February, 10 March, 15 March, 9 July, 15 July, and 16 July). These dates were chosen because there were large geomagnetic field disturbances and large numbers of scintillation events. The figure shows that larger disturbances in geomagnetic field lead to stronger scintillation events in general. Note that no amplitude scintillation events are considered because of the much smaller number of events and their lower intensity available in the data set.

Figure 12.

Max σφ occurrence probability versus geomagnetic field variation magnitude range. Data collected on 1 March, 10 March, 2 May, 28 May, and 1 November in 2011, and 15 February, 10 March, 15 March, 9 July, 15 July, and 16 July in 2012.

3.6 Percentage of Satellites Experiencing Simultaneous Scintillation

[37] An important measure of the impact of scintillation on satellite navigation is the number of satellites affected by scintillation simultaneously. This is because a navigation receiver can only generate PVT solutions when there is a minimum of four satellites available, and a larger number of satellites can improve the geometric dilution of precision and hence the PVT solution accuracy.

[38] Figure 13 is a plot showing the local geomagnetic field variations together with the number of satellites with elevation over 30° that were affected by scintillation (S4 ≥ 0.12 or σφ ≥ 6°) simultaneously on 15 July 2012. The percentage of satellites affected by scintillation was also calculated by dividing the number by the total number of satellites with elevation angles higher than 30°. The scintillation on this day was fairly intense, and during some time periods all satellites above the elevation mask were affected by scintillation. About 25% of the day, over 50% satellites were simultaneously affected by scintillation. In addition, it can be noticed that magnetic field disturbances and percentage of satellites affected are highly synchronized, which also confirm the results obtained in section 3.5.

Figure 13.

Magnetic field variation, and number and percentage of space vehicle (SV) signals experiencing scintillation simultaneously on 15 July 2012. Elevation mask is 30°. H, D, and Z are horizontal component, declination angle, and vertical component of local magnetic field vector variations.

4 Summary and Conclusions

[39] In this paper, statistical characterization of high-latitude ionospheric scintillation was presented on the basis of 50 Hz amplitude and phase data recorded by the GSV4004B receiver located in Gakona, Alaska. A conservative set of thresholds of 0.12 and 6° for S4 and σφ, respectively have been implemented to filter scintillation events from the background data. Based on the 5800 scintillation events extracted, several statistical distributions have been obtained and discussed which include maximum S4 and σφ distributions, scintillation event temporal and spatial distributions, and event duration distributions. The influences of local time, seasons, and solar activity on scintillation have also been investigated. These results provide a comprehensive view of when, where, and how scintillation occurs in the lower auroral region and can help with future scientific research on space weather and on robust GNSS receiver development.

[40] The major results discussed in this paper can be summarized as follows:

  1. [41] Based on the methods applied in this research, phase scintillation appears to be the dominant phenomena at high latitudes whereas amplitude scintillation is generally weak. Over 80% of observed events are phase scintillation only, while only 11% are amplitude only events. Nine percent are events with both amplitude and phase scintillation. Among all phase scintillation events, 50% of them have σφ > 10°. For amplitude scintillation, 90% of the events have S4 index under 0.25.

  2. [42] Phase scintillation events last much longer than amplitude scintillation events. The average phase scintillation events last 9.7 min, while the average amplitude scintillation events only last 3.7 min.

  3. [43] Scintillation events are most frequently observed during geographic or magnetic local nighttime. For Alaska, this corresponds to approximately UTC noon. Approximately 40% of scintillation occurred 2 h within UTC noon.

  4. [44] In the lower northern auroral zone where the receiver is located, scintillation intensity and occurrence frequency is primarily affected by solar activity and modulated by seasons. In spring and fall, scintillation is usually more intense, more frequent, and longer lasting than in summer and winter.

  5. [45] Scintillation occurs more frequently and more intense around the local magnetic zenith. According to the data analyzed, about 20% of scintillation events with duration longer than 5 min and maxS4 > 0.3 or maxσφ > 30° occurred within 10° elevation and 20° azimuth around the local magnetic zenith.

  6. [46] L2 P(Y) semicodeless tracking is much more fragile to scintillation than L1 tracking. For the receiver used, when σφ on L1 reaches 60° (or 1 rad), L2 signal having cycle slip or loss-of-lock is nearly 100%.

  7. [47] Scintillation event occurrence and event magnitude are strongly correlated with the variation rate and amplitude of local geomagnetic field at high latitudes. The correlation coefficients between the magnitude of local magnetic field disturbance and phase and amplitude scintillation indices are 0.87 and 0.73, respectively. On 15 July 2012 when large local magnetic field disturbance lasted about 15 h and the peak disturbance exceeded 1500 nT, over 50% of satellites in view were affected by scintillation for over 25% of time.


[48] This project is supported by AFOSR grant FA9550-10-1-0346 and AFRL grant FA8650-08-D-1451. Mike McCarrick of Marsh Creek provided the magnetometer data. The authors appreciate the support of the HAARP staff and the University Alaska Fairbanks Geophysical Institute for organizing and sponsoring HAARP campaigns and for the HAARP staff's continuous on-site support of the GNSS receiver data collection system setup.