Extreme medium-scale traveling ionospheric disturbances (MSTIDs) occurred at midlatitudes in East Asia during a geomagnetically active time on 10 November 2004. Using the Global Positioning System (GPS) observation data from Korean GPS reference stations, the characteristics of the MSTIDs on 10 November 2004 and their potential impact on GPS-based navigation systems in the Korean region are analyzed. The MSTIDs were first observed in the northeast part of South Korea at about 10:00 UT and propagated southwestward with successive wavefronts which extended from northwest to southeast. The peak-to-peak amplitudes of vertical total electron content (TEC) disturbances decreased from about 29 to 10 total electron content unit (1 TECU = 1016 el m−2), and the wavelengths lengthened from about 360 to 580 km from 12:53 to 14:38 UT. The propagation velocity of MSTID wavefronts was estimated using three nearby reference stations showing that velocity gradually decreased from about 254 m/s at 11:46 UT to 76 m/s at 21:26 UT. The ionospheric irregularities in small-scale regions accompanied by the MSTIDs were spatially and temporally varied from about 10:00 to 22:00 UT in response to the movement and intensity change of the MSTIDs. This event also generated anomalously large ionospheric spatial gradients which could cause unacceptable residual pseudorange errors for users of GPS augmentation systems. Frequent loss of the GPS signals, which occurred due to the intense ionospheric irregularities, could also degrade the continuity and availability of GPS-based navigation systems.
The ionosphere in nominal conditions varies smoothly and slowly over a large region with periodic patterns [Datta-Barua, 2004]. However, during geomagnetically active times, the Sun's highly energetic events (solar flares and coronal mass ejections) disturb Earth's ionosphere and lead to various anomalous ionospheric phenomena such as storm-enhanced densities, plasma bubbles, and traveling ionospheric disturbances (TIDs). The developments of these ionospheric phenomena depend upon geographical latitude and longitude, local time, season, geomagnetic activity, and other factors.
Medium-scale TIDs (MSTIDs), which involve plasma density fluctuations and propagate with wavelengths of several hundred kilometers, have been observed at midlatitudes with different types of instruments [e.g., Kubota et al., 2000; Herna'ndez-Pajares et al., 2006]. A likely source of MSTIDs during geomagnetically active periods is high-energy input from the Sun into the ionosphere-thermosphere system at high latitudes [e.g., Schunk and Sojka, 1996]. This could cause Joule heating in the auroral zone and generate equatorward atmospheric gravity waves (AGWs), producing MSTIDs [e.g., Buonsanto, 1999]. On the other hand, the origin of MSTIDs during geomagnetically quiet times may more likely be disturbances in the troposphere [Shiokawa et al., 2005]. The relationship between the occurrence of MSTIDs and the AGWs is not fully understood, but the AGWs may play an important role in generation of MSTIDs [e.g., Hines, 1960; Shiokawa et al., 2003].
It is reported that localized ionospheric irregularities (e.g., large ionospheric spatial gradients, plasma bubbles, or scintillations) caused by plasma instabilities accompany TIDs [Nishioka et al., 2009]. Ionospheric irregularities are harmful to Global Positioning System (GPS)-based navigation systems, including space-based and ground-based augmentation of GPS. One possible effect of ionospheric irregularities is an anomalously large ionospheric spatial gradient, which can pose an integrity threat to users of GPS augmentation systems [Datta-Barua et al., 2010]. In addition, when GPS signals travel into the small-scale regions where plasma density rapidly changes, it is scattered and diffracted. This leads to randomly rapid fluctuations in amplitude or phase of GPS satellite signals, which is known as ionospheric scintillation. This can cause loss of GPS signal tracking because deep signal fades from strong ionospheric scintillation break a GPS receiver's carrier tracking lock [Seo et al., 2009] and consequently reduce the continuity and availability of GPS-based navigation systems.
The ionosphere refracts electromagnetic signals such as those broadcast by GPS satellites, and this refraction effect is a delay in code-phase signal arrival time with respect to an identical signal passing through vacuum [Datta-Barua et al., 2010]. The total electron content (TEC), which is defined as the total number of electrons along the signal path from the GPS satellite to receiver, can be derived from these ionospheric delays. MSTIDs have been studied based on TEC data measured from the GPS observation data along with other observational instruments. Saito et al.  developed a two-dimensional ionospheric perturbations map derived from GPS observation data using a moving average filter in order to analyze characteristics of MSTIDs, such as direction of propagation, amplitude, wavelength, and velocity, in high resolution. Using the data collected from the GPS Earth Observation Network, Saito et al.  found characteristics of MSTIDs that were successive wavefronts which extended from northwest to southeast and moved in a southwestward direction with a peak-to-peak amplitude of about 1 TECU (total electron content unit, 1 TECU = 1016 el m−2) during the night of 8 August 1999. Kotake et al.  used worldwide GPS networks to reveal that activity of MSTIDs during the night is different from that during the day and thus has a longitudinal dependence.
Geomagnetic conditions were extreme in the early part of November 2004, and nighttime MSTIDs were observed in East Asia [Sahai et al., 2009; Nishioka et al., 2009]. Solar flares are sudden explosions near sunspots, and coronal mass ejections (CMEs) are large plasma releases from the Sun's corona. These events are known as causes of geomagnetic storms [Daglis, 2001]. The transit time from the Sun to Earth is 1 to 5 days [Owens and Cargill, 2004]. During the period of 1–7 November 2004, the Sun was very active with 11 M- and 2 X-class flares and 9 CMEs [Trichtchenko et al., 2007]. Based on this information, Earth's geomagnetic field could then potentially be disturbed 1–5 days after these solar events.
Figure 1 shows the variations of the Kp and Dst indices on 10 November 2004. The highest value of the Kp index was 8.7 at 06:00 UT on 10 November 2004, and this value indicates that the G-class event was actually G5 (extreme). A superstorm with a Dst index of −263 nT was observed at 11:00 UT on 10 November 2004. The MSTIDs which occurred in East Asia on 10 November 2004 are likely the result of the extreme geomagnetic activity caused by the solar flares and the CMEs at the Sun during the period of 1–7 November 2004.
While researchers have studied 10 November 2004 MSTIDs in East Asia, including the Japanese region, and have analyzed their characteristics of the spatial structure and movement of MSTIDs [Sahai et al., 2009; Nishioka et al., 2009], MSTIDs which occurred in the Korean region have not yet been comprehensively analyzed. In this paper, GPS observation data collected from Korean GPS reference stations are used to examine the MSTIDs in the Korean region on 10 November 2004. The results are compared to those observed in the Japanese region on the same day, and their potential impact on GPS-based navigation systems is analyzed. Section 2 describes the data used in this paper. In section 3.1, we examine the occurrence time of the MSTIDs in the Korean region. Section 3.2 investigates the latitudinal propagation of the MSTIDs by comparing vertical TEC (VTEC) fluctuations observed at two stations. The spatial structure of the MSTIDs is analyzed in section 3.3 using a two-dimensional VTEC perturbation map. Section 3.4 presents the results of MSTID velocity estimation. Section 4.1 shows the spatial and temporal variations of the VTEC fluctuations in small-scale regions accompanied by the MSTIDs. Section 4.2 discusses the potential impact of MSTIDs on GPS-based navigation systems, and section 5 draws conclusions.
GPS observation data, specifically dual-frequency (L1 at 1575.42 MHz and L2 at 1227.60 MHz) code and carrier-phase measurements collected from Korean GPS reference station networks, were used to analyze the characteristics of MSTIDs in the Korean region. Figure 2 shows the locations of South Korean GPS reference stations operated by the National Geographic Information Institute (NGII), the Differential Global Positioning System (DGPS) Central Office (DCO), and the Korea Astronomy and Space Science Institute (KASI) as of 2004. Ionospheric delays on the L1 signal were estimated using the long-term ionospheric anomaly monitor (LTIAM) developed by the Korea Advanced Institute of Science and Technology (KAIST). The LTIAM is designed to continually monitor ionospheric behavior by computing precise ionospheric delay estimates and ionospheric spatial gradients [Jung and Lee, 2012]. TEC data in the slant domain (i.e., along the actual path between satellite and receiver) that are derived from the delay estimates were then converted to equivalent vertical TEC (VTEC) data using a geometric mapping function or “obliquity factor,” for which an ionosphere thin shell model with a hypothetical height of 350 km is assumed [Misra and Enge, 2006]. The geometric mapping function M used for GPS augmentation systems [RTCA, Inc., 2008] is defined as
where RE is the radius of Earth (6378 km), hI is the assumed height of the ionospheric shell (350 km), and el is the elevation angle of the line of sight between a receiver and satellite. The obliquity factor has values between 1.0 at an elevation angle of 90° and just over 3.0 at that of 5°.
The Global Ionosphere Map (GIM) shows the distribution of VTEC on a global scale with a resolution of 2.5°, 5°, and 2 h in latitude, longitude, and time, respectively. This product is provided by the International Global Navigation Satellite System (GNSS) Service (IGS) Ionosphere Working Group in the ionosphere interexchange format and is available from four Global Data Centers: Center for Orbit Determination in Europe, Jet Propulsion Laboratory, European Space Agency, and Technical University of Catalonia [Hernández-Pajares et al., 2009]. The global trend of ionospheric behavior was analyzed with the GIM to investigate the possible expansion of geomagnetic equatorial anomalies to the Korean region (as will be shown in Figure 4).
3 Characteristics of MSTIDs
3.1 Time of Disturbance
On 10 November 2004, TEC fluctuations of MSTIDs were observed in the central part of Japan from about 10:00 to 22:00 UT [Nishioka et al., 2009]. To analyze the possible activity of these MSTIDs in the Korean region on the same day, the Daejeon station (36.4°N, 124.37°E; hereinafter referred to as DAEJ) located in the central part of South Korea was selected. Figures 3a and 3b show the time series of VTEC for all GPS satellites observed at DAEJ on 6 November 2004, a geomagnetically quiet day with a minimum Dst of 10 nT, and 10 November 2004, a disturbed day with a minimum Dst of −263 nT, respectively. The different colors of lines indicate VTEC data of each GPS satellite. While no rapid variation of the VTEC data was observed on 6 November, VTEC fluctuations started at about 10:00 UT and continued until about 22:00 UT on 10 November. This shows that DAEJ in the central part of South Korea observed the MSTIDs occurring in the Japanese region on the same day.
To better understand the overall ionospheric behavior, we analyzed the GIM from IGS and the regional ionosphere map of the Korean region constructed using GPS observation data from the Korean GPS reference stations with a sampling period of 30 s. For the regional map, the color contours were interpolated by determining a plane within each set of the three nearest VTEC values of ionospheric pierce points (IPPs) [Datta-Barua, 2004]. Each IPP is defined as a point where the line of sight vector from a satellite to a ground station intersects the spherical thin shell of the ionosphere assuming that the hypothetical height of the ionosphere is 350 km [Misra and Enge, 2006].
Figures 4a and 4b show the GIM and the regional ionosphere map at 16:00 UT when VTEC fluctuations were observed over all of South Korea. Note that the color scales of VTEC in the GIM are between 0 and 100 TECU, while those in the regional ionosphere map are between 0 and 30 TECU. Since ionization mainly occurs during daytime due to Earth's exposure to ultraviolet radiation from the Sun, enhanced VTEC around the geomagnetic equatorial region was observed far from the Korean region where 16:00 UT is around 1 A.M. in local time and moved westward as time progressed. However, irregularly structured features in the ionosphere with rapid VTEC variations from about 10 to 30 TECU within short distances were observed in the Korean region at this time as shown in Figure 4b. This indicates that the VTEC fluctuations observed at DAEJ (shown in Figure 3) were caused by localized ionospheric disturbances rather than the enhanced VTEC around the geomagnetic equatorial region.
3.2 Latitudinal (North-South) Propagation
Figure 5 shows VTEC differences at the Mokpo station (34.8°N, 126.4°E; hereinafter referred to as MKPO), which is located in the southwest part of South Korea, and the Inje station (38°N, 128°E; hereinafter referred to as INJE) which is located in the northeast part of South Korea. The VTEC fluctuations at MKPO (blue) and INJE (red) shown in this figure are the differences between two consecutive estimated VTEC values that are 30 s apart. Differences for the GPS satellites observed at these locations are plotted on horizontal lines. This representation was used to illustrate the start times of VTEC fluctuations observed at the two stations. At INJE, VTEC differences began to fluctuate from 09:53 UT, while those at MKPO were relatively quiet at this time. A manifestation of ionospheric irregularities at MKPO was observed from 10:47 UT. From this result of approximately 1 h of time delay of VTEC fluctuations between INJE and MKPO, it is presumed that the ionospheric irregularities occurred earlier at higher latitudes and then propagated to lower latitudes. On the same day in the Japanese region, VTEC fluctuations were observed earlier at higher latitudes and later at lower latitudes [Sahai et al., 2009]. In addition, the latitudinal propagation of the VTEC fluctuations caused by MSTIDs on 10 November in the Korean region was similar to the normal behavior of MSTIDs, which propagate from northeast to southwest [Hargreaves, 1992].
Joule heating from the coupling of the magnetosphere to the ionosphere in the auroral zone can generate MSTIDs during a geomagnetically active time [Shiokawa et al., 2005]. The ionosphere receives large energy inputs produced by a large amount of local ionization resulting from Joule heating [Syam et al., 2010]. The atmosphere at high latitudes is greatly affected by these energy inputs [Syam et al., 2010]. The AGWs caused by high-latitude disturbances travel to midlatitudes and produce equatorward TIDs [Bowman, 1990]. The AE index, which is a good indicator of increases in the Joule heating rate, was also high during the period of 8–10 November [Ahn et al., 1983; Sahai et al., 2009]. This propagation direction of VTEC fluctuations from higher latitudes to lower latitudes could possibly be affected by a southward movement of the MSTIDs from the auroral region.
3.3 Spatial Structure
Characteristics of MSTIDs including amplitude, wavelength, and the orientation of wavefronts are studied by observing spatial variations of VTEC perturbations within several hundreds of kilometers in this section. Figure 6 shows VTEC data (blue triangles), moving averaged VTEC data (red asterisks), and VTEC perturbations (green diamonds) of PRN 4 observed at the Cheonan station (36.5°N, 127.1°E; hereinafter referred to as CHEN). The hourly trend of VTEC was estimated from VTEC data using a 1 h moving average filter (called moving averaged VTEC data in this paper). The long-term trend of VTEC variation over time is well captured by the moving averaged VTEC data. The perturbations of VTEC data were estimated by subtracting the moving averaged data from the original (unaveraged) VTEC data.
After estimating ionospheric perturbations from VTEC data for all satellites at all stations, the perturbed VTEC data were mapped in the Korean region. To investigate the spatial VTEC structure of the MSTIDs, two-dimensional ionospheric perturbation maps [Saito et al., 1998] of the Korean region were plotted in Figure 7 from 10:35:00 to 14:38:30 UT on 10 November 2004. Note that the color scales of VTEC perturbations in Figures 7a–7d are between −10 and 10 TECU, while those in Figures 7e–7f are between −5 and 5 TECU. The red-colored region where VTEC perturbations were larger than approximately 10 TECU first approached the eastern part of the Korean region at 10:35 UT in Figure 7a and had moved westward by 11:07 UT in Figure 7b. These two figures show that the MSTIDs on 10 November propagated from east to west. Figure 7c reveals the two-dimensional VTEC perturbation structure of these MSTIDs. One positive peak and two negative peaks appeared successively as denoted with A, B, and B′, respectively. A wavelength of about 360 km was estimated using the distance between two negative peaks (B and B′) in Figure 7c, and the peak-to-peak amplitude was estimated as about 29 TECU at 12:53 UT. The wavefront stretched from northwest to southeast in Figure 7c and extended from about 30°N to 40°N geographic latitude in Figure 7d. However, it cannot be defined as the boundary of wavefronts because there are not enough IPPs to cover the regions that are above 40°N and below 30°N of geographic latitude.
Figures 7e and 7f show that the MSTIDs still had the same propagation direction and maintained the structure of the wavefronts at 14:38:30 UT. Two positive peaks and one negative peak are denoted with C, C′, and D, respectively, in Figure 7f. However, the wavelengths at 14:38:30 UT lengthened out to about 580 km, as estimated using the distance between the two positive peaks (C and C′) in Figure 7f, and the peak-to-peak amplitude was about 10 TECU. This suggests that the intensity of these MSTIDs decreases as the waveform propagates in time. The spatial structures of the MSTIDs on 10 November in the Korean region were almost the same as those of MSTIDs that appeared in the Japanese region on the same day. These characteristics of the MSTIDs observed by Nishioka et al.  include peak-to-peak amplitudes larger than 20 TECU, several wavefronts extending from northwest to southeast and a southwestward direction of propagation.
3.4 Propagation Velocity
Spatially linear semi-infinite ionospheric “fronts” with constant front velocity and large gradients induced by ionospheric storm are modeled to predict the maximum ionospheric-induced position errors that users of ground-based augmentation system may suffer [Datta-Barua et al., 2010; Jung and Lee, 2012]. Bang and Lee  developed an automated ionospheric front velocity estimation algorithm based on this model using a three-station-based method. Since the MSTIDs on 10 November 2004 also had approximately linear wavefronts as shown in the previous subsection, the velocity of MSTIDs was estimated using this algorithm.
Figure 8 describes the movement of one of an example MSTID-generated wavefront with parameters defined for velocity estimation. In this example, the MSTID wavefront extends from northwest to southeast and moves southwestward. The IPP moves northeastward, as shown in the figure. The parameters of the wavefront are defined as follows: (x1, y1), (x2, y2), and (x3, y3) represent the locations of the three observing stations. The peak delay times, tpeak _ 1, tpeak _ 2, and tpeak _ 3, represent the times at which the VTEC data observed at the three stations reach local maximum or minimum values [Bang and Lee, 2013]. The orientation angle, i, is the angle between the y axis and the wavefront (measured in a counterclockwise direction starting from the y axis), and the direction of IPP motion, α, is the angle between the x axis and the IPP moving direction (measured in a counterclockwise direction starting from x axis) [Bang and Lee, 2013].
Three stations, namely, Cheongsong (36.26°N, 128.03°E; hereinafter referred to as CHSG), Yecheon (36.39°N, 128.26°E; hereinafter referred to as YECH), and Kimcheon (36.08°N, 128.08°E; hereinafter referred to as KIMC), were selected to estimate the velocity of MSTIDs. We selected GPS satellites whose elevation angle is 80° or higher when seen from each of the three stations in order to observe MSTIDs which pass over a specific and relatively small area (i.e., near the center of South Korea). This will allow us to measure the velocities of successive wavefronts in series when they pass by the same area and thus observe the variation of MSTID velocity. Figure 9 shows five GPS satellites, PRNs 31, 7, 17, 9, and 18, and the observation times during which the elevation angles of those satellites were larger than 80° when seen at CHSG (red circles), YECH (blue squares), and KIMC (green asterisks).
Figure 10 shows the time series of VTEC data of PRN 9 observed at CHSG (red circles), YECH (blue squares), and KIMC (green asterisks). VTEC data in the shaded area indicate that the elevation angle of PRN 9 was larger than 80° at all three stations. It is seen that CHSG, YECH, and KIMC were affected by the same MSTID wavefront, since VTEC data observed at these stations exhibited a similar trend. The peak delay times, tpeak _ 1, tpeak _ 2, and tpeak _ 3, were estimated as 20:31, 20:40, and 20:47 UT at CHSG, YECH, and KIMC, respectively. This is an example of estimation of the peak delay times for PRN 9. Other peak delay times for PRNs 31, 7, 17, and 18 were obtained in the same manner. The other parameters, including orientation angle, i, IPP direction, α, IPP velocity, Vipp, and velocity in the normal direction of wavefront, Vn, were calculated using the methods described in Bang and Lee  and are summarized in Table 1. These results were used to estimate the MSTID velocities of successive wavefronts with respect to the ground, VMSTID, which is obtained by removing the component of the IPP velocity, Vipp, resulting from satellite movement [Bang and Lee, 2013].
Table 1. Results of MSTID Velocity Estimation
Estimation Time (tpeak _ 1 + tpeak _ 3)/2 (UT)
CHSG tpeak _ 1 (UT)
YECH tpeak _ 2 (UT)
KIMC tpeak _ 3 (UT)
Figure 11 shows the temporal variation of MSTID velocity, VMSTID. The time of estimation (denoted with diamond) are obtained by taking the average of first peak delay time, tpeak _ 1, and third peak delay time, tpeak _ 3. The velocities of successive wavefronts gradually decreased from 253.99 m/s at 11:46 UT to 75.66 m/s at 21:26 UT, while passing over (near zenith) the three stations.
4.1 Ionospheric Irregularities in Small-Scale Regions
It is known that TIDs and plasma instabilities make ionospheric irregularities (e.g., plasma bubbles, ionospheric spatial gradients, or ionospheric scintillations) [Nishioka et al., 2009] which can cause loss of satellite signal tracking. Spread F, which indicates the echo trace diffused by ionospheric irregularities in the F layer, was observed by ionograms earlier in time at higher latitudes and then at lower latitudes on 10 November in the Japanese region [Sahai et al., 2009]. While irregularity structures of the order of several hundreds of kilometers were investigated in section 3.3 to study the characteristics of MSTIDs, this section focuses on analyzing localized ionospheric irregularities within several tens of kilometers accompanied by the MSTIDs by measuring TEC variations. Ionospheric irregularities in small-scale regions can pose an integrity threat to users of GPS-based navigation systems (this will be addressed in the following subsection).
The spatial and temporal variations of ionospheric irregularities on this date were investigated using the 5 min rate of TEC index (ROTI), which is defined as the standard deviation of the rate of change of VTEC over 5 min. A 5 min ROTI is widely used to investigate ionospheric irregularities with a spatial scale of about 20 km when IPPs move around zenith with an approximate velocity of 60 m/s, assuming a hypothetical 2-D ionospheric shell height of 400 km [Pi et al., 1997; Nishioka et al., 2009]. In this paper, VTEC data for satellites with elevation angles larger than 45° were used assuming a hypothetical ionospheric shell height of 350 km.
After 5 min ROTIs for each PRN observed at each station were derived, 5 min ROTIs for all satellites at each station were averaged over time intervals of 2 h from 10:00 to 24:00 UT. For this paper, these averaged ROTIs over 2 h are called “2 h averaged ROTI,” and these values quantify the intensity of localized ionospheric irregularities within a few tens of kilometers every 2 h. The spatial features of the ionospheric irregularities in the Korean region were represented by creating continuous color contours of triangular planes determined by each set of three nearest 2 h averaged ROTI values at three stations.
Figure 12 shows the spatial and temporal VTEC variations between 10:00 and 24:00 UT on 10 November 2004. Note that the color scales of 2 h averaged ROTI in Figures 12d–12f are between 0 and 1 TECU/min, while those in Figures 12a–12c are between 0 and 2 TECU/min. The intense VTEC variations first appeared in northeast South Korea in Figure 12a. The stations in northeast South Korea had higher 2 h averaged ROTI values, with a maximum of 0.8784 TECU/min at the Kangreung station (37.5°N, 128.5°E; hereinafter referred to as KANR), than those in the southwest. The stations located along the northeast coast, such as KANR, Sokcho (38.3°N, 128.5°E; hereinafter referred to as SKCH), Jumunjin (37.5°N, 128.5°E; hereinafter referred to as JUMN), and Jukbyeon (37.0°N, 129.3°E; hereinafter referred to as JUKB), were more affected by ionospheric irregularities.
Figure 12b shows the most intense ionospheric irregularities in South Korea during this event. 2 h averaged ROTIs in the time interval between 12:00 and 14:00 UT were approximately twice as large as those in the previous time interval between 10:00 and 12:00 UT. The maximum value was estimated as 1.995 TECU/min at the Yangpyeong station (37.3°N, 127.3°E; hereinafter referred to as YANP). The red-colored region where 2 h averaged ROTIs was close to 2 TECU/min extended from northwest to southeast in the central part of South Korea. From the results of Figures 12a and 12b, ionospheric irregularities accompanied by MSTIDs also occurred in northeast South Korea and propagated southwestward. The movement of these irregularity structures in a scale of several tens of kilometers is similar in direction to the movement of MSTID wavefronts.
VTEC variations between 14:00 and 16:00 UT were less intense than those in the previous time interval, with a maximum value of 1.185 TECU/min at MKPO as shown in Figure 12c. 2 h averaged ROTIs of the stations located in southwest South Korea were about twice as large as those of stations such as SKCH, KANR, JUMN, and JUKB located in northeast South Korea. This indicates that the ionospheric irregularities moved farther southwestward.
In Figures 12d–12e, between 16:00 and 20:00 UT, the irregularities became calmer but not quiet. In the last time interval between 20:00 and 22:00 UT (Figure 12f), the irregularities were finally in a quiet condition with small 2 h averaged ROTIs (below 0.3 TECU/min) over the entire region of South Korea. These results show that the movement and intensity of ionospheric irregularities in small-scale regions are strongly correlated with those of MSTIDs.
4.2 Potential Impact on GPS-Based Navigation Systems
Various kinds of abnormal ionospheric phenomena affect the performance of GPS-based navigation systems. Especially, anomalously large ionospheric spatial decorrelation—within several tens of kilometers—may exist during ionospheric storm conditions. This may cause unacceptably large residual errors to users of differential GPS, such as space-based and ground-based augmentation systems, which broadcast differential corrections of range measurements and integrity information to users. We used the LTIAM tool [Jung and Lee, 2012] to examine anomalously large ionospheric gradients observed during the MSTID event on 10 November 2004. This software tool automatically identifies ionospheric spatial gradients which are larger than a certain threshold (50 mm/km in the slant domain in this paper) and output these as anomaly candidates.
Before being confirmed as true anomalies, these candidates should be manually validated to discriminate real ionospheric spatial gradients from faulty candidates caused by receiver glitches or postprocessing errors [Jung and Lee, 2012]. However, manual validation for a particular candidate may not be feasible if carrier-phase measurements contain excessive numbers of cycle slips, especially when the intensity of ionospheric irregularity is high. This could be the case for MSTID events during which ROTIs were high as shown in section 4.1. Even if not fully validated, the possibility that the candidate is a true anomaly is still significant knowing that intense VTEC variations caused by MSTIDs are present in data input to LTIAM. Thus, the anomaly candidates output from the LTIAM prior to manual validation was analyzed in this paper.
Figure 13 shows the number of candidates of ionospheric spatial gradients higher than 50 mm/km (blue bar) and mean of 2 h averaged ROTI from all stations (downward pointing triangles) in time intervals of 2 h. Note that the left-hand axis indicates the number of anomaly candidates, while the right-hand axis indicates the mean of 2 h averaged ROTIs in TECU/min. The mean of 2 h averaged ROTI represents the intensity of the ionospheric irregularities over all of South Korea during each 2 h time interval. A total of 573 anomaly candidates were automatically detected by the LTIAM on 10 November 2004. Most of these candidates were detected during the period of 10:00 to 22:00 UT when MSTIDs affected the Korean region. The largest number of anomaly candidates was detected in the time interval between 12:00 and 14:00 UT, and the mean of 2 h ROTI was also the highest in this time interval. Thus, the results imply that the intensity of ionospheric irregularities in small-scale regions accompanied by MSTIDs is highly correlated with the occurrence of large ionospheric spatial gradients.
The largest ionospheric spatial gradient, confirmed as a true anomaly via manual validation, was about 138.5 mm/km. It was observed between the Yeosu (34.7°N, 127.7°E; hereinafter referred to as YOSU) and Hadong (35.1°N, 127.4°E; hereinafter referred to as HADG) stations tracking PRN 31 at 14:18 UT. For users of ground-based augmentation system at a 200 ft decision height (as far as 6 km from the ground reference station), this spatial gradient can cause a residual error of 0.1385 m/km × 20 km (6 km + 14 km due to the memory of a code-carrier smoothing filter) = 2.77 m. This anomaly, if undetected and combined with the worst case conditions, can produce vertical position errors large enough to threaten users of GPS-based navigation systems [Lee et al., 2011; Seo et al., 2012].
Figure 14 shows the number of loss of lock indicators (LLIs), which denote loss of tracking on L1 or L2 signals from all visible GPS satellites at all stations in time intervals of 2 h. Each LLI event represents a loss of lock at a particular station-satellite pair, and the total number of LLIs over all stations for 2 h was plotted as one bar. If both L1 and L2 locks are lost at a particular station, this counts as one LLI. The number of LLIs over time intervals of 2 h exhibited a similar trend to that of the mean of 2 h averaged ROTIs in Figure 13. The largest number of LLIs also occurred in the time interval between 12:00 and 14:00 UT when VTEC variations were most intense.
Loss of receiver tracking caused by the ionospheric irregularities is another problem that degrades the performance of GPS-based navigation systems. Since many GPS-based navigation systems use both code and carrier-phase measurements, frequent loss of carrier lock reduces availability and continuity of navigation systems. In addition, simultaneous loss of multiple satellites leads to higher noise levels on code measurements due to frequent resets of carrier smoothing filters [Seo et al., 2009]. As shown above, a large number of LLIs occurred when intense ionospheric irregularities accompanied by MSTIDs were observed. This fact implies that MSTIDs could potentially degrade not only integrity but also the availability and continuity of GPS-based navigation systems. The intensity of MSTIDs and their impacts on GPS-based navigation systems are not fully understood, and thus, further study is required.
This paper investigated ionospheric disturbances which occurred during a geomagnetically active time in the Korean region on 10 November 2004 using GPS observation data. These phenomena were identified as MSTIDs whose characteristics, peak-to-peak amplitudes larger than 20 TECU, several wavefronts extending from northwest to southeast, and a southwestward direction of propagation, were almost the same as those observed in the Japanese region on the same day. MSTIDs started in the northeast part of South Korea at about 10:00 UT and continued until about 22:00 UT. An analysis of spatial structure revealed that these MSTIDs propagated from northeast to southwest, and successive wavefronts extended from northwest to southeast. The highest intensity of MSTIDs, with a peak-to-peak amplitude of about 29 TECU and a wavelength of about 360 km, was observed at 12:53 UT. Both the intensity and propagation velocity of the MSTIDs gradually decreased as the wavefronts moved in time.
Ionospheric irregularities in small-scale regions coinciding with MSTIDs were measured with ROTI, which exhibited the highest intensity during the period of 12:00 to 14:00 UT. The spatial and temporal variations of ROTI intensity were consistent with those of the MSTIDs. The intensity of ionospheric irregularity was highly correlated with the occurrence of large ionospheric spatial gradients. The results obtained in this study suggest that high-intensity MSTIDs can generate anomalous ionospheric gradients. Loss of GPS signals also frequently occurred during the MSTID events. While it is evident that anomalous ionospheric gradients and frequent loss of signals which occur concurrently with MSTIDs could be potential threats to users of space-based and ground-based augmentation systems, the knowledge we have is too limited to understand the full spectrum of actual impacts. More work is needed to quantify the magnitude of these threats and their likelihood of occurrence.
Many nations now operate their own GPS station networks, and these continue to expand. International collaboration would present an opportunity to maximize the use of all GPS observation data from dense and widely distributed stations, thereby helping to better understand the regional behavior of MSTIDs and support a more detailed threat assessment of how MSTID-related threats can impact GPS-based navigation systems. This information can be used to improve the performance of the existing ionospheric anomaly monitors implemented in space-based and ground-based augmentation systems.
The authors thank the National Geographic Information Institute (NGII), the DGPS Central Office (DCO), and the Korea Astronomy and Space Science Institute (KASI) for providing the Korean GPS observation data. The authors are also grateful to the Space Weather Prediction Center (SWPC) of the National Oceanic and Atmospheric Administration (NOAA) and the World Data Center for Geomagnetism at Kyoto University for the geomagnetic index data. The Global Ionosphere Map (GIM) was made available by the International GNSS Service (IGS), a service of the International Association of Geodesy and of the Federation of Astronomical and Geophysical Data Analysis Services. Moonseok Yoon was supported by KASI and the National Radio Research Agency (RAA) of the Republic of Korea. This research was supported by the KAIST Institute in 2013.