© K. Jacobsen et al., Published by EDP Sciences 2025

1 Introduction

On May 10th in 2024, the arrival of a Coronal Mass Ejection (CME) sparked what turned out to be the strongest geomagnetic storm in several decades. The month of May 2024 was a particularly active period with respect to space weather, with a double-digit number of X-class flares, several of which launched Earth-directed CMEs. The events at the Sun and their connection to the space weather impacts at Earth are described in Kwak et al. (2024). Kwak et al. (2024) also provides an analysis of effects in Korea and notes that aurora was captured by all-sky cameras there for the first time since the Halloween storm of 2003. Hajra et al. (2024) analyzed the interplanetary drivers of the space weather impacts. Tulasi Ram et al. (2024) investigated the mechanisms and impacts of the storm, with a particular focus on the compression of the magnetosphere. Hayakawa et al. (2025) presented a collection of important observational data describing the major features of the storm. The storm main phase spanned May 10 and 11, transitioning to a recovery phase early on the 11th. Strong impacts were still present throughout the early parts of the recovery on May 11th and 12th.

The storm had strong impacts on many parts of the Earth’s system. Pierrard et al. (2024) documented unprecedented changes to Earth’s radiation belts. Evans et al. (2024) reported remarkable changes of composition, temperature, and dynamics in the Earth’s thermosphere, while Mlynczak et al. (2024) presented significantly increased infrared emission from the thermosphere. Two unusually strong Forbush Decreases were observed during the storm (Mavromichalaki et al., 2024). The strength of the storm and the expansion of the auroral oval resulted in space weather impacts in regions that are not accustomed to such events. In Chile, Lazzús & Salfate (2024) analyzed the connection between the solar wind parameters and Pc5 pulsations measured by a ground magnetometer. Large-scale, large-amplitude anomalies of Vertical Total Electron Content (VTEC), as well as a shift in the location of the Equatorial Ionization Anomaly (EIA), were observed in India by Jain et al. (2025). In Mexico, Gonzalez-Esparza et al. (2024) took advantage of the unique opportunity for space weather and auroral studies using a large variety of instruments and methods. Karan et al. (2024) presented large-scale changes in the nightside ionosphere. The EIA southern crest was observed to move poleward and merge with the auroral region, which has not been observed before. Carmo et al. (2024) reported on ionospheric effects in the Latin American sector during the storm, including a super equatorial plasma bubble. Impacts over the Mediterranean sector, and in particular in Italy, were reported by Spogli et al. (2024). They attributed negative ionospheric storm effects to neutral composition changes, analysed enhanced ionospheric disturbances measured by the Rate-Of-TEC Index (ROTI) in connection to several different physical phenomena, and noted the very uncommon presence of aurora in the skies above Italy. As the event pushed the auroral oval far equatorward, a large part of Earth’s population had the opportunity to see the auroral lights, many for the first time. Grandin et al. (2024) presented a citizen science study of auroral activity through an online survey of almost 700 ordinary citizens from all around the world. Aurora was reported at locations even below 30 degrees magnetic latitude. Lockwood et al. (2025) puts the event into a historical context by comparing the auroral observations during extreme storms of the last 375 years and finding it to have the most extensive aurora since 1872.

The main focus of this study is scintillation activity at high latitudes. A variety of phenomena are linked to scintillations at high latitudes, with polar cap patches and particle precipitation being two main sources of plasma irregularities that result in scintillation (e.g., Moen et al., 2013; Jin et al., 2014; van der Meeren et al., 2015; Clausen et al., 2016; Jin et al., 2016; Spogli et al., 2016; Fæhn Follestad et al., 2020; Enengl et al., 2023, 2024; Miloch et al., 2024). On the dayside, scintillation occurs frequently in the cusp region (Prikryl et al., 2012; Oksavik et al., 2015). On the nightside, scintillation is most often found in the auroral oval region, with peak activity occurring as a plasma patch moves into a particle precipitation region (Jin et al., 2017). Scintillation has also been reported in connection with a Tongue of Ionization (ToI), in which high-density plasma is convected across the polar cap (van der Meeren et al., 2014).

Scintillation of Global Navigation Satellite System (GNSS) signals degrades the positioning results (Kintner et al., 2007), and earlier works on ionospheric disturbances at high latitudes have documented the impacts on GNSS services (e.g., Jacobsen & Schäfer, 2012; Jacobsen & Dähnn, 2014; Andalsvik & Jacobsen, 2014; Jacobsen & Andalsvik, 2016; Yang et al., 2020; Follestad et al., 2021; Paziewski et al., 2022).

As particle precipitation is one of the main sources of scintillation-inducing plasma structures and is directly related to currents flowing into and out of the ionosphere, some earlier works have studied the relation between ionospheric currents and scintillation. Jacobsen & Andalsvik (2016) studied the 2015 St. Patrick’s day storm and found that “GNSS disturbances, measured by ROTI, were most intense on the poleward edge of poleward-moving electrojet currents.” In another study of the same storm, with a more global view, Prikryl et al. (2016) also found the same relation, noting that “In relation to auroral electrojet currents, scintillation maps to strong EICs, particularly to the poleward side of the westward electrojet and to the poleward edge of the eastward electrojet current region.” Equivalent ionospheric currents (EIC) are a simplified representation of the true, complex currents within the ionosphere and are roughly the Hall current system, but may contain small contributions from Pedersen and field-aligned currents due to the non-uniformity of conductivity in the ionosphere. They model the ionospheric currents as if they were confined to a thin, conductive shell. Prikryl et al. (2021) further investigated the relation in a 4-case study and among other results found that with respect to vertical currents, the scintillation was particularly related to “boundaries between downward R1 and upward R1/R2 currents in the Harang discontinuity region.” While not the main focus of Prikryl et al. (2022), ionospheric currents and scintillation were briefly explored. Scintillation was observed to be collocated with a ToI being fragmented into plasma patches. In relation to the ionospheric currents, the phase scintillation in the auroral zone was found to be “associated with the westward electrojet and the poleward edge of the eastward electrojet.”

The high-latitude ionosphere response to the storm has already been studied by Themens et al. (2024), who focused mainly on the TEC parameter with supporting observations from ionospheric radars, ionosondes, and scintillation receivers. This study expands on the topic of scintillation by including data from a larger network of receivers, analyzing scintillation parameters in greater detail, and investigating connections to magnetometer-derived currents and satellite positioning performance degradations.

We note that at high latitudes, phase scintillation often occurs independently of amplitude scintillation (Alfonsi et al., 2011). The main reason for this is the dependence of the phase scintillation index on the plasma drift velocity. As the drift speed increases, the standard high-pass filter used for phase scintillation calculations does not fully exclude the impact of refractive effects (McCaffrey & Jayachandran, 2019; Wang et al., 2022). Thus, the phase scintillation index at high latitudes will often be a combination of refractive and diffractive phase variations. Separating their respective contributions is challenging, and there is currently no consensus on an alternative method for the calculation of the phase scintillation index.

2 Data sources

2.1 Scintillation receiver networks

Figure 1a shows the location of the scintillation receivers used in this study. The scintillation networks are:

• Red stars – Finnish Meteorological Institute (FMI). FMI has operated scintillation receivers since 2022. All receivers are Septentrio PolaRx5S units, and Septentrio’s RxTools software suite is used to calculate the scintillation indices and other signal parameters. During the May 2024 Mother’s Day storm, three receivers were operational. FMI’s receivers are part of the operational network that provides scintillation data for the PECASUS global space weather center for aviation.

• Green stars – Norwegian Mapping Authority (NMA). Consists of a mix of Septentrio PolaRxS and PolaRx5S receivers.

• Blue stars – Space Weather forecasting for Arctic Defense Operations (SWADO) network owned by the Danish Defense Acquisition and Logistics Organization (DALO) and operated by the Technical University of Denmark, Department of Space Research and Technology (DTU Space). The SWADO network consists of Septentrio PolaRx5S receivers sampling at 100 Hz and transmitting files with scintillation data to DTU servers.

• Purple stars – Canadian High Arctic Ionospheric Network (CHAIN) (Jayachandran et al., 2009). Consists of a mix of Septentrio PolaRxS and Novatel GSV4004B receivers. Details can be found at the CHAIN website.¹

Figure 1 Overview of scintillation receivers and magnetometers used in this study. (a) Locations of scintillation receivers used in this study. FMI in red, NMA in green, SWADO in blue, and CHAIN in purple. (b) The IMAGE magnetometer network in red. Magnetometer stations that were used for the North American sector are shown in blue.

All of the scintillation receivers operate at a sampling rate of 50 Hz or more. The phase and amplitude scintillation indices are calculated in the standard way (Van Dierendonck et al., 1993; Van Dierendonck, 1999). The phase scintillation index σ_ϕ is the standard deviation of the filtered phase, applying a sixth-order Butterworth high-pass filter with a frequency cutoff of 0.1 Hz to the phase data before the calculation of the index. For the phase scintillation from the Novatel receivers, a simple outlier removal was applied in which all phase scintillation values above 2 radians were removed. The amplitude scintillation index S4 is the normalized standard deviation of the detrended signal intensity, corrected for ambient noise. 350 km is used as the ionosphere altitude when calculating the ionospheric pierce point (IPP) of the scintillation measurements.

The S4 index data contains a high number of false positives at low satellite elevation angles. To avoid these, it has been filtered to include only data at elevation angles above 30 degrees. The σ_ϕ index data does not have the same issue, and thus includes data for all elevation angles above 5 degrees.

2.2 Magnetometer networks

Figure 1b shows the location of ground-based magnetometers used in this study.

• Red triangles – The International Monitor for Auroral Geomagnetic Effects (IMAGE) magnetometer network.²

• Blue triangles – Eleven different arrays in the North American (NA) sector. These arrays include AUTUMNX (Athabasca University THEMIS UCLA Magnetometer Network), NRCan (Natural Resources Canada), CARISMA (Canadian Array for Real-time Investigations of Magnetic Activity) (Mann et al., 2008), GIMA (Geophysical Institute Magnetometer Array), Technical University of Denmark (DTU) Magnetometer Ground Stations in Greenland, MACCS (Magnetometer Array for Cusp and Cleft Studies) (Engebretson et al., 1995), MagStar, THEMIS GMAG (Ground MAGnetometers) (Russell et al., 2008), and USGS (United States Geophysical Survey).

The magnetic field measurements are used to estimate the ionospheric equivalent current density. The same methods are applied to the two groups of magnetometers, but with some differences in grid configurations and other parameters. Note that the vertical currents from the IMAGE network processing are expressed as current densities (A/km²), while the vertical currents from the NA sector are expressed as currents flowing in the grid points (kA).

For several figures presented in Section 3.3.1, Section 3.3.2 and Section 3.3.3, a colored background shows the East-West current density in the Quasi-Dipole magnetic coordinate system. This has been chosen to have a better correspondence with the auroral electrojets than the geographic coordinate system. The transformation calculations were performed using the ApexPy software (van der Meeren et al., 2025), which implements the computations described in Richmond (1995) and Emmert et al. (2010).

2.2.1 Magnetometer processing – IMAGE

In May 2024, 47 stations located in northern Europe and Greenland (Fig. 1b) were providing data. After correcting the magnetometer data for any erroneous spikes and jumps, a 10-day sliding median baseline was subtracted from the data. The remaining variation magnetic field consists of an external part mainly due to ionospheric electric currents and an internal part due to induced telluric currents in the conducting ground. The ionospheric equivalent current density was calculated from the variation magnetic field using the two-dimensional (2-D) Spherical Elementary Current System (SECS) method (Vanhamäki & Juusola, 2020).

The ionospheric equivalent current density is a horizontal divergence-free sheet current density at 90 km altitude that fully reproduces the magnetic field below the ionospheric current sheet due to all external current systems. At high latitudes, it can be approximated to correspond to the divergence-free part of the horizontal ionospheric sheet current density, i.e., the component of the horizontal ionospheric currents that closes within the ionosphere. The field-aligned currents and the curl-free part of the horizontal ionospheric sheet current density that closes them are magnetically almost invisible on the ground. The curl of the ionospheric equivalent current density, called the vertical current density in this study, can be used as a proxy for the field-aligned current density under some conditions (see, e.g., Weygand & Wing, 2016; Juusola et al., 2023; Walker et al., 2024).

2-D SECS poles were placed in the IMAGE region on uniform grids with 0.5° latitude and 1° longitude resolution at 1 m depth and at 90 km altitude, and their amplitudes were determined by fitting the superposed magnetic field of the SECSs to the three components of the measured magnetic field. The layer at 1 m depth describes the telluric equivalent current density, thus allowing us to separate the disturbance magnetic field caused by them. The resulting ionospheric equivalent current density is most detailed and reliable in areas where the magnetometer station coverage is dense, such as northern Fennoscandia. In areas where there are large gaps in the station coverage, such as oceans, the equivalent current density only provides a very rough estimate of the large-scale divergence-free current system.

2.2.2 Magnetometer processing – NA sector networks

The ionospheric equivalent current density was calculated from the variation magnetic field using the two-dimensional (2-D) Spherical Elementary Current System (SECS) method with some minor differences from the IMAGE array calculations. The ionospheric equivalent current density is derived for an altitude of 100 km with a spatial resolution of 2.9 degrees in geographic latitude by 6.9 degrees in geographic longitude. This spatial resolution is based on the densest portion of the magnetometer arrays. See Weygand et al. (2011) for more details. The spatial resolution of the field-aligned-like currents is 1.45 degrees in geographic latitude by 3.45 degrees in geographic longitude. For each ground magnetometer station, a quiet time background must be removed. To calculate the quiet time background intervals of relatively smooth magnetometer data with a low standard deviation varying in length from an hour to a whole day, an automated routine is used for all three components of the magnetic field for nearly every day over a three-month period centered on the month of interest. These intervals are then averaged together and smoothed to create one 24 h quiet time background interval. Three months of quiet time data are used because it is difficult to obtain a quiet time background from the higher latitude stations where geomagnetic activity is frequently present, even when using the World Data Center for Geomagnetism’s five quietest days. Weygand et al. (2011) for more details.

2.3 Positioning service monitors

NMA operates a national network Real-Time Kinematic (RTK) (Frodge et al., 1994; Wübbena et al., 1996; Landau et al., 2002; Rizos, 2003) positioning service named “CPOS”. The service is based on the Trimble Pivot Platform (TPP) (version 4.1.3), which is a commercial software solution from Trimble Inc.. The Network Processor in TPP contains models for the troposphere, the ionosphere, and satellite orbits. Based on observations from the receiver network, it estimates the state of these models, which are then used to generate a data stream from a Virtual Reference Station (VRS) that is positioned close to a user. The user receiver receives GNSS observations from the VRS every second as if it were a physical station, but the observations have been calculated by TPP based on its internal modeling of reality. The user receiver then performs a differential positioning calculation relative to the VRS. The objective of the processing strategy is to eliminate/minimize the troposphere terms, ionosphere terms, and satellite orbit/clock errors from the observations before they are used to calculate a position. Details regarding TPP’s internal algorithms are not public information, but it can be considered as an industry-standard implementation of network RTK algorithms. The performance of the service is independently monitored by what we refer to as “RTK monitors”. These are survey-grade GNSS receiver/antenna equipment setups that connect to the service the same way a normal user. Positioning results from these are logged continuously. In this study, they are used to investigate the impacts of the disturbances on the positioning service. Additional information about CPOS and the monitor system can be found in Jacobsen et al. (2023).

2.4 Geodetic GNSS receiver networks

NMA monitors the state of the ionosphere using data from its national network of ≈ 300 geodetic receivers which cover the entire mainland Norway as well as the remote islands of Svalbard, Hopen, Bjørnøya, and Jan Mayen, and data from cooperating agencies in Denmark (Geodatastyrelsen and DTU), Sweden (Lantmäteriet), Finland (The Finnish Geodetic Institute (FGI) and Geotrim OY), and the Faroe Islands (Umhvørvisstovan). The main results are maps of Vertical Total Electron Content (VTEC) and Rate Of TEC Index (ROTI). Currently, these monitoring results are made publicly available through webpages.^{3, 4} In this study, the VTEC and ROTI data in Section 3.3.3 have been extracted from these maps. The NMA processing for the VTEC and ROTI data uses 350 km as the ionosphere altitude. The ROTI data are based on 1 second resolution data and calculated for 5-minute intervals.

VTEC data for the North American sector have been downloaded from the Massachusetts Institute of Technology (MIT) Madrigal database.⁵ These VTEC data use 350 km as the ionosphere altitude.

ROTI data for the North American sector have been downloaded from the website of the Institute for Space-Earth Environment Research (ISEE), Nagoya University.⁶ These ROTI data are based on 30-second resolution data and calculated for 5-minute intervals, using 300 km as the ionosphere altitude.

Several factors have an influence on ROTI values (Jacobsen, 2014). The dominant factor is the sampling rate used, with higher sampling rates leading to higher values and higher sensitivity to smaller-scale plasma structures. The use of data from different frequencies/signals and/or different receiver hardware is mainly important for the base noise level. During highly disturbed conditions, this will be a minor part of the total value. Both data sources used for ROTI in this study are based on equipment of comparable quality (referred to as “geodetic receivers”, meaning professional equipment). There may be some differences in base noise levels, but they are expected to be small. Because of this and because for this study we are mainly interested in highly disturbed conditions, these differences are considered to be not important for this study.

The important difference for this study is the sampling rates used for the ROTI data from NMA (1 s) and Nagoya (30 s). The Nagoya data will have lower values and be mostly influenced by larger-scale plasma structures, meaning much larger than scintillation-inducing scales. Those plasma gradients will still tend to be correlated with the occurrence of scintillation, as such gradients may be susceptible to plasma instabilities and/or are a result of particle precipitation. To distinguish between the two flavors of ROTI, they will be referred to as “ROTI (1 s)” and “ROTI (30 s)”.

2.5 SuperDARN

Ionospheric convection maps have been calculated and plotted based on observations of the Super Dual Auroral Radar Network (SuperDARN) (Greenwald et al., 1995; Chisham et al., 2007; Nishitani et al., 2019), using the SuperDARN Radar Software Toolkit (SuperDARN Data Analysis Working Group et al., 2022).

3 Observations

3.1 Solar wind

As the interplanetary sources of the solar wind structures and the development of the storm are well described in other studies, which are referenced in the introduction (see e.g., Fig. 2 in Spogli et al., 2024), we only recall the major features here. The storm started around 17 UT on May 10th. At this time, the solar wind magnetic field strength, speed, and dynamic pressure increased suddenly to unusually high values. Throughout the strongest part of the storm, on May 10th and 11th, the Interplanetary Magnetic Field (IMF) Bz is predominantly strongly negative, with shorter excursions to positive values. The solar wind speed increases to even higher values during this time, while the pressure exhibits significant variations but, on average, declines. These conditions are conducive to generating geomagnetic storms and substorm activity that will result in strong ionospheric disturbances.

3.2 Overview of scintillation

3.2.1 Overview of scintillation – Storm Onset

Figure 2 shows phase scintillation maps at a selection of times during the onset and early phase of the storm. The phase scintillation (σ_ϕ for the L1 frequency of GPS and Galileo satellites) is shown as colored dots at the ionospheric pierce points (IPPs), assuming an altitude of 350 km for the ionospheric single-layer model. The colored background is an interpolation of those values using a Gaussian radial basis function for the interpolation. Figure 3 shows SuperDARN convection maps at times corresponding to those in Figure 2. The colored dots with lines extending from them show the modeled convection velocity, with both the color and the length of the lines scaling with the convection speed. Contour lines are drawn for the electrostatic potential. Note that the modeling is most trustworthy near the areas in which there are observations (i.e., near the colored dots). The first map (Fig. 2a) shows the quiet state before the onset of the storm-induced scintillation. Figure 2b is right after the onset, showing scintillation activity between 65° and 70° North in Canada. In the convection map (Fig. 3b), there is a region of backscatter north of Alaska, but not much colocated with the scintillation. 10 min later (Fig. 2c), the affected area has expanded in longitude and moved ≈ 5° equatorward. The area of scintillation lies along the equatorward edge of the measurement points in the convection map. In Figure 2d, the most intense scintillation intense activity is located at the equatorward edge of the strongest convection flow (Fig. 3d), in the inflow region of the transpolar convection. This pattern is also seen in the following plots (Figs. 2e–2i and 3e–3i), but the activity is not constant in intensity. Figure 2h still shows activity below 60° N but does not contain strong scintillation, while in Figure 2i the stronger scintillation reappears. The cross-polar potential starts at about 20 kV and increases to more than 110 kV for the most active periods, staying in the range 80–120 kV for Figures 2d–2i and 3d–3i.

Figure 2 Phase scintillation (σϕ, in radians) maps at a selection of times before the onset of the activity and in the early phase of the storm.

Figure 3 SuperDARN convection maps, for times matching those in Figure 2.

3.2.2 Overview of scintillation – ToI

Figure 4 shows phase scintillation maps before, during, and after the ionospheric Tongue of Ionization (ToI) that has been noted in several other papers on this storm (see e.g. Themens et al., 2024; Hayakawa et al., 2025). The first map (Fig. 4a) shows the state before the formation of the ToI, the following three maps (Figs. 4b–4d) show the scintillation activity stretching across the polar cap in connection with the ToI, while the last two maps (Figs. 4e–4f) show the scintillation shortly after the ToI. Figure 5 shows SuperDARN convection maps at times corresponding to those in Figure 4. Total Electron Content (TEC) maps showing the ToI can be found in Themens et al. (2024). In Figure 4a, before the formation of the ToI, there are regions of scintillation activity both on the dayside over Canada and Alaska, and on the nightside in Norway and Finland, but not in the polar cap. The convection map (Fig. 5a) is greatly expanded and somewhat chaotic, but shows strong indications of fast flows over Canada. In Figure 4b, the ToI has formed, and scintillation is occurring in a broad channel crossing the polar cap over Greenland. The next two plots (Figs. 4c and 4d) also show this region of scintillation activity. The convection maps (Figs. 5b–5d) show an inflow region around the border of Canada and Alaska, being most clearly shown in Figure 5c. At 22:30 UT, there is a large shift in the IMF magnetic field. Themens et al. (2024) notes that 30 min after this shift, the polar cap patch activity is cleared over the course of an hour. Figure 4e is at 23:30 and shows that the polar cap scintillation has mostly disappeared, consistent with the disappearance of the ToI and polar cap patches. Over Canada, there is a region of strong scintillations in the inflow region of the convection pattern. Half an hour later, Figure 4f shows the same pattern of scintillation activity but with less intensity.

Figure 4 Phase scintillation (σϕ, in radians) maps before, during, and after ToI.

Figure 5 SuperDARN convection maps, for times matching those in Figure 4.

3.2.3 Overview of scintillation – May 11th

Figure 6 shows phase scintillation maps at a selection of times during May 11th, illustrating the variety in the location and shape of the areas affected by scintillation. Figure 7 shows SuperDARN convection maps at times corresponding to those in Figure 6. At 03 UT on May 11th  (Fig. 6a) there is scintillation activity visible at the low latitude edges of the measurement coverage. As reported in several other studies of the event, visible aurora and other disturbances were observed at low latitudes. It is likely that during the periods of maximum auroral oval expansion, there is also some scintillation activity at lower latitudes than those that are observable by the scintillation receiver networks used in this study. Likewise, the SuperDARN convection maps may struggle to accurately represent the convection if the convection pattern extends to the south of the areas covered by the radars. The next three plots (Figs. 6b–6d) show scintillation in the nightside, near dawn, moving poleward across Fennoscandia to the ocean area between Norway and Svalbard. Compared to the convection maps (Fig. 7f), the active area is along the edge of the convection contour line. We note that there is some uncertainty in those convection patterns, as there are few measurement points available to constrain the model. At 07:05 UT (Fig. 6e), strong scintillation is observed between 50° N and 60° N along almost the entire observed area, corresponding to the dawn flank return convection. Whether the same activity occurs at the dusk flank return convection is unknown, as the scintillation networks do not have coverage there at that time. At 12:20 UT (Fig. 6f), scintillation is observed in a band stretching from Finland to Iceland, with less intense activity west of Iceland. This roughly corresponds to the dayside convection inflow region (Fig. 6f). Half an hour later (Fig. 6g), a band of scintillation is also observed on the nightside stretching along the dusk flank return convection pattern (Fig. 7g). At 14:00 UT (Fig. 6h) a similar pattern is observed, with scintillation in the dayside convection inflow region near Iceland, and in both of the return convection flows (Fig. 7g). At 20:40 UT (Fig. 6i) there is a spike in scintillation activity in a broad region at the nightside. The convection maps do not contain any measurements at that time and appear to miss this part of the event.

Figure 6 Phase scintillation (σϕ, in radians) maps at a selection of times during May 11th.

Figure 7 SuperDARN convection maps, for times matching those in Figure 6.

The full set of scintillation maps is available in the Supplementary material as a movie file. (movie_Scint_20240510to20240512.avi)

3.3 Scintillation vs. ionospheric currents

3.3.1 Maps of scintillation and currents

Figure 8 shows an example of a combined map of Quasi-Dipole (QD) east-west ionospheric equivalent current densities and ionospheric phase scintillation. Positive (red) current values are eastward, while negative (blue) currents are westward. The arrows show the complete horizontal current densities. The scintillation is shown by colored dots, whose area scales linearly with the scintillation index value. While the scintillation color is capped at 0.5 radian, the size of the dots continues to scale beyond that. The main feature in Figure 8 is a broad westward current. Scintillation is predominantly observed within that current structure. The location of the current structure matches the return flow in the convection map (Fig. 6d).

Figure 8 The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map. This figure is for the same time as panel (d) in the Figures 6 and 7.

Figure 9 shows the same as Figure 8, but for the North American sector. There are two main features in Figure 9:

• A broad eastward current, in which strong scintillation is observed.

• Scintillation in connection with the ToI, stretching towards the top right of the plot.

Figure 9 Same as Figure 8, but for the North American sector. This figure is for the same time as panel (d) in the Figures 4 and 5.

The full set of map plots, as shown in Figures 8 and 9, is available in the Supplementary material. (CurrentsAndScint_Europe.zip & CurrentsAndScint_NA.zip)

3.3.2 Keograms – North America

Figure 10 shows keograms (i.e., time-vs.-latitude plots) of currents, scintillation, ROTI (30 s), and vertical total electron content (VTEC). The east-west current keogram is taken at a longitude of 99.575° W, the vertical current keogram is taken at a longitude of 101.3° W, while the scintillation keogram uses the mean observation values in a longitude interval of 95–105° W, and the ROTI/VTEC keograms use the mean observation values in a longitude interval of 98–102° W. The longitudes used for the current’s keograms are the current’s grid points closest to 100° W.

Figure 10 a) QD east-west (red-blue) current density keogram at 99.575° W. b) Vertical (red is downward) current keogram at 101.3° W. c) Amplitude scintillation index keogram for the mean value in the longitude interval of 95–105° W. d) Phase scintillation index keogram for the mean value in the longitude interval of 95–105° W. e) ROTI (30 s) keogram for the mean value in the longitude interval of 98–102° W. f) VTEC keogram for the mean value in the longitude interval of 98–102° W.

Figure 11a presents a zoomed-in view of the horizontal currents data shown in Figure 10, with an overlaid filtered selection of the scintillation data. Only scintillation values ≥ 0.3 radian is included in this plot. Figure 11b shows the same currents, but with ROTI (30 s) values overlaid. Only ROTI values ≥ 1.5 TECU/minute is included in this plot. The scintillation index and ROTI are both measures of phase variations, but due to the very different time resolution of the data used in their computation (50 Hz vs. $\frac{1}{30}$ Hz), they are sensitive to very different plasma structure scale sizes. Thus, both are included in the zoomed-in figure even though ROTI has a much more complete spatial coverage.

Figure 11 A zoomed-in view of panel (a) of Figure 10, with a) a filtered version of phase scintillation plotted on top, showing only values ≥ 0.3 radians. b) A filtered version of ROTI (30 s) plotted on top, showing only values ≥ 1.5 TECU/min. Small black dots are plotted when there is data coverage of σϕ/ROTI but with values below the thresholds.

There are several obvious similarities in the patterns of currents and scintillations in Figure 10, but the scintillation is not a one-to-one translation of the currents. For the time periods where there is a strong east-west current, the vertical currents are mostly distributed in two bands at the northern and southern edges of the horizontal current. The scintillation at times corresponds to the horizontal current, but at other times there are strong currents without scintillation. To further investigate the details of this, Figure 11 presents only the most active time period. The scintillation and ROTI keograms have been filtered to show only strong activity and then overlaid on the east-west current keogram in order to directly show the correspondence. Key observations include:

• Scintillation in the eastward electrojet current between 19 UT on May 10th and 01 UT on May 11th. There is a preference for being located at the poleward edge of the current region, but it is less clear than observed in previous studies. It is important to note that the electrojets extend to lower latitudes than the coverage of the scintillation measurements, so the lack of scintillation at 40–50° N is likely due to a lack of observability. The high ROTI values do extend within the electrojet to these lower latitudes in the evening of May 10th, indicating the continued presence of plasma structures as scales that may also produce scintillation-scale structures given suitable conditions for plasma instabilities.

• Scintillation at the highest latitudes (65° N and northward), mainly between 22 and 23 UT on May 10th, in connection with the ToI. The ROTI has a stronger response within the ToI than the σ_ϕ, indicating a large amount of large-scale plasma structures of which only some develop the smaller-scale structures that cause the phase scintillation index to increase. The enhanced plasma density of the ToI is clearly visible in Figure 10f. In the Supplementary material file DMSP_SelectedPlots.zip, there is a plot Map_PhaseScint1_and_DMSP_20240510_2205.png showing Defense Meteorological Satellite Program (DMSP) particle flux measurements in the area of the ToI, confirming that in this case, the particle precipitation is not a cause of the irregularities.

• Scintillation in the westward electrojet at 13–14 UT on May 11th. This is a part of the return convection, and the scintillation is likely related to plasma structures with a high flow velocity. This is consistent with the scintillation being located in the middle of the current area.

• The lack of scintillation, compared to what was observed in the previous bullet point, in the westward electrojet at 09–11 UT on May 11th. Only a few isolated scintillation measurements are observed. A likely explanation for the difference is that while there is a fast convection flow, the plasma is not as structured as in the electrojet flow observed later. The low to non-existent reaction in ROTI also supports this conclusion.

• Between 02 and 09 UT on May 11th there are several shorter periods where there is a reaction in the ROTI with little to no reaction in σ_ϕ. These are most likely periods in which there is some large-scale structuring of the plasma, but little to no generation of small-scale structures.

• On May 11th the ionospheric plasma density is depleted, not showing any sign of the normal daytime peak (Fig. 10f). For some periods during the day, there are smaller enhancements of plasma density colocated with enhanced ROTI. These are most clearly seen at the start of the day.

Figure 12 focuses on the time with the most intense S4 measurements. The background colormap shows the vertical currents, while the two panels show filtered versions of the amplitude and phase scintillation, respectively. Several time periods with different behaviour are observed:

• Between 17 and 19 UT on May 10th there are several shorter periods of currents alternating between upwards and downwards. The strongest observed phase and amplitude scintillation during this time is colocated with the most intense currents.

• The first long-lasting intense current structure starts at 19 UT on May 10th and weakens later between 22 and 23 UT. Scintillation is observed in the first half of this, but due to limitations in spatial coverage, any scintillation that might have occurred in its latter half could not be observed. The amplitude scintillation is strongest within the area of downward currents at 19–20:30 UT. Phase scintillation is observed around the peaks of the amplitude scintillation, but the strongest phase scintillation occurs a little later and is located in the middle between the two vertical currents. Unfortunately, due to the higher elevation angle filter for the S4 data, this lack of S4 observations may just be a result of a lack of observations below 52° N, and thus, we cannot make any conclusion based on the difference between phase and amplitude scintillation at that location and time.

• At 21:30–23:00 UT on May 10th both amplitude and phase scintillation are observed in connection with the ToI. The peaks of amplitude and phase scintillation are not perfectly colocated. As the activity is very scattered and the amplitude scintillation values are not very large, we are wary of overinterpreting this. If the observation is an indication of a real difference, a potential explanation is that the dominant sources of phase and amplitude scintillation have a different altitude profile.

• The second long-lasting intense current structure starts at 23 UT on May 10th and weakens at 01:30. In its first part, between 23 UT and midnight, scattered amplitude scintillation is observed within the downward current regions but the most intense values are located in the middle between the two vertical currents. The phase scintillation is also observed within the downward currents in connection with the amplitude scintillation there, but a much larger and stronger area of phase scintillation is centered in the middle. This most intense area of phase scintillation likely has a contribution from the refractive effects of convecting plasma structures in the electrojet.

• In the last part, from 00:30 on May 11th until the end of the plot, there are scattered amplitude scintillation observations at high latitudes. These are not associated with phase scintillation, ROTI, VTEC, or intense currents. As there are no signs of activity in any of these other measures, the amplitude scintillation could in this case be false positives, or it could be caused by specific local conditions that do not register in any of the other activity measures.

Figure 12 A zoomed-in view of panel (b) of Figure 10, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of phase scintillation plotted on top, showing only values ≥ 0.3 radians. Small black dots are plotted when there is data coverage of S4/σϕ but with values below the thresholds.

3.3.3 Keograms – Europe

Figure 13 shows keograms of currents, scintillation, ROTI (1 s), and VTEC. The current keograms are taken at a longitude of 20° E, the scintillation keograms use the mean observation values in a longitude interval of 15–25° E, and the ROTI/VTEC keograms use the mean observation values in a longitude interval of 18–22° E.

Figure 13 a) QD east-west (red-blue) current density keogram at 20° E. b) Vertical (red is downward) current density keogram at 20° E. c) Amplitude scintillation index keogram for the mean value in the longitude interval of 15–25° E. d) Phase scintillation index keogram for the mean value in the longitude interval of 15–25° E. e) ROTI (1 s) keogram for the mean value in the longitude interval of 18–22° E. f) VTEC keogram for the mean value in the longitude interval of 18–22° E.

Figure 14 presents a zoomed-in view of the horizontal currents data shown in Figure 13, with an overlaid filtered selection of the S4 and ROTI (1 s) data. Only S4 values ≥ 0.1 and ROTI values ≥ 4 TECU/min is included in this plot. The limit for ROTI in this case is greater than the limit applied for Figure 11 because these ROTI data are based on higher resolution measurements. In all cases, the objective of this filtering is to show the locations of peak activity. The limits are set high enough to filter out values seen in quiet and weakly disturbed states, and in such a way that the background currents map is sufficiently visible. As there is a modest amount of amplitude scintillation present, the limit for S4 excludes quiet-state values. Figure 15 presents a further zoomed-in view, with vertical currents and ROTI data filtered in the same way.

Figure 14 A zoomed-in view of panel (a) of Figure 13, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of ROTI (1 s) plotted on top, showing only values ≥ 4 TECU/min. Small black dots are plotted when there is data coverage of S4/ROTI, but with values below the thresholds.

Figure 15 A zoomed-in view of panel (b) of Figure 13, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of ROTI (1 s) plotted on top, showing only values ≥ 4 TECU/min. Small black dots are plotted when there is data coverage of S4/ROTI, but with values below the thresholds.

Using ROTI (1 s) as a proxy for the phase scintillation allows the use of thinner longitude slices and provides a much more complete spatial coverage, as it is calculated based on a much larger receiver network. While the ROTI (1 s) is not entirely equivalent to the phase scintillation index, it is considered to be a usable proxy (e.g., Makarevich et al., 2021), and is more than adequate for the purpose of tracking the region of active disturbances in the ionosphere. A simple visual comparison of the σ_ϕ and ROTI keograms confirm that they show the same general features. As before, there are vertical currents of opposite direction on the pole- and equatorward edges of intense electrojet currents, and while there are many similarities between the occurrence of scintillation/ROTI and strong currents, there are interesting differences to be found when investigating the details. The VTEC keogram shown in the bottom panel has several notable features:

• In the evening of March 10th, after 17 UT, a band of dense plasma is observed moving from north to south. Within this general plasma pattern are embedded three smaller intensifications moving northward. These intensifications coincide with enhanced scintillation/ROTI. The elevated plasma density likely results from both local precipitation and convected plasma structures. The TEC structures reaching the lower edge extend further south, as can be seen, e.g., in Figure 6b of Spogli et al. (2024). The intensifications of density and scintillation that rapidly move northward are consistent with intense particle precipitation from reconnection processes. In the Supplementary material file DMSP_SelectedPlots.zip is a selection of plots showing DMSP particle flux measurements in or close to areas of scintillation activity.

• At the high latitude part of the plot from about 22 UT, streams of plasma are observed flowing from north to south. These are remnants of the high-density plasma transported across the polar cap from the ToI on the dayside. Global TEC maps showing the plasma transport can be found in Themens et al. (2024).

• The plasma density on May 11th is generally much lower than the preceding and following days. This suppression of the VTEC level is referred to as a negative ionospheric storm and is a common feature during geomagnetic storm events (Danilov & Morozova, 1985; Buonsanto, 1999; Wang et al., 2010; Horvath & Lovell, 2015). Jin et al. (2025) studied the density depletion using both data from the Swarm spacecraft and ground-based GNSS receivers. Figure 6a of Spogli et al. (2024) shows that the significant plasma depletion was also observed at lower latitudes in Europe. They ascribe the depletion to neutral composition changes. As the depletion is not a result of nor a source of scintillation, this topic lies beyond the scope of the present study.

• An advantage of the suppressed VTEC level is that it makes it easier to measure other ionization effects during daytime. Thin lines of enhanced VTEC on the 11th coincide with occurrences of disturbances measured by σ_ϕ and ROTI, indicating energetic particle precipitation.

To further investigate the details in the relationship between the currents and scintillations, Figure 14 shows the filtered S4 and ROTI keograms on top of the horizontal current keogram. Some parts of the plot shows rapidly changing conditions in which it is challenging to come to strong conclusions other than noting that the scintillation occurs in connection to the strong currents. Even within the two main electrojet currents, there are several types of behaviour observed:

• Between 00 and 03 UT, the westward (blue) electrojet grows to an extent of 15 degrees of latitude. However, most of the area of currents is free from scintillation.

• In contrast, the westward electrojet between 03 and 07 UT coincides with broad regions of strong phase scintillation activity and scattered weak amplitude scintillation. The strongest amplitude scintillation occurs at ≈ 69° N at 06 UT, with S4 ≈ 0.2. We note that there is very poor data coverage for S4 above 70° N, meaning there could be unobserved amplitude scintillation during the strong currents between 05 and 07 UT. Plasma structures convecting at a high speed are likely responsible for a large part of the phase scintillation. The region in which the currents and scintillation occur shifts in a stair-case-like way, spanning ≈ 60–65° N between 04 and 05 UT, and ≈ 67–74° N between 05 and 06 UT. This is consistent with the location of the convection seen in the first four plots of Figure 7.

• Scintillations are observed in some parts of the eastward (red) electrojet. At 12–13 UT, it is located in the middle of the currents, while at 13:30–14:30 it lies on the poleward side of the currents. At 15 UT, scintillations are suddenly breaking the pattern by extending far out of the electrojet. This feature appears to be more strongly linked to vertical currents. Figure 15 provides a closer look at this. The feature at 15 UT coincides closely with a sudden, intense upward current. This indicates that these scintillations are caused mainly by energetic particle precipitation. Particle data from DMSP satellites confirm the presence of particle precipitation in the area (Plots are available in Supplementary material file DMSP_20240511_1455to1510.zip). Some amplitude scintillation is observed within the current area, with two hotspots approaching an S4 value of 0.2 (Fig. 15b).

3.4 Impact on positioning service

Figure 16 shows the time series of phase scintillation and positioning performance for selected receivers in the vicinity of Tromsø. The scintillation receiver TRO2 is co-located with the monitor receiver MTRM, while the other monitoring receivers are located at distances of 1.2 km (TM01), 11.8 km (TM04), and 11.5 km (TM05) from TRO2. MTRM is a Topcon TPS NET-G5 receiver, while the other monitor receivers are Leica GR50. All monitor receivers connect to the NMA’s network RTK service “CPOS”. For MTRM, TM01, and TM04, the connection is periodically broken, normally once per minute, in order to measure the “time-to-fix”. Thus, they will never have a fixed solution 100% of the time. The receivers of the two manufacturers respond differently to the interruption, leading to a different level of “normal” amount of fix solutions within a time interval. The time interval used for counting the number of fixed solutions is 10 min (600 s). One position is calculated per second, meaning that a continuous perfect fix solution yields a value of 600. The correction stream for the receiver TM05 is not broken, in order to test the difference between occasional and permanent connections by comparing with the results for TM04. For the results shown in the coordinate error panels, only results with fixed solutions have been included. When these are not available, the graph is empty. The number of satellites used for position calculation has also been counted in intervals of 10 min. The blue line in the plot is the mean number in the interval, while the intervals marked with a blue background color cover the 5th–95th percentile.

Figure 16 a, b: Phase scintillation observed at TRO2. The red line shows the mean value, while the colored background is a heatmap of the values. c, d: Error in the vertical coordinate, for MTRM (c) and TM01 (d). e, f: Number of fix solutions (red line) and number of satellites used for position calculation (blue line), for MTRM (e) and TM01 (f). Note that the normal level for the fix solution parameter is different for different receivers (more details in the text). g, h: Error in the vertical coordinate, for TM04 (g) and TM05 (h). i, j: Number of fix solutions (red line) and number of satellites used for position calculation (blue line), for TM04 (i) and TM05 (j).

The top row (Figs. 16a, 16b) contains phase scintillation time series measured in Tromsø, as a reference for the level of disturbance. It is repeated twice in order to line up in time with the plots below it. There is no noticeable increase in the position error graph of MTRM (Fig. 16c), nor does it have a drop in the number of fixed solutions (Fig. 16e). There are some occasional dips in the number of satellites used for positioning coinciding with peaks of scintillation activity, but as the number of satellites never approaches a worrying low number the impact on positioning solution is small. The results for MTRM show the maximum achievable performance and also serve as a check that there are no errors in the network RTK support system (i.e., if there is a problem at the other monitor receivers, it is caused by local or natural sources). The other receivers exhibit severe degradations of performance during the period of scintillation activity. TM01 experiences increased position error and frequent loss of a position solution. TM04 and TM05, which are further away, suffer long periods of no position solution.

4 Discussion

The observations of this study show several different kinds of behaviour during this one event, using indices that are sensitive to scintillation from diffractive effects (S4), a combination of the diffractive and refractive effects (σ_ϕ), refractive effects (ROTI) and plasma density gradients of non-scintillation-inducing scales (ROTI (30 s)). Equivalent ionospheric currents, SuperDARN convection maps, and VTEC data are used to provide further information about the state of the ionosphere and to relate the scintillation observations to physical regions and processes. GNSS positioning results from a monitoring network at ≈70° N shows the impact of the disturbed ionosphere on the end user results for a professional network RTK service.

4.1 Uncertainties/biases due to altitude differences

This study combines data calculated for different altitudes. The current data are calculated for 90 or 100 km altitude, while the GNSS-based data are calculated for an altitude of 350 km (300 km for the Nagoya ROTI data). As the data are defined on geographic grids, there will be differences in their magnetic coordinates. Plasma along magnetic field lines is strongly connected, with high correlation of densities and velocity. Vertical currents, if present, flow along magnetic field lines. At high latitudes, the magnetic field lines are close to vertical, meaning that the differences in magnetic coordinates as a function of altitude are less than they would be at lower latitudes. Calculating the horizontal differences between the magnetic coordinates at 100 and 350 km yields values of less than 1 degree latitude for the majority of the coverage area of the data used. The largest offsets are in the southernmost part of the European sector, reaching values of ≈1.4 degrees. Comparing these offsets to the grid resolution of the current grids, the offsets in the NA sector are much smaller than the grid resolution. In the European sector, the offsets are between 0 and 2 times the resolution. Combined with the property that current values at adjacent grid points tend to be highly correlated, the impact of remapping the currents to be magnetically conjugate to the scintillation data is negligible. This has been examined during the work on this study, but is not shown in this paper.

Themens et al. (2024) reported instances of significant plasma uplift during the storm, observed by the Eglin ionosonde (30.50 °N, 273.50 °E). The red line (“hmF2”) in Figure 1f in their paper shows these measurements, with peaks reaching 600 km altitude at 00–01 UT and 03–04 UT on May 11th. After the second peak, the altitude gradually decreased back to a normal level over the course of two hours. If ionospheric disturbances from this uplifted plasma layer were observed using a low elevation satellite and assumed to be at a 350 km altitude, there would be a considerable (up to a few 100 km) horizontal offset in its location. This may have affected the lowest latitude scintillation observations in Figure 10 at the times of the uplift, which in that case should be located further southward than they are. As the VTEC and ROTI data are based on a larger network of receivers that surround the plasma structures, those data would not be shifted in only one direction but instead would be shifted in a way that results in a blurring of the maps. High-altitude plasma was also later observed by the European Incoherent Scatter Radar (EISCAT), Svalbard Radar (ESR), and Poker Flat Incoherent Scatter Radar (PFISR) radars (See Fig. 5 in Themens et al., 2024), reporting “plasma with peak heights in excess of the ESR maximum altitude of 475 km”. From the radar data in the figure, it is seen that this concerns isolated patches of plasma, presumably split off from the Storm Enhanced Density (SED) plume. Dense plasma is observed from 300 km altitude to the maximum altitude observed (475 km) by the radar. As only a part of the altitude profile is known, it is not possible to know what the most suitable mapping altitude for GNSS data would be, but it would likely be higher than normal (>350 km). If the true altitude is significantly higher, the overall effect on the GNSS scintillation data would be a northward error in location for the lowest latitude data. For data in regions that are in the middle of the networks, the net effect would be a smoothing/blurring as the observations are shifted along the line-of-sight from each receiver. Concerning the scintillation observations in the ToI, which were shown by radar data to contain high altitude plasma patches, the scintillation does not appear as a continuous area but as many small, scattered patches of scintillation. The positioning of these is likely not exact. If someone were to attempt to match those data to individual plasma patches observed by other instruments, a careful remapping for each receiver-satellite link may be necessary. However, they are clearly related to the larger structure that is the ToI.

4.2 Scintillation in the auroral oval

It is generally known that the auroral oval is one of the major regions in which satellite signals are frequently disturbed, and that these disturbances have connections to plasma patches and particle precipitation (Weber et al., 1986; Basu et al., 1998; Kintner et al., 2007). This study adds to the body of knowledge, seeking to expand the documented impacts and the knowledge of the details of the connections.

The major part of ionospheric disturbances at high latitudes is closely connected to strong electrojet currents. A strong current is not in itself the source of the disturbances, as evidenced by the observations of some periods of strong currents for which there is no or little reaction in any of the disturbance indices. In order to disturb the satellite signals, plasma density variations at sufficiently small scales are required. These can be produced directly through energetic particle precipitation or through plasma instability processes.

The connection between scintillation and particle precipitation has been explored through the use of auroral emission observations as an indicator of the precipitation (Jin et al., 2015; Oksavik et al., 2015; Skjæveland et al., 2021; Enengl et al., 2023, 2024) and through direct measurements of the field-aligned currents (Fæhn Follestad et al., 2020). In this study, we have observed some instances of scintillation activity co-located with strong vertical currents (Figs. 12 and 15).

The major part of scintillation activity observed during the event was located within the auroral electrojets. These are areas of fast plasma flows that cause refractive effects to contribute to the σ_ϕ index as the standard phase filters will not remove all of the variation due to the movement of plasma across the line-of-sight. For parts of the observations, both phase and amplitude scintillation are observed simultaneously within the electrojets (Figs. 10 and 15), indicating that there is at least some scintillation from diffractive effects. However, σ_ϕ reacts more strongly than S4 and also over a larger area. This difference tells us that there is a contribution from refractive effects during high flow speeds and at times σ_ϕ is dominated by this contribution. For some other time periods, there is little to no scintillation observed during strong electrojets. At those times, the plasma is likely not structured at small scales and thus the observed variation is small even for high flow speeds. The main physical mechanisms to produce small-scale plasma structures are the Kelvin-Helmholtz Instability (KHI) (Keskinen et al., 1988; Kvammen et al., 2025) and the Gradient Drift Instability (GDI) (Basu et al., 1990). The KHI occurs when there is a velocity shear between regions of different plasma density, with a velocity perpendicular to the magnetic field. These conditions exist in the transition region from the fast flow of the electrojets to the slow flow outside of them. The KHI has also been found to operate in smaller flow channels that occur in the auroral/polar regions (Moen et al., 2013; Spicher et al., 2020; Kotova et al., 2025). The GDI occurs when an external force (e.g., an electric field) acts upon a plasma density gradient that is perpendicular to the magnetic field. Both KHI and GDI are capable of producing small-scale plasma structures (Burston et al., 2010), but they may also work more efficiently in concert by the GDI operating on the density gradients generated by the KHI (Carlson et al., 2007, 2008; Moen et al., 2013). The existence of scintillation-inducing plasma density variations co-located with larger-scale gradients has been verified by in situ measurements, consistent with the proposed mechanisms (Moen et al., 2012; Oksavik et al., 2012; Buschmann et al., 2025). A key finding of Oksavik et al. (2012) is that the growth rate of KHI is too slow to generate the smallest-scale irregularities but that the GDI is very efficient at creating them from “seed” gradients generated by particle precipitation or KHI.

In this study, we observe strong electrojets in which fast plasma flows provide suitable conditions for KHI. Convection observations from SuperDARN provide confirmation of the presence of fast flows, albeit not with complete coverage of direct observations. Plasma density gradients are provided by convection of high-density plasma from the sunlit atmosphere through the cusp across the polar cap and into the auroral oval and by particle precipitation. VTEC data (Figs. 10 and 13) show clear signs of the presence of high-density plasma patches in the ToI (but do not have the resolution to identify individual patches) and show signs of enhanced plasma density due to particle precipitation in the auroral oval region. Through KHI and GDI, the density gradients can be broken down into smaller-scale density structures. Although this study does not include direct measurements that prove that these processes are active, we observe the expected end result of a response in the disturbance indices, with the indices sensitive to larger scales reacting more strongly and over larger areas. In the case of fast flows, the contribution from refractive effects to the observed phase scintillation is increased. The observations show that the nature of the scintillation-inducing plasma structures and the balance of diffractive and refractive contributions vary with time and location. For some time periods, strong electrojets are observed without reactions in the disturbance indices, meaning that the plasma is smooth. The most likely reason is a lack of the initial density gradients that are needed for the plasma instability mechanisms. The VTEC map data are not detailed enough to confirm if this is the case or not.

4.3 Scintillation in the ToI

Scintillation has previously been related to ToIs (Prikryl et al., 2013; van der Meeren et al., 2014; Buschmann et al., 2025), and Themens et al. (2024) did note the presence of scintillation within the ToI of this event, as well as radar observations showing direct evidence of plasma patches within the ToI. The ToI extended from the NA sector across the polar cap. The enhanced plasma density of the ToI is clearly seen in the NA TEC data (Fig. 10), with some remnants of enhanced TEC still visible in the European TEC data (Fig. 13). The ROTI (30 s) shows a strong and scattered response (Figs. 10 and 11) along with reactions in both scintillation indices. The scattered response of the indices is consistent with small-scale plasma structures in connection with a series of patches within the ToI. The formation of the patches results in sharp density gradients at its edges, which may be further structured by the KHI and GDI processes as the patches convect. The patches may also be subject to particle precipitation as they travel through the cusp region, creating further density variations. After leaving the cusp region, patches may continue to develop irregularities as they cross the polar cap (Oksavik et al., 2010). Their plasma density will decay over time, so as they arrive at the nightside, their peak density will be less than it was at the start of their journey. At some point, the decay in plasma density will limit their potential as a source of irregularities. Due to this, their impact is greater shortly after the patch formation in the NA sector than their impact when arriving in the European sector after crossing the polar cap.

4.4 Impact on positioning service

Figure 16 shows positioning performance, using a network RTK service, for 4 receivers located at ≈70° N, 19° E in Norway. This means that the conditions shown in Figure 13 are highly relevant for these receivers.

The receiver TM01 represents a user fairly close to a network receiver. During the period of scintillation, there are frequent positioning errors of several tens of centimeters. There are many dips in the number of fixed solutions, but it rarely drops all the way to zero (Fig. 16f). The two main ways in which the signal degradation decreases the number of fixed solutions are cycle slips and phase noise. A fixed solution is reached when the receiver determines that it has successfully found an integer value for the phase ambiguity. A cycle slip occurs when the receiver is not able to keep a continuous track of the signal phase. When this happens, the phase ambiguity must be re-estimated. Frequent cycle slips will lead to more satellites being in a state of re-estimation, and can even make a satellite functionally unusable if new cycle slips occur before it is able to complete the phase ambiguity estimation. An increased level of noise in the phase observations will increase the time needed to perform an estimation of the phase ambiguity, as the uncertainty of the solution decreases more slowly. If the noise is too large, the receiver may never be able to reach a fixed ambiguity state. Even if a fixed solution is reached, phase noise will enter the positioning calculation and result in increased noise in the position. The result of the combination of these effects is that the positioning service quality (Fig. 16d) is very poor, with greatly increased errors and an unreliable solution that comes and goes and sometimes is gone for several consecutive minutes. A user with a critical need for positioning and no other option could get some residual use out of the service, as long as a reasonable level of accuracy is not needed.

TM04 and TM05 represent users at a moderate distance from the network receiver. As TM04 is periodically reset, it mimics a user who makes a measurement, disables the equipment, and then makes another measurement later. TM05, on the other hand, has a permanently active connection to the service and mimics equipment that is always active. During scintillation activity, both of them suffer a total loss of fixed solutions for intervals up to hours in length. Their position errors are not much greater than those of TM01, as the fixed solution is lost before the position error reaches extremely large values. The observation quality of TM04 and TM05 is the same as for TM01, so the difference in positioning performance is related to their distance from the RTK network. When starting the process to estimate the phase ambiguities, the receiver uses information received from the CPOS service to initialize parts of its state (i.e., the “a priori” state in its Kalman filters). The closer this information is to the actual value, the less time and measurements are needed to reach a stable solution. The accuracy of the information will be greatest close to the network receivers and gradually worse at a distance as the interpolated values diverge from the actual values. During ionospheric disturbances, the plasma density is less smooth than usual, and thus the interpolation errors tend to be larger than during quiet times (Jacobsen et al., 2023). From the onset of disturbances at 17 UT on May 10th to their end at 06 UT on May 12th, TM04 is barely able to reach a stable state for a few short time periods (Figs. 16g, 16i). The positioning service is, in practice, unavailable for TM04 for the entire 37 h duration. TM05 shows the advantage of having a continuous service during challenging conditions and is able to reach a stable state for up to a couple of hours in the intervals where scintillation activity lowers to near quiet levels (Figs. 16h, 16j). During these stable periods, the positioning service would be usable. However, unless the quiet periods can be predicted in advance and a user is very flexible regarding work time, this is not enough to be considered a useful service.

The fact that MTRM performs well and does not suffer any large losses of satellites and that the performance is strongly dependent on the distance from the network receiver indicates that the dominant cause of these problems are plasma structures at scales that are smaller than what the positioning service network server is able to accurately model, but for the most part not small enough to cause significant amplitude scintillations. This is confirmed by the ionospheric observations (Fig. 13) of strong phase variations (ROTI & σ_ϕ) but modest levels of amplitude variations (S4). The effects of such plasma gradients/structures on positioning have previously been explored during quieter conditions (Jacobsen et al., 2023; Sokolova et al., 2023). The same mechanisms are at play during this storm, but with much greater extent, duration, and intensity.

5 Conclusions

This study was enabled by the combined observational capability of four networks of scintillation receivers, providing an overview of scintillation in more than half of the northern auroral and polar regions. The observations of scintillation activity have been related to various physical processes occurring within and near the auroral oval through the analysis of estimated currents and plasma convection from multiple instruments. A summary of the main findings:

1. Phase scintillation was predominantly observed within the auroral electrojets. In the eastward electrojet, it showed a weak preference for the poleward edge of the current region. Phase scintillation within the electrojets is likely to have a large contribution from refractive effects due to high plasma flow speeds.

2. Amplitude scintillation was observed but with a lesser magnitude (S4 values up to 0.2) and extent than the phase scintillation. It showed a tendency to be more associated with vertical currents, but caution is needed in interpretation due to the extra elevation angle filter applied to remove false positives at low elevation, resulting in poor coverage of some areas of interest.

3. At times, strong electrojet currents were not associated with scintillation. If the plasma within the convection stream does not contain significant variations in density, it will not result in scintillation observations.

4. Phase and amplitude scintillation were observed within the Tongue of Ionization. Its distribution within the ToI suggests many smaller scintillation-inducing areas within the broader ToI, consistent with being related to a series of embedded plasma patches.

5. At times, scintillation was observed in connection with intense vertical current, indicating particle precipitation creating scintillation-inducing small scale plasma structures.

6. Positioning services were severely degraded during the event. The observations indicate that the dominant cause of these problems is plasma structures at scales that are smaller than what the positioning service network server is able to accurately model.

A general conclusion is that in the interpretation of auroral region scintillation it is important to include other measurements (e.g., estimated currents) in order to understand the environment of the source region of the scintillation, as there are multiple physical processes/conditions that can result in signal phase variations.

Acknowledgments

The authors thank the reviewers for their valuable suggestions, which led to substantial improvements in the paper. We thank the many different groups operating magnetometer arrays for providing data for this study, including:

• The institutes who maintain the IMAGE Magnetometer Array: Tromsø Geophysical Observatory of UiT the Arctic University of Norway (Norway), Finnish Meteorological Institute (Finland), Institute of Geophysics Polish Academy of Sciences (Poland), GFZ German Research Centre for Geosciences (Germany), Geological Survey of Sweden (Sweden), Swedish Institute of Space Physics (Sweden), Sodankylä Geophysical Observatory of the University of Oulu (Finland), DTU Technical University of Denmark (Denmark), and Science Institute of the University of Iceland (Iceland). The provisioning of data from AAL, GOT, HAS, NRA, VXJ, FKP, SIN, BOR, SCO, and KUL is supported by the ESA contracts number 4000128139/19/D/CT as well as 4000138064/22/D/KS.

• The THEMIS ground magnetometer network (Ground-based Imager and Magnetometer Network for Auroral Studies).

• The authors thank I.R. Mann, D.K. Milling, and the rest of the Canadian Array for Realtime Investigations of Magnetic Activity (CARISMA) team for data. CARISMA is operated by the University of Alberta and funded by the Canadian Space Agency. The Canadian Space Science Data Portal is funded in part by the Canadian Space Agency contract numbers 9F007-071429 and 9F007-070993.

• AUTUMNX magnetometer data was funded through the Canadian Space Agency / Geospace Observatory (GO) Canada program.

• The Canadian Magnetic Observatory Network (CANMON) is maintained and operated by the Geological Survey of Canada.

• The Magnetometer Array for Cusp and Cleft Studies (MACCS) array is supported by US National Science Foundation grant ATM-0827903 at Augsburg University and M. J. Engebretson, D. Murr, and E.S. Steinmetz.

• We would like to thank the Technical University of Denmark (DTU) magnetometer team for providing Greenland magnetometer data.

• We thank the Geophysical Institute, University of Alaska Fairbanks, and the Alaska Satellite Facility for providing Geophysical Institute Magnetometer Array (GIMA) data.

• The MagSTAR magnetometer array is supported by a US National Science Foundation grant NSF-1520864.

• We thank the U.S. Geological Survey (USGS) for providing ground magnetometer data.

The authors acknowledge the use of SuperDARN data. SuperDARN is a collection of radars funded by national scientific funding agencies of Australia, Canada, China, France, Italy, Japan, Norway, South Africa, the United Kingdom, and the United States of America. For the NA ROTI data: Global GNSS-TEC data processing has been supported by JSPS KAKENHI Grant Number 16H06286. GNSS RINEX files for the GNSS-TEC processing are provided from many organizations listed by the webpage http://stdb2.isee.nagoya-u.ac.jp/GPS/GPS-TEC/gnss_provider_list.html. We thank NOAA NCEI for DMSP data. Work by James M. Weygand was supported by NASA grants HSR-80NSSC18K1227 and SWO2R 80NSSC20K1364, NASA contract HPDE-80GSFC17C0018, NSF grant GEO-NERC 2027190, and NSF grant AGS-2013648 via subcontract from Augsburg University.

The editor thanks two anonymous reviewers for their assistance in evaluating this paper.

Funding

This work was supported by the Norwegian Space Agency (NOSA) (Project Number: 74SS2501).

Data availability statement

Scintillation data are available at https://doi.org/10.5281/zenodo.15045918 (Jacobsen et al., 2025). IMAGE data are available at https://space.fmi.fi/image. The code for the SECS method is available as a supplement to Vanhamäki & Juusola (2020). THEMIS GMAG, AUTUMNX, MACCS, GIMA, MagStar data are available from: http://themis.ssl.berkeley.edu/data/themis/thg/ CARISMA data are available from http://data.carisma.ca/ USGS and NRCanada data are available at https://intermagnet.org/ DTU magnetometer data are available at https://geo.phys.uit.no/ SuperDARN data is available through https://www.bas.ac.uk/project/superdarn/ or one of several other alternative sources. Software to process the data is available at https://github.com/SuperDARN/rst and https://pydarn.readthedocs.io/en/main/. Northern Europe VTEC and ROTI (1 s) map data is available at https://swe.ssa.esa.int/web/guest/rtim-federated or by contacting NMA. Global VTEC data are available at http://cedar.openmadrigal.org. Global ROTI (30 s) data are available at https://stdb2.isee.nagoya-u.ac.jp/GPS/GPS-TEC/. Positioning performance data used in this study are available as a Supplementary material file (PositioningData.zip). The file also includes positioning data from a few other receivers, which were not included in the figures. For further data from other events, contact NMA. DMSP particle data are available from https://www.ngdc.noaa.gov/stp/satellite/dmsp/index.html.

Supplementary material

movie_Scint_20240510to20240512.avi: Movie file of phase scintillation (σ_ϕ, in radians) maps for the entire event. Dots show individual scintillation measurements while the background color map is an interpolation of these.

movie_CurrentsAndScint_Europe_20240510_0000to1200.avi: Movie file of currents and scintillation in the European region for 2024-05-10 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_Europe_20240510_1200to2400.avi: Movie file of currents and scintillation in the European region for 2024-05-10 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_Europe_20240511_0000to1200.avi: Movie file of currents and scintillation in the European region for 2024-05-11 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_Europe_20240511_1200to2400.avi: Movie file of currents and scintillation in the European region for 2024-05-11 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_Europe_20240512_0000to1200.avi: Movie file of currents and scintillation in the European region for 2024-05-12 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_Europe_20240512_1200to2400.avi: Movie file of currents and scintillation in the European region for 2024-05-12 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240510_0000to1200.avi: Movie file of currents and scintillation in the North American region for 2024-05-10 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240510_1200to2400.avi: Movie file of currents and scintillation in the North American region for 2024-05-10 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240511_0000to1200.avi: Movie file of currents and scintillation in the North American region for 2024-05-11 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240511_1200to2400.avi: Movie file of currents and scintillation in the North American region for 2024-05-11 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240512_0000to1200.avi: Movie file of currents and scintillation in the North American region for 2024-05-12 00:00 to 12:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

movie_CurrentsAndScint_NA_20240512_1200to2400.avi: Movie file of currents and scintillation in the North American region for 2024-05-12 12:00 to 24:00. The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σ_ϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map.

References

All Figures

	Figure 1 Overview of scintillation receivers and magnetometers used in this study. (a) Locations of scintillation receivers used in this study. FMI in red, NMA in green, SWADO in blue, and CHAIN in purple. (b) The IMAGE magnetometer network in red. Magnetometer stations that were used for the North American sector are shown in blue.
In the text

	Figure 2 Phase scintillation (σϕ, in radians) maps at a selection of times before the onset of the activity and in the early phase of the storm.
In the text

	Figure 3 SuperDARN convection maps, for times matching those in Figure 2.
In the text

	Figure 4 Phase scintillation (σϕ, in radians) maps before, during, and after ToI.
In the text

	Figure 5 SuperDARN convection maps, for times matching those in Figure 4.
In the text

	Figure 6 Phase scintillation (σϕ, in radians) maps at a selection of times during May 11th.
In the text

	Figure 7 SuperDARN convection maps, for times matching those in Figure 6.
In the text

Figure 8 The background color map is the current density in the east-west (red-blue) direction of the Quasi-Dipole (QD) coordinate system. Arrows show the horizontal current density vector. Scintillation is plotted as colored dots, with a color scale capped at a value of 0.5 but dot sizes continuing to scale beyond that. Examples of dots for σϕ values of 0.1, 0.5, and 1.5 are shown to the right of the map. This figure is for the same time as panel (d) in the Figures 6 and 7.

In the text

	Figure 9 Same as Figure 8, but for the North American sector. This figure is for the same time as panel (d) in the Figures 4 and 5.
In the text

Figure 10 a) QD east-west (red-blue) current density keogram at 99.575° W. b) Vertical (red is downward) current keogram at 101.3° W. c) Amplitude scintillation index keogram for the mean value in the longitude interval of 95–105° W. d) Phase scintillation index keogram for the mean value in the longitude interval of 95–105° W. e) ROTI (30 s) keogram for the mean value in the longitude interval of 98–102° W. f) VTEC keogram for the mean value in the longitude interval of 98–102° W.

In the text

	Figure 11 A zoomed-in view of panel (a) of Figure 10, with a) a filtered version of phase scintillation plotted on top, showing only values ≥ 0.3 radians. b) A filtered version of ROTI (30 s) plotted on top, showing only values ≥ 1.5 TECU/min. Small black dots are plotted when there is data coverage of σϕ/ROTI but with values below the thresholds.
In the text

	Figure 12 A zoomed-in view of panel (b) of Figure 10, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of phase scintillation plotted on top, showing only values ≥ 0.3 radians. Small black dots are plotted when there is data coverage of S4/σϕ but with values below the thresholds.
In the text

Figure 13 a) QD east-west (red-blue) current density keogram at 20° E. b) Vertical (red is downward) current density keogram at 20° E. c) Amplitude scintillation index keogram for the mean value in the longitude interval of 15–25° E. d) Phase scintillation index keogram for the mean value in the longitude interval of 15–25° E. e) ROTI (1 s) keogram for the mean value in the longitude interval of 18–22° E. f) VTEC keogram for the mean value in the longitude interval of 18–22° E.

In the text

	Figure 14 A zoomed-in view of panel (a) of Figure 13, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of ROTI (1 s) plotted on top, showing only values ≥ 4 TECU/min. Small black dots are plotted when there is data coverage of S4/ROTI, but with values below the thresholds.
In the text

	Figure 15 A zoomed-in view of panel (b) of Figure 13, with a) a filtered version of S4 plotted on top, showing only values ≥ 0.1. b) A filtered version of ROTI (1 s) plotted on top, showing only values ≥ 4 TECU/min. Small black dots are plotted when there is data coverage of S4/ROTI, but with values below the thresholds.
In the text

Figure 16 a, b: Phase scintillation observed at TRO2. The red line shows the mean value, while the colored background is a heatmap of the values. c, d: Error in the vertical coordinate, for MTRM (c) and TM01 (d). e, f: Number of fix solutions (red line) and number of satellites used for position calculation (blue line), for MTRM (e) and TM01 (f). Note that the normal level for the fix solution parameter is different for different receivers (more details in the text). g, h: Error in the vertical coordinate, for TM04 (g) and TM05 (h). i, j: Number of fix solutions (red line) and number of satellites used for position calculation (blue line), for TM04 (i) and TM05 (j).

In the text