The interval 4–10 September 2017 was one of the most flare-productive periods of the solar cycle 24, producing strong-to-severe space weather episodes. Thus, on 6 September, the solar active region AR 30023 produced an X-9 level flare. The arrival of the associated coronal mass ejection produced severe geomagnetic storming on 7 and 8 September, preceded by two significant solar flare effects (Sfe) that could be seen in the lit hemisphere. In this article, we analyze the impact of these flares on the ionosphere as registered on the vertical incidence ionospheric sounder located at Ebre Observatory. We put the emphasis on the use of this instrument to detect solar flares by means of the absorption observed in its ionograms. We also analyze the impacts of these flares on the Earth's magnetic field and the temporal evolution of the second Sfe event using the technique of spherical elementary current systems, which allows tracing in detail the current system evolution during the Sfe lifetime. As expected, a sharp increase in current intensities occurred at the beginning. The peak of the disturbance was reached after a few minutes, when hard X-rays were dominant, while a slow decay followed the advent of soft X-rays and EUV rays. The modeled current systems appear abnormally displaced in longitude with respect to the subsolar point. The northern vortex shows up about 2 hr in advance of the subsolar point meridian, while the southern vortex is about 3 hr behind it. Both remain static, showing no significant shift over the whole episode. Although the event occurred during the equinox, a clear prevalence of the Northern Hemisphere was observed.
- We analyzed the effect of the solar flare of 6 September 2017 in the ionosphere and in the Earth's magnetic field
The ionosphere is an upper atmospheric layer formed by the ionization of atmospheric gases, mainly as a result of the incident electromagnetic radiation originating in the Sun. Depending on the dominant species and the ionizing radiation, this layer is stratified in different regions: D, E, F1, and F2. Under quiet conditions, the low-altitude electrons (D region) are produced by the Lyman-α line, as well as by EUV and X-ray solar radiation (e.g., Hargreaves, 1992).
Solar flares are powerful bursts of radiation coming out from an active region of the Sun. When they occur, the ionizing radiation is intensified, especially at the shorter-wavelength/higher-energy regimes. Harmful radiation from a flare cannot pass through Earth's atmosphere to physically affect life on the ground; however, an augmented ionization is produced at relatively low altitudes in the sunlit hemisphere. Enhancement of the electron density causes an increase of the high-frequency (HF) radio wave absorption (e.g., Davies, 1990), thus potentially affecting Global Positioning System and communications signals, among other disturbances (e.g., Blanch et al., 2013). Another remarkable consequence is the enhancement of the electric conductivity in the upper atmosphere, so electric currents in the ionosphere become more intense, producing magnetic variations on Earth that are known as solar flare effects or Sfe. Sfe are complex phenomena involving several physical processes at different layers of the atmosphere (Curto et al., 1994b). Sfe variations share many aspects with Sq variations, for example, both originate in the E ionospheric layer or dynamo region; both involve the creation of two vortices: one in the Northern Hemisphere and the other in the Southern Hemisphere; during solstice, the vortex in the summer hemisphere is prevalent over the one in the winter hemisphere, while in equinox time both are more balanced. But there are also net differences: Sfe currents have a deep vertical extension intruding partially into the D ionospheric layer; Sfe vortices are displaced with respect to Sq ones, etc. (Curto et al., 1994a).
Tracing the temporal evolution of the currents producing Sfe was already attempted by some authors (Van Sabben, 1961; Veldkamp & van Sabben, 1960). With a limited number of observatories, they only achieved quantitative descriptions of the current systems responsible for the Sfe disturbances. Later, Curto et al. (1994a), using data from 67 observatories, performed a global study of an Sfe event, seen at Ebre as a reversed Sfe. In this case, in the Northern Hemisphere the Sfe system was about 1 hr local time eastward of the Sq system and formed 4° higher in latitude. However, the way of drawing the path of the currents was rather simple and followed the classical method of Chapman and Bartels (1940).
A more sophisticated attempt was carried out by Gaya-Piqué et al. (2008). They analyzed the prompt Sfe associated with a strong X-ray flare that occurred on 5 December 2006 by using spherical cap harmonic analysis on a network of European magnetic observatories. The method presented some limitations, such as the difficulty to discriminate internal-external sources and the need of a careful selection of the size of the cap to get coherent results. The authors had to perform different tests, such as modifying the adopted temporal basis functions or the maximum spatial and temporal expansion degrees, all of which resulted in a somewhat cumbersome procedure if one tries to use it in a massive treatment of data, which is necessary for an automatic detection of Sfe events. We present here an analysis of the 6 September 2017 Sfe event by use of the spherical elementary current systems (SECS) method. Even though this technique also requires manual intervention in the election of some optimum parameters, such as the density and location of the current sources producing the magnetic observations, or the regularization parameter for the inversion, it presents some advantages compared to the common field continuation techniques (e.g., Fourier, spherical cap, or spherical harmonic expansions): (1) No fixed limitation of the spectral content has to be given for the whole analysis area, as it has to be done for the other techniques by truncation of a series expansion. (2) The locations of the elementary current systems can be chosen freely, such that they are most suitable with respect to the available measurement sites or the type of current system to be analyzed (Amm & Viljanen, 1999).
2 The SECS Method
SECS is an equivalent source method aimed at explaining the observed ground magnetic variations in terms of its current sources in the ionosphere and in the subsurface. Although the technique was developed by Amm and Viljanen (1999), the modeled current system was first constructed by Amm (1997) from the superposition of elementary currents originating from a network of poles. These elementary currents (and hence their superposition) are divergence-free because ground magnetometers are only essentially sensitive to the divergence-free part of the real current [A/m], which also consists of a curl-free part. This results in equivalent (rather than real) currents being effectively modeled. The intensity of the elementary currents is proportional to scale factors Ii which in turn are related to the total current (see, e.g., Vanhamäki et al., 2003; Marsal et al., 2017, for more details).
Note that flows clockwise around a maximum of ψ.
Although care must be taken with the election of some parameters, SECS has proven to be a robust and flexible method that allows a relatively simple computation of the equivalent currents. It has been used by McLay and Beggan (2010) and by Torta et al. (2017) for magnetic field interpolation purposes, by Weygand et al. (2012) to evaluate the consistency of SECS-derived equivalent currents with those from other types of measurements, by Juusola et al. (2015) in connection with geomagnetically induced currents, and by Marsal et al. (2017) to map the polar current systems associated with a geomagnetic sudden commencement, just to cite a few. The technique is used here for the first time to analyze the time evolution of the ionospheric currents during an important Sfe event.
The ground-based ionosonde is the most conventional equipment for measuring the vertical electron density profile. An ionosonde emits radio waves at different frequencies, usually from 1 to 20 MHz, and measures the time taken by the pulse to go up and down once reflected in the ionosphere. Both the altitude at which the pulse is reflected and the electron density can be deduced. It has been shown that ionosondes can be used to monitor the ionospheric response to solar flares (Handzo et al., 2014).
We have observed the effects of the 6 September 2017 solar flare in the lower ionosphere from the ionograms obtained with the Digisonde DPS-4D located at Ebre Observatory (EB040; 40.8°N, 0.5°E), where this event occurred around local noon. Data from this ionospheric station can be obtained from the Digital Ionospheric Data Base of the Center for Atmospheric Research of the University of Massachusetts, Lowell (Reinisch et al., 2004). We have used the Digisonde Ionogram Data Visualization/Editing Tool (SAO-X) to obtain the signal-to-noise ratio (SNR). This software is available at the website of the Center for Atmospheric Research (http://ulcar.uml.edu/), and a brief description is shown by Reinisch et al. (2005, and references therein). The SNR is computed by subtracting the most probable amplitude from the signal amplitude value, both in decibel units.
As regards the variations of the magnetic field, we have applied the SECS inversion technique to the mentioned Sfe event which, according to the International Service on Rapid Magnetic Variations (http://www.obsebre.es/en/rapid), had the maximum during the interval 11:55–11:59 UT. Data from the international real-time magnetic observatory network (http://www.intermagnet.org/data-donnee/data-eng.php) have been downloaded for geomagnetic observatories at both sides of the Atlantic Ocean. As the network is not homogeneously distributed, we have collected data from the Supermag network (http://supermag.jhuapl.edu/mag/?; Gjerloev, 2009, 2012) with the aim to cover the maximum area of influence of the Sfe. A total of 48 stations have been considered for this event.
Figure 1 shows the location of the magnetic observatories considered in this study. Our data set only includes magnetic observatories in the sunlit hemisphere. Stations located at high latitudes were not used because they are strongly influenced by auroral electrojets, and they are often disturbed by magnetospheric effects. Middle- and low-latitude areas are reasonably covered, although there is an asymmetry between the Northern and Southern Hemispheres as regards the distribution of observatories.
The Sun emitted two significant solar flares on the morning of 6 September 2017. The first one, classified as an X2.2 flare, peaked at 9:10 UT, while the second one was classified as X9.3 and peaked at 11:58 UT (Figure 2). The measurements were obtained from two gas-filled ion chambers on board the Geostationary Operational Environmental Satellite (GOES)-13 satellite, each measuring solar X-ray fluxes in different wavelength bands: 0.05–0.4 and 0.1–0.8 nm. Measurements in these bands have been made by National Oceanic and Atmospheric Administration (NOAA) satellites since 1974, and the design has changed little during that time period (Garcia, 1994).
Both flares erupted from an active region labeled AR 30023 in the Ebre Catalog (http://www.obsebre.es/en/sun-pictures), located at S09 W33. This region, combined with its neighbor region AR 30024, produced more than a dozen of midlevel M-class solar flares in 2 weeks. Both groups were mature and had a complex magnetic configuration amounting 36 sunspots. They were classified as e and f types according to the Zurich classification (Figure 3).
The X9.3 flare is the largest one so far in the current solar cycle, although it happened in the descending part of it. This is a phase when such eruptions on the Sun are increasingly rare, but historical records have shown that they can nonetheless be intense. In this case, the associated radiation created disturbances in the geospace which resulted in HF radio blackouts (https://spaceweather.gov).
When a solar X-ray flare occurs, the electron density in the lower ionosphere increases significantly throughout the illuminated hemisphere, potentially producing HF radio wave absorption. We are interested in the absorption suffered by the emitted signal in its journey through ionospheric reflection. This absorption can be observed directly from ionograms and also by analyzing the SNR. In this work we have analyzed the ionospheric response to the X-ray solar flare (X9.3) that occurred on 6 September 2017 from 11:53 to 12:10 UT, reaching its maximum at 12:02 (U.S. Air Force/NOAA Solar and Geophysical Activity Summary; ftp://ftp.swpc.noaa.gov/pub/warehouse/2017/). Figure 4 shows the ionograms before (Figure 4a), during (Figure 4b), and after (Figure 4c) the solar flare. In Figure 4a we can observe a typical daily ionogram where F2, F1, and E layers are clearly observed, along with an Es layer. The ionogram depicted in Figure 4b was obtained while the solar flare was active. We can observe no ionospheric traces due to the radio wave absorption produced by the increase of the electron density at low altitudes. Even if the solar flare ended at 12:10 UT, the effects in the ionosphere were observed during more than 3 hr, that is, from ionograms in the period from 11:55 to 15:10. Figure 4c shows the first ionogram recorded at EB040 station once the ionosphere recovered its regular behavior. As expected, we can observe (Figure 4c) the F2, F1, E, and Es layers again. Although not shown here, a similar effect was observed at the ionospheric stations of Dourbes in Belgium (DB049; 50.1°N, 4.6°E) and Pruhonice in Prague (PQ052; 50.0°N, 14.6°E).
Figure 5 shows SNR plots (when SNR is larger than 15 dB) for a quiet reference day (Figure 5a) and for 6 September 2017 (Figure 5b). The SNR plot for the reference day provides information about the minimum frequency of ionogram echoes, the deviant absorption between the E and F layers, possible radio interferences, and also about the better transmission frequencies per hour. It is possible to observe that, as expected, during the day it is possible to transmit at larger frequencies than during the night and that the minimum frequency is larger from 10 to 12 UT. Figure 5b shows the SNR plot for 6 September 2017 when several X-class solar flares occurred. This day, an X2.2 X-ray solar flare from 08:57 to 09:17 (with maximum activity at 09:10) and a X9.3 X-ray solar flare from 11:53 to 12:10 (with maximum activity at 12:02) happened (USAF/NOAA Solar and Geophysical Activity Summary; ftp://ftp.swpc.noaa.gov/pub/warehouse/2017/). From Figure 5b we can clearly observe the HF absorption produced by both X-class solar flares: the first interval of absorption of 1 hr duration and the second one of 3-hr duration as it was observed from the ionograms (Figure 4). The minimum frequency has also been affected, being larger when comparing with quiet days.
Regarding the impact on the Earth's magnetic field, the arrival of the coronal mass ejection associated with this flare produced severe geomagnetic storming on 7 and 8 September, while the subsequent high-speed streams produced further minor-to-moderate geomagnetic storm activity. In this study, however, we will focus on the effect of the flare radiation producing the Sfes.
On 6 September 2017, most European observatories detected two Sfe events due to their favorable location with respect to the Sun (e.g., Figure 6). In this study, we will concentrate on the Sfe that occurred near noon (http://www.obsebre.es/en/rapid) because it had a more distinct signal from the background noise. It was preceded by a magnetically quiet morning, with 3-hourly Planetary K-indices of 1 1 1 3 3 2 0 3 (https://www.gfz-potsdam.de/en/kp-index/). The flare producing this Sfe was optically classified as 2b. As regards radio flux, it had a peak of 3,200 sfu (1 sfu = 10−22 W·m−2·Hz−1) at the wavelength F122 cm (245 MHz) and a peak of 14,000 sfu at the wavelength F10.7 cm (800 MHz). Finally, a radio burst Sweep II/IV type occurred. According to NOAA, the flare started at 11:53 UT, had a maximum at 12:02, and ended at 12:10 UT (ftp://ftp.swpc.noaa.gov/pub/warehouse/2017/2017_plots/xray/).
Figure 7 shows an expansion of Figure 2 for the time of the big flare. A sudden increase in the band of short wavelengths (red line) dominates the first part of the event (11:56–12:10), while the second part (12:10–12:24) is shared between short and long (blue line) wavelengths with a slight dominance of the latter.
The amplitude of the Sfe was found at each particular station after removing the Sq field. This was obtained by linearly interpolating between the initial and the final Sfe time. With this subtraction, the core and crustal fields are removed, too. Sfes are assumed to be exclusively due to the ionospheric and induced currents associated with the solar flare; as a consequence, observatories in the nightside and in the auroral zone are assumed to not experience any Sfe, and thus we take a zero field contribution for them. They are sampled at 1-min intervals from the beginning of the Sfe event (11:55 UT) until its end time (12:24 UT), and they are used as input for the SECS computation.
- The X component (blue line) is positive (northward) in most observatories except for those in the north of the Northern Hemisphere vortex (e.g., STJ, HAD, and FRD). It has a smaller amplitude compared to the Y component (red line), which is clearly positive (eastward) in the American sector and negative (westward) in the European-African sector. These facts are coherent with the position of the center of the northern vortex in the middle of the Atlantic Ocean. The Z component (green) is also weaker than the Y component, and its uneven distribution makes it difficult to analyze its behavior because of its strong dependence on the Earth's conductivity.
- Most of the Sfe shapes are crochet-like as defined in the classical bibliography (Chapman & Bartels, 1940; Parkinson, 1983), with a short rise time and a long decay time.
- Observatories close to the area where the Sfe focus is expected to be (EBR and HAD) present Sfe amplitudes relatively small compared to those from other observatories far away from the focus (FRD, STJ, and TAM). This is because under the focus, currents with opposite senses cancel their magnetic effects.
Around noon, a peak of short duration with a sense contrary to the Sfe variation was detected in most observatories. This could have been produced by a small intrusion of a magnetospheric disturbance.
Using the mentioned data set that covers the sunlit hemisphere, we performed a SECS analysis. Following the precepts of Amm and Viljanen (1999), the grid spacing for the poles of the elementary current systems has been set to about one third of the average spacing of the input magnetic data in the area where the curl of the equivalent currents is expected to be concentrated, that is, in the (northern and southern) middle and low latitudes of the sunlit hemisphere. Since the distribution of data is nonuniform (see Figure 1), this has given rise to an irregular grid of poles. On top of that, we have added a background grid of sparsely distributed poles covering the globe. After some tests, the ϵ parameter has been chosen to be 0.04, that is, those singular values below 0.04 times the largest one have been neglected. This provides a reasonable trade-off between smoothness of the equivalent currents and data fitting. It is worthy to note here that the optimum ϵ parameter depends on the density and distribution of poles being used, as well as on the scale of the particular problem being treated.
Figure 9 presents a sequence of snapshots covering the whole duration of the event. On the left column, we show the evolution of the equivalent currents flowing at ionospheric E layer heights (110 km) when hard X-rays dominated, while the right column corresponds to times when soft X-rays and EUV rays dominated. In the second part, the current system decayed abruptly at the beginning, but then a slow decay followed, although with very weak intensities. During the whole duration of the event no significant drifting of the focus happened. Movie S1 (in the supporting information) shows the time evolution of the SECS-derived ionospheric current system. We emphasize here that we have placed SECS poles at ionospheric E layer heights (110 km), as well as at the subsurface (100 km depth) to account for induction, thus allowing for an external/internal separation of sources. The horizontal location of the latter poles is the same as those in the ionosphere. However, as we are interested in the primary sources of the magnetic field, we only display the modeled ionospheric equivalent currents.
Perhaps the most outstanding feature revealed by Figure 9 is that the Northern Hemisphere vortex clearly prevails over the southern one. This is so even if the event occurred during equinox, when a balance in the intensities of both hemispheres is expected (note that the latitude of the subsolar point is just about 6°N). The north-south asymmetry can result not only from the location of the subsolar point but also from tidal winds (e.g., Yamazaki et al., 2012). So the north-south asymmetry of the Sfe current system might be simply due to the north-south asymmetry of the background Sq current system. The northern curl presented an elliptic shape rather than a circular one, with its major axis in the east-west direction. At the moment of maximum, a total current of 230 kA was computed. Such a high intensity made the Sfe system comparable to that of the usual Sq (Yamazaki et al., 2011).
In the time of maximum intensities, the northern focus was located around 30°N, 10 LT and the southern focus around 30°S, 15 LT. These values contrast with those obtained by Curto et al. (1994a) using a long series of events, indicating that this event is a bit peculiar. In general, Sfe current systems are more synchronous with the subsolar point than the Sq system; however, in this case, both Sfe foci are far away from the subsolar point meridian (12 LT).
Due to its small amplitude, the intrusion of the magnetospheric disturbance observed at noon did not distort much the shape of the currents but perhaps compressed a bit the polar side of the northern loop.
After this work, it has been observed that the effects on the ionosphere last longer than the effects on the Earth's magnetic field. HF absorption lasted for more than 3 hr, while the Sfe—as observed in the magnetograms—only lasted for some tens of minutes. The radiation coming from the solar flare had a category of X for only a few minutes. However, the radiation after the maximum peak remained in category M for 3 hr. Thus, we can conclude that, for clear traces to be observed, Sfe requires a more intense solar flare (such as that produced by X-rays in category X), while HF radio absorption is more sensitive and can occur with fainter solar flares (category M).
From these results, we can conclude that ionospheric sounders can be very useful to detect and warn about possible HF communication blackouts in the vicinity of the sounder during solar flares. Future work can be done in this sense taking advantage of the existing ionospheric sounder networks such as Global Ionospheric Radio Observatory (GIRO), which provides data in near-real time. Thus, the observations presented in this paper can be automatized in near-real time to enable the development of a worldwide warning system for Sfe.
On the other hand, for the first time, the SECS method has been used to analyze the ionospheric currents associated with an Sfe. We have applied this technique to follow in detail the time evolution of the Sfe system on the occasion of the major solar flare event that occurred in solar cycle 24. Unlike other cases, the analyzed current system was very static, the shape of the current vortices remained stable, and their position did not shift substantially. We attribute this fact to the slow decay of the hard X-ray component, which spans even in the second part of the event, so the main characteristics of the first part remained basically unchanged. The Sfe current system was found away from the subsolar point meridian, with a faint southern focus noticeably shifted toward east. The northern vortex was clearly dominant; it was slightly shifted toward west, and its area presents a certain eccentricity in the zonal direction.
The SECS method is postulated as a useful tool for Sfe analysis, and it could be used to obtain complementary data for Sfe detection (Curto et al., 2017).
This work has been funded by the Spanish Government under the project CTM2014-52182-C3-1-P. E. B. contributed with research partly supported by Universitat Ramon Llull projects 2017-URL-IR2Q-028 funded by Generalitat de Catalunya. This paper uses data from the Roquetes Ionosonde which is owned by the Observatori de l'Ebre, and it is available through the Digital Ionogram Database network (DIDBase). The responsible Operations Manager is David Altadill. Also, the research results presented in this paper rely on the data collected at magnetic observatories worldwide, and we thank the national institutions that support them. We recognize the role of the INTERMAGNET program in promoting high standards of magnetic observatory practice. A list of the institutions participating in INTERMAGNET can be found at http://www.intermagnet.org/. In this study, these data were complemented by data from Supermag. Supermag is a worldwide collaboration of organizations and national agencies that currently operate more than 300 ground based magnetometers. SuperMAG provides easy access to validated ground magnetic field perturbations in the same coordinate system, identical time resolution, and with a common baseline removal approach. Finally, X-ray data were provided by the U.S. Department of Commerce, NOAA, Space Weather Prediction Center, and K indices were compiled by GFZ Potsdam (Germany) and guarded in the ISGI repository (http://isgi.unistra.fr/geomagnetic_indices.php). We thank all them.
|swe20768-sup-0001-2018SW001927-ms01.avivideo/avi, 10.7 MB||Movie S1|
|swe20768-sup-0002-2018SW001927-SI.docxWord 2007 document , 44.2 KB||Supporting Information S1|
Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.
- 1997). Ionospheric elementary current systems in spherical coordinates and their applications. Journal of Geomagnetism and Geoelectricity, 49(7), 947–955. https://doi.org/10.5636/jgg.49.947
- 1999). Ionospheric disturbance magnetic field continuation from the ground to the ionosphere using spherical elementary current systems. Earth, Planets and Space, 51(6), 431–440. https://doi.org/10.1186/BF03352247
- 2013). Space weather effects on Earth's environment associated to the 24–25 October 2011 geomagnetic storm. Space Weather, 11, 153–168. https://doi.org/10.1002/swe.20035
- 1940). Geomagnetism (p. 232). Oxford: Clarendon.
- 1994a). Solar flare effects at Ebre: Regular and reversed solar flare effects, statistical analysis (1953 to 1985), a global case study and a model of elliptical ionospheric currents. Journal of Geophysical Research, 99, 3945–3954. https://doi.org/10.1029/93JA02270
- 1994b). Solar flare effects at Ebre: Unidimentional physical, integrated model. Journal of Geophysical Research, 99, 23,289–23,296. https://doi.org/10.1029/94JA02070
- 2017). Automatic detection of Sfe: A proposal. Annales Geophysicae, 35, 799–804. https://doi.org/10.5194/angeo-35-799-201
- 1990). Ionospheric radio (p. 580). London: Peter Peregrinus.
- 1994). Temperature and emission measure from goes soft X-ray measurements. Solar Physics, 154(2), 275–308. https://doi.org/10.1007/BF00681100
- 2008). Equivalent ionospheric currents for the 5 December 2006 solar flare effect determined from spherical cap harmonic analysis. Journal of Geophysical Research, 113, A07304. https://doi.org/10.1029/2007JA012934
- 2009). A global ground-based magnetometer initiative. Eos, 90(27), 230. https://doi.org/10.1029/2009EO270002
- 2012). The SuperMAG data processing technique. Journal of Geophysical Research, 117, A09213. https://doi.org/10.1029/2012JA017683
- 2014). Ionospheric electron density response to solar flares as viewed by digisondes. Space Weather, 12, 205–216. https://doi.org/10.1002/2013SW001020
- 1992). The solar-terrestrial environment: An introduction to geospace—The science of the terrestrial upper atmosphere, ionosphere, and magnetosphere, Cambridge atmospheric and space science series. Cambridge: Cambridge University Press. https://doi.org/10.1017/CBO9780511628924
- 2015). High-latitude ionospheric equivalent currents during strong space storms: Regional perspective. Space Weather, 13, 49–60. https://doi.org/10.1002/2014SW001139
- 2017). Use of spherical elementary currents to map the polar current systems associated with the geomagnetic sudden commencements on 2013 and 2015 St. Patrick's Day storms. Journal of Geophysical Research: Space Physics, 122, 194–211. https://doi.org/10.1002/2016JA023166
- 2010). Interpolation of externally-caused magnetic fields over large sparse arrays using spherical elementary current systems. Annales de Geophysique, 28(9), 1795–1805. https://doi.org/10.5194/angeo-28-1795-2010
- 1983). Introduction to geomagnetism. Amsterdam: Elsevier. https://doi.org/10.1016/0090-3019(83)90027-7
- 1992). Numerical recipes ( 2nd ed.). Cambridge: Cambridge University Press. https://doi.org/10.1016/0140-6736(92)92491-W
- 2004). Automated collection and dissemination of ionospheric data from the digisonde network. Advances in Radio Science, 2, 241–247. https://doi.org/10.5194/ars-2-241-2004
- 2005). Recent advances in real-time analysis of ionograms and ionospheric drift measurements with digisondes. Journal of Atmospheric and Solar - Terrestrial Physics, 67(12), 1054–1062. https://doi.org/10.1016/j.jastp.2005.01.009
- 1961). Ionospheric current systems of ten IGY solar flare effects. Journal of Atmospheric and Terrestrial Physics, 22(1), 32–42. https://doi.org/10.1016/0021-9169(61)90175-1
- 2017). Improving the modeling of geomagnetically induced currents in Spain. Space Weather, 15, 691–703. https://doi.org/10.1002/2017SW001628
- 2003). One-dimensional upward continuation of the ground magnetic field disturbance using spherical elementary current systems. Earth, Planets and Space, 55(10), 613–625. https://doi.org/10.1186/BF03352468
- 1960). On the current system of solar flare effects. Journal of Atmospheric and Terrestrial Physics, 18(2-3), 192–202. https://doi.org/10.1016/0021-9169(60)90092-1
- 2012). Comparison between SuperDARN flow vectors and equivalent ionospheric currents from ground magnetometer arrays. Journal of Geophysical Research, 117, A05325. https://doi.org/10.1029/2011JA017407
- 2011). Application and validation of the spherical elementary currents systems technique for deriving ionospheric equivalent currents with the North American and Greenland ground magnetometer arrays. Journal of Geophysical Research, 116, A03305. https://doi.org/10.1029/2010JA016177
- 2011). Intensity variations of the equivalent Sq current system along the 210° magnetic meridian. Journal of Geophysical Research, 116, A10308. https://doi.org/10.1029/2011JA016632
- 2012). Sq current system during stratospheric sudden warming events in 2006 and 2009. Journal of Geophysical Research, 117, A12313. https://doi.org/10.1029/2012JA018116