SIMuRG: System for Ionosphere Monitoring and Research from GNSS

Currently, more than 6000 operating GNSS receivers deliver observations to multiple servers. Ionospheric data are derived from these measurements providing outstanding space coverage and time resolution. There are about 200 million independent measurements daily. Researchers need sophisticated software tools to deal with such a large amount of data. We present recent advances and products from the System for Ionosphere Monitoring and Research from GNSS (SIMuRG). Currently, SIMuRG provides the total electron content (TEC) variations filtered within 2–10 min, 10–20 min, and 20–60 min, the Rate of the TEC Index, the Along Arc TEC Rate index, and the vertical TEC. SIMuRG is an online service at http://simurg.iszf.irk.ru. The system can be used free of charge and allows calculating both maps and series for arbitrary time intervals and geographic regions. All the data products are available in the form of data or figures. We discuss the system and its geophysics applications.

Global navigation satellite systems (GNSSs) have become an international standard for studying the ionosphere. For example, GNSS data allow detailed studies of traveling ionospheric disturbances (TID). Afraimovich et al. (2000) suggested the interferometry technique based on GPS data to detect and study TID. Tsugawa et al. (2006) investigated large-scale TIDs (LSTIDs) in conjugate points to reveal the absence of magnetic connection along the magnetic field lines. Perevalova et al. (2008) found the ‘‘swirling’’ effect of LSTID propagation in the direction opposite to the earth’s rotation. Cherniak and Zakharenkova (2018) obtained a detailed specification of the ionospheric small- to large-scale disturbances associated with two major sources of the LSTIDs generation. Yang et al. (2017) suggested atomic decomposition for multi-TID detection. Hernández-Pajares et al. (2017) suggested ways to define the effects of TIDs on positioning accuracy and how to mitigate these effects in case of precise real-time kinematic positioning.

GNSS data allow detailed studies of the ionospheric response to solar flares. Afraimovich (2000) suggested a technique to estimate ionospheric response to a solar flare. Based on GNSS data, Le et al. (2013) showed that ionospheric responses to the flares are stronger at equinoxes than at solstices. This fact can be explained by the seasonal variation of the neutral density. Qian et al. (2011) showed that the ionospheric and thermospheric effects of a solar flare depend on the rise and fall rates of flux during the flare. Liu et al. (2011), based on GNSS data, provided an excellent review of solar flare effects on the ionosphere. GNSS also allows studying small effects in the ionosphere linked to the sun and to earth rotation. Afraimovich et al. (2009) confirmed the mechanism for generating the ionospheric disturbances by the solar terminator and showed the magnetohydrodynamic nature of their energy transfer. GPS data allow revealing not only ionospheric electron density decrease during a solar eclipse (Afraimovich et al. 1998) but also the generation of TID during solar eclipse shadow passage (Coster et al. 2017).

Many GNSS-based studies were devoted in connection to ionospheric response to earthquakes and tsunami, rocket launches, and explosions. The pioneering work by Calais and Minster (1995) showed the possibility of finding earthquake signatures in the ionosphere using satellite signals. Astafyeva et al. (2013) documented various ionospheric responses to different earthquakes. Jin et al. (2015) introduced GNSS ionospheric seismology and found significant coseismic and postseismic ionospheric anomalies from continuous GNSS observations. Pulinets and Davidenko (2014) used GNSS data for finding earthquake precursors. Savastano et al. (2017) developed a real-time system for monitoring tsunami based on single-frequency data. Based on GNSS, Calais and Minster (1996) found two sub-events perturbations after the rocket launch, which were two shock waves. The first wave propagated straightforward, while the other propagated along the horizontal atmospheric interfaces at 120 km. Lin et al. (2017) studied the most powerful SpaceX Falcon 9 rocket launch and found concentric traveling ionospheric disturbances. Perevalova et al. (2015) showed that a meteoroid explosion resulted in TEC disturbances generation that is also typical for the acoustic and gravity waves generated by earthquakes, explosions, and rocket launches.

Special attention in GNSS studies is paid to the coupling of atmosphere, ionosphere, and magnetosphere. Polyakova and Perevalova (2011) studied the region of about 2000–3000 km with enhanced TEC variations caused by tropical cyclones. Data of high spatial and temporal resolution allow finding concentric TID generated in the tropical cyclone area (Chou et al. 2017). Belakhovsky et al. (2016) showed that magnetospheric pulsations can result in variations in TEC. Demyanov et al. (2012) found a dependence of GPS losses-of-lock occurrence on magnetic field line orientation during super bubble event.

GNSS allowed detailed studies of geomagnetic storms effects on the dynamics of the ionosphere (Mendillo 2006), while ionosphere mapping (Hernández-Pajares et al. 2009) started a new era of climatological studies (Suresh and Dashora 2016). Ephishov et al. (2000) used such data to validate ionospheric models. Afraimovich et al. (2008) suggested a global electron content approach based on global ionosphere maps (GIMs) to study the dynamics of the ionosphere. Gulyaeva and Stanislawska (2008) proposed a new index of global ionospheric disturbance based on GIM, while Nesterov et al. (2017) suggested a disturbance index based on GNSS radio tomography. The GIM allowed investigating the planetary ionospheric dynamics during magnetic storms (Blagoveshchensky et al. 2018), including great statistical data for such events (Liu et al. 2017), as well as revealing the unusual local TEC enhancements even at mid-latitudes (Edemskiy et al. 2018).

The scientific community is developing software and databases for GNSS-based research using dense networks. There are several free-to-use software toolkits available for single-station TEC calculation. For example, IONOLAB (http://www.ionolab.org/) developed software for TEC calculation (Sezen et al. 2013) based on GPS data in RINEX 2 and 3 formats (Gurtner and Estey 2007). Seemala (2017) suggested software (http://seemala.blogspot.com) to estimate the vertical TEC using single-station GPS data (RINEX 2). Yasyukevich et al. (2015) provided software for GPS, GLONASS, Galileo, and BeiDou-based vertical TEC calculations (http://www.gnss-lab.org/).

There is a well-known DRAWING-TEC project (Tsugawa et al. 2018) that is a database (http://seg-web.nict.go.jp/GPS/DRAWING-TEC/) with global and regional (Europe, Japan, and North America) TEC maps. The maps are distributed in graphical format. The main purpose of the project is to study the medium-scale irregularities and their impact on GNSS. The OpenMadrigal project develops and supports an online database for geospace data (openmadrigal.org). It includes worldwide maps of absolute TEC in the form of data and figures in 5-min time resolution. The maps are calculated by the Massachusetts Institute of Technology Haystack Observatory (Rideout and Coster 2006). Madrigal TEC is incorporated in the GAMBIT software (http://giro.uml.edu/GAMBIT/), which is the browser for the real-time IRI model (http://giro.uml.edu/IRTAM/). The presented tools have both advantages (precalculated datasets) and disadvantages (predefined data products and low time resolution). So, we would like to have a system which includes all these advantages but not the disadvantages.

We assume that for common tasks in geophysical research, a system should be able to:

1. 1.

   calculate ionospheric (TEC-based) maps and series for arbitrary time intervals and geographic regions.

2. 2.

   treat measurements from various GNSS systems, including GPS, GLONASS, Galileo, BeiDou/Compass, SBAS, and multiple data servers.

3. 3.

   deal with data from private networks when the network owner cannot share it.

4. 4.

   provide data and visualization with at least 30-s resolution just out of the box.

5. 5.

   provide common ionospheric indexes.

Hence, our goal was to create a system, rather than a pure database, where a user can carry out data processing for specific tasks. The first item in the above list demands the system to be query-based, which also makes it easier to introduce new calculations. We have developed the SIMuRG—System for Ionosphere Monitoring and Research from GNSS (https://simurg.iszf.irk.ru/). The first step was completed in 2018 (Yasyukevich et al. 2018). However, we encountered some problems and shortcomings. We used only the 2–20-min and 20–60-min filtering. The first range includes both acoustic gravity waves (AGW) and internal gravity waves (IGW). That does not allow separating AGW and IGW effects in the ionosphere. The “slower” variations of 2–20 min prevent studying of 2–5 min variations in the power law spectrum of ionospheric irregularities. Also, the filtering procedure that we used leaves artifacts in TEC variations, thus preventing accurate analysis. The characteristics of the regular and irregular states of the ionosphere were also absent in the original system.

Here, we present an improved and debugged system. We introduce new data products that are widely used for geophysical analysis. We have implemented the Rate of the TEC Index (ROTI) and the Along Arc TEC Rate (AATR) index calculation and developed a new technique for vertical TEC estimation. We also introduced some technical features like API (application programming interface), a user interface for viewing of RINEX files data banks and supplementary information. Since observations cover thousands of kilometers for a single receiver, supplementary information like sounding geometry can help to locate an event recorded by a receiver. Below we provide a detailed description of current products and show examples of their application.

Spatial coverage of a geophysical instrument is one of the most important issues for research and applications. The global navigation satellite systems allow us to solve a wide variety of geophysical issues, and the more data we have, the more efficient and clearer solutions can be obtained. Installation of GNSS monitoring stations all around the world started in the early 1990s. Currently, more than 6000 stations provide data for open access. Figure 1 presents the locations of global GNSS network stations. Some stations are part of several networks, and their data are available on multiple servers. The bottom panel shows how many servers contain data for a given station. Shown on the top panel are stations that are included only in a single network, and the network provides quite unique data.

Worldwide GNSS receiver networks. The top panel shows stations whose data are located only on a single server. The bottom panel shows stations whose data populate multiple servers. The color code marks the number of servers that store data for a particular station

Full size image

There are several global networks managing data preparation, analysis, and provision. The most known network is managed by the International GNSS Service (IGS) (http://www.igs.org). The network is based on a voluntary contribution of over 200 self-funded agencies, universities, and research institutions in more than 100 countries. As of January 1, 2017, the IGS FTP server provides data from 2173 stations.

Another major GNSS network is run under UNAVCO, which is a nonprofit university-governed consortium that facilitates geoscience research and education using Geodesy (http://www.unavco.org). UNAVCO aims to advance and support geodesy. As of January 1, 2017, the UNAVCO FTP server provides data from 2395 stations. More than 80% of UNAVCO GNSS stations also provide data to the IGS as well.

The National Geodetic Survey (NGS), an office of NOAA’s National Ocean Service, manages a network of Continuously Operating Reference Stations (CORS) that provides GNSS data (https://www.ngs.noaa.gov/CORS/). As of January 2017, the CORS network counts 1685 stations, hosted by over 200 different organizations.

The densest regional GNSS network is managed by the Geospatial Information Authority of Japan (GSI, former GEONET network). It comprises over 1300 stations with an average spatial resolution of about 20 km. GSI is used for crustal deformation monitoring and GNSS surveys in Japan (http://terras.gsi.go.jp/). Data are available over FTP after registration. At present, the GSI data are not handled in SIMuRG.

Système d’Observation du Niveau des Eaux Littorales (SONEL) is a part of the Global Sea Level Observing System (GLOSS) developed under the auspices of the IOC/UNESCO. SONEL includes a GNSS network aimed at providing continuous measurements of sea and land levels at the coast for studying long-term sea level trends. As of 2017 January 1, the SONEL server provides data from 777 stations (ftp://sonel.org/).

GNSS network under the Space Geodesy Group of Korea Astronomy and Space Science Institute (KASI) has been operating the first International GNSS Service (IGS) Global Data Center in Asia and Oceania region since 2006 (www.spacegeodesy.re.kr). The group aims at global geodesy. The primary task is to find the small variations in the earth’s shape and its rotation velocity. The center’s FTP server provides the data of IGS stations and 9 stations of KASI GNSS Network (ftp://nfs.kasi.re.kr/).

Geoscience Australia incorporates approximately 100 CORS of several GNSS networks, such as the Australian Regional GNSS Network (ARGN), South Pacific Regional GNSS Network (SPRGN), and AuScope Network. The stations are in the Australian and South Pacific regions. In total, the network provides data from 238 stations since 1993. All the data are available free of charge over FTP (ftp://ga.gov.au/).

EUREF Permanent GNSS Network is a voluntary federation of over 100 self-funding agencies, universities, and research institutions. The network is operated under the aegis of the International Association of Geodesy. The data from 300 stations in Europe are available at ftp://igs.bkg.bund.de/EUREF/obs/.

New Zealand network GeoNet is operated by the Earthquake Commission (EQC), GNS Science, and Land Information New Zealand (LINZ). The GeoNet aims at monitoring geological hazards in New Zealand (https://www.geonet.org.nz). Among other geophysical instruments, it hosts 191 GNSS stations located in the Pacific Ocean and provides data over FTP. The first station of the network was installed at the University of Otago in 1995.

Brazilian Network for Continuous Monitoring of the GNSS (RBMC, former LPIM network) is a regional network established in 1996 and includes more than 130 stations (https://www.ibge.gov.br/en). Data are available over FTP (ftp://geoftp.ibge.gov.br/).

The first stations of the Northern California earthquake data center (NCEDC) network have provided data since 1991 (http://www.ncedc.org/ftp/pub/gps/rinex/). The network monitors crustal deformation across the Pacific-North America plate boundary and in the San Francisco Bay Area for emergency earthquake alert. The primary goal is earthquake hazard studies. The network includes stations from Bay Area Regional Deformation (BARD) and U.S. Geological Survey (USGS) Earthquake Science Center networks. For now, NCEDC provides data of 65 stations.

The South Africa GNSS base station network (TrigNet) consists of 54 GNSS stations that are located in South Africa, developing a grid with station separation from 80 to 300 km. The network provides data with a time resolution up to 1 Hz (ftp://trignet.co.za).

National Observatory of Athens (NOANET) network was established for the monitoring of crustal deformation and its relation to seismicity in Greece and has provided continuous GNSS measurements since 2006. Currently, it consists of 51 stations, of which 18 are permanent (http://www.gein.noa.gr/services/GPS/noa_gps.html).

The Canadian High Arctic Ionospheric Network (CHAIN) (Jayachandran et al. 2009) is a set of ground-based radio facilities in the Canadian High Arctic, within the polar cap mostly, aimed at understanding of polar cap processes (http://chain.physics.unb.ca/chain/). The GNSS network of CHAIN consists of 25 high sample rates (up to 50 Hz) GNSS Ionospheric Scintillation and TEC monitors. Data are available over FTP after registration.

There are several regional networks in Russia. The network EFT-CORS (https://eft-cors.ru/) provides GNSS data via the interactive form and allows registered users to get data from more than 400 stations. The largest Russian network HIVE (https://hive.geosystems.aero/) is managed by Industrial Geodetic Systems Company (located in Omsk, Russia) and includes more than 500 stations, mostly located in the European part of Russia. After registration, the system provides the data in RINEX or RTK formats with a time resolution up to 1 Hz. Networks HxGN SmartNet (http://smartnet-ru.com/) and RTKNet (http://rtknet.ru/) include more than 300 and about 250 stations, respectively, and provide data after a subscription procedure. We are negotiating on the data exchange. Some of the GNSS stations located in Russia also provide data to global networks.

In this section, we describe different types of calculated products and their background. SIMuRG calculates 5 types of GNSS TEC-based data:

1. 1.

   TEC variations (phase measurements) of 2–10 min

2. 2.

   TEC variations (phase measurements) of 10–20 min

3. 3.

   TEC variations (phase measurements) of 20–60 min

4. 4.

   Rate of TEC Index (ROTI) and ROTI-like indexes

5. 5.

   Adjusted TEC

ROTI is a well-known and widely used type of data. Adjusted TEC is an estimation for the vertical TEC having a high temporal and spatial resolutions. The variations types are chosen to match major physical processes: 2–10 min variations correspond to the acoustic gravity wave band, and the 10–20 and 20–60 min variations correspond to medium-scale and large-scale TIDs. All the calculations are performed for a continuous measurement series. Data series with a gap bigger than a typical time step are treated independently. In addition to the data product listed above, SIMuRG also stores metadata that might be helpful for research beyond standard calculations. They are satellite azimuth and elevation, phase and pseudorange TEC, UT, and satellite and station name.

TEC variations

All the data are calculated from dual-frequency phase measurements. The dual-frequency phase combination allows calculating so-called slant TEC I_S along the receiver–satellite line-of-sight. We use a 10° elevation cut-off for all the data, which means low elevation data are not stored in the database. We correct the initial raw phase TEC series to eliminate losses of phase lock, which appears as jumps in the TEC series. TEC variations are calculated based on the centered moving average (CMA) with a given window. The CMA filter is a finite impulse response filter with a smooth transfer function. So the filter passes frequencies only in the indicated range and drops frequencies beyond the range. Before CMA filtering, we detrend TEC data using spline smoothing (Dierckx 1975). The cubic smoothing spline estimate f_S of the function f (over the class of twice differentiable functions) is defined to be the minimizer of:

where Y_i = f(x) + ε_i is the ith set of n observations in [a; b], ε_i are independent zero mean random variables, and λ defines the trade-off between fidelity to the data and smoothing. A positive smoothing factor is used to choose the number of nodes. The number of nodes will be increased until the smoothing condition is satisfied:

where w_i is the weights for spline fitting; they all are supposed to be equal in our case. We use λ = length of Y, which should be good if w_i is an estimate of the standard deviation of Y_i.

Variations obtained after detrending and CMA filtering depend on elevation, which is an effect of geometry rather than physical phenomena. To reduce geometrical dependence of variations amplitude, we convert so-called slant TEC variations dI_S to equivalent vertical variations dI_V as

with the F_K denoting the modified single layer mapping function (Schaer 1999):

where θ is an elevation, h_max is the thin-layer ionosphere altitude (506.7 km), R_E is the earth radius (6371 km), and α is a correction factor (0.9782).

ROTI and AATR

Rate of TEC Index (ROTI) is an index for the intensity of small-scale irregularities (Pi et al. 1997), which is calculated as a root-mean-square of the rate of TEC (ROT),

where ΔI denotes the TEC change over time Δt, which is a temporal resolution of measurements (typically 30 s). ROTI calculation is performed on the 5-min time interval. ROTI for a particular time includes ROT from ± 2.5 min interval adjacent to this time.

Along with ROTI, we calculate the ROTI-like index Along Arc TEC Rate (AATR) (Juan et al. 2018), which is normalized by a mapping function and does not depend on elevation angles as significantly as ROTI.

Adjusted TEC

The adjusted TEC is an estimation for the absolute vertical TEC. There are different techniques on how to estimate absolute vertical TEC. In previous studies, we used Taylor expansion to recover absolute TEC (Yasyukevich et al. 2015). However, inaccuracy of bias estimation can result in different bias at spatially close stations. Rideout and Coster (2006), for example, use 3 stations simultaneously to improve the interstation biases estimation. Absolute TEC estimation requires significant computation power. Following Hernández-Pajares et al. (2018), we decided to use ionosphere maps to estimate the phase TEC bias. The maps provide a uniform basis for all the stations. In contrast to them, we use only one TEC value obtained from GIM for each ‘receiver-satellite’ TEC series.

The algorithm for calculating the adjusted TEC is as follows. For each continuous slant TEC series, we find the minimal TEC value I_min, and the corresponding elevation θ_min and observation time t_min. We find the ionospheric pierce point (IPP) for the receiver–satellite line-of-sight at t_min. For t_min and the IPP coordinates, we calculate TEC from GIM maps (I_GIM) as a time and space trilinear interpolated value. We use 15-min UQRG maps (Roma-Dollase et al. 2018) to narrow the data gap and to decrease the error caused by time interpolation. From the vertical I_GIM, we obtain the slant I_bias using the reversed modified single layer mapping function, I_bias= I_GIM· F_K(θ_min). I_bias is applied to the entire slant TEC series (Fig. 2) for which it was calculated, I_i′ = I_i + I_bias − I_min. Finally, we convert slant to vertical I_V by direct transformation, I_V= I_i′/F_K(θ_i). Thus, we obtain a series of quasi-vertical TEC, cleaned from geometry and bias effects. Data obtained this way have the same reference (GIM), are of high time resolution, and preserve variations.

Slant TEC and elevation curves. I_bias is estimated to resolve phase TEC ambiguity. θ^b_s is the elevation at minimal TEC

Full size image

The scheme for absolute TEC calculation from geostationary satellites differs slightly. Since measurements from geostationary satellites are characterized by motionless ionospheric pierce points (Kunitsyn et al. 2016), the TEC series can be long term and are not affected by satellite passage. Thus, to obtain absolute vertical GEO-TEC I_GEO series, we can and should use long-term continuous series. For this, we follow Hernández-Pajares et al. (2018). We calculate I_bias for every I_GEO measurement. Since the pierce point is roughly static, there is only a time dependence of corrections I_bias (t_i)= I_GIM (t_i) · F_K, where F_K is a constant here. We use the 10-day sliding window to calculate one value of I_shift to be applied for the entire 10-day series so that:

Then, I_shift is applied to the series, and the vertical I_VGEO is obtained as I_VGEO = I_GEO + I_shift. As new data become available, the sliding window is shifted, and the procedure is repeated. Since data are distributed in one-day files, the sliding window is moved by 24 h. However, due to high noise level (Kunitsyn et al. 2016) of I_GEO we do not include them in the maps currently.

The SIMuRG is a system consisting of several independent modules (Fig. 3). Such a structure allows users to install and apply a module with minimal dependencies. Users might be interested in specific calculations (core module), or data management (DB module), or data visualization (entire SIMuRG). The DB module might be quite useful in understanding data availability. The data availability report is done by going throughout all FTP, which may cause significant traffic. The system is task-based, which means that a query for data processing is formed by one part of the system and executed by another one. The modules of the system are:

General system architecture

Full size image

• simurg_core: core functionality, including procedures to download GNSS data, to read RINEX files, to calculate and to process the data, to store and visualize the results;

• simurg_web: web-interface including interactive plots and API (see SIMuRG web-site for documentation) of the system;

• simurg_db: the database storing metadata and queries. Actual data are stored as properly designed dataset in files (see simurg_core);

• simurg_flow: “glue” module forming system data flows.

The SIMuRG allows third-party services addition, which is marked as External Services in Fig. 3. Such a service is required for complex processing or may require additional hardware not installed in the SIMuRG server. A service is linked to the SIMuRG by means of message exchange using JSON API. The service should process a request, which contains data, metadata, and options of how data should be handled. After the request is completed, the processing result is grabbed by the SIMuRG. This approach allows distributed computations and guarantees that SIMuRG will not overflow by calculations preventing new requests. For a user, services are hidden and the SIMuRG is like a “front end” for them. For now, there are two candidates for such services. The first is the map features detector based on deep learning, which hardly employs GPU computations. The second is the D1-GPS algorithm implementation for ionospheric irregularity velocity field retrieval on a global scale (Afraimovich et al. 2001).

The data products of the SIMuRG are represented in two main ways: map and series. A map is data at corresponding geographic points, i.e., multiple IPP for a certain moment. Vice versa, a series is data for a single station–satellite pair versus time for an arbitrary time range. The data presentation is contained in core and web modules of SIMuRG. For web presentations, we introduce interactive plots that allow manipulation just in a web browser.

SIMuRG uses mixed documented (MongoDB) and binary (HDF—hierarchical data format) database. This grants redundancy: If either part is lost, there is still a possibility of data recovery. The RINEX files information is preserved in MongoDB in order to quickly get a data availability report. RINEX data are stored in HDF in order to get the map and series data fast. In terms of data retrieval from SIMuRG, the HDF allows a tenfold file size reduction compared to ASCII.

Each user request falls into the following tasks:

• Download files, including observational and navigational RINEX files, global ionospheric maps from different networks.

• Calculate series of phase and pseudorange TEC based on two-frequency measurements.

• Estimate elevations and azimuths for corresponding lines-of-sight.

• Detrend and filter data to produce TEC variations and indexes.

• Store data.

• Create maps of data products. The time resolution of the maps is 300 s in standard mode and 30 s for approved requests (please contact us if you need high resolution). The user sets the elevation cut-off for the requested data.

• (optionally) request an external service for data processing

After the request is completed, a user receives an email and can download the processed data, both as an image and HDF file.

SIMuRG provides global and regional data. Examples of a global distribution are presented in Fig. 4. The snapshots present different types of TEC variations for the 18:40:00 UT 2015 June 22 geomagnetic storm (Astafyeva et al. 2016): 2–10 min, 10–20 min, and 20–60 min (from top to bottom). The storm was accompanied by a broad range of ionospheric irregularities (Reiff et al. 2016). The coastlines, the magnetic equator, and magnetic parallels are shown in the background. The darkened region shows the night side. A user can specify whether to show these options when creating a request. The plots are of printing quality and can be used directly. Maps with the distribution of TEC variations show the appearance of enhancement on the auroral oval boundary (top and middle panels). After storm onset, LSTID can be observed on the maps. The most effects are observed on the 10–20-min TEC variations maps.

Snapshots at 18:40 UT of TEC variations products for the June 22, 2015, magnetic storm: 2–10 min (top), 10–20 min (middle), and 20–60 min (bottom)

Full size image

Figure 5 shows the regular (adjusted TEC, top) and irregular (ROTI, middle) ionosphere structure. The bottom panel shows stations used for calculations. One can observe an increase in ROTI values (bottom panel). After storm onset, the auroral oval spreads to the mid-latitudes (see a movie in the supplementary material video 1). It agrees with the fact that the lower boundary of the convection region expanded toward low latitudes (Singh and Sripathi 2017). The absolute TEC shows global scale structures, including the region of an equatorial anomaly with strong TEC enhancement. Also, the top panel shows high TEC values in the pre-sunset mid-latitude Australian sector.

Snapshots at 18:40 UT of regular (adjusted TEC, top) and irregular (ROTI, middle) ionosphere structure for the June 22, 2015, magnetic storm. The bottom panel shows the map of GNSS stations used

Full size image

Figure 6 shows an example of TEC series. The top panel presents 10–20-min TEC variations obtained from station IRKJ for June 22, 2015, 16:30–21:30 UT. The variation curve for each satellite is shifted along the Y-axis to coincide with the satellite number to show the variations clearer. The variations reveal an increase in the amplitude after 19:30 UT. The observed effect has a 1-h lag from the storm onset. Analyzing the dynamics of 10–20-min TEC variations (Supplementary material video 2), there is a large-scale traveling ionospheric disturbance propagating equatorward from the auroral region. The delay is a time for the LSTID to reach the mid-latitudes. The interactive interface allows choosing the satellite under interest and analyzes different types of variations and experiment geometry. The combined analysis of maps and series allows a better understanding of the physical process. The bottom panel shows the sounding geometry. It helps better understanding of the spatial distribution of the variations. The geometry is IPP curves plotted over the globe. Plots of the geometry are given and are specific for every series query in the SIMuRG.

Series of 10–20-min TEC variations (top) and IPP (bottom) during the June 22, 2015, magnetic storm for different “GPS satellite–IRKT” line-of-sights

Full size image

Figure 7 shows the 2–10-min TEC variation maps for GPS (top) and GLONASS (bottom) for the June 22, 2015, magnetic storm. The panels actually show coverage for each system. The coverage relates to the number of receivers recording the GPS signal (top) and the GLONASS signal (bottom); all receivers record GPS signals, but only some of them record GLONASS, and even fewer stations record Galileo. Another factor is the orbital inclination. It can be seen that despite a larger number of measurements by GPS, a user can miss out on effects at high latitudes, which can be seen by GLONASS. On the bottom panel, in the GLONASS data there is a plasma depletion in the auroral oval region at about 48°N. In GPS data (top panel) due to low elevations for these data, the effect is hidden. Adding GLONASS data allows increasing coverage.

Maps of TEC variations for GPS-only (top) and GLONASS-only (bottom) data. The example is for 18:35 the June 22, 2015, magnetic storm

Full size image

To free the geophysical research process from routine data processing and leave more time for analysis, we introduce an online System for Ionosphere Monitoring and Research from GNSS (SIMuRG). SIMuRG serves as a processing system for RINEX and other GNSS data, which also calculates common and original data products. The results are provided as plots for express estimation and as data for user-specific processing. The data can reproduce most of the known ionospheric effects caused by earthquakes, magnetic storms, solar flares, solar terminator, and solar eclipses. We believe the system could assist researches in studying a whole variety of ionospheric events and also in analyzing the lithosphere–atmosphere–magnetosphere–solar wind coupling.

SIMuRG can be used for:

1. 1.

   Studying the large-scale and medium-scale traveling ionospheric disturbances.

2. 2.

   Studying small-scale irregularities and their dynamics.

3. 3.

   Analyzing space weather and the impact of ionospheric disturbances on radio system operation.

4. 4.

   Investigating auroral oval processes and dynamics.

5. 5.

   GNSS data integrity control.

6. 6.

   Studying coupling of sun, earth magnetosphere, ionosphere, and lithosphere.

The main advantages of the SIMuRG are:

1. 1.

   High time (up to 30 s) and space resolution.

2. 2.

   TEC variations in three ranges of periods. 2–10 min band corresponds to acoustic gravity wave periods, 10–20 and 20–60 to medium-scale and large-scale TIDs.

3. 3.

   Data presentation as series or/and maps.

4. 4.

   API for automatic usage of SIMuRG capability.

5. 5.

   Handling the data from all the existing systems (GPS, GLONASS, Galileo, BeiDou, SBAS).

6. 6.

   Capability to study global or regional processes.

7. 7.

   High performance and scalability.

From to the latest test, we obtained performance parameters. The routine calculation for a day, consisting of about 4000 files, it takes about 3 h to download the files and calculate all corresponding data. The 288 maps, based on a 5-min time step, require about 5 h to prepare the data and create the figures and movie. A one-day processing results in 15% utilization of processor for 8 h. Hence, 3 “data days” are processed in one real day which takes 15% of processor time, leaving enough resources for user requests.

We invite everybody to use SIMuRG (http://simurg.iszf.irk.ru). All the described data products can be found there. We would appreciate any comments and suggestions, as well as bug reports.

The authors thank A.S. Shabalin for his help at the beginning of the research, A.A. Kustavinova for the Logo made for the system, and Dr. R.V. Zagretdinov for his help on GNSS networks in Russia. This work was performed under the Russian Science Foundation Grant No. 17-77-20005. In the research, we used GNSS data obtained and stored by Scripps Orbit and Permanent Array Center, UCSD, IGS (Dow et al. 2009), Bundesamt für Kartographie und Geodäsie (BKG) Data Center IGS, Wuhan University, Crustal Dynamics Data Information System (CDDIS), Korea Astronomy and Space Institute, National Geodetic Survey (Data from NOAA’s National Geodetic Survey, Continuously Operating Reference Station network of Global Navigation Satellite System reference stations that is referenced here is publicly available in multiple data protocols online at https://www.ngs.noaa.gov/CORS/data.shtml), the UNAVCO facility, the Geospatial Information Authority of Japan (GSI), Système d’Observation du Niveau des Eaux Littorales (SONEL), the EUREF Permanent Network Services, Geoscience Australia, the New Zealand GeoNet project, the State GPS network of the Republic of Bulgaria, Institut Geographique National, Instituto Geográfico Nacional, Instituto Tecnológico Agrario de Castilla y León, Geodetic Data Archiving Facility, Instituto Brasileiro de Geografia e Estatística, Canadian High Arctic Ionospheric Network, and Angara Common Use Center (http://ckp-rf.ru/ckp/3065) operating under budgetary funding of Basic Research program II.16.

Affiliations

1. Institute of Solar-Terrestrial Physics of Siberian Branch of Russian Academy of Sciences, PO Box 291, Irkutsk, Russia, 664033

   Yury V. Yasyukevich, Alexander V. Kiselev, Ilya V. Zhivetiev, Ilya K. Edemskiy, Semen V. Syrovatskii, Boris M. Maletckii & Artem M. Vesnin

2. Institute of Atmospheric Physics, Czech Academy of Science, Boční II, 1401, 14131, Prague, Czech Republic

   Ilya K. Edemskiy

3. M.V. Lomonosov Moscow State University, Moscow, Russia

   Semen V. Syrovatskii

Corresponding author

Correspondence to Yury V. Yasyukevich.

Publisher's Note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Reprints and Permissions