Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions

Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions

Accepted Manuscript Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions T.L. Gulyaeva, F...

1MB Sizes 3 Downloads 51 Views

Accepted Manuscript Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions T.L. Gulyaeva, F. Arikan, I. Stanislawska PII: DOI: Reference:

S0273-1177(17)30227-2 http://dx.doi.org/10.1016/j.asr.2017.03.039 JASR 13173

To appear in:

Advances in Space Research

Received Date: Revised Date: Accepted Date:

13 July 2016 19 March 2017 25 March 2017

Please cite this article as: Gulyaeva, T.L., Arikan, F., Stanislawska, I., Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions, Advances in Space Research (2017), doi: http://dx.doi.org/10.1016/j.asr.2017.03.039

This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

Earthquake aftereffects in the Equatorial Ionization Anomaly region under geomagnetic quiet and storm conditions

T.L. Gulyaeva1, F. Arikan2, and I. Stanislawska3

1

IZMIRAN, Kaluzskoe Sh. 4, Troitsk, Moscow 108480, Russia, [email protected]

2

Department of Electrical and Electronics Engineering, Hacettepe University, Beytepe, Ankara 06800, Turkey, [email protected] 3

Space Research Center, PAS, Barticka 18-A, Warsaw, Poland, [email protected]

Abstract In addition to multi-scale spatio-temporal trends that shape the ionosphere variability, the ionosphere responds to the disturbances that are solar, geomagnetic and seismic in origin. In this study, postseismic ionospheric disturbances are investigated retrospectively from 1999 to 2015 using two different sets of ionospheric maps of the F2 layer critical frequency, foF2. One set of foF2 maps is obtained by assimilating Global Ionospheric Maps (GIM) of Total Electron Content (TEC) into IRIPlas model (IRI-Plas-foF2). Another set of hourly foF2 maps is obtained using PRIME-251 mapping technique (PRIME-foF2) by the assimilation of ionosonde foF2 data into IRI-CCIR model. The geomagnetic storms affecting the ionosphere are determined with relevant thresholds of geomagnetic AE, aa, ap, ap(τ) and Dst indices. It is observed that more than 60% of the earthquakes occur in the Equatorial Ionization Anomaly (EIA) region within the belt of geomagnetic latitudes ±40 N and geographic longitudes 90 to 190 E. The co-seismic foF2 disturbances, DfoF2, are identified for the cells of the map if an instant foF2 value is outside of pre-defined bounds of foF2 median () and standard deviation (), μ±1σ, in the map fragment of 1,000 km radius around the earthquake hypocenter. The results of positive ionospheric disturbances, DfoF2p, and negative disturbances,

DfoF2n, in the EIA region during the 12 h after earthquake differ with respect to geomagnetic quiet and storm conditions, nighttime and daytime, magnitude and depth of the earthquake. The maximum spatial variability (for more than 50% of map cells in the vicinity of hypocenter) is observed with positive disturbances (DfoF2p) for the earthquakes that occurred during daytime at a depth of 70 to 300 km.

Key words: Seismo-ionospheric coupling; Equatorial Ionization Anomaly, EIA; foF2, TEC, Space Weather, Earthquake, Risk, Threat

1.

Introduction.

Ionosphere is the ionized region of atmosphere whose spatio-temporal trends define a general structure that is well studied almost over a century. In addition to the variability due to climatological and positional trends, ionosphere is also influenced by solar, geomagnetic, gravitational and seismic activities (Pulinets and Liu, 2004; Ding et al., 2007; Afraimovich and Astafyeva, 2008; Hegai et al., 2011; Kutiev et al., 2013). The so-called ‘disturbances’ in the ionosphere can be local, regional or global and their duration can last from minutes to days (Afraimovich et al., 2004; Hernandez-Pajares et al., 2006). These kinds of secondary variability can cause a serious threat for both Space and Ground Based Augmentation Systems (SBAS and GBAS), communication and direction finding systems and navigation and positioning systems (Stankov et al., 2009; Kim et al., 2015). In order to mitigate the signal disruptions or disturbances, the ionosphere needs to be monitored and the threats should be identified (Lognonné et al., 2006; Lee et al., 2008; Jakowski et al., 2012; Bergeot et al., 2014; Yoon et al., 2014). The detection and identification of threats should be based on risk assessment which should involve comprehensive retrospective statistical analysis (Nemec et al., 2009; Katamzi et al., 2012; Li and Parrot, 2013). 2

The deviations from the spatio-temporal trends can be monitored through the two observable parameters of ionosphere, namely, the F2 layer critical frequency (foF2), that is proportional to the peak electron density NmF2, and the Total Electron Content (TEC) (Ding et al., 2007; Tsugawa et al., 2007; Hegai et al., 2011; Gulyaeva, 2012; Gulyaeva et al., 2013). The solar and geomagnetic storms affect the earth’s ionosphere and plasmasphere on a global or regional scale while the seismic activity is usually classified with respect to the location, date, time, magnitude and depth of the earthquake. In the literature, there have been a few seismo-electromagnetic phenomena which are already found to be in statistical correlation with earthquakes (EQs), including but not limited to the ionospheric perturbations both in the lower ionosphere (Hayakawa et al., 2010) and in the upper ionosphere (Liu et al., 2006, Ryu et al., 2016); the geomagnetic field variations (Khachikyan et al, 2008; 2012); and the TEC as given in (Jin et al., 2015). A pattern of spatial variation of seismicity proves to be linked with the main geomagnetic field structure, namely, the largest amount of earthquakes occur in the regions where the geomagnetic Z-component of the Solar Magnetospheric (GSM) coordinate system reaches large positive value (Khachikyan et al., 2008; 2012). The co-seismic ionospheric disturbances should be distinguished from the effect of geomagnetic storms which could intensify or suppress the EQ related variability in the ionosphere and magnetosphere (Karatay et al., 2010; Devi et al., 2014; Guglielmi et al., 2015; Gulyaeva and Arikan, 2016; Gulyaeva et al., 2016). Comparison of earthquakes with the equatorial ring current disturbances has shown that the earthquakes occurred during the Disturbance Storm Time (Dst) storms comprise 13% of the total number of more than 79,000 earthquakes M5.0+ that occurred between 1964 and 2013 (Gulyaeva, 2014). While the severity of a geomagnetic storm is defined by the Dst index which serves as a standard measure of the energy transfer from the solar wind to the ring current within the magnetosphere (Sugiura, 1963), there are other geomagnetic indices specifying impact on the ionosphere under quiet or disturbed conditions as well (Deminov et al., 2013).

3

The latitudinal propagation of the foF2 disturbances has been earlier pointed out with the high variability zones near the equatorial anomaly crest (Fotiadis and Kouris, 2006).

The global

interconnection of the magnetosphere and ionosphere disturbances is confirmed by the correlation found between the variation in two independent processes occurring at widely separated regions in space, namely, the equatorial ring current intensity and the behavior of the ionospheric densities at high-latitudes (Yadav and Pallamraju, 2015). The clear annual asymmetry of TEC due to the cumulative earthquake impact on the nighttime ionosphere has been revealed earlier in (Gulyaeva et al., 2014) for latitudes below -15° N along the magnetic meridian of 270° E which is co-located with the eastern border of the Equatorial Ionization Anomaly (EIA) enhanced seismic activity region which deserves special consideration. Recent studies of EQ-related ionospheric effects observed with DEMETER satellite have revealed that the intense mid-latitude earthquakes contribute to EIA enhancement, and that the ionosphere can be disturbed even 10 days before the actuation of the earthquake (Ryu et al., 2016) whereas the post-seismic ionospheric effects are the less evident. In another case study, the postseismic fast propagating travelling ionospheric disturbances have been detected which quickly travel away from the epicenter of the M9.0 Tohoku earthquake along the main island of Japan with a speed of 2.3-3.3 km/s, accompanied by sequences of concentric circular TEC wavefronts (Liu at al., 2011). These results are supported with the seismic precursor and aftereffect detections in the ionosphere using Global Positioning System (GPS) TEC data both from single stations or a network (Karatay et al, 2010; Klimenko et al., 2011; Xia et al., 2011; Gulyaeva and Arikan, 2016). The aim of the present study is to find out new empirical evidence of the earthquake related foF2 disturbances under quiet space weather conditions and geomagnetic storms. Here, the criteria for specification of EQs under ‘storm’ and ‘quiet’ geomagnetic conditions are extended including relevant thresholds for geomagnetic AE, aa, ap, ap(τ) and Dst indices. The region of focus is the EQ-dense EIA area within the belt of magnetic latitudes ±40 and geographic longitudes from 90 to 190 E which 4

occupies only 20% of the total earth’s surface while more than 60% of EQs of M6.0+ have occurred there during six recent decades. Our study is dedicated to the investigation of the post-seismic disturbances during the 12 h after the EQ occurrence in the EIA zone under geomagnetic quiet and storm conditions. For this purpose, the earthquake data are divided into two main sets, namely, the earthquakes that take place during quiet conditions or geomagnetic storm conditions. This grouping allows us to distinguish the EQ impact on the ionosphere under the particular magnetosphere state unambiguously. The geomagnetic storm conditions are investigated with respect to the increase in ionization (positive storm conditions) or decrease in ionization (negative storm conditions). The earthquakes are classified according to the location, time (night or day), season, magnitude and depth of the event. While the ionosphere variability has been investigated in the numerous studies using GIM-TEC global maps (Suresh and Dashora, 2016; Gulyaeva and Arikan, 2016; and references therein), it is the first attempt in the present study to make statistical investigation of EQs with the foF2 global maps. The foF2 peak is located at 250 to 400 km over the earth which is much closer to the EQs sources than the GPS-derived TEC which integrates the ionosphere and plasmasphere plasma density at 65 to 20,200 km. Hence, one could expect the greater EQs effect on foF2 than on TEC measured through the GPS sub-orbital altitudes. The investigation is carried out with the F2 layer critical frequency (foF2) maps similar to the method used in (Gulyaeva and Arikan, 2016). A forward moving median in a 15 day sliding window is computed for each cell on a foF2 map along with its standard deviation. The difference between the 15th day value of foF2 for the chosen hour and its median, DfoF2, is divided to the computed standard deviation for the same cell and hour. This ratio is categorized with respect to a V index, whose value can vary between -4 to 4, indicating the variability from severe negative disturbance to severe positive disturbance, respectively, as given in (Gulyaeva and Arikan, 2016). The number of cells around the earthquake hypocenter within 1,000 km radius is collected according to the distribution of detected disturbance and an efficiency value is computed. The co-seismic and post5

seismic impact and energy of earthquakes, computed as efficiency index, are followed within the 12 h period after the EQ occurrence. The classification of efficiency index with respect to the type of foF2 map; the state of the ionosphere; the type of the geomagnetic storm; location, magnitude, time and depth of the earthquake between 1999 and 2015 forms a comprehensive retrospective statistical basis for risk and threat assessment of ionospheric weather in the EIA zone. In Section 2, the method of analysis is summarized. Section 3 consists of the description of the data sets that are used in this study and the results are presented in Section 4.

2. Method of Statistical Data Analysis.

In this section, the method of data analysis that is used to compute the efficiency of earthquake aftereffects from foF2 maps is introduced. The foF2 map provides the value of foF2 for a given region in a certain spatial resolution in latitude and longitude. For each latitude and longitude interval, which is denoted as ‘cell’, an hourly value of foF2 is assigned. The structure and composition of foF2 maps used in this study will be provided in Section 3. The foF2 value of cell m, for day d, and hour l is designated as xm;d(l). According to Liu et al. (2006), the recurrence time of the M ≥ 5 earthquakes is 14.2 days. It is also shown in Karatay et al. (2010) that the earthquakes have to be monitored within a 15 day window prior to the earthquake. Therefore, in order to find the references background value of foF2, the first step is to compute the sliding median of every successive 15 days of foF2 at each cell of the map and for each hour. The foF2 sliding median is defined by a -15 day moving window (including the day of EQ and -14 preceding days), and this median value is assigned to the last day of the window, i.e., to the fifteenth day of the window. We use such type of “forward” median approach because it has a potential for development of forecasting model similar to those in (Gulyaeva et al. 2013; Muchtarov et al. 2013; Gulyaeva and Arikan, 2016): 6

m;d d (l )  median( xm;d (l )...xm;d (l )) s

i

i

(1)

s

where m denotes the cell index and l represents the hour index. di = d-14 and ds = d are the initial and final days in the sliding window period, respectively. In the second step, the deviation from the median value is calculated using the standard deviation, , as:



(2)

where NT is the total number of days between day ds and di. The measurements in calculation of  coincide with those that have been already used for determination of the median. An interval within one standard deviation around the median accounts for approximately 68% of the dataset, while two and three standard deviations account 95% and 99.7%, respectively. The measure of foF2 variability can be expressed as deviation from the median  (DfoF2) normalized by the standard deviation  for NT number of days prior to and during day d as:



(3)

The final step of the algorithm is the categorization of foF2 variability by V(foF2) index derived from DfoF2/ value for each cell m, day d and hour l, according to the thresholds similar to those applied to TEC maps as given in (Gulyaeva and Arikan, 2016). The V(foF2) index represents 7

the integer magnitude of foF2 variability with respect to the median within the sliding window of 15 days in terms of  grades. The magnitude of variability changes from Vn = -4, extreme negative foF2 anomaly, to Vp = +4, extreme positive foF2 anomaly, in steps of V = 1. When DfoF2 < ±1, the ionosphere is defined to be in undisturbed or quiet state. If an instant foF2 value is outside of predefined bounds of  ±1, then foF2 disturbance or anomaly is detected (Akhoondzadeh, 2015). One of the major discriminants of earthquake effects on the ionosphere is the spatial range of the disturbance. Typically, geomagnetic storms affect large portions of globe after the anomalous changes in IMF-B, global electric currents and have patterns that can be recognized in the geomagnetic field. The co-seismic and post-seismic disturbances in the ionosphere can be observed locally or regionally depending on the type, magnitude and depth of the earthquake as indicated in various studies. In this study, Efficiency factor (Ef) of the ionosphere response to impact of earthquakes is defined by the ratio of number of the instant foF2 values which fall outside of μ±1σ (computed for each cell, each day and each hour of the map) to the total number of the foF2 cells in the vicinity of EQ hypocenter with radius of 1,000 km. Efficiency of the negative ionosphere disturbances, Efn, is represented by relative density of occurrence of the extreme negative indices, DfoF2n/ ≤ -1, on the specified fragment of a map. Similarly, the efficiency of extreme positive disturbances, Efp, corresponds to the relative density of occurrence of the extreme positive indices, DfoF2f/ greater than +1 around the EQ hypocenter(s) on the map or series of EQs on the relevant maps (where the total number of cells in the vicinity of EQ on the maps is defined as mtot) as given below:

fn

(4)

fp

where mDfoF2n and mDfoF2p denote the total number of cells whose DfoF2 value is outside of the defined bounds of μ±1σ for the extreme negative and extreme positive disturbances, respectively. The

8

method of analysis outlined above is applied to the data described in Section 3, and the results are presented in Section 4.

3.

Description of Data Used in the Study.

The data that are used in this study can be categorized into the following groups:

3.1. Earthquake Data Set. The most extensive earthquake data set between 1964 and 2015 is provided by the Northern California Earthquake Data Center of the Advanced National Seismic System (ANSS) (NCEDS, 2014). This data set is extended back to the period from 1953 to 1963 with catalogue of the intense global earthquakes (Golubeva, 1972). The composite catalogue of earthquakes prepared by ANSS is a world-wide earthquake catalog which is generated by merging the master earthquake catalogs from contributing ANSS member institutions and then removing duplicate events, or non-unique solutions for the same event. EQ events with uncertain depth of hypocenter (D=0) are excluded from the analysis. We use the monthly and annual data for earthquakes of magnitude M6 to M10 from the NCEDS catalogue from January, 1999, to December, 2015, during the period where the hourly GIMTEC maps are available for the 23rd and 24th solar cycles. The EQ occurrence for the different ranges of the hypocenter depth in the Pacific region is provided in detail by Levin and Sasorova (2012). The hypocenter depth (D) is classified into three groups as the shallow depth, where D70 km (D1); the descent depth, where 70
9

The region selected for the present study is the earthquakes-dense Equatorial Ionization Anomaly (EIA) region within the belt of geomagnetic latitudes ±40 N and geographic longitudes 90 E to 190 E which are close to magnetic longitudes between 160 and 270 E.

3.2. Specification of Quiet and Storm Conditions Using Geomagnetic Indices.

The magnetosphere state is designated as ‘storm’ if at least one of the following criteria is satisfied (Gonzalez et al., 1994; Gulyaeva and Arikan, 2016):

AEmax ≥ 500 nT ; aamax > 45 nT; apmax > 30 nT; ap() > 18 nT; Dstmin ≤ -30 nT

(5)

Otherwise, conditions will be considered as ‘non-storm’ or ‘quiet’. The above conditions should be fulfilled both for the nearest UT hour (or 3 h UT interval) following the earthquake and the nearest pre-earthquake hour (or 3 h UT interval) to capture storm or sub-storm impact at the time of EQ event. Here, AEmax is the auroral electrojet AE value for two near-EQ hours, aamax is the maximum mid-latitude aa index value for a given and preceding 3 h intervals; apmax is the maximum ap value for a given and preceding 3 h intervals. The ap() is the mean weighted value of ap index as given in (Wrenn, 1987):

ap() = (1 - )(ap0 + ap-1 + ap-2 2 + … )

(6)

with the characteristic time T = 11 h or  = exp(-3/T) which is approximately 0.76; ap0, ap-1,… are ap values at a given time in preceding 3 h intervals. The Dst min is the minimum disturbance storm time value for two hours including the EQ occurrence. All the above indices are expressed in nanoTeslas

10

(nT). The periods of storms and sub-storms may occur at all latitudes from the pole (AE index) through the mid-latitudes (aa, ap and ap() indices) to the equator (Dst index). The results of comparison of EQs M6+ under geomagnetic quiet and storm conditions for the global distribution (EQ-g, n1 number of events) and the selected EIA area (EQ-a, n2 number of events) for 1953-2015 are given in Table 1. The occurrence of EQs at the EIA area consists of almost 60% of the total number of global earthquakes. The percentage of occurrence is 57% for shallow depth and increases to 84% for the deep depth. These proportions are kept the same both for geomagnetic quiet conditions and storms the latter presenting 20% of the total number of EQs. A similar ratio has been observed for the earthquakes that occur under ‘storm’ conditions which amount to 20% of all earthquakes. The mean energy emitted by the earthquakes in the EIA belt (within ±40 of magnetic latitude) is presented in Table 2 for four ranges of Richter magnitude (M) under geomagnetic quiet and storm conditions based on EQs data set for 1953-2015. Here, two subsets are provided: E1(n1) for 160 E to 270 E in magnetic longitude, and E2(n2) for the rest EIA longitudes, where n1 and n2 indicate number of EQ events for each subset. Both the number of EQs and the emitted energy are larger in the EIA area under consideration in the present study as compared with the rest EIA longitudes.

3.3 The Coordinate System Used in the Analysis.

Ionospheric plasma drift and the consequent TEC and electron density change in the topside ionosphere and plasmasphere are bound to the earth’s magnetic field, so both the geomagnetic coordinate system and geographic coordinate system are used throughout this study. Figure 1a presents the magnetic coordinate frame for the global spatial percentage distribution of the earthquakes with M6+ that took place during the ‘non-storm’ or ‘quiet’ state of ionosphere; and Figure 1b indicates the similar information for the ‘storm’ time earthquakes observed during 1953-2015. The thin white lines 11

outline the EIA area within the magnetic latitudes of ±40°, and geographic longitudes 90 E to 190 E which are close to magnetic longitudes between 160 and 270 E of the EQ-dense events. Crosses are placed at the EQ hypocenters. The EQs data set from 1953 to 2015 exhibits the region of enhanced seismic activity at the Equatorial Ionization Anomaly belt within ±40 of geomagnetic latitudes and 90 E to 190 E of geographic longitudes. This unique EIA area representing 20% of the earth’s surface serves as a basis for 60% of earthquakes of Richter magnitude M6+. The EQ data set is separated into two subsets of EQs which occur either under geomagnetic quiet conditions or storms. The total global set of EQs of M6+ for 1953-2015 consists of 7,067 events of which 4,247 events occurred in the EIA EQ-dense area including 20% of ‘storm’ subset under used storm-time definitions. For the period of 1999-2015 there are 1627 EQs of M6+ in the EIA area where 1327 events took place within the quiet subset and 300 EQs within the ‘storm’ subset.

3.4. Description of the foF2 maps Used in the Analysis. As mentioned in the introduction section, the F2 region critical frequency, foF2, is one of the determining parameters of ionosphere. Therefore, the investigation in this study is based on the spatiotemporal distribution of foF2, i.e., foF2 maps. Two different sets of foF2 maps are used in this study. The first set of maps is obtained by assimilating TEC values from the Global Ionospheric Maps (GIM), into the International Reference Ionosphere (IRI) extended to Plasmasphere model (IRI-Plas-foF2) as discussed in detail in (Gulyaeva et al., 2011; 2013; Arikan et al., 2015). The IRI represents monthly averages of electron and ion densities and temperatures and the total electron content in the altitude range of 50 km – 2000 km (Bilitza et al., 2017). The IRI-Plas model extends the altitude range of IRI to the GPS orbital height of 20,200 km. In process of TEC assimilation by IRI-Plas code, the IRI-Plas algorithm is complemented with a feedback loop in order to produce an instantaneous NmF2 (proportional to the square of foF2). 12

The first run of IRI-Plas code produces the median (quiet) TECmed using the IRI-CCIR prediction of median foF2med (NmF2med) and IRI-CCIR peak height hmF2med. Then, the second run of IRI-Plas code with TECgps input is used to convert it to an instant foF2 (NmF2) fitted to the GPS signal’s travel through the ionosphere and plasmasphere. At this stage it is assumed that the ratio of TECmed to the measured TECgps is proportional to the ratio of the instant foF2 (NmF2) to foF2med (NmF2med). The above algorithm is applied to each cell of GIM-TEC map to produce IRI-Plas-foF2 map in the IONospheric EXchange (IONEX) format. The IRI-Plas-foF2 map products derived from GIM-TEC with the IRI-Plas model proved to be reliable for the investigation of the morphological features of the single crest phenomenon at the longitude of 120° E for the geomagnetic quiet days in the Equatorial Ionization Anomaly (EIA) region (Huang et al., 2014). We have utilized GIM-TEC provided by Jet Propulsion Laboratory of California Institute of Technology (JPL) in IONEX format. The vertical TEC is modeled by JPL in a solargeomagnetic reference frame using bi-cubic splines on a spherical grid. A Kalman filter is used to solve simultaneously for instrumental biases and vertical TEC on the grid as stochastic parameters as given in (Manucci et al. 1998). The IONEX global map consists of 5183 grid values between -87.5° N to 87.5° N in steps of 2.5° in latitude, -180° E to 180° E in steps of 5° in longitude, available for science and applications since start of GIM-TEC products in mid-1998 (Hernandes-Pajares et al., 2009). For the purpose of the present study, the source GIM-TEC maps are converted to geomagnetic coordinates binned in -87.5° N to 87.5° N in steps of 2.5° in geomagnetic latitudes, and 0° E to 360° E in steps of 5° in geomagnetic longitude using the International Geomagnetic Reference Field (IGRF) model. The second set of foF2 maps is obtained using the PRIME251 mapping procedure (PRIMEfoF2) provided by Space Research Center of Poland (Stanislawska et al., 1996; 2000). These maps are produced with Kriging technique by assimilation of the ionosonde foF2 data into IRI model adjusted to the ITU-R (CCIR, 1983) median foF2 from 1999 to 2006. The mapping procedure among the data13

sparse isolated measurements of more than 60 ionosondes worldwide is based on the characteristic variability of foF2 and constructing and modelling semivariogram, i.e. the function that illustrates the differentiation of the parameter value depending on the distance between the measurements. The coordinate dependent weight (scaling factor) is introduced in Kriging interpolation due to anisotropy of the semivariogram which is systematically different in different azimuths. The scaling factor is modelled in terms of foF2 correlation distances between the measurements which are usually less in latitude than in longitude under quiet and storm conditions (Stanislawska et al., 1996; Stanislawska and Gulyaeva, 2015). In the present study, these foF2 maps series are extended from 2007 to 2015 by incorporating the ionosonde foF2 data into updated PRIME251 software and the global hourly PRIME-foF2 maps in IONEX format have been produced. In a case of missing ionosonde foF2 observation, the missing value is filled by prediction according to the method described in (Gulyaeva et al., 2008). The above sets of data are used in the analysis that is discussed in the next section.

4. Results The statistical analysis method given in Section 2 is applied to both sets of foF2 maps described in Section 3. In the first step, for each earthquake of M6+, and for each map cell in the defined EIA region, 15 day sliding window hourly median, the standard deviation and DfoF2 are calculated as given in Eqn. 3, 4, 5, respectively. The V(foF2) index is computed for each cell within the 1,000 km neighborhood in the region of interest centered at the earthquake hypocenter [e e]. The analysis of map for a rectangular region defined by (i, i) to (s, s) is provided by Gulyaeva et al. (2013, Appendix A) for the increments in  and  given as  and , respectively. However, the region around the EQ hypocenter with radius of 1,000 km is not a simple rectangular region but it is represented by sections of 16 cells comprised of a square of 4 × 4 cells in latitude and longitude, superimposed by a rectangle of 8 × 2 cells in latitude and longitude surrounding (e, e). The time of 14

the earthquake and the state of the ionosphere are also registered using the solar zenith angle, seasons and the solar and geomagnetic indices given in Section 3. The DfoF2 is further classified as positive and negative and the efficiency is computed according to Eqn. 4. An example of V(foF2) maps is provided in Figure 2 for the M7.1 EQ that took place at Honshu, Japan, at a depth of 292 km, on 9 August, 2009, at 10:55:55 UT which corresponds to the local nighttime of 20:11 LT with solar zenith angle of  = 106. The coordinates of the hypocenter are [33.17 N, 137.94 E] geographic, and [24.4 N, 207.4 E] geomagnetic (indicated with a star on the maps). In Figure 2, the V(foF2) values are derived using the PRIME-foF2 map set. In Figure 2a and Figure 2b, V(foF2) values for -2 h preceding the earthquake and -1 h before the earthquake are provided, respectively. Figure 2c corresponds to V(foF2) map that is drawn 4 min after EQ and Figure 2d is the map for the same parameter 1 h after EQ. The fragment of the map within the radius of 1,000 km surrounding earthquake hypocenter (Figure 2b, points) is determined by  ≤ e ± 10,  ≤ e ± 7.5. This EQ event took place at solar minimum (sunspot number SSN = 5) during quiet geomagnetic conditions with AE = 377 nT, aa = 20 nT, ap = 6 nT, ap = 4.4 nT, Dst = -15 nT. Therefore, the regions of the denser plasma clouds with enhanced foF2 (red sections) and plasma troughs with depleted electron density, foF2 (blue sections) are attributed directly to the EQ effect. As it can be observed from Figure 2a, the disturbance in the ionosphere started as a gradient from severe positive to severe negative deviation 2 hours prior to the earthquake. A strong enhancement one hour prior to the earthquake is followed by a depletion during the time of the earthquake as given in Figure 2b and Figure 2c, respectively. One hour after the earthquake, another strong enhancement around the earthquake hypocenter is observed in Figure 2d. The ionospheric disturbance that started at nighttime under geomagnetic quiet conditions at solar minimum is observed in large areas exceeding 1,000 km radius from the hypocenter following the propagation of seismic waves registered with seismographs worldwide.

15

In order to compare the performance of sensitivity to the ionospheric variability of the two sets of foF2 maps, IRI-Plas-foF2 maps and the ionosonde PRIME-foF2 maps, given in Section 3, the Root Mean Square (RMS) deviation is computed between the hourly foF2 value XEQ, and the -15 day median, X in the vicinity of hypocenter at the nearest hour after earthquake as



(7)

The above equation is applied to the two sets of foF2 maps and the deviations from the cell median within 15 day sliding windows are grouped according to the ionospheric state, earthquake location, depth, magnitude, energy, time of the earthquake; and finally, according to the solar activity. Results of evaluation of a post-seismic variability of foF2 with IRI-Plas-foF2 maps and PRIME-foF2 maps are presented in Figure 3 and Table 3. The daytime is defined with solar zenith angle  < 90 and the nighttime is defined with solar zenith angle  > 90. The season is specified as the equinox for months March, April, September and October, the summer in the North (South) hemisphere for the months May, June, July, August (November, December, January, February), and the winter in the North (South) hemisphere for the months November, December, January, February (May, June, July, August). The solar cycle dependence of RMS deviations between the EQ-induced foF2 variability and the median follows from Figure 3 with growing deviation towards the solar maximum plotted with 12monthly smoothed solar radio flux F10.712. Some antisymmetric trend between RMS and F10.712 flux peak of solar cycles 23 and 24 is observed which deserves further study with longer time series of foF2 maps in order to resolve contradictions in the ionospheric parameters trends derived from the limited time periods of observations (Laštovička, 2013; Gulyaeva et al., 2016). Annual number of EQs is plotted in the bottom panels which show large number of EQs under quiet geomagnetic conditions 16

with their majority towards the solar minimum (Gulyaeva, 2014). Under the storm conditions the number of EQs tends to a minimum towards the low solar activity due to low geomagnetic storm occurrence at those years. It follows from Figure 3 and Table 3 that RMS for EQ events are dominant under storm conditions when EQ effects are enhanced by foF2 variability due to storm. RMS in Table 3 for EQs is dominant during equinox season and during the nighttime when larger F2 peak electron density variability occurs. It can also be observed from Table 3 that IRI-Plas-foF2 maps are more sensitive in indication of earthquake related disturbances and geomagnetic disturbances compared to PRIME-foF2 maps due to the efficiency of GIM maps in monitoring TEC variability. Therefore, the efficiency of earthquakes given in Eqn. 4 is calculated using IRI-Plas-foF2 maps. In Figure 4, the temporal variation of the efficiency of earthquakes, Ef, in Eqn. 4, is provided for the 12 h period after the earthquake occurence. The daytime earthquakes ( < 90) and nighttime conditions ( > 90) for both ‘storm’ subset and ‘quiet’ conditions are grouped according to the positive and negative efficiency values for a period 1999 to 2015 based on IRI-Plas-foF2 maps. Symbol S+ in the graphs stands for the positive ‘storm’ E fp, Q+ for ‘quiet’ time E fp, S- for the negative ‘storm’ E fn, and Q- for the negative quiet Efn. In general, all statistical results for the quiet and storm conditions confirm existence of earthquake aftereffects. Efficiency of EQ impact on foF2 variability is greater than 15% in all cases which means that on average 15% of deviations of foF2 around the EQ hypocenter area exceeds predefined threshold ±1. In some individual EQs, the foF2 disturbances in the sense defined in the present study could be missing in the EQ predefined area within 1,000 km radius from the hypocenter but the overall Efficiency results of the EQ impact on the ionosphere from below are comparable with geo-Efficiency of the external solar and interplanetary sources (Hapgood, 2011). The most important outcome of results in Figure 4 is that the Efficiency of EQs for the positive DfoF2p disturbances under storm condition (S+) is dominant as compared with other results. Peak of Efp at storms by daytime (and as a consequence for the total data set) occurs at the nearest integer hour 17

(t = 0) after EQ up to Efp = 50%. So, under daytime conditions the EQ impact on the ionosphere during the storm peaks just after EQ onset time which is gradually suppressed afterwards during the next 12 h to the persistent level of other conditions presumably due to the effective daytime photoionization of plasma species. As distinct from the daytime conditions, the nighttime ionosphere displays the enhanced positive DfoF2p results for nighttime storm, Efp, which is about twice enlarged as compared with the quiet conditions and the negative ionosphere storm disturbances (Figure 4, right panel) which is kept about the same level during the 12 h post-seismic period after the EQ onset. One can assume that the nighttime ionosphere does not have relevant resources to suppress the post-seismic positive disturbance effects during 12 h after the event. It has been pointed out by Ryu et al. (2016) that the normalized equatorial density enhancement is sensitive and proportional to both the magnitude and the hypocenter depth of an earthquake with the dominant effect for the EQs with relatively shallow depth. Our results for EQ Efficiency during 19992015 period are plotted in Figure 5 for three levels of depths: (1) the shallow depth D1 (left column); (2) the descent depth D2 (middle column); (3) the deep depth D3 (right column). Subdivision to three classes of EQs depth is applied similar to Figure 4 to the total data set (Figure 5, upper panel), daytime (middle panel) and nighttime data (lower panel). The enlarged vertical scale for the daytime conditions (middle panel) is highly noticable. Figure 5 demonstrates dominant Efficiency of EQs on foF2 disturbances during daytime (middle panel) for the 2nd class of the descent depth (middle column). Just after the onset of EQ, the daytime positive storm Efficiency Efp = 60% for the descent depth D2, the next effective are daytime positive storm results Efp = 50% for the shallow depth D1. There is no significant positive storm Efficiency for the daytime deep depth D3. The nighttime positive storm Efficiency reaches its peak E fn = 40% 5 h after the EQ onset. After the peak, all results for the positive storm Efficiency decrease towards the mean values.

18

5. Conclusion

In this study, the structural changes of ionosphere are investigated with respect to disturbances in the F2 layer peak electron density under quiet conditions in the geomagnetic field and storms in their relation to the earthquakes of Richter magnitude from M6 to M10, in EIA region between 1999 and 2015. The F2 layer critical frequency values are obtained from both the IRI-Plas-foF2 maps with assimilation of GIM-TEC by IRI-Plas model and PRIME-foF2 maps constructed by assimilating the ionosonde foF2 measurements by IRI-CCIR maps. The analysis is based on the deviation of hourly foF2 value from its reference median which is computed within a sliding window of 15 days prior to the earthquake date. The proposed disturbance indicator, DfoF2, is computed along with the standard deviation from the median. Negative or positive deviations of foF2 from the pre-defined bounds of ±1 are used as basis of ionospheric variability after the earthquake occurrence. The spatial influence of earthquakes is investigated through the foF2 variations in the vicinity of hypocenter within the radius of 1,000 km. The temporal variability is observed through 23rd and 24th solar cycles, with respect to ionospheric seasons, nighttime and daytime occurence of earthquakes and within 12 h after the earthquake event. The earthquakes are grouped according to their location, magnitude, and depth. This comprehensive statistical study provides a very firm database for assessment of risk and threat due to ionospheric weather in EIA region for co- and post-seismic disturbances. It is observed that the earthquake aftereffects can be felt as severe positive enhancements or negative depletions within 1,000 km neighborhood of the earthquake hypocenter for earthquakes with magnitude larger than M6 on the Richter Scale. The RMS deviations from the reference median indicate that the time and depth of the earthquake and seasonal variations of ionosphere are also contributors to co-seismic ionospheric perturbations of the state of the ionosphere. The efficiency of the disturbances increases towards the solar cycle maximum.

19

It is found that the IRI-Plas-foF2 maps with GIM-TEC assimilation are more sensitive in indication of disturbance conditions. According to the efficiency of EQ impact computed using IRIPlas-foF2 maps, the 15% to 60% ionospheric disturbance in the fragment of the map surrounding the EQ hypocenter confirms the coupling between seismic activity and the F2 layer peak electron density. The post-seismic positive disturbances indicated by DfoF2p are more dominant in the fragment of the map in the vicinity of EQ occurrence as compared to the negative disturbances DfoF2n under storm conditions. The level of positive ionospheric disturbances, DfoF2p, and negative disturbances, DfoF2n, in the EIA region during 12 h after the EQ reveals the difference for geomagnetic quiet and storm conditions, daytime and nighttime EQ occurrence, emitted energy and depth of EQ. The daytime earthquakes in the EIA region have a higher efficiency of positive storm conditions and the aftereffects can be felt within a 1,000 km neighborhood during 12 h after the earthquake. The maximum positive efficiency based on DfoF2p is observed for more than 50% of map cells in the vicinity of EQ hypocenter at daytime for the descent depth of EQs in the range of 70-300 km (D2). The EQ-induced peak of the positive storm DfoF2p value is observed at the first integer UT hour after EQ for daytime. Although the value decreases linearly during the 12 hours after the daytime earthquake, the ionospheric enhancement is retained approximately at a constant level after the nighttime earthquakes. The most important conclusion of the present study is that the efficiency and magnitude of the disturbances in the EIA region indicate that the risk of ionospheric weather increases when M6+ earthquakes occur during daytime along with positive type ionospheric storms. The aftereffects can be observed during the 12 h period after the EQ at distances more than 1,000 km in extend. This study also indicates the importance and necessity of retrospective statistical classification of threat levels in development of proper mitigation techniques for GBAS systems. This study will be further extended to other regions of the earth which are susceptible to earthquakes with magnitudes 4 to 7 and the risk assessment will be carried out to determine the impact area of lower magnitude earthquakes. 20

Acknowledgements The

GIM-foF2

maps

from

1999

to

now

are

provided

by

IZMIRAN

at

http://ftp.izmiran.ru/pub/izmiran/SPIM/Maps/foF2/ and IONOLAB at http://www.ionolab.org/. The PRIME251 foF2 instantaneous maping software is provided by SRC at http://rwc.cbk.waw.pl/. The ionosonde

foF2

maps

for

1999-2006

are

provided

by

SRC

at

ftp://ftp.cbk.waw.pl/idce/grid/fof2_maps_99_06/. The ionosonde foF2 data for PRIME maps are provided by the International Space Environment ISES Service (http://www.spaceweather.org/), UK Solar System Data Centre (http://www.ukssdc.ac.uk/wdcc1/data_menu.html), Australian Space Weather

Data

Service

(http://www.sws.bom.gov.au/World_Data_Centre/1/3),

(http://wdc.nict.go.jp/IONO/HP2009/ISDJ/index-E.html), (http://spidr.ionosonde.net/spidr/logoff.do),

IZMIRAN

NOAA Ionospheric

SPIDR Weather

NICT, Server,

Japan USA

service,

RF

(http://www.izmiran.ru/sewrvices/iweather/). GIM-TEC data used for production of IRI-Plas foF2 maps are provided by the Jet Propulsion Laboratory of California Institute of Technology (JPL) at (ftp://sideshow.jpl.nasa.gov/pub/iono_daily/). Catalogue of the Advanced National Seismic System (ANSS) is provided by the Northern California Earthquake Data Center (NCEDS) at http://www.nceds.org/anss/. The solar radio flux is provided by Space Weather Sevice, Canada at (ftp://ftp.geolab.nrcan.gc.ca/data/solar_flux/daily_flux_values/fluxtable.txt). The geomagnetic indices are provided by WDC for Geomagnetism at (http://wdc.kugi.kyoto-u.ac.jp/). This study is partly supported by TUBITAK EEEAG 115E915. We are grateful to the Reviewers for the constructive comments and helpful suggestions.

References. Afraimovich, E.L., Astafyeva, E.I., Voyeikov, S.V., 2004. Isolated ionospheric disturbances as deduced from global GPS network, Ann. Geophys., 22, 47-62, DOI:10.5194/angeo-22-47-2004. 21

Afraimovich E.L., Astafyeva E.I., 2008. TEC anomalies – local TEC changes prior to earthquakes or TEC response to solar and geomagnetic activity changes? Earth Planets Space, 60, 961-966. Akhoondzadeh, M., 2015. Firefly algorithm in detection of TEC seismo-ionospheric anomalies. Adv. Space Res., 56, 1, 10–18, DOI:10.1016/j.asr.2015.03.025. Arikan, F., Sezen, U., Gulyaeva, T.L., Cilibas, O., 2015. Online, automatic, ionospheric maps: IRIPLAS-MAP. Adv. Space Res., 55(8), 2106-2113. Astafyeva, E.I., Afraimovich, E.L., 2006. Long-distance traveling ionospheric disturbances caused by the great Sumatra-Andaman earthquake on 26 December 2004. Earth Planets Space, 58, 10251031. Bergeot, N., Chevalier, J.M., Bruyninx, C., Pottiaux, E., Aerts, W., Baire, Q., Legrand, J., Defraigne, P., Huang, W., 2014. Near real-time ionospheric monitoring over Europe at the Royal Observatory of Belgium using GNSS data. J. Space Weather Space Climate, 4, A31. Bilitza, D., Altadill, D., Truhlik, V., Shubin, V., Galkin, I., Reinisch, B., Huang, X., 2017. International Reference Ionosphere 2016: from ionospheric climate to real-time weather predictions. Space Weather, 15, 2, 418-429, DOI: 10.1002/2016SW001593. CCIR, 1983. CCIR Atlas of Ionospheric Characteristics. Comite Consultatif International des Radio Communications Rep. 340. Geneve, International Telecommunication Union. Deminov, M.G., Deminova, G.F., Zherebtsov, G.A., Polekh, N.M., 2013. Statistical properties of variability of the quiet ionosphere F2-layer maximum parameters over Irkutsk under low solar activity. Adv. Space Res., 51, 702-711, DOI: 10.1016/j.asr.2012.09.037. Devi, M., Barbara, A.K., Oyama, K-I., Chen, C-H., 2014. Earthquake induced dynamics at the ionosphere

in

presence

of

magnetic

storm.

Adv.

Space

Res.,

53,

609–618.

DOI:10.1016/j.asr.2013.11.054. Ding, F., Wan, W., Ning, B., Wang, M., 2007. Large scale traveling ionospheric disturbances observed by GPS total electron content during the magnetic storm of 29-30 October 2003. J. Geophys. Res., Space Phys., 112, A6. 22

Fotiadis, D.N., Kouris, S.S., 2006. A functional dependence of foF2 variability on latitude. Adv. Space Res., 37, 1023–1028, DOI:10.1016/j.asr.2005.02.054. Golubeva, N.V., 1972. Catalogue of the intense global earthquakes of M≥6.0 from 1953 to 1967. Institute of Physics of the Earth, Moscow, 164p. (in Russian). Gonzalez, W.D., Joselyn, J.A., Kamide, Y., Kroehl, H.W., Rostoker, G., Tsurutani, B.T., Vasyliunas, V.M., 1994. What is geomagnetic storm? J. Geophys. Res., Space Phys., 99, 5771-5792. Guglielmi, А.V., Lavrov, I.P., Sobisevich, A.L., 2015. Geomagnetic Sudden storm commencements and earthquakes. Solar-Terr. Phys., 1, 1, 98-103. DOI: 10.12737/5694. Gulyaeva, T. L., Stanislwska, I., Tomasik, M., 2008. Ionospheric weather: cloning missed foF2 observations for derivation of variability index. Ann. Geophys., 26, 315-321, DOI:10.5194/angeo26-315-2008. Gulyaeva, T.L., Arikan, F., Stanislawska, I., 2011. Inter-hemispheric imaging of the ionosphere with the upgraded IRI-Plas model during the space weather storms. Earth, Planet Space, 63, 8, 929-939, DOI:10.5047/eps.2011.04.007. Gulyaeva, T.L., 2012. Empirical model of ionospheric storm effects on the F2 layer peak height associated with changes of peak electron density. J. Geophys. Res., Space Phys., 117(A2). Gulyaeva, T.L., Arikan, F., Hernandez-Pajares, M., Stanislawska, I., 2013. GIM-TEC adaptive ionospheric weather assessment and forecast system. J. Atmos. Solar-Terr. Phys., 102, 329-340. DOI: 10.1016/j.jastp.2013.06.011. Gulyaeva, T.L., Arikan, F., Hernandez-Pajares, M., Veselovsky, I.S., 2014. North-South components of the annual asymmetry in the ionosphere. Radio Sci., 49, 485-496. DOI: 10.1002/2014RS005401. Gulyaeva, T.L., 2014. Association of seismic activity with solar cycle and geomagnetic activity. Develop. Earth Sci., 2, 14-19. Available from: http://www.seipub.org/des/.

23

Gulyaeva, T., Arikan, F., 2016. Statistical discrimination of global post-seismic ionosphere effects under geomagnetic quiet and storm conditions. Geomatics, Natural Hazards, Risk, DOI: 10.1080/19475705.2016.1246483. Gulyaeva, T., Arikan, F., Stanislawska, I., 2016. Persistent long-term (1944-2015) ionospheremagnetosphere associations at the area of intense seismic activity and beyond. Adv. Space. Res. DOI:10.1016/j.asr.2016.11.022 Gutenberg, B., Richter, C.F., 1956. Earthquake magnitude, intensity, energy and acceleration. Bull. Seismol. Soc. Amer., 46, 2, 105-145.. Gutenberg, B., Richter, C.F., 2010. Magnitude and energy of earthquakes. Annals Geophys., 53, 7–12. Jakowski, N., Béniguel, Y., De Franceschi, G., Pajares, M.H., Jacobsen, K.S., Stanislawska, I., Tomasik, L., Warnant, R., Wautelet, G., 2012. Monitoring, tracking and forecasting ionospheric perturbations using GNSS techniques, J. Space Weather Space Climate, 2, A22. Hapgood, M.A., 2011. Towards a scientific understanding of the risk from extreme space weather. Adv. Space Res., 47, 2059–2072, DOI:10.1016/j.asr.2010.02.007. Hayakawa, M., Hobara, Y., 2010. Current status of seismoelectromagnetics for short-term earthquake prediction, Geomatics, Natural Hazards Risk, 1, 2, 115-155, DOI: 10.1080/19475705.2010.486933. Hegai, V.V., Legen'ka, A.D., Kim, V.P., Georgieva, K., 2011. Wave-like perturbations in the ionospheric F2-layer observed after the Ms 8.1 Samoa earthquake of September 29, 2009, Adv. Space Res., 47, 11, 1979-1982. Hernandez-Pajares, M., Juan, J.M., Sanz, J., 2006. Medium-scale traveling ionospheric disturbances affecting GPS measurements: Spatial and temporal analysis, J. Geophys. Res.. Space Phys.,111, A7. Hernandez-Pajares, M, Juan, J.M., Sanz, J., Garcia-Rigo, A., Feltens, J., Komjathy, A., Schaer, S.C., Krankowski, A., 2009. The IGS VTEC maps: a reliable source of ionospheric information since 1998. J. Geodesy, 83, 263–275.

24

Huang, L., Wang, J., Jiang, Y., Huang, J., Chen, Z., Zhao, K., 2014. A preliminary study of the single crest phenomenon in total electron content (TEC) in the equatorial anomaly region around 120E longitude

between

1999

and

2012.

Adv.

Space

Res.,

54,

2200–2207.

DOI:10.1016/j.asr.2014.08.021. Jin, S., Occhipinti, G., Jin, R., 2015. GNSS ionospheric seismology: Recent observation evidences and characteristics. Earth-Sci. Rev., 147, 54–64, DOI:10.1016/j.earscirev.2015.05.003. Karatay, S., Arikan, F., Arikan, O., 2010. Investigation of total electron content variability due to seismic and

geomagnetic disturbances

in

the

ionosphere.

Radio

Sci.,

45,

RS5012.

DOI:10.1029/2009RS004313. Katamzi, Z.T., Smith, N.D., Mitchell, C.N., Spalla, P., Materassi, M., 2012. Statistical analysis of travelling ionospheric disturbances using TEC observations from geostationary satellites, J. Atmosph. Solar-Terr. Phys., 74, 64-80. Khachikjan G.Ja., Stikhamaya, G.G., Stikhamiy, A.P., Solonitsina, N.F., 2008. Spatial distribution of earthquake epicenters and geomagnetic declination angle. Inland Earthquake, 22, 264-270. Khachikyan, G., Inchin, A., Lozbin, A. , 2012. Spatial Distribution of Seismicity: Relationships with Geomagnetic Z-Component in Geocentric Solar Magnetospheric Coordinate System. Int. J. Geosciences, 3, 1084-1088, DOI:10.4236/ijg.2012.35109. Klimenko, M.V., Klimenko, V.V., Zakharenkova, I.E., Pulinets, S.A., Zhao, B., Tsidilina, M.N., 2011. Formation mechanism of great positive TEC disturbances prior to Wenchuan earthquake on May 12, 2008, Adv. Space Res., 48, 3, 488-499, DOI:10.1016/j.asr.2011.03.040. Kim, M., Choi, Y., Jun, H.S., Lee, J., 2015. GBAS ionospheric threat model assessment for category I operation in the Korean region, GPS Solutions, 19, 3, 443-456. Kutiev, I., Tsagouri, I., Perrone, L., Pancheva, D., Mukhtarov, P., Mikhailov, A., Lastovicka, J., Jakowski, N., Buresova, D., Blanch, E., Andonov, B., 2013. Solar activity impact on the Earth's upper atmosphere, J. Space Weather Space Climate, 3, A06, DOI:10.1051/swsc/2013.028. 25

Laštovička, J., 2013. Are trends in total electron content (TEC) really positive?, J. Geophys. Res. Space Physics, 118, 3831–3835, DOI:10.1002/jgra.50261. Lee, C.C., Liou, Y.A., Otsuka, Y., Chu, F.D., Yeh, T.K., Hoshinoo, K., Matunaga, K., 2008. Nighttime medium-scale traveling ionospheric disturbances detected by network GPS receivers in Taiwan, J. Geophys. Res.. Space Phys., 113, A12. Levin B.W., Sasorova, E.V., 2012. Seismicity of the Pacific Region: global feature detection. Moscow, Janus-K, 307p. (in Russian). Li, M., Parrot, M., 2013. Statistical analysis of an ionospheric parameter as a base for earthquake prediction. J. Geophys. Res., Space Phys., 118(6), pp.3731-3739. Liu, J.Y., Chen, Y.I., Chuo, Y.J., Chen, C.S., 2006. A statistical investigation of preearthquake ionospheric anomaly. J. Geophys. Res., 111, A05304, DOI: 10.1029/2005JA011333. Liu, J.Y., Chen, C.H., Lin, C.H., Tsai, H.F., Chen, C.H., Kamogawa, M., 2011. Ionospheric disturbances triggered by the 11 March 2011 M9.0 Tohoku earthquake. J. Geophys. Res., 116, A06319. DOI: 10.1029/2011JA016761. Lognonné, P., Artru, J., Garcia, R., Crespon, F., Ducic, V., Jeansou, E., Occhipinti, G., Helbert, J., Moreaux, G. & Godet, P.E. 2006, Ground-based GPS imaging of ionospheric post-seismic signal, Planet. Space Sci., 54, 5, 528-540. Mannucci, A.J., Wilson, B.D., Yuan, D.N., Ho, C.M., Lindqwister, U.J., Runge, T.F., 1998. A global mapping technique for GPS-derived ionospheric TEC measurements. Radio Sci., 33, 565–582. Mukhtarov, P., Andonov, B., Pancheva, D., 2013. Global empirical model of TEC response to geomagnetic activity. J. Geophys. Res., Space Phys. 118, 6666-6685. DOI:10.1002/jgra.50576. NCEDC, 2014. Northern California Earthquake Data Center. UC Berkeley Seismological Laboratory, Dataset. DOI: 10.7932/NCEDC. Němec, F., Santolík, O. & Parrot, M., 2009, Decrease of intensity of ELF/VLF waves observed in the upper ionosphere close to earthquakes: A statistical study. J. Geophys. Res., Space Phys., 114(A4). 26

Pulinets, S.A., Liu, J., 2004. Ionospheric variability unrelated to solar and geomagnetic activity/ Adv. Space Res., 34, 1926-1933, DOI:10.1016/j.asr.2004.06.014. Ryu, K., Oyama, K-I., Bankov, L., Chen, C-H., Devi, M., Liu, H., Liu, J-Y., 2016. Precursory enhancement of EIA in the morning sector: Contribution from mid-latitude large earthquakes in the north-east Asian region. Adv. Space Res., 57, 1, 268-280, DOI: 10.1016/j.asr.2015.08.030. Stankov, S.M., Warnant, R. & Stegen, K. 2009, Trans-ionospheric GPS signal delay gradients observed over mid-latitude Europe during the geomagnetic storms of October-November 2003, Adv. Space Res., 43, 9, 1314-1324. Stanislawska, I., Juchnikowski, G., Cander, Lj., 1996. The Kriging method of ionospheric parameter foF2 instantaneous mapping. Annali Geofis., 39, 4, 845-852. Stanislawska, I., Juchnikowski, G., Hanbaba, R., Rothkael, H., Sole, G., Zbyszynski, Z., 2000. COST 251 recommended instantaneous mapping model of ionospheric characteristics — PLES, Phys. Chem. Earth, Part C: Solar, Terrestrial & Planetary Science, 25, 4, 291-294, DOI:10.1016/S14641917(00)00019-2. Stanislawska I., Gulyaeva T.L., 2015. Ionospheric W index based on GNSS TEC in the operational use for navigation systems. Satellite Positioning: methods, models and applications, Chapter 7, 131-148, DOI:10.5772/59902. Sugiura, M., 1963. Hourly values of equatorial Dst for IGY, NASA Rept. X-611-63-131, GSFC. Greenbelt, Maryland. Suresh, S., Dashora, N., 2016. Climatological response of Indian low-latitude ionosphere to geomagnetic storms, J. Geophys. Res. Space Physics, 121, 4880–4897, doi:10.1002/2015JA022004. Tsugawa, T., Otsuka, Y., Coster, A. J. & Saito, A. 2007, Medium scale traveling ionospheric disturbances detected with dense and wide TEC maps over North America, Geophys. Res. Lett., 34, 22. Wrenn, G.L., 1987. Time-weighted accumulations ap() and kp(). J. Geophys. Res. 92, 10125-10129. 27

Yadav, S., Pallamraju, D., 2015. On the coupled interactions between ring current intensity and highlatitude ionospheric electron density variations. J. Atmos. Solar-Terr. Phys., 125, 50-58, DOI:10.1016/j.jastp.2015.02.006. Xia, C., Wang, Q., Yu, T., Xu, G., Yang, S., 2011. Variations of ionospheric total electron content before three strong earthquakes in the Qinghai-Tibet region. Adv. Space Res., 47, 506-514, DOI: 10.1016/j.asr.2010.09.006. Yoon, M., Lee, J., 2014. Medium-scale traveling ionospheric disturbances in the Korean region on 10 November 2004: Potential impact on GPS-based navigation systems, Space Weather, 12, 173–186, doi:10.1002/2013SW001002.

Table 1 Occurrence of earthquakes (EQ) M6+ under geomagnetic quiet conditions and storms between 1953 and2015. Earthquakes are grouped according to the three classes of epicenter depth, D, namely: D170 km; 70
Quiet conditions Storm conditions Total D1 D2 D3 Total~% D1 D2 D3 Total~% EQ-g, n1 4299 930 407 5636~80% 1085 234 112 1431~20% 7067 94 861~20% 4247 EQ-a, n2 2459 585 342 3386~80% 618 149 60.1 57.0 63.7 83.9 60.2 60.1 EQ-a,% 57.2 62.9 84.0

Table 2 The mean energy (E, in J), of earthquakes in the EIA belt within ±40 of magnetic latitude for four ranges of Richter magnitude (M), under geomagnetic quiet and storm conditions. E1(n1) for EQ between 160 E and 270 E in magnetic longitude, E2(n2) for the rest EIA longitudes, where n1 and n2 indicate number of EQ events between 1953 and 2015.

M

n1

Quiet conditions E1 n2

E2 28

n1

Geomagnetic storms E1 n2 E2

6.0:6.5 2314 1.18E+14 867 6.5:7.0 723 6.90E+14 301 7.0:7.5 246 3.63E+15 87 7.5:9.0 103 6.35E+16 42 Total 3386 2.42E+15 1297

1.21E+14 609 1.16E+14 211 6.72E+14 168 6.67E+14 81 3.80E+15 56 3.66E+15 34 5.72E+16 28 1.30E+17 11 2.34E+15 861 4.69E+15 337

1.21E+14 6.54E+14 3.44E+15 2.86E+16 1.51E+15

Table 3 RMS (MHz) deviation of foF2 (in MHz) during 1 h after EQ from median  for IRI-Plas (RMS1) and PRIME (RMS2) maps for the EIA area within 1,000 km radius around EQ hypocenter.

1 2 3 4 5 6

RMS1 0.0032 0.0049 0.0041 0.0056 0.0054 0.0056

Quiet Storm Conditions RMS2 EQs RMS1 RMS2 EQs 0.0038 1327 0.0107 0.0096 300 Total 0.0056 664 0.0162 0.0145 155 Daytime 0.0051 663 0.0140 0.0126 145 Nighttime 0.0065 398 0.0204 0.0141 79 Winter 0.0060 475 0.0155 0.0150 143 Equinox 0.0070 454 0.0219 0.0205 78 Summer

Figure Captions.

Figure 1. The zones of enhanced seismic activity for earthquakes M6+, in percent, relative to the total number of earthquakes for 1953-2015 in geomagnetic coordinates frame. (a) Earthquakes under quiet space weather conditions; (b) Earthquakes during the geomagnetic storms. EIA area of enhanced seismicity within ±40 of geomagnetic latitude at the geographic longitudes from 90 E to 190 E is indicated with thin lines. The crosses denote the EQ hypocentres along the tectonic plate boundaries. Figure 2. Parametric maps of V(foF2) index derived from PRIME-foF2 maps for the earthquake at Honshu, Japan, with M7.1, on 9.08.2009 at 10:55:55 UT (a) -2 h prior EQ at 09:00 UT; (b) -1 h prior EQ at 10:00 UT, with the EQ surrounding area of 1,000 km radius indicated by points; (c) +4min after EQ at 11:00 UT;(d) 1 h after EQ at 12:00 UT. EQ-dense EIA area borders are indicated with thin lines at magnetic latitudes ±40 and geographic longitudes 90 to 180 E. Earthquake hypocenter is shown with white star. Figure 3. Histogram of annual RMS deviation of foF2 at EQs from median foF2 (IRI-Plas, in white, PRIME, in green) under quiet geomagnetic conditions (left) and storms (right). 12-month smoothed 29

monthly solar radio flux, F10.712 is indicated with a dashed blue line. The number of annual EQs is provided in the bottom panels. Figure 4. Efficiency of the seismic impact on the ionosphere for 12 h after EQ occurrence with DfoF2 disturbances for nighttime, daytime and the total data set under quiet conditions and during the geomagnetic storms. Figure 5. Efficiency of the seismic impact on the ionosphere for 12 h after EQ occurrence with DfoF2 disturbances for three classes of depth D1, D2, D3 at EQ hypocenter, for nighttime, daytime and the total data set and for the EQ under quiet conditions and during the geomagnetic storms.

Fig.1a

30

Fig. 1b

Fig. 2

31

Fig. 3

Fig. 4

32

Fig. 5

33