DESCRIPTION:

The April 1, 2014 M 8.2 earthquake occurred at shallow depths just offshore of the Chilean coast. The location and mechanism of the earthquake are consistent with slip on the plate boundary interface (megathrust) between the Nazca and South America plates. The Nazca plate subducts eastward beneath the South America plate at a rate of about 67 mm/yr at the latitude of the earthquake.

 

 

Figure 1: Earthquake epicenter and Dobrovolsky area.

 

 

The mainshock was preceded by nearly 2 weeks of increase seismicity. On 15 March, the area was reactivated with 11 events of ML 2.6 to 4.6, followed on 16 March by the first big foreshock of Mw 6.7 (at ~40 km south of the main shock hypocenter).

 

ANALYSES:

  1. SEISMOLOGY

 

i) Accelerated Moment Release (AMR) & Revised-AMR

AMR & R-AMR are techniques that are based on the hypothesis that when a large EQ is approaching, there is an increment of the rate of seismicity and/or the mean magnitude of EQs. The Accelerating Moment Release (AMR) method was developed by Bowman et al. (1998) and is based on the hypothesis that prior to a large earthquake the stress field in the vicinity of the next rupture increases in such a way that one can observe an acceleration of the background seismicity following a power-law function. The change in the seismicity rate produces a regional increase in the cumulative Benioff strain, which is a measure of the cumulative seismic energy of the seismic events considered prior to a mainshock.

The proposed revision of the AMR method (so called R-AMR) is based on the introduction of a “reduced” Benioff strain for the earthquakes of the seismic sequence where, for the same magnitude and after a certain distance from the main-shock epicentre, the closer the events the more they are weighted. In the revised version the ordinary expression of Benioff strain is modified as follows:

 

where G(Ri) is an appropriate attenuation function with distance Ri.

The seismic data for AMR and R-AMR analysis are extracted from USGS Catalogue starting from the 1st January 5 years before the main shock, in a circular area centered in the main-shock epicentre which radius equals the Dobrovolsky’s radius.

 

Figure 2: Results from AMR analysis, USGS catalog (Iquique EQ-Chile, April 2014).

 

Figure 3: Results from R-AMR analysis, USGS catalog (Iquique EQ-Chile, April 2014).

 

AMR is able to detect the acceleration before the main-shock but only in a very poor way, in comparison with R-AMR, as can be shown in Figure 2 and Figure 3 respectively.

The application of the R-AMR methodology provides the best results in detecting the precursory seismic acceleration, when compared with those found by ordinary AMR technique, especially when using USGS catalog.

 

 

  1. SWARM – GEOMAGNETISM

Two different approaches were developed to search EQ precursors using geomagnetic Swarm data: MASS method and Wavelet method. Both analyses are based on Level 1B MAGxLR Swarm product.

 

i) MASS algorithm (Magnetic Swarm anomaly detection by Spline analysis)

The main goal of this analysis will be the detection and identification of anomalies, i.e. an unusual behavior of the considered Swarm magnetic data (Level 1B).

The MASS algorithm selects satellite tracks in the area of interest (Dobrovolsky Area, DbA) and then differentiates the magnetic field components -in NEC frame- and scalar F data. Thus, data were interpolated by a cubic spline and the residual between original track and spline was computed. The algorithm produces plots of both vectorial and scalar components of the Earth magnetic field, projected on the geographical map (with satellite track, EQ epicentre and DbA); see Figure 4.

The MASS algorithm performs also an automatic research of anomalies estimating the Root Mean Square (RMS) of the track between -50° and + 50° geomagnetic latitude. The data recorded at high latitudes are cut to avoid auroral effects. It also computes the rms in a small moving window of 3.0° latitude-width and compares it with the RMS of the whole track. An output file with the indication of how many windows have rms>kt*RMS (with kt=threshold) for each track. A filter on geomagnetic conditions was applied in order to select tracks characterized by quiet conditions (|Dst|≤20 nT and ap≤10 nT).

 

 

 

Figure 4: Example of anomalous track detected by MASS algorithm (Sat_ A, 5 March 2014), Iquique-Chile EQ.

 

ii) Wavelet Analysis

Wavelet analysis were used to investigate time-frequency content of geomagnetic signals, complementing time series analysis unrevealing spectral content variations.

For each day within the time interval of interest, all the tracks passing through the Dobrovolsky’s area (DbA) are selected and extracted according to the external conditions measured by the Dst and ap indices, as performed by MASS algorithm. Then the first time derivative for each component of the tracks was performed and the trend was removed by means of a spline function. Each individual residual signal is then the object of the successive wavelet analysis.

We applied the Continuous Wavelet Transform (CWT), with the Morlet wavelet, in order to characterize the signals we suspect could be related to the lithosphere-ionosphere coupling.

The Wavelet spectral analysis seems to show the existence of two different anomalous families characterized by distinctive behaviors. The first family contains two  twin signals -shown in Figure 5 -  that are clearly symmetric with respect to the magnetic equator, and, if occur in daytime, must be necessarily a product of the driving forces exerted on ionosphere particles, in particular of the “equatorial fountain”. Thus, we can exclude those kind of signals from the class of anomalies possibly related to the LAIC.

 

 

Figure 5: Example of “twin-signals” resulting from Wavelet Analysis, Chile EQ.

 

 

Figure 6: Example of “single short impulse” resulting from Wavelet Analysis, Chile EQ.

The second family of anomalies is characterized by single short impulses (see Figure 6). These events happen very close to the epicenter of the mainshock and at different local times (even night time, thus excluding the direct solar influence), and almost in quiet magnetic conditions. Nevertheless, the origin of those impulsive anomalies cannot be clearly explained, but the potential correlation with seismicity is worth investigating.

 

  1. SWARM – IONOSPHERE from Satellite

The Swarm data related to the study of ionospheric background were used to tune the search for earthquake-related anomalies. Satellite-based data for the ionospheric characterization of the EQ events are mainly those referred to the LP (Langmuir Probe) instrument aboard the SWARM satellites, which determine local properties of the plasma, such as temperature and density by measuring the collected current due to electrons and ions. The electron density Ne is the relevant parameter used for the ionospheric characterization of the EQ events. Two different methods were developed to analyze Swarm ionospheric electron density data: NeSTAD and NeLOG.

 

i) Method I: NeSTAD

The NeSTAD (Ne Single Track Anomaly Detection) analysis has been applied to the Swarm constellation data as described in D3203-Updating Chapter 3. In particular, LP and IBI data available in the period from 2 March to 1 May 2014 have been used to derive the track anomaly parameters. The NeSTAD has been initialized with the mild outliers mode (k=1.5) and with an “excess area” parameter equal to 0.1.

Then, to tag the interesting track anomalies for this particular event, the following criteria have been applied:

  • R>Rthr=0.85 i.e. only the strongly identified anomaly are taken into account, and standard deviation of the filtered track below σthr=0.1 or standard deviation of the filtered track above σthr=0.1, independently of R value.

Only morning tracks have been selected (02-06 LT), to remove the impact of the equatorial fountain during the day and to minimize the impact of the plasma bubble formation during night-times quiet ionospheric conditions (absolute value of Dst in the considered day not exceeding 20 nT).

Figure 7 shows the cumulative number of the tagged anomalies through the NeSTAD algorithm for Alpha (top panel), Bravo (middle panel) and Charlie (bottom panel). The black dashed line indicates the day in which the Chile M8.2 EQ  occurred. The red boxes indicate the days in which disturbed geomagnetic conditions have been recorded and that are not included in the tagging process. The two blue dashed lines indicate the time interval during which the given Swarm satellite covered the DbA of the event at morning time.

 

Figure 7 - Cumulative number of the anomalies identified through the NeSTAD algorithm for Swarm satellite Alpha (top panel), Bravo (middle panel) and Charlie (bottom panel). The black dashed line indicates the day in which the Chile M8.2 EQ occurred. The red box indicates the days in which disturbed geomagnetic conditions have been recorded. The two blue lines indicate the time interval during which the given Swarm satellite covered the DbA of the event at morning local time.

Figure 8 shows an example of tagged anomaly.

 

Figure 8 - Example of an identified anomaly with the NeSTAD algorithm and tagged as interesting for the Chile M8.2 EQ. Tagged anomaly refers to Swarm Alpha satellite, between 08:57 and 9:14 UT (morning LT) on 14 march 2014. NeSTAD parameters are reported in the top part of the figure. Orange lines indicate the position of the dip equator and of the isolines at ±15° and ±20°. The blue arrow indicates the direction of the satellite.

 

ii) Method II: NeLOG

The NeLOG method performs an automatic search of ionosphere electron density anomalies from Swarm data. The source data are the Swarm Advanced data products, in particular “Provisional Plasma dataset”, obtained by Langmuir Probes of EFI instrument.

NeLOG evaluates the logarithm of the electron density and compares each sample with a fit made by a 10th degree polynomial. A sample is classified as anomaly if the residual value exceeds a given threshold kt times the standard deviation of residual of the fit. A track is selected as anomaly track if it has more than 10 anomaly samples in DbA with the constraint of quiet geomagnetic condition (|Dst|<20 nT and ap<10 nT).

For the 2014 Chile EQ a threshold of 4.0 is selected after different analysis with threshold of 3.0, 3.5, 4.0 and 4.5. In this case study, tracks are selected within a mean local time between 22 and 6. An example of detected anomaly track is given in Figure 9.

 

 

Figure 9: Example of detected anomaly track by NeLOG (Sat A,29 March 2014) before Iquique-Chile EQ.

 

  1. IONOSPHERE Ground-based

Ionosondes

Ionospheric sounding is performed using a HF radar system known as ionosonde. The ionosonde output is usually presented in the form of a graph called ionogram. The ionospheric parameters or characteristics of typical scientific interest extracted from the ionogram are the heights, local plasma densities, and critical (reflection) frequencies of the F2, F1, and E peaks.

The ionospheric station used to analyse the Iquique-Chile EQ is the Tucumán Ionospheric Observatory (Latitude: 26.9° S, Longitude: 294.6° E, San Miguel de Tucumán, Argentina). It plays a particularly important role both for its position near the southern peak of the equatorial anomaly, and for being one of the few ionospheric monitoring stations in South America.

The ionosphere ground-based analysis for this event method is a multiparameter approach to earthquake analysis (Korsunova and Khegai - 2006, 2008). Figure 10 shows an example of ionospheric anomalies found before the earthquake occurrence on 01/04/2014.

 

Figure 10: The ionospheric anomaly of the 13.01.14 using observed Δh’Es, δfbEs, and δfoF2 variations and 3-hour ap.

In the considered day, geomagnetic conditions were mostly unsettled to quiet. Such conditions, in the Antarctic region, reached minor storm levels early in the UT day and were due to a large solar wind speed. The ap index reached its maximum value (15nT) between 03 and 06 UT, i.e. at the beginning of the day, later in the day it was very low. The AE index reached the maximum value of 136 nT. By consequence, this ionospheric anomaly seems to be not related to geomagnetic or auroral activity and can be a candidate for LAIC-related anomaly.

 

GNSS

The so-called “single station GNSS analysis” is applied to the TEC data acquired by the IGS GNSS receiver installed at Arequipa (Areq, 16.27 S, 71.29 W). GNSS TEC related products are provided to characterize the ionospheric environment during the preparation phase of the selected seismic events. In this analysis calibrated vTEC data are selected according to the geographic coordinates, time and geomagnetic conditions and interpolated by using the natural neighbours technique over a regular grid. Local TEC background maps are then computed as mean value of the vTEC in each point of the regular grid (lat, lon) of the map at each hour HH. Such background is subtracted from each actual Map(lat, lon, dd, HH, MM), to evaluate the maps of the residuals. Starting from the maps of the residuals, the detection of anomalies is performed using a threshold based on the standard deviation.

Another approach applied the method of Ensemble Empirical Mode Decomposition (EEMD) to vTEC data acquired by the igge IGS receiver from 25th March to 02nd April. Such method allows separating the oscillation components of a signal from a global trend. Subtracting the trend from vTEC (considered as a time varying signal) allows highlighting the “anomalous” behavior of the vTEC signal, excluding the effect due to the photoionization and other usual sources of vTEC variations (e.g. seasonal, time of the day, etc….).

The final step of the analysis is performed by applying the Continuous Wavelet Transform (CWT) to the oscillation component of the vTEC signal in order to construct a time/frequency representation (wavelet spectrum) of the signal.

 

 

Figure 11: Wavelet Spectrogram of the oscillation component of the vTEC signal from iqqe station from one month before the EQ.

The spectrogram highlights the presence of a large power associated with a well-defined oscillation ranging between 0.12 to 0.16 mHz (from about 100 to 140 minutes) from 16th March 2014 at 21.30 UT to 17th March 2014 at 05.30 UT confirming the effects of the foreshock on vTEC.

 

 

CONCLUSIONS:

In this section we have been resumed and described the different approaches applied in the study of earthquake precursor events.

Both Accelerated Moments Release (AMR) and R-AMR were successfully applied to 2014 Chile EQ (M8.2), and the R-AMR seems to provides better results in detecting precursory seismic acceleration with respect to ordinary AMR methodology.

The analysis of Swarm magnetic data in which long trend term was removed by cubic-spline fit gives a useful method to identify geomagnetic anomalies in Swarm data.

Finally the estimation of ionospheric background shows that Swarm and GNSS data are in agreement, confirming the possibility to use a cross-linked criteria for the detection of anomalies related to the preparatory phase of a seismic event.