Ultra Low Frequency ( ULF ) European multi station magnetic field analysis before and during the 2009 earthquake at L ’ Aquila regarding regional geotechnical information

This work presents ground based Ultra Low Frequency (ULF) magnetic field measurements in the frequency range from 10–15 mHz from 1 January 2008 to 14 April 2009. In this time period a strong earthquake series hit the Italian Abruzzo region around L’Aquila with the main stroke of magnitudeM = 6.3 on 6 April 2009. In the frame of the South European Geomagnetic Array (SEGMA), a European collaboration runs ULF fluxgate instruments providing continuously magnetic field data recorded in midand south Europe. The main scientific objective is the investigation of signal variations due to seismic activity and the discrimination between other natural and human influences. The SEGMA station closest to the L’Aquila earthquake epicenter is L’Aquila observatory located in the epicenter region. For the scientific analysis we extract the nighttime period from 22:00–02:00 UT and determine the power spectral density (PSD) of the horizontal ( H ) and vertical ( Z) magnetic field components and the standardized polarization ratio ( Z) over (H ). To discriminate local emissions from global geomagnetic effects, data from three SEGMA stations in distances up to 630 km from the epicenter region are analyzed and further compared to the independent global geomagnetic ∑ Kp index. Apart from indirect ionospheric effects, electromagnetic noise could be originated in the lithosphere due to tectonic mechanisms in the earthquake focus. To estimate Correspondence to: G. Prattes (gustav.prattes@oeaw.ac.at) the amplitude of assumed lithospheric electromagnetic noise emissions causing anomalies in the PSD of the ( Z) component, we consider magnetotelluric calculations of the electric crust conductivity in the L’Aquila region. Results found at L’Aquila observatory are interpreted with respect to the lithosphere electrical conductivity in the local observatory region, the ∑ Kp index, and further in a multi station analysis. Possible seismic related ULF anomalies occur ∼2 weeks before the main stroke.


Introduction
Since two decades many authors have performed measurements over a wide frequency range and applied different signal processing methods and reported about electromagnetic effects related to earthquakes.Seismic ULF phenomena possibly related to earthquakes have been studied since Fraser-Smith et al. (1990).The polarization method based on the ratio of vertical and horizontal magnetic field intensity has been introduced by Hayakawa et al. (1996) and improved by Ida et al. (2008).The standardized polarization method was used by Masci et al. (2009) to investigate seismic events in the L'Aquila region.ULF anomalies related to seismic activity in a multi station approach have been investigated by Kopytenko et al. (1993) and Hayakawa et al. (2007).
G. Prattes et al.: Ultra Low Frequency (ULF) European multi station magnetic field analysis In the frame of the SEGMA network, the basic polarization method was applied in Prattes et al. (2008).Further, Villante et al. (2010) performed a classical polarization analysis with magnetic field data recorded before and during the L'Aquila earthquake series 2009.In the present work we show results obtained with the improved standardized polarization method applied on recorded data at several geomagnetic stations belonging to the SEGMA network spread over mid-and south Europe in a time period of 15 months, including the strong seismic active and further seismic weak time periods in the area of L'Aquila.
After a short performance summary of the SEGMA fluxgate magnetometer including a geographical overview of the SEGMA chain, we focus on the L'Aquila earthquake series 2009.Concerning signal processing the standardized polarization method is applied to four SEGMA stations.The improved standardized polarization method has justified that both the B Z and the B H components can be treated equally.Results in a first five months time period of weak seismic activity in the observatories local area show that data quality of the analyzed stations is comparable with regard to background noise and human influence.In a second period the station comparison is extended to the time span of the L'Aquila earthquake series with LAQ observatory being located in the main earthquake stroke region, whereas the SEGMA stations (CST, NCK) are located in (420 and 630) km distance to L'Aquila acting as comparison stations.This temporal splitting enables one to compare electromagnetic anomalies in the ULF range emitted in the local epicenter region.
According to theory there are three main sources of ULF noise emission: (i) geomagnetic effects, (ii) man-made noise, and finally (iii) seismic activity.Geomagnetic events are expected to be observed at all stations, and the K p index gives an opportunity to compare the measurements to a planetary index.Experience shows that recorded data are less influenced by man-made noise during nighttime, so for the current analysis we extracted the four hour period from 22:00-02:00 UT.Instrumental housekeeping data has been used to eliminate days with data gaps or instrument dependent errors due to e.g.time synchronization.
To characterize magnetic field anomalies related to seismic activity in the ULF range we distinguish between two major effects: (iii 1 ).The direct electromagnetic effect.A source of wideband electromagnetic noise is assumed in the earthquake focal zone.Higher frequency emissions are more attenuated due to a lower skin depth, while the lower frequencies, especially in the ULF range, can penetrate the Earth's lithosphere.This behavior leads to an increase of the polarization ratio due to an increase of B Z .The ULF band from 10-50 mHz was found to be mainly affected, and authors report about ULF anomalies related to earthquakes around particular frequencies of 10 mHz.The measured ULF anomalies described in this work are possibly related to the direct electromagnetic effect.Observing several frequency sub bands from 10-500 mHz showed that anomalies turn out most significant in the sub band from 10-15 mHz.The source of emission is interpreted by Molchanov et al. (1998) due to microfracture electrification or the opening of an ensemble of cracks in the focal zone of the earthquake.Microfractures have sizes in the range from 10 −4 -10 −1 m −2 , and the number of cracks is estimated in the order of 10 6 -10 7 m −3 generated on a time scale of approximately microseconds.The velocity of the opening is approximately 10 3 m s −1 , which is in the range of seismic waves.Electromagnetic ULF investigation related to lithosphere processes and earthquakes have been performed by Kopytenko et al. (1994).
(iii 2 ).Indirect seismic effects: Strong seismic activity can lead to Atmospheric Gravity Waves (AGWs), leading to turbulence in the lower ionosphere.This process causes a depression of ULF waves going down from the magnetosphere and leads to a decrease of B H on the Earth's surface.

Instrument, database, and seismic conditions
In this section we describe: (i) the magnetic field fluxgate instrument called CHIMAG, (ii) the SEGMA array with respect to the L'Aquila earthquake, and (iii) the strong seismic activity in the L'Aquila region 2009.

Instrument
SEGMA has heritage from the CHInese MAGnetometer (CHIMAG) fluxgate magnetometer chain with the primary goal to investigate magnetic pulsations in the ULF range.The vital parameters of the high temporal resolution 3-axes fluxgate magnetometer are the measurement range of ±512 nT, the compensation field of 60 000 nT in X and Z, and ±30 000 nT in Y direction.The accuracy is 8 pT at a temporal resolution of 1 Hz, the highest possible sampling frequency is 64 Hz.The 3-axes magnetometer measures in X (positive northward), Y (positive eastward), and Z (positive towards the centre of the Earth) direction.The horizontal components X and Y build up as described in Eq. ( 1), (1) A detailed description of the instrument can be found in Magnes et al. (1999) andSchwingenschuh et al. (2000).

Database
Figure 1 shows a map of the SEGMA network with 5 operating stations.Indicated by yellow markers, the observatories Bulgaria, shown by a white marker, is not evaluated in the frame of this work but is introduced here for completeness as part of the SEGMA chain.SEGMA ULF data are available with high reliability during the last ten years.This multi station setup enables one to distinguish between seismic active and quiet time periods in that area.Table 1 contains the SEGMA ULF stations with geographical coordinates, corrected geomagnetic coordinates, and the distances to the L'Aquila earthquake epicenter region.

The L'Aquila earthquake series 2009
From 30 March 2009 to 9 April 2009, US Geological Survey (USGS) recorded 25 earthquakes in the region around L'Aquila within a magnitude range from 3.6 ≤ M ≤ 6.3.The main stroke was recorded on 6 April 2009, 01:32:39 UT in a depth of 10 km and geographical coordinates 42 • 33 , 13 • 33 .The disastrous earthquake series gives the opportunity to perform ULF investigations with the SEGMA chain related to seismic activity.The main stroke had the magnitude M ≥ 5, and more than 25 earthquakes belonging to the series have been recorded in the local area of the L'Aquila ULF SEGMA station in a short time period.This condition is promising to observe ULF anomalies related to seismic activity.2 show the magnitude-distance-depth distribution of the L'Aquila earthquake series.
In Table 2 all seismic events of the earthquake series with distances less than 10 km to L'Aquila earthquake epicenter and depth in the range from 5-18 km are summarized.

ULF analysis related to seismic activity
Basically, many physical processes are related to seismic activity.In this section we emphasize electromagnetic wave emission due to processes in the Earth's lithosphere.Pressure variations in the upper lithosphere lead to the generation of microfractures that are related to wideband electromagnetic emission in the earthquake focus zone.High frequency components are damped strong due to low skin depth (low pass filter function of the lithosphere).The lower frequencies can penetrate the Earth's crust without significant attenuation.So the chance to observe an electromagnetic earthquake signature is significantly increased in the ULF range than in other higher frequencies.An important requirement to study electromagnetic manifestations related to earthquakes is knowledge about the lithosphere pre-condition in the local area of the observation point.Main interest is on the electrical parameters and especially the electrical resistivity of ground layers.Palangio et al. (2007)

Signal processing methodology
We extracted the local midnight period from 22:00-02:00 UT for further analysis.This nighttime period is less influenced by man-made artificial noise than daytime periods.Each of these segments is subjected to a Fast Fourier Transform (FFT) analysis.The selected sampling frequency is f S =1 Hz, which gives a Nyquist frequency of  sub bands to extract the frequency band f = 10-15 mHz.Further, the power spectral densities (PSD) of the individual component (S H Day , S ZDay , S H Day , S ZDay ) in (nT 2 Hz −1 ) were determined as described in Eqs.

S H Day
where B H (ω) and B Z (ω) are the Fourier transformed components in the sub band f.To get better statistics the PSDs are averaged as shown in Eq. (3).
where N indicates the sample size.
To obtain monthly PSD statistics (monthly mean and standard deviation), all S H Day (ω) and S ZDay (ω) spectra contributing to a month are processed (excluding days with bad data) and averaged similar to Eq. ( 3).
The improved polarization analysis normalizes each geomagnetic field component as described in Eq. ( 4): where µ H Month and µ ZMonth are the monthly mean PSD values and σ H Month and σ ZMonth are the monthly standard deviation PSD values.The standardized polarization is determined as follows: The standardized polarization ratio increases if (i) the vertical magnetic PSD increases and/or (ii) the horizontal magnetic PSD decreases.Experience shows that polarization analysis results are robust if a significant variation of H Day is observed and H Day ≥ 0.1.

Standardized polarization analysis results
This section compares the four stations CST, RNC, NCK, and LAQ during a (i) seismic quiet time period of five months.Further, (ii) long term results for the entire 15 months and (iii) results focused on the L'Aquila seismic active period for the three stations CST, NCK, and LAQ are presented; and (iv) short term results for L'Aquila are discussed.

Summary and Outlook
The advanced standardized polarization analysis applied in this work shows a series of ULF polarization events in the frequency range from 10-15 mHz at L'Aquila observatory about 2 weeks before the Abruzzo 2009 earthquake series.The observed events are due to an increased B Z PSD at L'Aquila observatory, while the stations CST and NCK show no anomalies in this time period.Multi point results during    seismic quiet periods do not show events like this, and further long term results indicate that these events are unique during the extended time span of 15 months.We conclude that the increased B Z PSD at L'Aquila has a local source of emission and is possibly related to seismic activity in the region.The ground electrical resistivity profile has been used to estimate the skin depth for ULF variations, in particular for the 10-15 mHz range, and support an argument for ULF studies near the earthquake epicenter.The observed anomalies before the L'Aquila earthquake series lasted several days and were locally emitted in or near the L'Aquila region.They occurred during the seismic active period.The classical polarization method has not shown an evidence of polarization events in relation to seismicity reported in Villante et al. (2010).The SEGMA chain with multiple stations in various distances to L'Aquila gives the opportunity for further analysis.Apart from the direct electromagnetic effect described in this work, which could lead to ULF events related to earthquakes, the indirect seismic effect is suggested to be analyzed in the frame of SEGMA during the L'Aquila earthquake period.Further, the analyses are suggested to be extended and compared to a wider ULF frequency range up to 500 mHz and even higher frequencies, e.g. the Very Low Frequency (VLF) range (see Rozhnoi et al., 2009).Concerning ULF studies, a detailed analysis of the horizontal components X and Y (see Du et al. 2002) will be performed to find the direction of ULF emission.A more detailed background noise level and local ULF source of emission estimation could help to clarify if the emitted signals are of true seismic origin.

Figure
Figure 2 and Table2show the magnitude-distance-depth distribution of the L'Aquila earthquake series.In Table2all seismic events of the earthquake series with distances less than 10 km to L'Aquila earthquake epicenter and depth in the range from 5-18 km are summarized.

Fig. 2 .
Fig. 2. Magnitude of seismic events (top panel) and distance to L'Aquila observatory during 30 March 2009 and 9 April 2009.The occurrence of the main stroke is indicated with an arrow.The magnitude of the seismic events is shown in the upper panel, the distances of earthquakes epicenters to L'Aquila in the lower panel.

Fig. 3 .
Fig. 3. K p and standardized daily horizontal (blue) and vertical (red) PSD from 1 August 2008 to 31 December 2008 for CST, NCK, RNC, and LAQ station in a seismic quiet time period.

Fig. 4 .
Fig. 4. Five day running mean of K p , magnitude of the seismic events with the earthquake swarm in April 2009, and standardized polarization from 1 January 2008 to 14 April 2009 for CST, NCK, and LAQ station.LAQ station shows anomalies above and below 2σ eleven days prior the impending earthquake series.The earthquake series main stroke is indicated with an arrow in the second panel.

Figure 5 Fig. 5 .
Figure 5 Fig. 5. K p and standardized daily horizontal (blue) and vertical (red) PSD from 1 January 2009 to 14 April 2009 for CST, NCK, and LAQ station.Days in the colored shaded time period are probably influenced by ULF anomalies due to the impending seismic activity.The L'Aquila earthquake series main stroke is indicated with an arrow.

Figure 6 Fig. 6 .
Figure 6 Fig. 6.Top K p , magnitude of seismic events and distance to L'Aquila, standardized polarization from 1 January 2009 to 14 April 2009 and 5 day running mean of standardized polarization for CST, NCK, and LAQ station.Time periods probably affected by ULF events due to seismic activity are color shaded.

Table 1 .
reported a calculation of the ground resistivity profile for L'Aquila station by means of the single station magnetotelluric tensor evaluation.Several layers from 0-50 km depth show an increase of the electrical resistivity at d ∼ 2 km depth from SEGMA ULF chain and distances to L'Aquila erathquake.

Table 2 .
Seismic events with less than 10 km distance to L'Aquila EQ extracted from USGS.The main stroke is indicated in bold letters.

period from 1 August 2008 to 31 December 2008
Day and vertical Z Day PSD for a five month time period is depicted.Data from the observatories CST, NCK, RNC, and LAQ are analyzed from 1 August 2008 to 31 December 2008.The upper panel shows the K p geomagnetic index.The horizontal PSD is well correlated for CST, NCK, and RNC with the K p index, with the exception of 8 November 2008 to 22 December 2008 for LAQ observatory.Global geomagnetic effects causing high PSD values are found in the horizontal PSD, e.g. during the period from 30 September 2008 to 5 October 2008.Geomagnetic effects are not distinct in the vertical PSD except for the time periods 11 August 2008 to 14 August 2008, 2 September 2008 to 9 September 2008, and on 6 December 2008.The probability to observe locally emitted anomalies, e.g.artificial noise or effects in the lithosphere, is higher in the vertical PSD.The single component results show that the vertical and horizontal PSDs are comparable in terms of background noise level.