IONOSPHERIC SIGNATURES OF SEISMIC EVENTS AS OBSERVED BY THE DEMETER SATELLITE

IONOSPHERIC SIGNATURES OF SEISMIC EVENTS AS OBSERVED BY THE DEMETER SATELLITE M. Parrot and F. Lefeuvre LPC2E/CNRS, 3 A Av Recherche Scientifique 45071 Orleans cedex 2 France lefeuvre@cnrs-orleans.fr URSI KEYWORDS: seismic activity, ionospheric signatures, EM wave field perturbations, ion density perturbations, electron density perturbations ABSTRACT: Observations made by the CNES/DEMETER satellite, from March 2004 to December 2010, are used to evaluate the efficiency of ionospheric signatures of seismic events. Ionospheric signatures are defined. They are applied to observations made for a series of Chile earthquakes including the 27 February 2010 one (M 8.8). Case studies and statistical studies are used to point out precursor signals. Points to be solved to get reliable precursory signals are discussed. 1. INTRODUCTION Since the beginning of the eighties, scientists are looking for a way to predict seismic events from perturbations in the EM (electromagnetic) wave field measurements (generally below ~20 khz) or/and perturbations in the plasma parameters of the ionosphere. The EM wave field measurements are conducted at ground and/or on-board satellites. The plasma parameters measurements are generally performed on low altitude satellites. Strictly speaking, those perturbations are not ionospheric signatures of seismic events. Several physical mechanisms or/and noise sources may produce most of them. However, provided proper checks are made, the apparent relevance of numerous published observations (see for instance Molchanov and Hayakawa, 2008; Hayakawa et al. 2010) seems to indicate that they cannot be pure coincidence. Up to the recent years, the lack of a large and reliable data base prevented to draw clear conclusions about possible associations with earthquakes of perturbations in the EM wave fields and ionospheric plasma parameters. However, with the multiplication of VLF/LF ground-based stations and with the launch in 2004 of the CNES DEMETER satellite (Parrot, 2006), the situation is becoming clearer. The present paper deals with wave and plasma measurements performed on-board the DEMETER satellite only. Observations were made from August 2004 to December 2010 (the satellite operations were stopped on 9 December). As information, from August 2004 to October 2009, observations were performed above 17366 earthquakes with magnitudes larger than or equal to 4.8. After all observations have been be analyzed, one may expect a data base of 1600 to 1700 such events. The aim of the present paper is to give an idea of studies in progress in the area. The plan of the paper is as follows. The DEMETER data set is presented in section 2. Ionospheric events often used as ionospheric signatures of seismic events are defined in section 3. Section 4 is devoted to observations made above a series of Chile earthquakes including the 27 February 2010 one (M 8.8) Points to be solved to get reliable precursory signals are discussed in section 5. 2. DATA SET The DEMETER (Detection of Electro- Magnetic Emissions Transmitted from

Earthquake Regions) satellite was assigned three main scientific objectives: (i) to study the ionospheric disturbances in relation to the seismic and to volcanic activities, (ii) to survey the ionospheric disturbances in relation with the anthropogenic activity, and to contribute to the understanding of the generation mechanism of these disturbances, (iii) to study the Earth electromagnetic environment. One is concerned here only by the first objective. The satellite was launched on June 29 2004, on a quasi Sun-synchronous polar orbit, with an inclination of 98.23 and an altitude of about 710 km. The altitude was changed to about 660 km in December 2005. The operations were stopped on 9 December 2010. The satellite performed 14 orbits per day and measured continuously between -65 and +65 of invariant latitude. Due to its specific orbit, DEMETER was always located either shortly before the local noon (~ 10:30 Local Time) or the local midnight (~ 22:30 Local Time). The elements of the scientific payload used for identifying ionospheric signatures of seismic and volcanic activities are described in Berthelier et al. (2006a, 2006b), Lebreton et al. (2006) and Parrot et al. (2006). They consist of: three electric (ICE) and magnetic (IMSC) sensors covering the frequency rang running from ~ DC to 18 khz, a Langmuir probe (ISL) used to determine the local electron temperature (Te) and the local electron density (Ne) an ion analyzer (IAP) used to determine the local temperatures (Ti) and densities (Ni) of the H +, He + and O + ions. The satellite had two main modes of operation: a burst mode, which provided waveform data, and a survey mode which provided averaged power spectra. 3. IONOSPHERIC EVENTS USED AS IONOSPHERIC SIGNATURES An example of ionospheric events used as ionospheric signatures is given in figure 1. The data were recorded on November 19, 2007 between 05:13:00 and 05:21:00 UT, i.e. 10 days before the 29 November 2007 earthquake occurring in the Chile area (lat -36 85, long 262 35) at 03:26:23 UT (M 6.3, depth 10 km). Going from the top panel to the bottom panel, the following signatures are observed. (a) ELF wave field - Spectrogram of an electric field component up to 2 khz. ELF waves are observed around ~ 05:17:50 UT. Probably due to the sensor sensitivity, no observations are made on the magnetic components. (b) ULF waveform The perturbation at ~ 05:17:50 UT shows that the ELF event is a ULF/ELF event, and that there is probably a second one at ~ 50 s later, (c) Electronic density (Ne) A perturbation is observed around ~ 05:17:50 UT. (d) Electron temperature (Te) No clear observation. (e) Ion density (Ni) Pic in O+ density at ~ 05:17:50 UT (f) Ion temperature (Ti) a minimum is found at ~ 05:17:50 UT The (g) panel provides the space and time positions of the earthquakes epicentres which occur close to the track of the satellite orbit. The red triangle at 05:17:22 UT indicates the closest approach to the earthquake epicentre.

(a) (a) (b) (b) (c) (c) (d) (d) (e) (e) (f) (f) (g) (g) Figure 1 ionospheric signatures (see section 3) It must be noted, that, when the DEMETER burst mode is on, the simultaneous measurement of the ELF wave field components allows to perform inverse ray-

tracings, and so to check whether wave field observations are generated close to the seismic region or not (Santolik et al., 2006). 4. ANALYSES OF A SERIES OF EARTHQUAKES Two different studies of ionospheric perturbations related to Chile earthquakes are reported here. The first one (Zhang et al. 2011) is relative to a series of 45 Chile earthquakes, of magnitude larger than 6.0, observed by the DEMETER satellite from 8 August 2004 and 26 March 2010.The second one (Parrot et al., 2011) is focused on the 27 February 2010 Chile earthquake of magnitude 8.8. It must be noted that, due to an operational problem the DEMETER burst mode was off for most of the events used for the two analyses. As a consequence the waveforms are averaged onboard and only two wave field components are transmitted to the ground. This both limits the sensitivity of the wave parameters and prevents inverse ray tracings allowing to check if the EM field received on board came from a seismic region or not. The Zhang et al. (2011) study was focussed on the investigation of the ULF/ELF electric field data. However, with the DEMETER survey mode, the only available data were the VLF spectrum data (f > 125 Hz) and the ULF electric field (F < 20 Hz). Taking into account of the fractal structure of the electromagnetic signals before strong earthquakes (see Immura et al., 2010 and references therein) the authors searched for ionospheric anomalies using a new parameter, called LEFI (Logarithm Electric Field Intensity), and derived from ULF/ELF electric field perturbations. After having processed the local nigh time data during 5 days (including 3days before and 1 day after) around all earthquakes with magnitude above 6.0, they found that precursory anomalies were recorded before 2/3 of the earthquakes group. Moreover they pointed out an evolutional feature, which appeared in the outer area in the beginning, and gradually moved towards the central area. The Parrot et al. (2011) study was conducted from the full set of available DEMETER data (ULF/ELF, Ne, Te, Ni, Ti). The authors searched for precursory signals from 7 February 2010, i.e. 20 days before the M 8.8 27 February 2010 Chile earthquake. They found non negligible perturbations in the ionospheric parameters from 20 days to 10 days before the earthquake, the most important ones being observed on 14, 15 and 18 February. The most stable ionospheric signature was the one observed on Ni. The existence of a 10 to 20 days precursor signal was confirmed by a statistical study of latitude variations versus the triggering of the precursory anomalies in ion density, for 2007, 2008, 2009 and 2010. It must be noted that long term precursors are also observed on the 29 September 2009 Samoa earthquake (M 8.1). This was also the case for the 26 December 2004 Sumatra earthquake (M 9.3) for which, due to operation problems, they are no DEMETER data at the time of the earthquake. Coming back to the 27 February 2010 event, one may note that, if the source of the EM wave fields had been well established (which supposes burst mode data, then inverse ray tracing), information contained in successive precursory signals could have been used to have an idea on the location of the future epicentre. Considering that there is a ~8 longitudinal shift between two successive satellite paths over the same geographical region, and knowing the geographical position of the Chile trench, one could use a triangulation procedure to identify a potential epicentre region. However, the satellite crossing epicentre regions during ~3 minutes/day only, one cannot be sure to have the most relevant information. 5. DISCUSSION The main result reported here is probably the existence of a 10 to 20 days precursory signal derived from perturbations in the ion density. To confirm it, one may either do the same type of analyses for other major earthquakes or analyze in a more detailed way the statistical study of latitude variations versus the triggering of the precursory anomalies in ion density. Although the survey mode ULF/ELF data are not ideal to look for precursory signals, one could also use the Zhang et al. (2011) approach to check if the 10 to 20 days precursory signal is also seen on the wave data.

The Zhang et al. (2011) detection of precursory anomalies before 2/3 of the earthquakes group, observed over 5 days (including 3days before and 1 day after) is also a very interesting result. However, analyses over long time periods are needed to see if they may be interpreted in terms of short term or long term precursory signals. To this regards, due to time distances between a pass of the satellite above the region of a future epicentre and the triggering of the earthquakes, none of the studies reported here have allowed to check the existence of a decrease of wave intensity 0-4 hours before the time of the main shock of shallower earthquakes. This last result was pointed out by Nemec et al. (2009) using more than 3.5 years of measurements performed at about 1.7 khz. REFERENCES Berthelier, et al., 2006a. ICE, the electric field experiment on DEMETER. In: Planetary And Space Science, 54, pp. 456-471. Berthelier et al. 2006b. IAP, the thermal plasma on DEMETER. In: Planetary And Space Science, 54, pp. 487-501. Hayakawa et al.,2010. A statistical study on the correlation between lower ionospheric perturbations as seen by subionospheric VLF/LF propagation and earthquakes. In: Journal of Geophysical Research, Volume 115, Issue A9, CiteID A09305 Immamura et al., 2010. Fractal analysis of subionospheric LF propagation data and consideration of the lithosphere- atmosphere-ionosphere coupling. In: Nat. Hazards Earth Syst.Sic. 10, pp 901-906. Lebreton et al., 2006. In: Planetary And Space Science, 54, pp. 472-486. Molchanov et al., 2008. Seismo electromagnetics and related phenomena: History and latest results. Terrapub, Tokyo. Molchanov, O.A., 2011. Underlying mechanism of precursory activity from analysis of upward earthquake migration. In: Nat. Hazards Earth Syst. Sci., 11, 135-143 Nemec et al., 2009. Decrease of intensity of ELF/VLF waves observed in the upper ionosphere close to earthquake: a satellite study, Journal of Geophysical Research, 114, doi:10.1029/2008ja013972 Parrot, M., 2006. Preface Special issue of Planetary and Space Science «DEMETER». In: Planetary and Space Science, 54, pp 411-412. Parrot et al., 2006. The magnetic field experiment IMSC and its data processing on board DEMETER: Scientific objectives, description of the firs results. In: Planetary and Space Science, 54, pp 441-455. Parrot et al., 2011. Statistical analysis of the ion density measured by the satellite DEMETER in relation with the seismic activity. In: Earthquake Science, submitted. Santolik et al.,2006. Analysis methods for multi-component wave measurements on board the DEMETER spacecraft. In: Planetary and Space Science, 54, 5, pp 512-527. Zhang et al., 2011. ULF/ELF ionospheric electric field and plasma perturbations related to Chile earthquakes. In: Advances in Space Research, 47, pp 991-1000