Enhanced ULF radiation observed by DEMETER two months around the strong 2010 Haiti earthquake

In this paper we study the energy of ULF electromagnetic waves that have been recorded by the satellite DEMETER, during its passing over Haiti before and after a destructive earthquake. This earthquake occurred on 12/1/2010, at geographic Latitude 18.46o and Longitude 287.47o, with Magnitude 7.0 R. Specifically, we are focusing on the variations of energy of Ez-electric field component concerning a time period of 100 days before and 50 days after the strong earthquake. In order to study these variations, we developed a novel method that can be divided in two stages: first we filter the signal keeping only the very low frequencies and afterwards we eliminate its trend using techniques of Singular Spectrum Analysis, combined with a third-degree polynomial filter. As it is shown, a significant increase in energy is observed for the time interval of 30 days before the strong earthquake. This result clearly indicates that the change in the energy of ULF electromagnetic waves could be related to strong precursory earthquake phenomena. Moreover, changes in energy were also observed 25 days after the strong earthquake associated with strong aftershock activity. Finally, we present results concerning the comparison in changes in Energy during night and day passes of the satellite over Haiti, which showed differences in the mean energy values, but similar results as far as the rate of energy change is concerned.

Most papers have been devoted to studies of ELF / VLF waves (Parrot, 1983;Gokhberg et al., 1983;Larkina et al., 1983Larkina et al., , 1989Parrot et al., 1985Parrot et al., , 1989 , 1992). For these reasons we chose to focus on space based ULF EM emissions to study their energy changes during a long period before and after a strong EQ. Our study was based on the analysis of measurements from the DEMETER satellite around the deadly earthquake of Haiti on January 12, 2010.

Data analysis and results
For the estimation of ULF signals we used data derived from DEMETER's satellite  there is no significant variation in the waveform corresponding to the period of seismic quiescence. Therefore, we assume that the anomaly in the waveform (Fig.2a) which we are interested in, can be attributed to a precursory earthquake signal.
In order to estimate the energy of the possible pre-Earthquake signal we focus on the signal shown in Figure 3a, which corresponds to the perturbed waveform (Fig.   2a). On this signal a low pass frequency filter is applied, keeping frequencies lower  In order to further highlight these increases in the mean energy, we estimate the mean value of energy of pre-earthquake signals per 25 days as it shown in Figure   4b. We observe that for the time interval 50-100 days before the main event, the mean    Figure 6 presents results of the mean value of energy per 25 days for the preearthquake signals that are recorded by the DEMETER satellite during night (Fig. 6a) and day passing (Fig. 6b) over Haiti. Figure 6a is the same as Figure 4a and has already been described. Figure 6b shows the mean value of energy for the signals recorded by DEMETER satellite during its day passing (Down Orbits) over Haiti for The comparison of Figures 6a, 6b shows that the energy perturbation of preseismic day-passing signals is much smaller than night-passing ones. As it is found for the time interval of 100 days before the earthquake the mean value of energy during the day-passing is 0.005 (mV/m) 2 while for the night-passing is 0.13 (mV/m) 2 , namely 26 times greater. We think that this difference could be due to the strong ionization of the Ionosphere during day time that causes great attenuation in the preseismic signals. However, the results in Figures 6a and 6b are very similar as far as the rate of energy change is concerned. In both figures, the energy of the preearthquake signals increases at the same rate regarding the time before the earthquake.
This result indicates the efficiency of the applied method since it can reveal precursory phenomena in data concerning the day orbits, although the energy changes are very weak. Figure 7a represents the average energy of the pre-earthquake signals recorded by the satellite during night-passing, for 100 days before the main earthquake as well as for the aftershock signals for 50 days. As we can see, there is a significant increase in the mean energy recorded a month before the main earthquake which remains at the same levels approximately for 25 days after, while consecutively decreases gradually.
In Figure 3a). For filtering we used Singular Spectrum Analysis, combined with a third-degree polynomial filter. c) Τhe square power of pre-Earthquake signal shown in Fig. 3b.   Figure 5b is similar to Figure 4b, but in this case we have replaced the values of energy that are greater than 0.1 (mV/m) 2 with zero. As it is shown in this figure the most significant change of energy is observed for the time interval of 0 -25 days before the earthquake. Figure 6. Results of the mean value of energy per 25 days for the pre-earthquake signals that are recorded by the DEMETER satellite during night (Fig. 6a) and day passing (Fig. 6b) over Haiti. In both figures, the energy of the pre-earthquake signals increases at the same rate regarding the time before the earthquake. Figure 7. a) The average energy of the pre-earthquake signals recorded by the satellite during night-passing, for 100 days before the main earthquake as well as for the aftershock signals for 50 days. b) The average energy per 25 days of the observed signal during the night, for 100 days before and 50 days after the earthquake.