Anomalous geoelectrical and geomagnetic signals observed at Southern Boso Peninsula, Japan

Geoelectrical and geomagnetic fluctuations are considered the end product of several geophysical phenomena. In particular these signals measured in seismically active areas can be attributed to stress and strain changes associated with earthquakes. The complexity of this problem has suggested the development of advanced sophysticated methods to investigate the heterogeneous nature of these fluctuations. In this paper, we analyzed the time dynamics of short-term variability of geoelectrical potential differences and geomagnetic fields obsereved at Kiyosumi (KYS), Uchiura (UCU), and Fudago (FDG) stations, located in the southern part of Boso Peninsula, one of the most seismically active areas in Japan. Anomalous changes in electric and magnetic fields are obeserved in mid-night on October 6, 2002. the anomalous signals observed on October 6, 2002 are different from those originated from the train and other cultural noises according to the investigation on preferred directions of geoelectric field. The investigation of simaltaneous geomagnetic field changes suggest that the source of the electromagnetic change might be generated by underground current because of the polarity pattern oberved at KYS, UCU and FDG. Therefore, electrokinetic assumption under the ground seems one of the possible solutions for the generation of anomalous signals. It is important to understand the ULF electromagnetic environment for the study on the preparation process of crustal activity and systematic understanding both electromagnetic and seismic phenomena. Mailing address: Dr. Katsumi Hattori, Graduate School of Science, Chiba University, 1-33, Yayoicho, Inage-ku, Chiba 263-8522, Japan; e-mail: hattori@earth.s.chiba-u.ac.jp () Now at: Marine Works Japan Ltd., Japan.

Figure 1 shows a map of our ULF geoelectromagnetic stations at Boso Peninsula.This region is located in one of the most active seismic zones in Japan.There is a geoelectromagnetic sensor array with intersensor distance of about 5 km.Torsion-type magnetometers with three components and two horizontal geoelectric fields are measured by pairs of electrodes, whose distance is about 30-50 m (Hattori et al., 2004b) are in operation.
The observed ULF geomagnetic and geoelectric potential data are consist of 1) signals originated from the external source field associated with the solar-terrestrial interactions such as geomagnetic pulsations and geomagnetic storms, and their inductive field, which appears simultaneously in the global (hundreds of km) scale; 2) the regional (a few tens of km) signals such as artificial noises associated with the leakage current from DC-driven trains, and earthquake-related signals; and 3) local (less than a few kms) signals around the sensors.The signals associated with the crustal activity are very weak in general, and therefore the signal separation is of critical importance.
In this paper, we apply Interstation Transfer Function (ISTF) (Harada et al., 2004a,b) method to the observed ULF geomagnetic and geoelectrical potential data to remove global variation mentioned above.We use geomagnetic data observed at Kakioka Magnetic Observatory, hereafter KAK, (26.23°N, 140.19°E),Japan Meteorological Agency, as the remote reference data.We found anomalous variations in ULF geomagnetic and geoelectrical potential data in midnight on October 6, 2002 (JST).We also investigated the direction of signal arrival with the use of geoelectrical potential data observed at local midnight in order to identify the chracteristics of signals among background noises, intense transient artificial noises such as the leak currunt of DC driven trains, and the other signals using the potential gradient and we have analyzed signal characteristics at stations.

ULF electromagnetic observation
Figure 1 shows the map of stations at Southern Boso Peninsula.There is an array network with intersensor distance of about 5 km.These stations are Uchiura (UCU), Kiyosumi (KYS), and Fudago (FDG).At each site, three components of geomagnetic field and two horizontal geoelectrical potential difference components are recorded with 50 Hz sampling (Hattori et al., 2004b).The clock is synchronized by GPS.At the reference site KAK, three geomagnetic fields are measured with 1 Hz sampling rate.Therefore, the data down sampled to 1 Hz have been used in this paper.Figure 2a-c indicates the configuration map of three stations.
The anomalous variations in geomagnetic and geoelectrical potential difference data are found at midnight.Figure 3a-f shows geomagnetic and geoelectric variations in the midnight during 00 h-04 h LT.For geomagnetic data (fig.3a-c), the reference data are also plotted and it is smooth, on the other hand lower three curves simultaneously have transient variations in each component.Figure 3d-f shows geoelectrical potential differences at FDG, UCU, and KYS station, respectively.The strange transient changes in geomagnetic and geoelectrical potential difference data detected simultaneously.Around the station, intense artificial noises are originated from DCdriven train as shown in fig. 1, no DC-driven trains were passing through around the stations at this moment.These strange signals seldom observed at three stations simultaneously in midnight (after 01:00-03:00 JST).There are only seven times by eyes for analyzed four years data from 2000 to 2003.

Removing global external variations with InterStation Transfer Function (ISTF)
In order to enhance the strange variation clearly, the InterStation Transfer Function (ISTF) method with wavelet transform has been performed (Harada et al., 2004a,b).The ISTF meth- terrestrial interaction and their ideal inductive variations of the geoelectrical potential differences at the site using the reference data.Therefore, the residuals between the observed and estimated variations at the site provide only regional and local variations and earthquake-related phenomena are included.We use geomagnetic data observed at Kakioka Magnetic Observatory, Japan Meteorological Agency, as the remote reference data.
We applied the ISTF method to the data shown in fig.3a-f.In computation, we use the ISTFs estimated in Harada et al. (2004a) for geomagnetic components and those for electromagnetic component for Harada et al. (2004b).The results at KYS station are shown in fig.4a-e, where panel a, b and c-e are corresponding to the geoelectrical potential and geomagnetic variations, respectively.The strange change is well understood after reducing the global external variation.Especially, it is obvious in geomagnetic variation.Similar results have been obtained for UCU and FDG.

Direction finding of geoelectrical potential data and polarity of geomagnetic data
Baselines for potential measurements are not oriented to the real north-south and east-west direction as shown in fig.2a-c.Therefore using an adequate rotation, observed values are projected to the actual NS and EW directions and this computation gives the gradient of potential or preferred orientation of the geoelectric field.The procedure is formulated by following equations: .V x and Vy correspond to the geoelectric fields of EW and NS component, resepectively.Ex and Ey mean the observed electric fields (potential/ baseline length) at a site.θ and ω are angles between the baselines and north direction.After plotting a Vx-Vy map, the distribution of observed Vx-Vy shows a not random but an elliptical pattern, in general.The direction of the line of apsides indicates the apparent direction of signal arrival around the site at the time.A geoelectric field projected to an arbitary direction φ from the north direction is given by φ is corresponding to the most sensitive direction.This approach can be used for a noise reduction method.After the prefered direction of known signals such as a leak current of DC driven trains is determined, the observed data are projected to the orthogonal to the preferred direction and less fluctuation data set is reproduced, that is SNR is improved (Nagao et al., 2000;Uyeda et al., 2000).
The direction φ also shows the steepest gradient of additive potential difference in a short period.That is, if we assume the additive potential is generated by a point charge, the direction φ indicates the lication of the point charge.
In this paper, directions of geoelectrical signal arrival is investigated based on the latter concept at first. Figure 5a The characteristics of background noises at each station are also evaluated.Figure 9a-c shows the result and the direction seems to be oriented to the nearby village, radio wave equipment antenna and/or along the geographical condition such as a local valley, a stream or a river.Furthermore, we pay attention to polarities of geomagnetic changes at three stations.As shown in fig.6a,b, the changes in geoelectrical potential differences and geomagnetic fields record exactly same time.The geomagnetic data at Kanozan station (KNZ), which is operated by Geological Survey Institute Japan and is located at 20 km northern part from our stations,

Discussion and concluding remarks
In this paper, the time series variation of geoelectrical potential and geomagnetic data was measured at KYS, UCU, and FDG, located at the southern part of Boso Peninsula, which is one of the most seismically active areas in Japan.
The directions of signal arrival were investigated to understand background and anomalous behaviors.External intense signal reduction with use of the interstation transfer function method have been also examined.These analyses have shown the capability to discriminate the signal sources.Especially, direction finding analysis of time series data has shown the effective source separation.
It is safe to say that the anomalous signals observed on October 6, 2002 are different from those originated from the train and other cultural noises according to the investigation on preferred directions of geoelectric field.Although the effective direction finding method does not exist for ULF geomagnetic field, the investigation on amplitude for quasi-static fields provides the consistent.
The investigation of simultaneous geomagnetic field changes suggest that the source of the electromagnetic change might be generated by underground current because of the polarity pattern oberved at KYS, UCU and FDG.Therefore, electrokinetic assumption under the ground seems one of the possible solutions for the generation of anomalous signals.In order to evaluate this hypothesis, the electrical underground structure around the stations should be determined.The observed signals at three stations should be explained clearly from the point of waveforms (amplitude and phase) by means of model computations based on the structure and assumed source region.
The anomalous variations in geoelectrical potential differences and geomagentic fields observed on October 6, 2002 during the active period of 2002 Boso slow slip event (Ozawa et al., 2003).Figure 12a,b shows the map of our sta- This event was recorded by GPS deformation network.Figure 3a-f shows the results of GPS deformation measurement and there is a few cm displacement in early October.It is reported that the estimated displacement at the source region under the ground could be about 10 cm and the converted magnitude could be M w 6.6 (Ozawa et al., 2003).This crustal activity is very unique because it is large and it occurred just below our small array network station.In addition, there starts a small swarm activity at the edge of the slow slip region on October 6, as shown fig.12a,b.
It is not clear but there is a possibility that the anomalous signals might correlate with the slow slip event.Because the observation has been in operation since February 2000.Although daytime data are not used for the analysis because of contamination of train noises unfortunately, nighttime data are good in SN.We could not identify any similar change with large amplityde even at nighttime.The slow slip event is reported that the converted magnitude is greater than 6 and our station is just above the source region.During the active period of the slow slip event the stress might be complicated in the region and this may cause the unusual hydrological system.Then, the anomalous current is driven and anomalous signals due to electrokinetic scheme might be detected at the stations.
To identify waveform and/or to estimate direction of arrival are an important role in monitoring the ULF electromagnetic environment in seismic areas and for understanding the preparation process of crustal activity.

Fig. 1 .
Fig. 1.The map of geoelectric and geomagnetic stations in Southern Boso Peninsula, Japan.KYS, UCU, and FDG indicate the stations.The railway lines are also shown.

Fig. 7 .
Fig. 7.Estimated directions of arrival for the strange signals on October 6, 2002 (described in fig.6a,b), using preferred orientation investigation in geoelectrical data.The numbers correspond to those in fig.6a,b.

Fig. 8 .
Fig. 8.Estimated directions of arrival associated with DC driven train.The preferred orientation approach is adopted for geoelectrical data.The numbers correspond to those in fig. 7.

Fig
Fig. 9a-c.The estimated directions of arrival for the background noises.a) FDG; b) KYS; and c) UCU.The actual configuration of these stations is taken in account for this figure.
changes on October 6, 2002.On the other hand, there is no clear relation between train noises and data observed at Kanozan station.This result also suggest that the anomalous signals on October 6 are not train noises.The signals during the time interval 01:30-01:35 on October 6, 2006 (JST) are shown in fig.10.The polarity of vertical component in KYS station is different from the other two stations (UCU and FDG).Taking account of polarity of horizontal component, it is reasonable to assume these signals are generated under the ground due to the current flow based on the electrokinetic process.That is, the source region under the ground could be estimated inside the triangle area and the direction of the fluid flow could be indicated by an arrow as shown in fig.10.On the other hand, the polarity pattern a few hours later at three stations is presented in fig.11.The polarity is coherent among three station and the current flow under the ground may be changed as in fig.11.

Fig
Fig. 12a,b.The 2002 Boso Slow Slip event and our stations.a) The rectangular region corresponds to the source region of the 2002 Boso Slip event.Vectors indicate the data of GPS deformation and small circles are epicenters.A broken circle indicate the small swarm activity on October 6, 2002.b) The variation of GPS deformation beween Ohgata and Ohara (afterOzawa et al., 2003).