Monitoring of changes in electric field along the İznik Mekece fault in the western part of North Anatolian fault zone

Abstract

The self-potential (SP) difference between two electrodes in electrical contact with thetopsoil has been observed at two stations: Dlrazali (DRZ) and <;erke~li (CRS). These stationsare on the iznik-Mekece Fault in the western part of the North Anatolian Fault Zone (NAFZ).The choice oflocations was based on magnetic profile measurements, proximity to the iznikMekeceFault and distance from known electrical and magnetic disturbances. Observationshave been made since 1990 and these have been searched for possible anomalous changesbefore earthquakes that occurred in the study area.From October 1990 to July 1992, lead electrodes were used in both the EW and NSdirections. The electrode span was 40m and the electrodes were buried at a depth of 1m. InNovember 1992, two pairs of electrodes were set up parallel to one another but of differentlengths (40m and 80m) in both the EW and NS directions in an attempt to increase thereliability of the observed data. The new electrodes are of lead and ceramic Pb-PbC12.Throughout the measurement period we have faced problems with the maintenance of thecables connected to the electrodes and particularly with instrumental failures. Therefore, wehave some gaps in our data set. A new data-acquisition system has been designed by Varol(1994) in order to gather uninterrupted, more precise and reliable data to solve the instrumentalproblem. This system has been tested in the field for a few months and in the stage of researchand development.In order to analyze the data, first of all, wild values and discontinuities were removed anda 6-hour running mean filter was applied to the SP data. Plots of smoothed SP data were madefor an interval of 25 days before to 5 days after each earthquake that occurred within - 200kmof the observing sites during the measurement period. After examining the SP plots during theinterval before earthquakes, a common pattern was detected in some records of the DRZstation. This pattern starts 15±3 days before the earthquake, then flattens out and is followedby a decrease up to the time of the earthquake.The whole of the DRZ data have been searched for similar patterns in order to investigatewhether this pattern is indeed an earthquake precursor or not and 20 examples were found.Nine of them (45%) were followed by an earthquake and this is considered to be anunexpectedly high value.Finally, the probability of an earthquake occurring within the time intervals based on`standard deviation` of the time lag between a `sharp rise` and the following earthquake(±3days) by chance (Pc) and by prediction (pp) was calculated to estimate the significance levelof the precursor. Then it is found that the prediction exceeds chance by at least a factor of twoin all cases.