Current progress in using multiple electromagnetic indicators to determine location, time, and magnitude of earthquakes in California and Peru (Invited)

Since ultra-low frequency (ULF) magnetic anomalies were discovered prior to the 1989 Loma Prieta, Ca. M7.0 earthquake, QuakeFinder, a small R&D group based in Palo Alto California has systematically monitored ULF magnetic signals with a network of 3-axis induction magnetometers since 2000 in California. This raw magnetometer data was collected at 20-50 samples per sec., with no preprocessing, in an attempt to collect an accurate time history of electromagnetic waveforms prior to, during, and after large earthquakes within 30 km. of these sensors. Finally in October 2007, the QuakeFinder team observed a series of strange magnetic pulsations at the Alum Rock, California site, 14 days prior to M5.4 earthquake. These magnetic signals observed were relatively short, random pulsations, not continuous waveform signals like Pc1 or Pc3 micropulsations. The magnetic pulses have a characteristic uni-polar shapes and 0.5 sec. to 30 sec. durations, much longer than lightning signals. In May of 2010, very similar pulses were observed at Tacna, Peru, 13 days prior to a M6.2 earthquake, using a QuakeFinder station jointly operated under collaboration with the Catholic University in Lima Peru (PUCP). More examples of these pulsations were sought, and a historical review of older California magnetic data discovered fewer but similar pulsations occurred at the Hollister, Ca. site operated by UC Berkeley (e.g. San Juan Bautista M5.1 earthquake on August 12, 1998). Further analysis of the direction of arrival of the magnetic pulses showed an interesting “azimuth clustering” observed in both Alum Rock, Ca. and Tacna, Peru data. The complete time series of the Alum Rock data allowed the team to analyze subsequent changes observed in magnetometer “filter banks” (0.001 Hz to 10 Hz filter bands, similar to those used by Fraser-Smith in 1989), but this time using time-adjusted limits based on time of day, time of year, Kp, and site background noise. These site-customized limits showed similar increases in 30 minute averaged energy excursions, but the 30 minute averages had a disadvantage in that they reduced the signal to noise ratio over the individual pulse counting method. In other electromagnetic monitoring methods, air conductivity instrumentation showed major changes in positive air-borne ions observed near the Alum Rock and Tacna sites, peaking during the 24 hours prior to the earthquake. The use of GOES (geosynchronous) satellite infra red (IR) data showed that an unusual apparent “night time heating” occurred in an extended area within 40+ km. of the Alum Rock site, and this IR signature peaked around the time of the magnetic pulse count peak. The combination of these 3 indicators (magnetic pulse counts, air conductivity, and IR night time heating) may be the start in determining the time (within 1-2 weeks), location (within 20-40km) and magnitude (within +/- 1 increment of Richter magnitude) of earthquake greater than M5.4