Regional applicability of earthquake forecasts using geoelectric statistical moments: Application to Kakioka, Japan

Electromagnetic anomalies have become promising for short-term earthquake forecasting. One forecasting algorithm based on statistical moments of geoelectric data was developed and applied in Taiwan. The objective of our research was to investigate such a reliable, rigorously testable algorithm to issue earthquake forecasts. We tested the applicability of the forecasting algorithm and applied it to geoelectric data and an earthquake catalog in Kakioka, Japan with a long-term period of 26 years. We calculated the variance, skewness, and kurtosis of the geoelectric data each day, determined their anomalies, and then compared them with earthquake occurrences through the forecasting algorithm. We observed that the anomalies of variance, skewness, and kurtosis significantly precede earthquakes, suggesting that the geoelectric data distributions deviate from normal distributions before earthquakes. Furthermore, the forecasting algorithm can select robust optimal models and produce explicit forecasting probability for two-thirds of all experimental cases. Therefore, we concluded that the forecasting algorithm based on statistical moments of geoelectric data is universal and may contribute to short-term earthquake forecasting.