We analyzed an extended time series of Schumann Resonance recordings with two multi-parametric statistical methods, the generalized linear Logistic Regression—LogReg and the non-linear Random Forest—RF, in order to test their potential for earthquake prediction within a narrow time-space-magnitude window of 48 h, 250 km from our observing site, and events higher than magnitude 4 of the Richter scale. The LogReg method identified the power of the signal within our 10-min recording intervals as the main seismic precursor parameter. The RF method obtained promising results that will improve with continuous enrichment of the running data sample with new data. We conclude that a systematic analysis of Schumann Resonance recordings may lead to satisfactory levels of seismic prediction.

我们使用两种多参数统计方法（广义线性Logistic回归LogReg和非线性随机森林RF）分析了舒曼共振记录的扩展时间序列，以测试其在狭窄时空内的地震预测潜力-距我们观测站点250公里的48小时震级窗口，且事件高于里氏震级4级。LogReg方法将在10分钟记录间隔内的信号功率确定为主要地震前兆参数。RF方法获得了令人鼓舞的结果，随着使用新数据不断丰富运行数据样本，该结果将得到改善。我们得出结论，对舒曼共振记录进行系统的分析可能会导致令人满意的地震预测水平。