Detection of artificial anomalies in the Earth's infrasonic and magnetic fields

This paper investigates a method for automatically detecting artificial anomalies in infrasonic and magnetic records using the Isolation Forest algorithm.The effectiveness of this approach for detecting infrasonic wave generation and reflection events is analysed.The results demonstrate that the method can identify anomalies with minute-level precision; however, only 20% of the detected events are true anomalies, indicating excessive algorithm sensitivity.The proposed approach allows refinement of event timing predicted by theoretical models and can be improved in future work.