A Wind Noise Detection Algorithm for Monitoring Infrasound Using Smartphone as a Sensor Device

Infrasound monitoring is promising for early warning systems to mitigate damage of disaster. However, wind noise contains the same frequency components as infrasound does, and they need to be separated. To achieve this purpose, a wind noise detection algorithm is proposed. Unlike conventional methods that typically use two microphones, the proposed method assumes that one pressure and one acoustic sensor is available. This assumption comes from a requirement that a smartphone is used as a sensor device. Wind noise is detected as anomaly detection of the microphone signal, using extreme value distribution. Comparing with the data obtained by an anemometer, it is shown that the proposed method successfully determines time periods where wind noise exists under a practical environment, depending on the condition of wind.