11:00 AM - 1:00 PM
[MTT45-P02] Construction of automatic detection system for Infrasound events using multiple observations and evaluation of weather impacts
Keywords:infrasound, weather, lightning, wind, rain
In this study, we constructed an automatic event detection system using threshold and multisite observations with multiple infrasound sensors installed around the Kami Campus of Kochi University of Technology (Kami, Kochi Pref.). Infrasound observation can detect various events such as tsunamis, volcanic eruptions, thunders, etc. However, the observation always detects various waveforms such as noise caused by wind, or opening/closing doors of a room if they are installed indoors. Therefore, it is difficult to detect any events automatically, and the development of a system has been required. Previous research has confirmed that weather conditions such as rainfall and wind speed can affect infrasound measurements, but such studies has been insufficient to determine the extent of the effect.
Here, SAYA ADX III-INF04LE (hereafter INF04) has been used as the infrasound sensor. In the single-site threshold detection, a group of signals exceeding the threshold value is detected from the dataset observed at each observation site and considered as an event. However, with this method alone, even locally large signals such as noise due to wind or the opening/closing of doors are also identified as events. Therefore, after having the single-site threshold detection at three or more sites, the detection time difference is to be calculated as simultaneous detection at three sites. In case of the same event identification, the maximum time difference is given when the signal propagates in a straight line at the speed of sound, so when the time difference is within that range, the event can be detected as the same one.
In order to evaluate the effect of weather conditions on infrasound observations, we conducted simultaneous observations of infrasound, wind speed, and rainfall in an outdoor observation hut on the Kami campus. The meteorological sensors used were an anemometer (S-WSB-M003) and a rain gauge (S-RGB-M002) manufactured by Onset Computer Co., U.S.A.
The observation period was 308 days, from October 24, 2020 to October 12, 2021 (excluding March 22 to May 6, 2021), when the three sites were operating simultaneously. At threshold of 1 Pa, thousands of events exceeding the threshold were detected at each observation site. Using the times of all the detected events, the simultaneous detection at three sites were performed, and 11 events were finally detected. Of these 11 events, 9 were thought to be thunderstorms, and the remaining 2 were estimated to be other noise events with severe wind or unknown source.
On the other hand, a statistical study of the variation of the maximum infrasound signal in HF band (0.1 to 50 Hz) of INF04 with the maximum instantaneous wind speed showed that, for example, at a threshold value of 1 Pa, the maximum was exceeded the threshold when the wind speed was higher than 5 m/s. A comparison of the measured rainfall and the infrasound LE band (~0.1 Hz) on a day of heavy rainfall showed that when local heavy rainfall occurred in the vicinity of the observation sites, there was a corresponding decrease of about 200 Pa at three sites, while there was no significant decrease at another observation site about 8 km away.
In conclusion, we have developed an automatic event detection system using the INF04 infrasound sensors, which is being operated by our laboratory in Kochi Pref., based on threshold values and the simultaneity of multiple observation sites, and have been able to detect a total of 11 events for the 308-day dataset. In the simultaneous meteorological observation with infrasound, we were able to confirm the relationship between wind speed and observed values of infrasound, as well as that between rainfall and infrasound.
Here, SAYA ADX III-INF04LE (hereafter INF04) has been used as the infrasound sensor. In the single-site threshold detection, a group of signals exceeding the threshold value is detected from the dataset observed at each observation site and considered as an event. However, with this method alone, even locally large signals such as noise due to wind or the opening/closing of doors are also identified as events. Therefore, after having the single-site threshold detection at three or more sites, the detection time difference is to be calculated as simultaneous detection at three sites. In case of the same event identification, the maximum time difference is given when the signal propagates in a straight line at the speed of sound, so when the time difference is within that range, the event can be detected as the same one.
In order to evaluate the effect of weather conditions on infrasound observations, we conducted simultaneous observations of infrasound, wind speed, and rainfall in an outdoor observation hut on the Kami campus. The meteorological sensors used were an anemometer (S-WSB-M003) and a rain gauge (S-RGB-M002) manufactured by Onset Computer Co., U.S.A.
The observation period was 308 days, from October 24, 2020 to October 12, 2021 (excluding March 22 to May 6, 2021), when the three sites were operating simultaneously. At threshold of 1 Pa, thousands of events exceeding the threshold were detected at each observation site. Using the times of all the detected events, the simultaneous detection at three sites were performed, and 11 events were finally detected. Of these 11 events, 9 were thought to be thunderstorms, and the remaining 2 were estimated to be other noise events with severe wind or unknown source.
On the other hand, a statistical study of the variation of the maximum infrasound signal in HF band (0.1 to 50 Hz) of INF04 with the maximum instantaneous wind speed showed that, for example, at a threshold value of 1 Pa, the maximum was exceeded the threshold when the wind speed was higher than 5 m/s. A comparison of the measured rainfall and the infrasound LE band (~0.1 Hz) on a day of heavy rainfall showed that when local heavy rainfall occurred in the vicinity of the observation sites, there was a corresponding decrease of about 200 Pa at three sites, while there was no significant decrease at another observation site about 8 km away.
In conclusion, we have developed an automatic event detection system using the INF04 infrasound sensors, which is being operated by our laboratory in Kochi Pref., based on threshold values and the simultaneity of multiple observation sites, and have been able to detect a total of 11 events for the 308-day dataset. In the simultaneous meteorological observation with infrasound, we were able to confirm the relationship between wind speed and observed values of infrasound, as well as that between rainfall and infrasound.