Tornado Infrasound Observations

Perkins Tornado on May 11, 2017
Storm and Tornado Details
On May 11, 2017 a line of storms to the west of the OSU Infrasound Array included a large hail-producing supercell. At 2013 UTC the supercell produced an EFU tornado near Perkins, Oklahoma (35.97, -97.04). The confirmed tornado was located 18.7 km south-by-east of the OSU Infrasound Array. The official tornado path length and damage width were 0.16 km (0.1 miles) and 46 m (150 ft), respectively. There were live news reports of a possible second tornado after the first confirmed tornado, but it was never confirmed due to the storm being rain wrapped and no low-level radar coverage. Below are time traces of the air temperature, relative humidity, wind speed, and speed of sound from ground level MESONET sites (PERK, STIL, MARE) as well as a weather station located near the OSU Infrasound Array (DML). Also shown is a satellite image of the storm and a map of the locations of these sites relative to the reported tornado and the closest radar (KTLX).

Infrasound Signal
The time trace from two of the infrasound microphones are shown below with the time in minutes relative to the reported tornado (11 May 2017, 2013 UTC). This shows that at approximately 8 minutes before the tornado report (assumed to be tornadogenesis) a significant infrasound signal appears well above the background levels observed during the previous 2 hours. Note that during these previous 2 hours there were multiple storm cells in the region as illustrated in the satellite image. Approximately 20 minutes following the tornado report there is a second infrasound event that is similar to the original.

Since a tornado is a non-repeated event (though there is a potentially periodic signal produced), it is treated as a nonstationary single record. Using the trial-and-error approach for such a signal from Bendat & Piersol (2000), it was determined that an averaging period of 100 seconds was appropriate. This reduced the random errors while minimizing the bias errors. Using this period, the correlation between microphones, bearing angle of received signal, and the power spectral density was examined. As shown below, during both events their was a significantly higher correlation between microphones relative to outside of these events, which was used to identify the start and end of events for subsequent analysis.

Due to cross-talk issues between two of the microphones an exact bearing angle could not be determined, but assuming the received signals are plane waves there are only two possible solutions. Thus it is possible to determine if the received signals are consistent with what is expected if it were produced from the tornado of interest. The signals were bandpass filtered between 5-50 Hz, which is the expected frequency range (including harmonics) from a tornado. Below is a plot of the resulting bearing angle estimates (blue diamonds) compared with the front edge, center, and trailing edge of the storm core that produced the tornado. Also, the red dashed lines denotes the relative angle that the confirmed tornado was reported. This shows that the bearing angle of the received signal is consistent with the movement of the storm core that produced the tornado.

Infrasound Spectra
Spectral analysis of both events shows that they have nearly identical spectra, that there is a broadband rise in infrasound relative to before and after the tornado, and there is a peak that forms at ~8 Hz (with some harmonic content). Below are sound pressure spectra during the first infrasound event (-8 < tr < 10 min) compared with the sound pressure spectra before the tornado formed. Thin black lines correspond to spectra from individual 100 sec intervals, and the thin white line is the mean with a 1/50th decade filter applied. Also shown is comparison of the pressure power spectra during events 1 and 2 as well before and after these events.

Stillwater Funnel Cloud on Oct 7, 2018
On Oct 7, 2018 between 2335 and 2350 UTC there were two independent sightings (with video) of a funnel cloud forming near the north side of Stillwater, OK. Video recorded by Dr. James Kidd in Stillwater (36.1874N, 97.0543W) looking southeast shows the funnel cloud and can be view online ( Infrasound levels were low throughout the day until this funnel cloud formed (see time trace of entire day below). While the correlation between microphones was high during this event, no peak in the power spectra formed.

Bendat, J. S. and Piersol, A. G. (2000). Random Data: Analysis and Measurement Procedures. 3rd Edition (John Wiley & Sons, New York), 457-477.