Page 48 - 2016Spring

Infrasound from Tornadoes
The first published accounts of sound recordings from a tor- nado were by Arnold et al. (1976). These data were restricted to the frequency range from 100–2,000 Hz. All measure- ments were collected using low-quality analog equipment, which led to significant distortion of the measured signal. The main findings were that the measured signal was broad- band in character, with a roll-off above 500-Hz, and the band start frequency was associated with the lower frequency cut off of the measuring equipment.
More recently Bedard (2005) has reported on measurements of infrasound by a four-element array of sensors in the 0.5– 10 Hz band from tornadic thunderstorms and concluded that the 0.5–2.5 Hz band contained the maximum correla- tion between sensors. Note that this result is dependent on the sensor separation, the atmospheric noise field at the time of the measurements, and characteristics of the sensors (such as its frequency response and noise floor), as well as on the characteristics of the infrasound signal at the array. It was suggested that this signal was associated with the tornadic activity, but the mechanisms for its creation remain unclear. For example, Abdullah (1966) proposed a mechanism based on radial oscillations of the tornado vortex; however, Shecter (2012) demonstrated that this mechanism was implausible. Other possible mechanisms include turbulent flow associ- ated with the tornado vortex and interactions between the tornado and the ground.
If tornados were known to produce infrasound or low-fre- quency signals, and if these signals had characteristic fea- tures, infrasound arrays could be used to augment existing early warning systems and possibly significantly improve the safety of individuals living in high-risk areas for torna- dos. However, previous results do not unequivocally estab- lish that tornados produce infrasound nor do they provide any characteristics of the signal that are unique to tornados that have touched down. As part of a program on hazards monitoring that were funded by NOAA, the National Center for Physical Acoustics (NCPA) and Hyperion Technology Group Inc. conducted a series of infrasound sensor deploy- ments in Oklahoma over the summer of 2011. The primary interest in this deployment was to address both of those is- sues.
Data Collection
The NCPA, in collaboration with Hyperion Technology Group Inc., collected data for two tornado outbreaks on May 24, 2011, and June 11, 2011. The sensors used for these mea- surements were a new class of digital infrasound sensor (Fig-
46 | Acoustics Today | Spring 2016
Figure 4. Digital NCPA sensor (for scale, floor tiles 1 foot × 1 foot in size). The sensor is GPS syn- chronized, 1000 samples/second, (22.5 noise free bits).
ure 4) developed at the NCPA. These sensors have a built-in 24-bit digitizer with GPS time syncing and 802.11/b wireless connectivity.
The NCPA sensors have a unique capability to measure sig- nals from infrasound to low-frequency acoustics (0.015-500 Hz). The frequency response of this sensor was from 0.001- 100 Hz (Figure 5), and its gain was set so that the maximum transducible pressure amplitude was about 90 Pa. For this exercise, the sensor sampling rate was set to 1,000 samples/ second.
Figure 5. Nominal sensitivity curve for the NCPA microphones used in this deployment. This sensitivity function has an asymptotic value of 25 mV/Pa, a pole at 6.4 mHz, and a zero at the origin.
Oklahoma Campaign
Oklahoma was targeted for this project because the storm- chasing season lasts for a shorter period. As mentioned pre- viously, the relative flat terrain coupled with the absence of standing forests offers improved visibility and a reduced risk of unintentional tornado intercepts. One of the objectives of this deployment was to obtain broadband signals from tornados by placing sensors in the region of tornadogenic storms. In some cases, sensors were placed within a few ki- lometers of the path of very large tornados (EF4 and EF5), but this was not a primary goal of the project.
The second and primary objective was to deploy region- al arrays northeast of the tornadic convective storm. The northeast direction was chosen because this is the direction