Sounds of Hurricane Juan

By Dennis F. Jones
Defence R&D Canada - Atlantic
November 14, 2003

As Hurricane Juan made its way north toward the 40th parallel on Saturday, September 27, 2003, I began to monitor the satellite images every few hours. The predicted path of this well-developed Atlantic storm included crossing Nova Scotia. The opportunity to perform a once-in-a-lifetime acoustics experiment began to take root in my mind. As a professional acoustician/physicist I wanted to obtain a record of the sounds produced by Juan's hurricane force winds to share with all Nova Scotians.

By Sunday, September 28, 2003, Juan was on an imminent collision course with my hometown. As the hurricane approached from the south, I set up a microphone facing northwest near a patio door on the second floor of my house. With this orientation I was able to keep the door cracked open throughout the entire hurricane without any rain or wind coming into the house. Although the microphone is equipped with a foam windscreen, high winds could cause overloads that would degrade the quality of the recording. I also placed the microphone on a soft cushion to isolate it from any wind-induced house vibrations.

The microphone was an Audio-TechnicaŽ AT825 stereo field microphone with a frequency response of 30-20,000 Hz and a dynamic range of 102 dB. The microphone output was fed into a TEAC TASCAM DA-P1 Digital Audio Tape (DAT) Recorder with a 20-20,000 Hz frequency response and a dynamic range of 90 dB. These key specifications ensured that all frequencies audible to the human ear were recorded and that both quiet and loud sounds were captured without distortion.

Anticipating that the power would be lost and that I would be recording in the dark, I fully charged the DAT battery and kept a watch and flashlight nearby. I started recording at 2130 ADT Sunday evening and stopped at 0138 ADT Monday morning. Whenever I observed a significant event I would simply speak into the microphone. I also called out the time periodically. To save on DAT battery life, I turned off the recorder for periods of time. In total, I recorded about one hour of good quality data with only a few overloads caused by extreme wind gusts that buffeted the patio door and microphone.

The rain began just prior to 2130 ADT and the winds were picking up with occasional gusts, gusts that were already causing trees to break in our neighbourhood. The 2.5-second spectrogram shows the 2-20 kHz broadband lines of a large branch tearing off a tree trunk. The sound is less than half a second in duration and is similar to the noise made by someone crunching on a crisp potato chip. The wind noise around our home was enhanced by the presence of several maple trees still sporting most of their leaves. These leaves began to shred into small pieces and many of them littered the deck and stuck to the windows. The width of the horizontal blue band in time series 1 represents the amount of noise made by the wind about two and a half hours before landfall. The thicker the blue band the noisier the winds. One minute of data is shown. The peak at 0.5 minutes is a strong wind gust.

Over the next two hours several sounds were recorded including stronger winds than I have ever experienced before, rain pounding on the metal patio flashings, more broken trees and possible low-frequency strumming from power cables (40-200 Hz). There was also a narrowband sound at 200 Hz and another at 6kHz with 12 and 18 kHz harmonics. These could be Helmholtz-type vibrations caused by wind passing over the openings of air-filled cavities somewhere around the house. The power went out in my neighbourhood at 2320 ADT but there was still a distant glow of lights to the northwest. Lightning was observed at 2336 ADT but thunder was neither heard nor recorded. The house began to sway quite noticeably at 2353 ADT and the lights to the northwest went out at 0010 ADT when Hurricane Juan made landfall.

The wind noise at landfall is shown in time series 2. Comparing with time series 1 it is evident that the noise amplitude has significantly increased. Since much of Dartmouth was in or near the eastern eyewall of the hurricane, time series 2 represents some of the strongest winds in the storm. The wind gusts are longer in duration and often occur in quick succession like the four large peaks in the first half minute of time series 2.

At 0115 ADT the noise and motion of the house were a bit much for my family and so they moved downstairs to the main level to sleep. A bright flash of light was observed at 0135 ADT to the northwest just above the houses across the street. There was no sound heard or recorded. I discussed this flash with Manning Smith, System Control Supervisor at Nova Scotia Power Inc. and he told me that there are two active lines there that could have been the source of the flash: the 113H-443 25kV feeder line from the Dartmouth East substation (113H) and the L5011 69kV transmission line that connects the Farrell Street substation (99H) with the Imperial Oil substation (58H).

As Hurricane Juan made its way across the province toward the Minas Basin the winds began to diminish. Time series 3 shows the wind noise about one and a half hours after landfall. The wind is still gusty and stronger than the wind in time series 1, presumably because the eye of the hurricane is one hour closer to Dartmouth. Note that the peak near 0.6 minutes in this time series includes my voice as I make comments on the DAT recorder.

As with many scientific experiments, collecting the data is often easier than understanding the data. In the months ahead I will be looking at the data to determine in more detail the sources of the various sounds recorded.

Spectrogram: Broadband sound of a breaking tree branch.

Time Series 1: Juan 2.5 hours before landfall - strong winds with tree-breaking gusts.

Time Series 2: Juan at landfall - very strong winds with longer duration gusts.

Time Series 3: Juan 1.5 hours after landfall - strong winds with tree-breaking gusts.