Several months ago I splurged for an infrasound detector, with the intention to establish a baseline in several homes on Amherst Island before the wind turbine project becomes operational and then compare them with the levels afterwards.  All of the homes I’m interested in are built close to the shoreline and I figured that waves would be the main contributor.  While I don’t have the ability to record wave action, I do record wind speed and direction from the nearby Kingston airport.  So far (at least on Amherst Island in a rural environment; in my town it varies more with human activity) it seems that wind itself is the main contributor, with direction (and subsequent wave action) not seeming to have much of an effect.  I use the word “seem” as I’m still looking over the data, and will be updating my postings as I gain any new insights.

The home where the detector is now installed is on Amherst’s South Shore, facing Lake Ontario.  Visible from the home, across a sliver of Lake Ontario, is the Wolfe Island Wind Project (WIWP), with the nearest of its 86 turbines 13.8 km away.  I wondered if part of the infrasound I was measuring was due to the WIWP, but there was no way for me to separate out the effects of the wind and the project.

And then serendipity struck!  On December 20th an ice storm swept through the area and the WIWP went offline until December 27th, while the wind kept blowing.  And I was recording the entire thing!  There have been several articles written about infrasound from wind turbines, but I am unaware of any studies that involved a shutdown of anything more than a few minutes, let alone over a week.

Now the numbers.  For the entire month (698 samples, the weather station at the airport shuts down overnight sometimes), the average noise level (in mPa) was 297 with an average wind speed of 10.16 mph.  For the non-iced period (562 samples while WI was running normally) the averages were 306 and 10.18.  For the iced-over period (133 samples, I left 3 transitional hours off) the averages were 262 and 10.14.  The increment comes to 44 mPa, an increase of 17%.  The dB equivalents are 82.3 to 83.7, an increase of 1.4 dB.  Now 1.4 dB doesn’t sound like a lot, but recall that this is an average over an entire week.  Kinda like an average temperature deviation of 2 F doesn’t sound like a lot; however it is enough to get us to remark on it being either colder or warmer than usual.

To show the difference in a more meaningful manner, I charted the noise levels vs. wind speeds (above 4 mph – there’s very little noise below that speed) for the two periods.  Here’s the charts.  First, with the WIWP shut down.  The slope of the line indicates that each mph increase in wind speed produces an average of 41.8 additional mPa of infrasound.

Second, with the WIWP operating.  The slope is now 54.5.

These results fly in the face of several studies that claim that wind projects produce very little infrasound.  Wind proponents will, of course, find all sorts of reasons why these measurements are invalid: my science was wrong, my equipment was wrong, my biases were wrong, the winds were wrong, something, anything.  What they won’t do, of course, is cooperate enough to allow an independent researcher to run a similar experiment.

Some details.  The detector was an Infra-20, with a frequency range of 0.1-20 Hz.  It was located in the same place the entire month of December (in an upstairs loft overlooking a large glass-fronted great room, which may explain the relatively high averages).  I used a home-written Perl program on an XP/Dell computer to record the 50 samples/sec that the Infra-20 produces (their Amaseis wasn’t reliable enough – my code has been surprisingly stable; I missed much less than 1% of the samples for all reasons combined during the entire month).  Every hour I’d calculate the average noise level and save that for input to a spreadsheet.  If you’d like the spreadsheet or the program let me know.