A study by a retired NY State Department of Environment staffer has found that in one community where two new wind farms are planned, “worst case” atmospheric conditions can occur up to 30% of nights in summer and fall, peaking at over 40% of nights in early summer.  The wind farms in Cape Vincent are moving forward in the permitting process based on noise modeling submitted by the wind developers, which predicts minimal impact on neighbors thanks to an average background ambient noise level of 45dB.  This study found that on nights with little wind at ground level, actual ambient sound in this rural area is generally below 35dB, and in many areas, drops to 25dB or lower for much of the night.  Also, and most importantly, the study used standard predictive measures (including wind differential at two near-ground heights, daytime solar radiation, and night time cloud cover) to estimate how often the winds at turbine hub height would be high enough to turn the turbines on, even as the wind at ground level remained low – the situation that often triggers the worst night time noise complaints near wind farms.  The sobering result was that such nights, which create noise issues for neighbors far beyond those predicted by the simpler noise modeling used during permitting, could be a regular occurrence for most of the summer and fall.  After taking noise measurements at a wind farm currently operating in a nearby town – which found levels similar to those predicted and allowed in current Cape Vincent planning – the author notes that the vast majority of Cape Vincent homes will be close enough to hear the turbines easily on these “worst case” nights, with a third  of local households likely to experience objectionable noise levels.

It is important to note that these levels of audibility and possibly objectionable sleep disruption will still occur on a minority of nights; indeed, on an annual basis, only  around 15% of nights. And, on those nights, just a portion of the population (those closest to the turbines) will be affected. This perspective is part of what has led to our current acceptance of relatively small setbacks from homes (often 1000-1500 feet). Our social decisions about wind farm siting will need to grapple with the question of how much disturbance we consider acceptable – in this case, a third to half of the residents of Cape Vincent may be looking at troubling noise levels for a third to half of summer and autumn nights. To protect most residents from most of the noise intrusions (by keeping noise levels to 5-10dB above the true local ambient level) would require setbacks of 1km or so (between a half and three-quarters of a mile), which would likely limit the Cape Vincent area to more like 30-50 turbines, rather than the planned 200.

Click on to go below the fold and read a more detailed summary of the research paper, which first appeared on AEI’s new science research page.

Clifford P. Schneider. Measuring background noise with an attended, mobile survey during nights with stable atmospheric conditions. Internoise 2009. [DOWNLOAD PAPER(pdf)]

This well-designed study, by a retired New York State Department of the Environment staffer, sheds light on several key questions surrounding standard noise assessments of wind farms. Most importantly, it quantifies the extent of one of the key atmospheric components of excessive wind farm noise, finding that stable night time atmospheres may occur two-thirds of the time in the summer and fall, with wind high enough at turbine height to trigger them into action on 30% of nights, increasing to 40% in June and July. On these “worst case” nights when ground air is still but turbines are active, turbine noise is likely to be significantly louder than local background ambient noise from late evening until the beginning of the pre-dawn bird chorus. Schneider’s study also included a brief but useful test of whether a quick but systematic mobile sound survey can be used instead of set of arbitrarily-chosen monitoring sites (addressing questions about whether arbitrary site selection accurately measures local averages) – he did 10-minute recordings at 21 sites, one every mile along two rural roads, and compared the results to 5 baseline sites chosen arbitrarily around the same town.

This study took place in the township of Cape Vincent, New York, which sits at the confluence of the St. Lawrence River and Lake Ontario, where two wind farm proposals would place 200 turbines in town, covering the majority of its land area. Local and state ordinances require sound to be within 5dB of average ambient noise levels; the wind developer’s sound level report based their projected compliance on a day-night average of 45dBA. By contrast, Schneider’s study focused in on the times when other wind farms have generated the most problematic noise impacts: nights when wind is low at ground level but high enough at hub height to trigger turbines to turn on. In these conditions, night-time background ambient sound levels are very very low, and turbine noise can dominate the nighttime soundscape. The results are striking, to say the least.

Night-time sound levels were at or below the lowest that the researcher’s meters could record, 25dBA, for most of the late-night hours at 4 of the 5 baseline recording locations, and under 30dB from 9pm-4:30am at three; one site received shore waves and dropped below 35dB for just a few hours in the middle of the night. Average readings along the mobile monitoring route (which was traversed over the course of three nights) were similar, and generally ranged from 25-32dB, with one site at 36dB. By contrast, measurements near a wind farm in the region, also measured at several (10) locations, showed the quietest times of night (L90) ranged from 35-43dB, and the average (LEQ) from 36-45dB. Bottom line: wind turbine sounds are likely to be 10-15dB louder than background ambient sound on these “worst case” nights in Cape Vincent.

But how often do such nights occur? In what is likely to be the section of the study with them farthest-reaching impact, the results suggest that turbines could be operating (hub height winds >4 m/s) on nights with very still, quiet ground conditions (ground wind <1.5 m/s) for 42% of nights in June and July, and for a total of 30% of nights from June through October. This lines up well with the only other study that I know of that specifically addressed stable night air conditions, the van den Berg study that found such conditions occur about a third of the time in the Netherlands, with similar levels suspected in many temperate zones. Using New York State DEC standards for predicting human reaction to noise, which is based on increasing levels of industrial noise above background ambient, this study suggests that (counter to the developer’s prediction that no residents will experience noise more than 5dB above ambient, i.e., loud enough to trigger complaints), the vast majority of households in town will find the noise “very noticeable” (9-14dB above ambient) on these nights, with 34% finding it “objectionable” (84 homes at 14-19db above ambient) and a further 19% finding it “very objectionable to intolerable” (48 homes at 19-25dB above ambient).

It is important to note that these levels of audibility and possibly objectionable sleep disruption will occur on a minority of nights; indeed, on an annual basis, probably around 15% of nights. And, on those nights, just a portion of the population will be affected. This perspective is part of what has led to our current acceptance of relatively small setbacks from homes (often 1000-1500 feet). Our social decisions about wind farm siting will need to grapple with the question of how much disturbance we consider acceptable – in this case, a third to half of the residents of Cape Vincent may be looking at troubling noise levels for a third to half of summer and autumn nights. To protect most residents from most of the noise intrusions (by keeping noise levels to 5-10dB above the true local ambient level) would require setbacks of 1km or so, which would likely limit the Cape Vincent area to more like 30-50 turbines, rather than the planned 200.

Two other results are worth mentioning. As mentioned above, the systematic mobile sound assessments matched closely with those made at arbitrary locations, suggesting that either method is reliable. The author notes that the arbitrary locations that he chose were well away from buildings, and that a systematic mobile survey with a random start avoids possible problems of subjectivity in site selection. He notes: “I could have increased the measured SPL (at the arbitrary locations) if I had located equipment closer to homes, barns, and roads, and if I had picked nights with moderate winds (ed. note: or had not chosen to focus on low-wind nights).” Further, he notes that the attended metering used in the mobile survey allows identification and documentation of noise intrusions, some of which might be of interest in assessing either true ambient or noise conditions with turbines operating (i.e., a site with many cars passing will show elevated sound levels, compared to one with fewer; this increased noise could either lead to over-estimate of quiet times at these sites, or improperly suggest that turbines are the primary source elevating sound levels at a site.) Finally, this study’s measurements of dBC (low frequency sound) found substantial variability at different sites, more variable than audible sound, including several sites where the difference between dBA and dBC approached 20dB; this suggests that later measurements with turbines operating may not necessarily suggest that the turbines are the source of such variability.