The Wind Blew the Other Way, and the Complaints Came Anyway

Saturday’s Lynyrd Skynyrd concert at the Ford Amphitheater was the most strongly upwind evening in three years of records, and it drew the largest response the resident-run noise map has ever logged on a night the wind was blowing the other way. It was also the second show running that the Independent’s own sound forecast ranked as quiet. The paper has since tested seven further weather measures to find what the forecast is missing. None of them worked.

Thirty-Two Reports on a Saturday Night

Lynyrd Skynyrd played the Ford Amphitheater on Saturday, September 19, with a 7:30 p.m. start. Thirty-two reports reached the resident-run noise map for that evening and the following day, putting the show among the six largest nights of the fifty-five performance dates the map has covered since it opened in September 2024.

One of the thirty-two is almost certainly a coordinate error — a report placed 22.6 miles east carrying text and a decibel value identical to another at 6.7 miles north. Thirty-one are used for every distance and direction figure below, and those thirty-one come from roughly twenty-six distinct locations. Residents reached for superlatives:

“My windows are closed fan blowing tv loud I STILL hear and FEEL it. Absolutely loudest ever.”

“It’s the loudest I’ve ever heard. We usually can’t hear it inside. It is keeping my kids up. Outside is like a full concert. I can understand every word.”

“Loudest concert to date since opening- could hear every word sung”

“Live 3 miles south of the venue and can very clearly hear this. It’s keeping my kids awake. I’ve never been bothered by the noise before.”

Two reports carried a decibel figure: 73 dBA at 1.2 miles east, and 45 dBA at 6.7 miles north. Neither names an instrument, and neither can be treated as a measurement.

The Most Upwind Night in the Record

What makes the evening unusual is not the volume of reports. It is that the wind was against them.

Averaged across the event window, the weather model puts the component of the wind carrying sound toward the complaint field at −5.66 meters per second — negative meaning away. That is the most strongly upwind of all fifty-five performance dates in the record. The next two are a Jason Isbell show in April 2025 and an Alison Krauss show in August 2026. Between them they drew no reports at all.

Twenty-five other nights in the record were upwind. Their median response is one report. The worst any of them had produced was twenty-six, at a Foreigner concert in October 2024. Saturday produced thirty-two.

The U.S. Air Force Academy airfield, three miles south of the venue, agrees with the model and was not consulted by it.

The wind blew out of the north for the whole evening, between 3.5 and 7.4 meters per second. The only direction it carried sound strongly was due south. Exactly one report came from the south.

So the explanation offered in this newspaper’s September 6 piece does not rescue this night. That piece noted the model projects wind onto a single fixed bearing, and suggested it may have been pointed the wrong way. On Saturday the flow was against the reporting neighborhoods on every bearing tested, on the model and on the station alike. The sound traveled crosswind and upwind.

The Forecast Said Low Again

Run against Saturday’s actual weather through the published model code, the sound forecast scores the show 1 — “low–moderate.” When the forecast was made, the fifteen nights in that band carried a median of two reports and a worst of twenty-six. Saturday beat the band’s ceiling and now sets it.

This is the second consecutive show the tool has called low and the second it got wrong. September 5 scored 1 and drew twenty-four.

Honesty requires noting that the band already held one anomaly. Its previous worst — the Foreigner show — was itself an upwind night. A band whose ceiling is set by an unexplained evening is a band with a known weakness, and Saturday widened it.

Adding one night moved the published statistics more than a single date should. Folding September 19 into the record drops the correlation between the wind term and complaint volume from +0.47 to +0.40, and the held-out correlation between forecast and outcome from +0.45 to +0.39. Temperature inversion, already weak, falls out of even marginal significance. A tool that can be moved that far by one Saturday is a tool with fifty-five observations behind it, and that is worth a reader knowing.

The Reports Moved East

The geography also shifted. Reports came from ten points east, ten northeast, five north, four southeast, one south and one northwest. The vector-mean bearing for the night is 66 degrees — east-northeast — against the roughly 20 degrees the model watches, and against the north-northeast arc that has carried most of the record. That shift is large enough to have moved the model’s own bearing by a degree and a half in a single week, the biggest weekly move yet recorded.

They also came from further out. Median distance was 2.5 miles, and 26 percent came from four miles or more, against a 7.1 percent baseline across every other night in the record. Reports arrived from beside Pine Creek High School, from three miles south, and from a cluster between 1.1 and 1.9 miles due east.

One artifact belongs in the open. Six of the thirty-one sit within about a hundred yards of each other at six miles northeast, carrying fragmentary text — “9-10,” “Loud,” “9:40,” “Loud WOW.” That is one household filing six times at nudged coordinates, working around the map’s rule of one report per exact location. The thirty-one usable submissions represent roughly twenty-six distinct places, and both numbers are given here rather than one standing in silently for the other.

A Test That Came Back Negative

Saturday had one feature September 5 did not: a textbook late-evening stable layer.

Skies cleared at nine and the boundary layer collapsed from 325 meters to 55 over four hours while the inversion deepened. The complaint timestamps follow it: 9:03, 9:15, 9:40 and 11:22 p.m. The Independent has tested inversion before and found it weak, but what it tested was the average across the show. A late-forming inversion under clearing skies is a different variable, and it had not been tested.

So it was — on seven measures, each defined before any of them was computed, across the fifty-three performance dates that preceded September 5. The two nights that prompted the test were held out of it, so that they could not vouch for the hypothesis they had suggested.

Nothing cleared that threshold. Nothing cleared a conventional 0.05 either. The late-evening inversion beat the show-average version by nine thousandths of a rank correlation, which is noise. Boundary-layer height came out with the wrong sign — a lower, more stable layer should carry more sound to homes, and the measured relationship leans very slightly the other way.

Three further checks, all negative. The late inversion adds almost nothing once wind is accounted for: controlling for the wind component it falls to +0.21, while wind controlling for it holds at +0.46. Among the nights the model scored 0 or 1 — the ones it gets wrong — the correlation is +0.07, which is flat. And it would not have flagged either night that prompted it: September 5 and September 19 both sit at a late inversion of +1.2 °C, the 55th percentile of the distribution, against a median of +0.95. Both are ordinary.

The measure is weak, it is not additive to wind, it does not separate the nights the model misses, and it does not describe the two nights that motivated it. It has not been added. Shipping it would have made the tool look more sophisticated without making it more accurate.

Where the Explanation Now Points

Every meteorological variable this newspaper can compute has now been tested against these two nights: surface wind, wind at 80 meters and its peak hour, inversion by four definitions, boundary-layer height and its collapse, cloud cover, temperature. September 5 and September 19 sit near the middle of the distribution on all of them. Whatever produced those evenings is not in the weather record.

That leaves two candidates, and this newspaper can measure neither from outside. One is the source — program content and operating level. The other is reporting behavior: a venue under litigation, a ballot initiative in circulation and a newspaper writing about both will draw more reports than the same sound did a year ago, and nothing in a self-selected record can separate that from a louder night.

The venue’s own monitors recorded both evenings. The Independent put five questions to VENU Holding Corp. about September 5 on September 6, asking what was different that night, what the remote stations recorded and how the pre-show ambient reference was calculated, with a stated deadline of September 11. The company has not responded, then or since. No questions about September 19 had been put as this was written.

The complaint record has its own limits, and one of them moved this week. Two reports present in the September 13 pull are gone from the live data, both from households that filed again on September 19 and 20 from the same coordinates. Because the map permits one report per exact location, a repeat reporter does not add to the record — they overwrite themselves, and an earlier night quietly loses a complaint. The King Gizzard & the Lizard Wizard show of August 2025 has dropped from fifteen reports to fourteen, and Sammy Hagar’s August 2026 show from two to one, for reasons that have nothing to do with either evening. Every per-night count in this record is a lower bound that erodes as its reporters return.

The live forecast page now carries a dated note recording both misses, the test and its result. Nothing else about it has changed, because nothing tested justified a change. The next show is Friday, September 25.

Sources & further reading

The Independent’s own coverage: Sound Reached Six Miles on a Nearly Windless Night — the September 5 show, and the first miss · At the Ford Amphitheater, the Complaints Track the Wind — the three-season analysis the forecast came out of · The Independent Builds a Sound Forecast for the Ford Amphitheater — how the model was built and validated · Ford amphitheater sound forecast — the live tool, updated twice daily.

Complaint data: Ford Amphitheater noise map, all 825 records pulled September 20, 2026.

Weather: Open-Meteo historical forecast archive, hourly, at the venue coordinate · Iowa Environmental Mesonet ASOS archive, hourly observations from the U.S. Air Force Academy airfield.

Venue: Ford Amphitheater events calendar · VENU Holding Corp.

Methodology

Complaint counts come from the resident-run noise map, a self-selected, anonymous, advocacy-aligned record of complaints rather than a measurement of sound. It cannot be read as decibels. Reports are credited to the show on their own date where one exists, otherwise to the previous evening’s; the count of thirty-two includes ten dated September 20, the day after the concert, and the venue’s own events calendar lists no September 20 performance. Thirty-two is the figure the tracker carries, because the report placed 22.6 miles east survives every documented cleaning rule. The Independent treats it as a longitude typo — its latitude, text and decibel value are identical to a report 6.7 miles north — and excludes it from every distance and direction figure, which is why those rest on thirty-one.

Distances and bearings are measured from a venue coordinate the Independent derived by trilateration from residents’ own stated distances, accurate to about 0.09 miles. Records beyond 25 miles, records at null coordinates, records their authors marked for exclusion and legacy migrated records are removed before any figure is computed; 825 raw records clean to 793.

Wind figures are hourly. Model values come from the Open-Meteo historical forecast archive at the venue coordinate, best-match model, grid cells roughly two miles across; the observations are hourly METAR reports from the Air Force Academy airfield, converted from knots. The downwind component is the projection of the wind onto the bearing from the venue to the receptors, positive when sound is being carried toward homes. The bearing the forecast uses is the vector-mean direction of all cleaned complaint locations, currently 19.9 degrees, and it drifts each time the record is regenerated. Figures for September 19 are averaged over the event window the tracker itself uses, the advertised set time through three hours later. Scores and tiers were produced by running the published model code against the hourly record, not calculated by hand.

The seven-measure test used Spearman rank correlations with a tied-rank correction and two-sided p-values from the Fisher z transform, across the fifty-three performance dates preceding September 5. Those two nights were held out because they suggested the hypothesis; including them would have let the test vouch for itself. The 0.0071 threshold is a Bonferroni correction for seven simultaneous comparisons. That seven-measure test used a fixed 7 to 11 p.m. window across all nights, which reproduces the tracker’s stored per-event wind figures at a rank correlation of 0.97. All seven measures and their definitions were written down before the numbers were computed, and all seven are reported above whatever the result.

Resident comments are reproduced exactly as submitted, including spelling and punctuation.