The Measurement Gap

Why the tool we use to measure aircraft noise was not designed to protect health

For fifty years, aviation noise policy has been built on a single metric: the day-night average sound level, or Ldn. It was a reasonable tool for its time and purpose. But the science of how noise harms the human body has advanced considerably since the 1970s — and the gap between what we now know and what we currently measure has real consequences for communities living beneath flight paths.

Discrete events drive health harm — not averages

The epidemiological research conducted over the past three decades consistently finds that discrete noise events — not average levels — are the primary driver of the health outcomes associated with aircraft noise.

The stress physiology is straightforward. A single aircraft overflight above 65 dBA triggers a measurable cortisol response. Research shows cortisol requires a minimum of 15 minutes to return to baseline after such an event. If the next overflight arrives before recovery is complete, stress hormones compound. Over months and years, this chronic physiological stress is the mechanism by which aircraft noise produces cardiovascular disease, hypertension, and metabolic dysfunction. A 24-hour average is blind to whether that recovery ever happened.

What Ldn was designed to do — and what it does well

Ldn — the day-night average sound level — was developed in the early 1970s by the U.S. Environmental Protection Agency (EPA) to compare noise exposure across environments and inform land use planning. It averages all sound energy over a 24-hour period, applying a 10 dB penalty to nighttime events to acknowledge their greater disturbance potential. The FAA adopted it as the standard for aviation noise assessment in 1981.

Where Ldn genuinely works is in long-term cumulative exposure tracking. The epidemiological studies that established the link between aircraft noise and cardiovascular disease — HYENA (Health Effects of Noise Near Airports, European Commission, 2002–2006), RANCH (Road Traffic and Aircraft Noise Exposure and Children's Cognition and Health, European Commission, 2001–2005), and others — were largely built on Ldn-derived exposure estimates, and they produced real, replicable findings. The dose-response curves showing increased hypertension risk above 55 dB Ldn are not artifacts of a bad metric. They reflect the fact that chronic disease accumulates over years of exposure, and an integrating metric captures that load in ways that a single event count cannot. Ldn also enables cross-community and longitudinal comparison — a common currency that allows researchers to ask whether noise near BWI is worse than near O'Hare, or whether it has changed over a decade.

It is worth being precise about where Ldn fails and where it does not. For continuous or near-continuous noise sources — industrial facilities, heavily trafficked highways, occupational environments — where no meaningful quiet interval exists between noise events, cumulative energy averaging is an appropriate tool. The body's primary exposure mechanism in those environments is sustained energy load, and Ldn captures that load reasonably well. OSHA's time-weighted average for occupational noise is essentially the same concept and remains appropriate for its intended application.

The failure is specific: Ldn is the wrong tool for impulsive or event-driven noise in residential environments where meaningful quiet intervals exist between events. Aircraft noise is the canonical case. Each overflight is a discrete acoustic event, separated from the next by an interval during which the body has the physiological opportunity — or is denied the opportunity — to recover. In that structure, what matters is not the average energy load but the event count, the peak level, and whether recovery was possible before the next event arrived. Averaging conceals all three. The metric was designed for a different problem.

The studies that built the case were event-aware all along

There is an important detail about those foundational epidemiological studies that is rarely discussed: the Ldn values used in HYENA, RANCH, and similar research were not computed from continuous steady-state noise. They were computed from aircraft event logs — records of individual flights with individual noise levels. The researchers had the event data. They then summarized it into an average for the statistical model.

The averaging in Ldn-based health research was a modeling choice, not a physical reality. The underlying exposure was always event-driven. Researchers who have returned to those same datasets and analyzed event-level variables — peak levels, event counts, inter-event intervals — have found stronger dose-response relationships than the averaged models produced.

Averaging conceals what communities actually experience

Aircraft noise is not a steady hum. It is a series of discrete events — each with a distinct onset, peak, spectral character, and duration — separated by intervals of relative quiet. The health effects most documented in modern research are triggered by those events: the cortisol spike from a startle response, the sleep fragmentation from a nighttime overflight, the compounding stress when recovery time is too short. None of these mechanisms are captured by a 24-hour average.

A neighborhood experiencing 40 sharp 75 dBA overflights per day and a neighborhood exposed to steady 58 dBA road traffic can produce identical Ldn values. The physiological experience — and the health risk — are not remotely comparable. The average is mathematically accurate. It simply does not measure what causes the harm.

This is not a flaw in the mathematics of Ldn. It is a category error — applying an averaging metric to an event-driven phenomenon. The metric was calibrated against community annoyance surveys from the 1970s, not against modern cardiovascular and cognitive health endpoints. If you recalibrated the regulatory threshold of 65 dB Ldn using current health evidence, it would move significantly downward — and the compatibility contours that have governed residential development near airports for decades would expand significantly outward.

Communities falling below the 65 dB Ldn threshold are not necessarily safe. They may simply be experiencing harm that the current measurement framework cannot see.

The consequences are not abstract. Land use planning built on Ldn contours has determined where residential development is permitted near major airports for forty years. Families have purchased homes in areas the contour maps designated as compatible while experiencing dozens of overflights per day that exceed 70 dBA, disrupt sleep, and produce the physiological harm documented in the peer-reviewed literature. The contour said compatible. The body said otherwise.

Event-based measurement is not intractable — it just hasn't been tried

The conventional objection to event-based community noise monitoring is practical: deploying calibrated sound level meters continuously across a community is expensive, technically demanding, and produces data that is difficult to summarize for policy. That objection made sense when continuous monitoring required dedicated hardware at every measurement point.

It does not make sense anymore. The smartphone in nearly every resident's pocket contains a microphone, a GPS receiver, a high-performance digital signal processor, and a network connection. With the right software — software that implements the same IEC 61672 A-weighting and ISO 532 psychoacoustic analysis used in professional instruments — a distributed community monitoring network becomes feasible at a fraction of the cost of traditional approaches.

TrueNoise has demonstrated this on a minimal budget. A small number of monitoring sessions near BWI Airport, using a consumer microphone and a calibrated iPhone app, produced a richer characterization of community noise exposure — event counts, psychoacoustic profiles, recovery deficits, aircraft-level detail — than most Ldn studies ever achieved. The point is event-based characterization is feasible now.

Two approaches compared

The path forward

The communities living beneath flight paths have been telling regulators for decades that the noise contour maps do not reflect their experience. The reason is not that the maps are drawn incorrectly — it is that the metric the maps are built on was calibrated against annoyance surveys from an era before the physiological mechanisms of noise-induced disease were understood, and that it has not been updated to reflect fifty years of subsequent research.

The people who built and maintain those frameworks were not acting in bad faith. They were using the best available tools. Now we have new tools.

The path forward is to add event-based, psychoacoustic measurement to the regulatory toolkit — not as a replacement for existing approaches, but as the necessary complement that the health evidence has always demanded. Communities deserve planning decisions made with data that reflects what actually happens to the human body beneath those flight paths.

That is what we are building toward.

Sources and references