Investigation: +2 dB Approach Offset Direction and Scope

Investigation date: 2026-04-12 Status: Complete — corrected in PR #105 GitHub issue: #104

1. Full calculation trace

Source of truth

The +2 dB offset exists at exactly one location in the codebase:

File: data/noise/easa/icaoToEasaMap.ts Lines: 700 and 708 (after PR #103 attribution comments were added) Value: +2 (additive, in dB) Applied to: faaMeasurement.lamax1000ft

takeoffDb: faaMeasurement.lamax1000ft,
approachDb: faaMeasurement.lamax1000ft + 2,

What the input represents

lamax1000ft is described in the interface (data/noise/faa/helicopterMeasurements.ts:33) as "Maximum A-weighted sound level at 1000ft." Analysis of the FAA data shows it correlates most closely with the flyover EPNdB measurement:

The tight ~3.5 dB offset from flyover EPNdB (vs. 8.2 dB from approach EPNdB) strongly indicates lamax1000ft is a flyover/level-flight measurement, not a takeoff or approach measurement.

How the offset is applied

getEASANoiseProfile() (line 664) is called when any code path needs a noise profile. For helicopter types with FAA ROSAP data:

• takeoffDb is set to lamax1000ft (the flyover measurement, used as a takeoff proxy)
• approachDb is set to lamax1000ft + 2 (flyover + 2 dB, intended as approach proxy)

What the output feeds into

Every downstream consumer uses the same pattern: direction === 'arrival' ? profile.approachDb : profile.takeoffDb. This feeds into:

Scope of the offset

The offset applies only to helicopters — specifically, the ~9 helicopter types that have FAA ROSAP data (R22, R44, R66, S76, EC35, A109, B06, B407, AS50, plus B429, B212, B412, S92, S61, EC45, EC55, A139, R66, H60 depending on which is the most recent measurement per type). Jets and fixed-wing aircraft are never affected.

2. Comparison to authoritative sources

The EASA convention in this codebase

All 47 EASA-derived profiles in icaoToEasaMap.ts use the convention approach < takeoff:

Zero exceptions. Every single entry has approach quieter than takeoff.

Physical acoustics basis

For fixed-wing aircraft and jets: approach is quieter because engines are at reduced thrust during descent. Takeoff requires maximum or near-maximum thrust, producing more noise. This is well-established and uncontroversial.

For helicopters: the situation is more complex due to Blade-Vortex Interaction (BVI) noise. During approach, rotor blades can intersect tip vortices from preceding blades, producing sharp impulsive noise (Schmitz & Yu, J. Sound & Vibration 109(3), 1986; Brentner & Farassat, Progress in Aerospace Sciences, 2003). BVI is strongest at specific descent angles and speeds.

However, BVI affects EPNdB more than LAmax. EPNdB (Effective Perceived Noise) heavily penalizes tonal and impulsive content — exactly what BVI produces. LAmax (maximum A-weighted level) simply measures the peak broadband level and is less sensitive to BVI's spectral characteristics.

The FAA ROSAP data in this codebase confirms this:

• In EPNdB: approach is +3.7 to +4.7 dB louder than takeoff (BVI effect captured)
• In LAmax (which lamax1000ft represents): the BVI effect is attenuated

What AEDT / INM do

FAA AEDT and the legacy INM do not use a generic offset. They use Noise-Power-Distance (NPD) curves from the ICAO ANP database with separate curves per operating mode (approach, departure, flyover). The approach-departure difference is aircraft-specific and varies with thrust setting, speed, and altitude — not a fixed constant.

What ECAC Doc 29 does

Same approach: separate NPD curves per mode. No generic offset.

3. Verdict

The +2 dB offset is applied in the wrong direction relative to the codebase's own convention.

The reasoning:

1. lamax1000ft is a flyover measurement (confirmed by EPNdB correlation analysis).
2. The code assigns takeoffDb = lamax1000ft. In the EASA convention used everywhere else in the codebase, takeoff is the louder value. Using a flyover value for takeoff already slightly understates takeoff noise (flyover ≈ takeoff − 0.4 dB in EPNdB on average, so this is minor).
3. The code assigns approachDb = lamax1000ft + 2. This makes approach louder than takeoff by 2 dB.
4. Every other profile in the codebase (47 EASA entries) has approach quieter than takeoff by 2–4 dB.
5. The +2 offset therefore produces approach values that are 4–6 dB higher than they would be if the EASA convention were followed (the swing from −2 to +2 is a 4 dB net error).

The offset should be negative (approach quieter than flyover/takeoff), not positive.

Confidence level: high. The evidence is internally consistent: the EASA convention, the physical acoustics (reduced thrust on approach for total sound level), and the FAA data (where approach EPNdB is higher but LAmax1000ft — a flyover measurement — does not include approach-specific conditions) all point the same direction.

The one counterargument — that BVI makes helicopter approach noise genuinely louder — applies to EPNdB (which penalizes impulsive content) but does not reverse the LAmax relationship. The EASA profiles for the same helicopter types in this very file already account for this by making approach 2 dB quieter than takeoff, not louder.

4. Impact analysis

Aircraft types affected

Only helicopters with FAA ROSAP data. After the PR #102 unification, these are the types where getEASANoiseProfile() triggers the FAA override path. The 9 most common at KJPX:

Direction of error

The site overstates approach noise for all affected helicopter arrivals by ~4 dB. This is a consistent bias: every helicopter arrival at KJPX has its source level inflated by 4 dB above where the EASA convention would place it.

4 dB is perceptually significant — roughly a 2.5× perceived loudness difference.

Operational impact

At KJPX during summer season, approximately 30–50% of daily operations are helicopter arrivals (the primary noise concern). A rough estimate:

• ~10–20 helicopter arrivals per busy summer day
• Every one of these is overstated by 4 dB in all noise displays, trails, footprints, DNL calculations, and high-noise event counts

What users see

• Noise trails: helicopter arrival segments show noise colors as if 4 dB louder than the model should compute
• DNL: helicopter arrival SEL values are inflated, pushing DNL higher
• High-noise events: flights that would be just under the 65 dB threshold at a given distance may be incorrectly counted as high-noise events
• Noise footprint corridors: approach corridors for helicopters are wider than they should be

5. Options considered

Option A: Remove the offset entirely

Use lamax1000ft for both takeoff and approach. Simple but doesn't differentiate the two phases.

Option B: Flip the sign (change +2 to −2)

Aligns with the EASA convention. Simple mechanical change.

Option C: Use per-type EASA static values for approach

Use FAA measurement for takeoff, EASA-derived value for approach. Mixed provenance.

Option D: Restrict to helicopter-specific BVI-aware offset

Retain a positive offset for BVI. Defensible in EPNdB but inconsistent with LAmax convention.

Option E (selected): Flip the sign to −2 and document the convention

Aligns with the EASA convention (approach ~2 dB below takeoff). The −2 dB offset matches the helicopter-category average from the EASA data (−2.3 rounded). Simple, consistent, defensible.

6. Resolution

Option E was approved and implemented in PR #105. The offset was changed from +2 to −2, with a detailed attribution comment documenting:

• That lamax1000ft is a flyover measurement (per correlation analysis)
• That the −2 dB offset represents the EASA convention (approach quieter than takeoff/flyover)
• The history of the original +2 value and why it was incorrect