Method v1.1

How we count flights and estimate noise

Every number on OverMe comes from the rules on this page. If a rule changes, the version number changes and the change is logged at the bottom.

What counts as an overflight

A flight counts for a neighborhood when, at its closest point, it is within 1.5 nautical miles of the neighborhood's center and below 5,000 feet above the ground. That is about 1.7 statute miles, or 2.8 km. The distance is measured in a straight line through the air, so altitude is part of it. A plane 1 nautical mile up and directly overhead is 1 nautical mile away.

• One pass, one count. We follow each aircraft's track and find the point where it came closest to the center. A new pass starts after 10 minutes with no signal from that aircraft.
• Gaps are filled in. Our archive keeps fewer points on straight, steady flight. We draw a straight line between points up to 3 minutes apart to find the closest approach.
• Ground traffic is removed. We drop positions under 50 feet above the ground or slower than 50 knots, except for helicopters. That removes taxiing aircraft.
• Altitude is above the ground. Aircraft report pressure altitude. We correct it with that hour's typical gap between GPS and pressure altitude, then subtract the ground elevation. Expect about ±150 feet of error.
• Night means 10 p.m. to 7 a.m. local time. Low means below 1,500 feet at the closest point.

Today each neighborhood is a circle around a named center point. Street-level areas built on a grid of hexagonal cells (the open H3 system, about 0.7 km² per cell) are being added. They use the same distance and altitude rule, measured from each cell's center.

Where the data comes from

Most aircraft broadcast their GPS position, altitude, speed and identity about twice a second. This signal is called ADS-B. US rules have required it in most controlled airspace since 2020. Anyone with a small radio receiver can pick it up.

• The ADSBiq network. Home receivers run by volunteers, plus partner and community ADS-B networks that share data with ADSBiq. All feeds are merged into one archive with one position per aircraft per second. Positions that fail basic physics checks (impossible jumps in place or speed) are flagged and left out.
• Our own share. Around Palm Beach International, ADSBiq's own receivers supplied 37 to 46 percent of low-altitude positions in summer 2026. The rest came from partner networks.
• Aircraft type. Each aircraft's transponder address is matched to its type in a public aircraft database. 97 percent of counted flights matched a known type class. The rest are classed from the category the aircraft broadcasts, or shown as "unknown."
• Live view. The "right now" map uses the same network, updated every second. It is for watching. Counts and scores come only from the archive.

How we check coverage

A receiver has to hear an aircraft to count it. Near the ground, buildings, trees and the curve of the Earth block the signal. So the lowest altitude we can see in an area depends on how close a receiver is. Coverage is the biggest single reason a count could be too low.

Around Palm Beach International, coverage reaches very low. This table shows how low the lowest 1 in 20 passes near each neighborhood center were, and how many passes were below 1,000 feet, across 133 days. Arrivals and departures tracked a few hundred feet off the ground show that receivers hear aircraft that low here.

We test coverage two ways today:

• Remove a source and see what changes. From Aug. 14 to Sept. 28, 2026, ADSBiq's own receivers were missing from the archive by mistake. When we restored them, daily counts rose only 1 to 4 percent. The other networks had already seen nearly every flight.
• Look at the empty hours. Hours with no flights should match the airport's quiet hours (about 1 to 5 a.m. at PBI). Empty hours at busy times would point to a gap. We have not found any.

The coverage badge on each page

Every neighborhood and street page shows "Coverage here: good, fair or limited." It comes from a coverage map of hexagonal cells (about 0.7 km² each) within 15 nautical miles of 107 US airports, rebuilt from the last 14 days of tracks. For each cell and its six neighbors we measure two things: how often low tracks continue without a gap of more than a minute, and the lowest altitude at which aircraft are reliably seen.

• Good: at least 90% of low tracks are continuous, and we see aircraft down to about 1,000 feet (2,000 feet where no track was lost).
• Fair: at least 75% continuous, and seen down to about 2,500 feet. Most flights are counted; some very low ones may be missed.
• Limited: anything else. Counts here may be too low, especially for low flights.
• Not rated: too little low traffic to judge, or outside the area mapped so far.

Where coverage is limited, one home receiver nearby usually fixes it. See adsbiq.com/join.

How estimated noise is computed

Nothing on OverMe is a microphone reading. For each flight we estimate two numbers at the neighborhood center: the peak level (Lmax, in A-weighted decibels) and the total sound energy of the pass (SEL).

1. A reference level by aircraft class

Each class gets a typical peak level for a departure 1,000 feet away. These are class averages, calibrated against measured flights near Palm Beach International (see validation). They are not per-model certification data.

Arriving aircraft, with engines near idle, get −8.0 dB. Aircraft in level flight get −3.1 dB. Climbing or descending means more than 300 feet per minute at the closest point.

2. Distance

Sound spreads out and fades with distance, and the air absorbs some of it. We use the slant distance at closest approach, with a floor of 200 feet, and one fitted term that covers both effects:

That is about 6.7 dB less for each doubling of distance. The 22.4 comes from the same monitor data as the reference levels.

3. Sound exposure (SEL)

A slow, close pass lasts longer, so it delivers more sound energy than a fast one with the same peak. SEL adds that up, where v is ground speed:

The daily score

The daily score adds up every counted flight's sound energy over 24 hours, the way the FAA's day-night average sound level (DNL) does. Flights between 10 p.m. and 7 a.m. count as if they were 10 dB louder.

A score 10 points higher means ten times the sound energy. Most people hear that as roughly twice as loud.

The score is "DNL-style," not an official DNL. It covers only aircraft we counted inside the circle. It leaves out other sounds. It is not comparable to the FAA's modeled noise contours or to a calibrated noise monitor.

Outside the US. Europe uses Lden instead. It splits the day into day (7 a.m. to 7 p.m.), evening (7 to 11 p.m., +5 dB) and night (11 p.m. to 7 a.m., +10 dB). When OverMe covers European areas, scores there will use Lden weighting and say so. All areas today are in the US.

Known limits and error ranges

• Noise is calibrated at one airport. Near PBI, a single flight's estimated peak level is typically within about 3 dB of a calibrated outdoor monitor. It can be off by more for piston planes, helicopters and flights more than 1 nautical mile away. The calibration comes from six monitors in West Palm Beach and Palm Beach; elsewhere, expect larger errors until we compare with monitors there. The model has no per-model noise tables, no terrain effects, no weather, no buildings, and no indoor-versus-outdoor difference. Comparisons (day to day, before and after a change, one neighborhood to another) are sturdier than any single number, because most errors repeat the same way.
• Some aircraft don't broadcast ADS-B. It is not required in all airspace. Some helicopters, older light aircraft, gliders and many military flights are missed. Low, hovering helicopters are our weakest case.
• Coverage gaps. Very low flights far from a receiver can be missed. See coverage.
• Position error. Closest approach can be off by a few hundred feet where points are far apart. Altitude is about ±150 feet around PBI. Street-level areas elsewhere don't yet have the pressure correction, so expect about ±300 feet there.
• Data delay. Daily figures are not same-day. Our archive closes each day about 33 hours after it ends (UTC). Until then, yesterday is shown from a faster preliminary feed and replaced with the final numbers automatically. Every page says which day it covers.
• History has gaps. The record has 133 days: Oct. 1, 2025 to Nov. 9, 2025; June 30, 2026 to Sept. 29, 2026; Oct. 1, 2026 to Oct. 1, 2026. Summer is the quiet season at PBI. Don't assume winter looks the same.

Privacy filtering

• Private aircraft appear by type only. We show a flight number only when it is a public airline flight number (like JBU1234). Private and charter registrations are never shown.
• No aircraft IDs are kept. The OverMe database stores no transponder addresses or registrations. The live map strips them on our server before anything reaches your browser.
• FAA privacy programs. Owners can ask the FAA to limit tracking of their aircraft (LADD) or use a private address (PIA). Today we read those flags from a public, community-maintained aircraft database, and suppress the flight number of any flagged aircraft. We are adding the FAA's own LADD data feed. Because we hide every non-airline identity anyway, a blocked aircraft is never named on OverMe.
• No people. We never show a receiver's location or anyone's address. An address you search is used once to find your neighborhood and is not stored. See the privacy policy.

Validation

We checked OverMe against two things it does not use: microphones and the FAA's own operation counts.

Against calibrated noise monitors

Six Class-1 sound level meters near PBI are run by residents and published at pbinoise.com. They are independent, resident-run monitors, and their public data made this check possible. We are grateful for it. From June 30 to Sept. 29, 2026, the monitors logged 56,988 aircraft noise events. We matched 52,115 of them (91%) to a flight in our data, by aircraft identity and time. At closest approach, the two systems placed the aircraft within 33 meters of each other (median).

For 48,481 matched flights inside our counting rule, leaving out events the monitors flagged for wind or rain:

• By aircraft class (v1.1 bias, then typical error): Airliner +0.0 and 2.9 dB; business jet -0.1 and 3.6 dB; unknown type -0.1 and 3.2 dB; turboprop -0.2 and 5.1 dB; heavy jet -0.1 and 2.4 dB; piston -0.6 and 5.8 dB; helicopter -0.8 and 6.1 dB. Piston planes and helicopters are few and vary most.
• Distant flights. Beyond 1 nautical mile we still read low. A monitor only logs a flight that stands out from background noise, so the distant flights it does log are the unusually loud ones. These are under 2% of matches.
• Flights louder than 65 dB, day by day. Our daily count tracks each monitor's count: correlation 0.96 at Everglades Island, 0.92 at Historic El Cid, 0.91 at Prospect Park, 0.90 at Tarpon Bay, 0.89 at Flamingo Park, 0.88 at Poinciana Park. We count more of them, because we include every pass and a monitor counts only what it can pick out of background.

Against FAA tower counts

Our daily count of flights near the airport (within 4 nautical miles and below 2,500 feet) matches FAA tower operations (ATADS) at a correlation of 0.98 over 49 days. We count about 4% more, mostly low transits and helicopters that the tower count leaves out.

What this means

Our counts are measured. Our decibel figures are estimates, good to about 3 dB per flight near PBI. They are best for comparing days, places and before/after changes. They are not a substitute for a calibrated meter. The full validation note, with every table, is available from [email protected].

Versions and changes

This is method v1.1. The version is printed in every data download. When we change a rule, we raise the version and log the change here.

Data notes

Changes to the underlying data, not the method:

• Oct. 2, 2026. ADSBiq's own receivers were missing from the archive for Aug. 14 to Sept. 28, 2026. They were restored and those days recounted. Daily counts rose 1 to 4 percent; estimated scores moved by 0.5 dB or less in every neighborhood.

Found a problem? See our corrections policy.

We count planes; you decide.