Runway Headings, Night Trains, and the Plant Nearby

A listing calls it a quiet residential street. The federal grade crossing inventory says thirty freight trains pass the end of that street after dark, on two main tracks, and that horns are not silenced there. Both statements describe the same address. Only one of them was checked.

Noise is the category where a scouting trip lies to you most easily. You visit on a Saturday afternoon, the wind is out of the south, and the airport is landing on the runway that keeps traffic away from the neighborhood. Eleven minutes on the sidewalk, nothing. Freight, meanwhile, is nocturnal by design, because railroads move it when the passenger slots are free. What you need is a record of the year, not a sample of an afternoon. Three federal systems keep one.

Runway numbers already tell you the flight path

Before any noise map, read the runway designators. The FAA's Aeronautical Information Manual defines them precisely: the runway number is the whole number nearest one-tenth of the magnetic azimuth of the centerline, measured clockwise from magnetic north (AIM 2-3-3). So Runway 17/35 is a strip lying on a 170/350 degree line. Aircraft using it are either descending toward a heading of 170 or climbing out on 350. Extend that line a few miles both ways on a map and you have drawn, roughly, where the traffic is.

The letters matter too. L, R and C mark parallel runways — the difference between one stream of aircraft over a neighborhood and two.

Designators are printed on the airport diagram, published free in FAA's Digital Airport/Facility Directory on a 56-day cycle. The Chart Supplement entry also names the calm-wind runway — the direction used when nothing forces the choice, a decent proxy for the default pattern.

What this will not tell you is which end is in use tonight. That is a wind question, and it flips. A house two miles off the departure end of one runway sits under climbing aircraft on some days and under nothing at all on others.

The contour map that exists for about sixty airports

Under 14 CFR Part 150, an airport operator may prepare a Noise Exposure Map showing contours at 65, 70 and 75 DNL. The rule's own term is the yearly day-night average sound level, averaged over a full year of days, and the nighttime period it works in runs 2200 to 0700 local, weighted ten decibels heavier than the daytime hours in the formula at Sec. A150.205. The night penalty sits inside the metric rather than being applied to it afterward.

The consequential part is Table 1 in Appendix A, the land use compatibility table. Read on 20 August 2026, it marks "residential, other than mobile homes and transient lodgings" as compatible below 65 and non-compatible in every band above it. Mobile home parks get a flat N above 65 with no footnote attached. Schools carry the same cells as housing. Hospitals, churches, auditoriums and concert halls get a 25 in the 65-70 band, meaning they are generally compatible only where the structure achieves 25 dB of outdoor-to-indoor noise level reduction. Offices are compatible straight through 70 before they need anything.

The footnote on those residential cells is the part worth reading twice. Note (1) says that where a community decides housing must be allowed anyway, building codes should reach 25 to 30 dB of outdoor-to-indoor reduction, then adds that ordinary construction already provides about 20 dB, so what is really being asked for is 5 to 15 dB over standard, and that such designs "normally assume mechanical ventilation and closed windows year round." The compatible version of a house inside a contour is a house you keep shut. The table also carries FAA's own disclaimer that its designations are not a federal determination that any land use is acceptable or unacceptable; that call belongs to local authorities.

Two things before you go hunting for the map.

It is voluntary. FAA's Airport Noise Compatibility Planning Information page, last updated 8 January 2026, collects the links that exist, and counting the airport identifiers on it on 20 August 2026 gets you to roughly sixty nationwide. Your candidate town's airport is probably not among them.

And the dates run old. FAA's state-by-state Part 150 status pages list, for Colorado, a Noise Compatibility Program approved 12 August 2008 at Centennial Airport (APA) in Denver with a revised Noise Exposure Map found in compliance 1 August 2017 — and, for Pueblo Memorial, a map in compliance since 10 September 1992. The Colorado page itself carries a last-updated stamp of August 2022. A 1992 contour describes a fleet that has since turned over twice.

So use the contour as a floor. DNL 65 is an annual average; it cannot tell you that a single 2 a.m. cargo departure rattles a window, because one event a night across a year averages down to almost nothing. It answers whether an address sits in the zone FAA calls incompatible with housing, not whether something will wake you up.

For that, go to the airport rather than to FAA. Many larger ones run a public flight-track and noise-monitor replay — search the airport name plus "noise office" or "flight tracker" — where you pick a date and watch the actual tracks over a neighborhood. FAA's own noise complaint and inquiry portal is a complaint channel, not a lookup tool. That an airport staffs a noise office at all is itself a signal about what the neighbors have been living with.

Thirty trains after dark, and whether they blow the horn

This is the best data of the three, and almost nobody uses it. FRA keeps an inventory record for every highway-rail crossing in the country, and the public front door is the Rail Crossing Locator, which accepts a street address. The same records sit behind an open feature service published in the USDOT National Transportation Atlas Database; the file I queried on 20 August 2026 was compiled 21 July 2026 and holds 242,108 crossings, of which 126,132 are public and at grade.

The fields that matter for a house: TotalDaylightThruTrains and TotalNighttimeThruTrains, MaximumTimetableSpeed, NumberOfMainTracks, and WhistleBan.

Eight public at-grade crossings came back for Flagstaff, Arizona — one town, one railroad, BNSF's Seligman Subdivision on the transcontinental main line — queried 20 August 2026. Six of them, in order along the line from west to east:

The two left out are Enterprise Road, also silenced round the clock, where the timetable speed drops to 10 mph, and one record flagged public whose street name field reads only PRIVATE. Every crossing also has its own inventory page on FRA's safety data site — Cosnino Road's is crossing 025125W, and the ID at the end of that address is the only part that changes.

Same line, same traffic, radically different nights. Every one of those crossings sees about sixty trains a day, with the nighttime column carrying half of them. The four with a ban sit between mileposts 344.3 and 340.5, the stretch running through town. At Cosnino Road, seven miles further east at milepost 333.2, there is no ban, and none again at Angell Turnoff eleven miles beyond that. A buyer comparing two houses eight miles apart on the same railroad is comparing two different products.

What "not silenced" costs is specified in regulation. Under 49 CFR Part 222, the engineer sounds two long blasts, one short and one long, beginning 15 to 20 seconds before the crossing and continuing until the locomotive occupies it, with a hard cap of a quarter mile in advance for trains moving faster than 60 mph. 49 CFR 229.129 requires the horn itself to produce a minimum of 96 and a maximum of 110 dB(A) measured 100 feet in front of the locomotive. Thirty night trains times roughly twenty seconds of that, at every crossing on the corridor, is the thing you are actually deciding about.

Nationally these bans are scarce. Of the 126,132 public at-grade crossings, 4,191 carry a round-the-clock one — about one in thirty. Entries reading Partial are rarer again: 172 among those public at-grade crossings, 205 across all 242,108 records in the file. Do not assume a town has one because it looks like the kind of town that would.

Three traps in this file. The first is that WhistleBan is a field on a self-reported inventory form, not a legal status, and it does not speak the regulation's vocabulary. Part 222 defines a New Partial Quiet Zone as one where horns are not routinely sounded between 10 p.m. and 7 a.m.; pre-rule and intermediate partial zones may cover some other stretch of evening or night, and a new zone of either kind has to run at least half a mile of right-of-way. A crossing coded Partial in the inventory is therefore not automatically a Part 222 partial quiet zone, and the two records can disagree. FRA publishes a separate list of quiet zone locations by city and state on its railroads.dot.gov site, and the public authority that established a zone is the one that can confirm its hours.

The second trap is that the file splits trains into a daylight column and a nighttime column without stating, in its published data dictionary, where the boundary between them falls. Thirty night trains is a real count of something. It is not a count of trains between two clock times you picked yourself.

The third is vintage. The Angell Turnoff Road record carries a highway traffic count — thirty vehicles a day — stamped 1982. Its train counts are current; its road count is from the Reagan administration. Read each field's date separately instead of trusting the record as a unit.

The obvious move is to look up the crossing nearest the property and stop there, and it is the wrong unit. The horn does not start at the crossing: § 222.21 puts it 15 to 20 seconds ahead of one, and never more than 1,320 feet ahead above 60 mph. A train working a corridor repeats the sequence at every crossing on that corridor, so a house sitting between two of them hears both. Pull every public at-grade crossing within about a mile and read them as a set.

What ECHO's circle actually counts

EPA's Enforcement and Compliance History Online will draw a radius around a point and return every regulated facility inside it. The headline count is the least useful number it hands back. Two circles, both two miles, both queried 20 August 2026, both centered on a coordinate printed here so you can run them again:

The totals sit within a fifth of each other, and the hazardous-waste row runs backwards from what you would guess: the mountain town has more RCRA registrations than the petrochemical one. RCRA sweeps in any business generating hazardous waste at all — repair shops, dental practices, print shops, university labs — so in an ordinary neighborhood that column is mostly a census of small businesses.

The columns that actually separate these two circles are the last two, and the names behind them separate them further. Flagstaff's three facilities with a quarter of non-compliance on the books are Northern Arizona University, the Rio de Flag wastewater plant and a Target store. Deer Park's nine include Shell Chemical at twelve quarters out of twelve, plus Lubrizol, Oxy Vinyls and Praxair. Two lists of similar length, describing different lives.

So: ignore the total, read the TRI and Clean Air Act counts, then open the individual records behind those two. ECHO also returns a total-penalties figure for the circle; leave it alone unless you are willing to work out which facility, which statute and which year each dollar came from, because the number by itself will not tell you.

For chemicals specifically, EPA's TRI Toxics Tracker accepts a street address directly and defaults to a 10-mile radius, expandable to 100 if nothing turns up nearby. It filters by year, chemical and industry sector, which is how you learn whether the one reporting facility down the road releases a solvent to air or a metal to water. Different things to live beside.

Three lines on the sheet instead of three adjectives

None of this decides anything by itself. It converts "seems quiet" into cells you can compare across two candidate addresses, the same way a commute becomes a range rather than a number once you measure it properly.

The air line is two lookups and one absence: the runway designators and the bearing they imply, the offset between the address and the extended centerline, and whether a Part 150 contour exists for that airport at all. For most airports it does not, and that blank is a finding rather than a gap.

Rail is a list, not a number — every public at-grade crossing within a mile, each with its night count, its timetable speed and its whistle ban status, because the corridor and not the nearest crossing is the unit. Ground is two counts from ECHO, TRI reporters and Clean Air Act permits inside two miles, and then the individual records behind whichever of those is not zero.

Every cell wants two dates on it: the day you pulled it and the day the file behind it was compiled. The crossing numbers above came out of a file compiled 21 July 2026. The Centennial contour a Denver buyer gets pointed at was found in compliance in 2017, and Pueblo's in 1992.

Then go stand there, once after dark and once early. That is the real payoff of the lookups: they tell you when to be on the sidewalk and what to listen for, so a visit confirms or refutes something specific instead of producing one more adjective.

It belongs to the housing line in the seven-number frame, and pairs with the demographic pull in the Census tables worth opening. Those tell you who lives around an address. This tells you what passes over and beside it.

Sources, read or queried 20 August 2026: 14 CFR Part 150 for the DNL definition and Appendix A Table 1, 49 CFR Part 222 for horn timing and quiet zone categories, and 49 CFR 229.129 for horn sound levels, all via eCFR; FAA AIM 2-3-3 for runway designators; FAA Airport Noise Compatibility Planning Information and the state-by-state Part 150 status pages; FRA Rail Crossing Locator, the per-crossing inventory pages on FRA's safety data site, and the underlying USDOT National Transportation Atlas Database Railroad Grade Crossings service, compiled 21 July 2026, queried directly for Flagstaff and for the national counts, with its published data dictionary for field definitions; EPA ECHO facility search, queried by coordinate and two-mile radius at the two points named above; EPA TRI Toxics Tracker.

The FAA, the FRA and the EPA compiled these files for their own purposes, and rating a house was not among them. Noise contours exist for compatibility planning, the crossing inventory for grade-crossing safety, ECHO's counts for enforcement priorities. Read each for the job it was built to do and it holds; read it as a verdict on one address and it does not. A verdict of that kind takes an acoustician, an air-quality specialist and a night actually spent on the sidewalk — none of which is a thing a query returns. The queries themselves are public and will run for any address. What counts as too loud stays yours to decide.