Methodology

We use FCC Antenna Structure Registration records whose status is Constructed. Registrations that are granted but not yet built, or that have been terminated, are excluded: someone searching for structures near them wants the ones that exist. That leaves 141,542 structures.

Records outside the 50 states, the District of Columbia and Puerto Rico are excluded, because the geographic hierarchy this site is built on does not cover the other territories.

Height is the overall height above ground level as filed, which includes antennas and lightning rods on top of the supporting structure. That is why a tower's overall height can exceed the height of the tower itself.

The registration also records ground elevation and overall height above sea level. Those three figures are internally consistent in all but 22 of 197,269 records, which is one of the checks we use to confirm the fields are being read correctly.

Heights are filed in metres. Feet are converted at 0.3048 m per foot and rounded.

Height ranks are calculated within a state, county or place across all constructed registrations in that area — not against a sample. A percentile of "taller than 96%" means that 96% of the other registered structures in that state are shorter.

Comparisons against a "median for this type" use the median height of registrations sharing the same structure type in the same state, and are only shown when that group has at least 20 members. Below that the median is too noisy to say anything useful.

Distances use the haversine formula on a spherical earth of radius 6,371 km. At the scales on this site the difference from a proper ellipsoidal calculation is well under the accuracy of the filed coordinates themselves.

Bearings are true bearings, given as 16-point compass directions.

The "radio horizon" figure is d ≈ 4.12 × √h, with height in metres and distance in kilometres. This is the standard 4/3-earth-radius approximation, which accounts for the way the atmosphere refracts radio waves slightly around the curve of the earth.

It is a geometric ceiling, not a coverage prediction. It assumes flat, unobstructed terrain and says nothing about transmit power, frequency band, antenna pattern, or what is in the way. Real usable range is normally much shorter. We show it because height is the one variable the public record actually gives us, and because it makes the relationship between height and reach concrete.

Structures are assigned to the place named in the registration where that name matches a Census place in the same state, and to the nearest Census place centroid otherwise. Where a name match would put a structure more than 60 km from that place, we treat it as a same-name-different-place error and fall back to the nearest centroid. Each page shows the distance to the place it is assigned to, and describes a structure as being "near" rather than "in" a place when it sits more than 8 km from the centre.

Per-capita figures use Census Population Estimates for 2025. They are omitted entirely — not shown as zero — where no population figure exists. That affects Census Designated Places and Puerto Rico's zonas urbanas, which the population estimates programme does not cover.

The prose on each structure page is generated from that structure's own figures using a fixed set of sentence patterns, chosen deterministically from the registration number. Every statement in it is a restatement of a number shown elsewhere on the same page. Where the data does not support a particular statement, the sentence is omitted rather than filled with a placeholder.

No large language model writes or rewrites these summaries, and nothing on the site is generated from sources other than the datasets listed on the data sources page.

If a figure here disagrees with the FCC's own record, the FCC is right and we have a bug. The registration number on each page is the key to look it up at the source. Pylon Plans does not modify, interpolate or estimate the underlying filings.