Telecom Tower REITs data: the federal registry, the observed grid and the listed tape, in one catalog. · Head-to-head
OpenCellID Bulk Downloads vs Data.gov Antenna Tower Datasets Catalog
Which telecom tower reits data: the federal registry, the observed grid and the listed tape, in one catalog. data fits your job: OpenCellID Bulk Downloads, or Data.gov Antenna Tower Datasets Catalog. API, files, or your warehouse. Daily, weekly, or hourly.
OpenCellID Bulk Downloads
Data.gov Antenna Tower Datasets Catalog
Where the fields line up
No shared field names. These two answer different questions.
| Field | OpenCellID Bulk Downloads | Data.gov Antenna Tower Datasets Catalog |
|---|---|---|
radio | documented | not in this set |
mcc | documented | not in this set |
net | documented | not in this set |
area | documented | not in this set |
cell | documented | not in this set |
unit | documented | not in this set |
lon | documented | not in this set |
lat | documented | not in this set |
range | documented | not in this set |
samples | documented | not in this set |
changeable | documented | not in this set |
created | documented | not in this set |
What each contains
Pick by fit, not by loyalty.
| OpenCellID Bulk Downloads | Data.gov Antenna Tower Datasets Catalog | |
|---|---|---|
| Position | `lat` / `lon` decimal degrees on every row - exact GPS when `changeable=0`, measurement average otherwise | Point geometry inside each municipal or state layer, one hop beyond the catalog card |
| Time stamping | `created` and `updated` Unix seconds per cell, inside a rolling 18-month window | `dataset_last_updated` date plus `catalog_last_checked` audit timestamp per record |
| Radio technology | `radio` enum: GSM, UMTS, LTE or CDMA, with `net` for the network code | None - inventory layers catalogue structures, not transmitting equipment standards |
| Site identity | `cell` plus `area` and `unit`: cell ID, location/tracking area, scrambling code or PCI | `title` and `organization` naming each inventory; no per-structure identifier in the card dictionary |
| Signal evidence | `samples`, `averageSignal` and `changeable` recording measurements behind each cell | None - closest analogue is `views_last_month`, which measures catalog interest, not RF activity |
| Publisher attribution | Implicit - one project, stewarded by Unwired Labs under the OpenCelliD name | Explicit per result: `organization` plus an `org_type` badge for Federal, State, County or City |
| Descriptive text | Field semantics documented in the project's schema reference rather than carried per row | `description_snippet` truncating each dataset's own notes onto the card |
| Fitness metadata | `range` estimating each cell's coverage radius in metres | `search_relevance` scoring each card against the query terms |
What each does better
OpenCellID Bulk Downloads
Global reach and uniform schema. Coverage spans countries worldwide, partitioned by mobile country code, with per-country cell totals published alongside - the United Arab Emirates alone shows 209,465 logical cells (111,996 UMTS, 52,836 LTE, 44,621 GSM, 12 NR), Albania 19,237, Afghanistan 16,186. Whatever market your tower REIT thesis touches, the same fourteen columns describe it; nothing in the American public-inventory world crosses a border.
Signal-grade attribute depth. Each row carries more than a pin on a map: range estimates the cell's coverage radius in metres, samples counts the measurements behind it, changeable separates survey-grade positions from measurement-derived ones, and averageSignal preserves the observed strength. That is enough to model coverage footprints, spot gap zones against a competitor's footprint, or weight confidence before joining cells to owned-tower portfolios - the kind of work catalog cards cannot support.
Self-cleaning recency. The 18-month observation window means every export reflects cells actually seen recently, with per-cell created and updated timestamps letting you distinguish long-standing sites from transient ones. For portfolio screening across dozens of countries at once, that consistency is the product.
the Data.gov Antenna Tower Datasets Catalog
Physical structures, officially attested. These are permitting authorities publishing their own registers: Connecticut Siting Council telecommunications towers and antennas current to July 2026, Los Angeles small cell nodes attached to streetlight poles current to August 2026, Loudoun County telecom towers, San Francisco's existing and proposed commercial wireless facility sets. When a question is 'which structures exist and who permitted them', an inventory beats an inference every time - see antenna structure registration for the federal counterpart regime these local layers complement.
Sub-federal discovery no national registry offers. Because Data.gov harvests agency and municipal publishers directly, this query surfaces city and county tower point files that never enter FCC-style tower registration - exactly the tier where small cell densification actually happens. Eight results sound thin next to millions of cells until you remember each result is an entire maintained inventory, most offering several file renditions apiece.
Honest provenance metadata. Every card names its publishing organization and level of government, carries the underlying modification date, and shows the catalog's own link-check timestamp - so you can tell a living layer from a frozen one. That candour cuts both ways: San Francisco's existing-facilities set is explicitly flagged as no longer maintained, and two of the eight matches are tangential federal studies (a NIST 3.5 GHz coastal sensor placement simulation and DOI radar imagery of the Ivanpah solar tower) rather than tower inventories at all. Curation is still required; at least the catalog admits it.
The verdict
Verdict: sample both, pick by fit. Let the unit of analysis decide.
If you are verifying structures - due diligence on a specific American market, counting permitted verticals in a county, grounding a density narrative in permit-grade records - the Data.gov Antenna Tower Datasets Catalog reaches the official inventories, provided you accept the pruning: eight matches, some frozen, two off-topic. Analysts usually start with OpenCelliD for breadth and pull the catalog for the few markets where an authoritative register changes the answer. Either way, cut each sample to your actual markets before judging.
Sample both, pick by fit. See OpenCellID Bulk Downloads · See Data.gov Antenna Tower Datasets Catalog
Fair questions
Do logical cells from OpenCellID correspond one-to-one with physical towers?
No, and treating them as if they do is the classic mistake. An OpenCellID row is a logical cell identified by radio standard plus mcc, net, area, cell and unit - one physical structure can host several sectors or technologies, and measurement-derived positions can sit slightly off the mast. Aggregate cells to a spatial tolerance before comparing counts against a physical inventory such as those surfaced by the Data.gov antenna tower query.
Which dataset covers more geography?
OpenCellID, decisively: coverage spans countries worldwide, partitioned by mobile country code, with per-country cell totals published - from hundreds of thousands of cells in dense markets to thousands in smaller ones. The Data.gov antenna tower results are United States-only, skewed to the states and municipalities that publish their own tower GIS layers: California cities, Connecticut, Loudoun County Virginia.
Which of the two is more current?
Different kinds of current. OpenCellID keeps a rolling 18-month observation window with per-cell created and updated timestamps, so exports always reflect recently observed cells. The Data.gov results are a snapshot over continuously harvested catalog metadata: individual layers range from a 2019 NIST study to Los Angeles small cells touched in August 2026 and Connecticut Siting Council filings current to July 2026 - while San Francisco's existing-facilities set is flagged as no longer maintained. Check the vintage that matters for your market rather than trusting either wholesale.
What fields do the two datasets actually share?
Position and time, conceptually. Both encode latitude and longitude - OpenCellID as typed lat/lon decimal columns per cell, the catalog through point geometry inside the layers its records lead to - and both carry timestamps, per-cell Unix created/updated moments on one side versus dataset_last_updated and catalog_last_checked audit dates on the other. Beyond that the dictionaries diverge: radio identifiers and measurement attributes versus organization, government level and format metadata.