CelesTrak SATCAT Catalog Search & Downloads

Every tracking app you have ever opened is standing on this table. The Online Satellite Catalog - SATCAT - is the master registry of tracked Earth-orbiting objects maintained by Dr. T.S. Kelso under the CelesTrak name, and where position feeds answer where is it now, SATCAT answers the questions that survive longer than an ephemeris: whose the object is, when it launched, whether it is still up, and if it came down, when.

The scale is the argument. The boxscore current as of 21 August 2026 counts 70,355 catalogued objects: 19,894 payloads on orbit, of which 16,885 are active, plus 14,975 pieces of debris on orbit. Each row carries seventeen fields - identity in three forms (common name, COSPAR designator, NORAD catalog number), object class, owner code, launch date and site, decay date where applicable, period, inclination, apogee, perigee, radar cross-section and status flags.

The ledger opens at the beginning: NORAD catalog numbers 1 and 2 are both dated 4 October 1957 - the SL-1 rocket body that lifted it, and Sputnik 1 itself. Sixty-nine years of launches and decays sit in one consistent row shape.

Around the rows, the family publishes derived views that answer management questions directly: the SATCAT Boxscore splitting the population on orbit versus decayed by country and organization, a Launch Boxscore by year and site, an Analyst Satellite Catalog holding objects excluded from the main count, and a Discrepancy Report comparing this registry against Space Track. Datadory scores this record 9 out of 10 against a catalog mean of 7.81, with field definitions verified during research. Get a sample of this dataset and the dictionary ships with it.

The first three catalog numbers, flat exactly as they arrive:

# Rows 1-3 - catalog numbers 1 through 3, exactly as delivered
object_name     : SL-1 R/B
object_id       : 1957-001A
norad_cat_id    : 1
object_type     : R/B
ops_status_code : D
owner           : CIS
launch_date     : 1957-10-04
launch_site     : TYMSC
decay_date      : 1957-12-01
period          : 96.19
inclination     : 65.10
apogee          : 938
perigee         : 214
rcs             : 20.42

object_name     : SPUTNIK 1
object_id       : 1957-001B
norad_cat_id    : 2
object_type     : PAY
ops_status_code : D
owner           : CIS
launch_date     : 1957-10-04
launch_site     : TYMSC
decay_date      : 1958-01-03
period          : 96.10
inclination     : 65.00
apogee          : 1080
perigee         : 64
orbit_center    : EA
orbit_type      : IMP

object_name     : SPUTNIK 2
object_id       : 1957-002A
norad_cat_id    : 3
object_type     : PAY
ops_status_code : D
owner           : CIS
launch_date     : 1957-11-03
launch_site     : TYMSC
decay_date      : 1958-04-14
period          : 103.74
inclination     : 65.33
apogee          : 1659
perigee         : 211
rcs             : 0.08

# The derived layer - SATCAT Boxscore aggregates, current as of 2026-08-21
catalogued_objects : 70,355
payloads_on_orbit  : 19,894    (active: 16,885)
debris_on_orbit    : 14,975

Read rows one and two together, because they belong together: catalog number 1 is the SL-1 rocket body that carried Sputnik 1 to orbit, launched the same day from the same site - TYMSC, Baikonur - and both were gone within months, decay dates stamped December 1957 and January 1958. Row three, Sputnik 2, rode higher: apogee 1,659 km against Sputnik 1's 1,080, and it stayed up until April 1958.

Then notice what is missing, because the gaps are honest. Sputnik 1's row carries no RCS value - radar cross-section is published where known and left blank where not, which is exactly what a measured quantity should do. The owner code reads CIS, the code family applied to Soviet-heritage objects, a reminder that these codes are conventions worth resolving once and keeping.

The bottom block is the derived layer doing its own work: the boxscore publishes the catalog's arithmetic - 70,355 catalogued, 19,894 payloads on orbit with 16,885 active, 14,975 debris - so headline counts arrive computed rather than waiting on a tally pass over seventy thousand rows.

Seventeen documented fields per object, definitions verified against the published layout during research, doing four jobs.

Identity: OBJECT_NAME, OBJECT_ID (the COSPAR designator, YYYY-NNNAAA) and NORAD_CAT_ID. Three keys per row sounds redundant until the day you need them - names collide and change, designators encode launch year and piece, and the NORAD number is the integer that joins every other orbital dataset in existence.

Classification and ownership: OBJECT_TYPE (PAY, R/B, DEB, UNK) separates payloads from rocket bodies and debris in one character, and OWNER carries the country or organization code - states like US, CIS and PRC alongside operators like IRID, SES and EUTE, which is what makes fleet-attribution queries possible without a corporate-name-matching exercise.

Lifecycle: LAUNCH_DATE and DECAY_DATE bracket each object's orbital career, LAUNCH_SITE records where it started, and OPS_STATUS_CODE compresses current state - + active, - inactive, D decayed - into one filterable character. Isolating the 16,885 active payloads is a predicate, not a derivation.

Geometry and observability: PERIOD, INCLINATION, APOGEE, PERIGEE describe the orbit in minutes, degrees and kilometres; RCS adds radar cross-section in square metres where measured. DATA_STATUS_CODE, ORBIT_CENTER and ORBIT_TYPE round out the quality and classification layer - including central-body codes that reach past Earth orbit to cislunar, lunar and solar-centered regimes.

The core is the seventeen-column object table. Around it, the catalog family's derived views map onto your use case at sampling:

• SATCAT Boxscore expansion - objects on orbit versus decayed, split by country and organization, resolved into dimension rows ready to sit beside the object table.
• Launch Boxscore - the same arithmetic by launch year and launch site, which is a cohort table for anyone studying cadence.
• Owner-code resolution - the compact codes joined out to readable country and organization names, so the table reads for humans as well as machines.
• Analyst Satellite Catalog - the objects kept out of the main count, preserved as a visible side table rather than deleted from history.

Name the ones your models need when you scope the sample; column naming locks down for your pipeline at the same moment.

Geography - global by construction. Every tracked Earth-orbiting object from every launching state and operator appears in the same table, organized by object rather than by territory, which is precisely what makes cross-country comparisons honest: the United States, China, Russia and every operator constellation share one row shape.

Temporal - from Sputnik 1, launched 4 October 1957, through the present catalog. Both endpoints of each object's career are dated - launch at the start, decay at the end - and decay stays blank while the object remains in orbit, so the blank is information, not absence.

Granularity - one row per catalogued object, roughly 70,355 as of 21 August 2026, spanning payloads, rocket bodies and debris. No rollups, no country averages in the base table: every downstream aggregation builds from the atomic unit, and the boxscore supplies the pre-computed headlines when you want them.

Set against Datadory's wider catalog - 1,744 datasets across 159 viable industries, average quality 7.81 - this slice scores 9/10, ranking seventh on our best cable-satellite datasets list while sitting in the building products pool it shares.

API, files, or your warehouse. Daily, weekly, or hourly.

Pick the channel your team already works in: REST lookups for interactive slices, flat files sized for overnight loads, or a direct pipe into Snowflake, BigQuery or Redshift. Every delivery ships with the full seventeen-field dictionary, the owner-code resolution and validation samples, so your engineers validate the shape before anything recurring starts.

Cadence follows the question. Weekly builds the entry-and-exit log - new launches arriving, decays stamping themselves - which is the longitudinal record nobody keeps for you. Hourly rarely pays for itself against a registry whose rows describe careers measured in years.

Name the owners, object classes or date windows when you request the sample and it comes back shaped to them.

• Debris-population studies - object class and radar cross-section on tens of thousands of rows give population models their denominator and their observability term. See data scientists use cases.
• Constellation churn analysis - launch-to-decay histories since 1957 quantify attrition rates per operator generation, the activity signal behind space-economy theses; investors and quants use cases picks up the method.
• Broadcast-fleet attribution - owner codes plus orbital geometry isolate the geostationary arc, answering whose transponders are up there without a name-matching project.
• Space-activity reporting - boxscore counts by launching state are the citable numbers in policy and news coverage; journalists and academics use cases explains why provenance matters here.
• Pipeline plumbing - NORAD_CAT_ID as the universal join key lets product builders wire registries to positions and elements once and reuse the spine everywhere.
• Fleet-watch alerts - newly catalogued payloads arrive as dated rows with owner attached, a watchtower for anyone tracking capacity additions.

Journalists, academics and students rank highest at relevance 3 - this is the citable public catalogue, with boxscore summaries built for quotation and methodology documented well enough to footnote. Data scientists and ML engineers hold relevance 3 as well: typed rows, designator filters and lifecycle dates form clean panels for debris-population and lifetime models.

Developers and data-product builders take relevance 3 for the schema itself - a stable, documented table keyed on the one integer every orbital system shares. At relevance 2, investors and quant researchers read churn and attrition out of launch-to-decay pairs. Competitive intelligence teams sit at relevance 1 with a narrow but real job: watching the ledger for newly catalogued payloads entering rival constellations.

Honestly: this record landed in the Building Products pool through its cable-and-satellite tag, and the vertical around it is specification directories and certification registries. That mismatch is a feature for anyone working the demand chain - the satellite industry buys building products too, and this registry is the asset register behind every ground station, teleport and antenna farm the rest of the slice describes.

Within orbital data, SATCAT owns the registry layer. CelesTrak Current GP Element Sets (NORAD TLE / SupGP) holds the propagation-grade orbital elements - the where-it-will-be layer, scoring 10/10 in our catalog. Jonathan McDowell's General Catalog of Artificial Space Objects (GCAT) carries the engineering biography - mass, bus, manufacturer, geosync slot longitudes. N2YO Live Real-Time Satellite Tracking Directory positions 35,103 objects right now, and the UCS Satellite Database freezes 7,560 operating satellites with deep attributes at May 2023 as a citable baseline.

They stack rather than compete: SATCAT supplies the keys and the lifecycle ledger the others hang from. The head-to-head sits on our SATCAT vs GCAT comparison.

Four things, stated plainly.

First, blanks are measurements too. RCS is empty where nobody measured it, DECAY_DATE is empty while the object stays up, DATA_STATUS_CODE varies by row. Model the missingness as what the catalog knows rather than imputing courageously.

Second, owner arrives as codes. US, CIS, PRC, IRID, SES, EUTE and friends are compact conventions, not full legal names - the resolution table comes with the delivery, and corporate-family questions (one operator, many codes) deserve a pass before any rollup.

Third, totals are dated snapshots. The 70,355 figure is the 21 August 2026 boxscore; the ledger moves as objects launch and decay. Cite totals with their date, and let a recurring cadence accumulate the movement the snapshot cannot show.

Fourth, units travel with the columns. Period in minutes, inclination in degrees, altitudes in kilometres, cross-sections in square metres. Join on the unit, not on habit - and note the orbit codes reach beyond Earth (cislunar, lunar, solar-centered regimes), so filter ORBIT_CENTER before assuming everything in the table circles this planet.