How Satellite Tracking Works: TLEs, SGP4 and Orbits

Satellite tracking works by prediction rather than reception. A tracker downloads a compact description of each satellite's orbit, called an element set, and runs it through a standard mathematical model (SGP4) to calculate where the satellite should be at any moment. The dot you see moving across a satellite map is the output of that calculation, refreshed many times a second in your browser.

That is very different from how aircraft and ships are tracked. Planes and vessels broadcast their own GPS positions over radio, which is why flight maps can show a position that is only seconds old (see how ADS-B flight tracking works). Satellites have no equivalent public beacon. Their orbits are measured by ground radars and telescopes, fitted to a model, and published.

This guide explains where the orbit data comes from, what is inside a two-line element set, how SGP4 works at a practical level, what NORAD catalog numbers are and why they changed in 2026, how ISS pass predictions and Starlink "trains" work, and where the numbers stop being trustworthy.

Open the live satellite tracker

What is satellite tracking?

Satellite tracking is the process of determining where an object in orbit is now and where it will be later. Professional tracking is done by networks of radars and optical sensors that observe objects and refine their orbits. Public tracking, the kind you do on a website or app, reuses the results of that work.

There are two layers to keep separate. The first is orbit determination: measuring an object repeatedly and solving for the orbit that best fits those measurements. The second is orbit propagation: taking a known orbit and projecting it forward or backward in time. Public trackers only do the second part. They depend on someone else having done the first.

Where does satellite orbit data come from?

The public satellite catalog is maintained by the US Space Force's 18th Space Defense Squadron, which detects, tracks and identifies artificial objects in Earth orbit and publishes the results through Space-Track.org. NASA describes the squadron as the unit responsible for that catalog maintenance. Space-Track requires a free account and acceptance of a user agreement.

CelesTrak republishes the same general perturbations (GP) data in ready-made groups such as space stations, Starlink, weather satellites or the last 30 days of launches. Most hobby and web trackers, including GodsView AI, pull from CelesTrak. CelesTrak states that its GP data updates once every two hours and asks automated clients not to download the same data again before it changes.

What GodsView uses

GodsView loads satellite orbits from CelesTrak and refreshes them about once an hour, which is free to view on the map. The full list of feeds and refresh rates is on the data sources and methodology page.

What is a TLE (two-line element set)?

A TLE, or two-line element set, is a fixed-width text format that packs one satellite's orbit into two 69-character lines, often preceded by a name line. The format dates from the punch-card era, which is why every field sits in exact columns. Space-Track's documentation lists the column layout.

Here are the fields that matter most when you read one:

The epoch is the most important field for anyone reading a satellite map. It tells you how old the measurement behind the prediction is. A TLE with an epoch from this morning is fresh. One from three weeks ago may put the satellite well away from where it really is.

The mean motion is a quick sanity check. An object completing about 15 to 16 revolutions per day is in low Earth orbit. NASA notes that the ISS makes 16 orbits of Earth every 24 hours. A value near 1.0 means a geostationary or geosynchronous orbit that matches Earth's rotation.

How does SGP4 turn a TLE into a position?

SGP4 (Simplified General Perturbations 4) is the analytical model used to generate and use TLEs. It accounts for the main forces that bend a real orbit away from a perfect ellipse: Earth's slightly flattened shape, atmospheric drag on low satellites, and, for higher orbits, the gravitational pull of the Sun and Moon.

The details matter for one reason. TLE values are mean elements, fitted specifically for SGP4. They are not a snapshot of the satellite's exact position and velocity. If you feed them into a different orbit model, you get a worse answer, not a better one. The reference implementation most software uses today comes from the paper Revisiting Spacetrack Report #3 (Vallado, Crawford, Hujsak and Kelso, 2006), which published compatible, freely usable source code.

How a tracker uses SGP4, step by step

1. Download the latest element sets for the satellites you want to show.
2. Pick a time, usually "now" according to your device clock.
3. Run SGP4 to get position and velocity in an Earth-centered inertial frame.
4. Rotate that into Earth-fixed coordinates to account for Earth's rotation.
5. Convert to latitude, longitude and altitude and draw the point on the map.
6. Repeat for future times to draw the ground track or predict passes.

Because this runs in your browser, a tracker can animate thousands of satellites smoothly without downloading new positions each second. The only network traffic is the periodic element-set refresh.

How accurate is satellite tracking?

Accuracy depends mostly on two things: the age of the element set and the orbit. In a NASA Ames study, Levit and Marshall reported that TLEs propagated with SGP4 showed range errors growing at roughly a kilometer per day, and that their improved method cut the growth rate to about 100 meters per day (arXiv 1002.2277). Real error varies widely by object.

Low satellites are hardest. Atmospheric drag changes with solar activity, so low objects can drift off their predicted tracks faster during active space weather. Satellites that maneuver, such as the ISS during a reboost or Starlink satellites adjusting their orbits, also drift from older predictions until a new element set is published.

For a map view, this level of accuracy is more than enough: a few kilometers is invisible at global zoom. For pointing a narrow-beam antenna or timing a photo of a satellite crossing the Moon, use the freshest data you can and check the epoch.

What is a NORAD catalog number, and why did it change in 2026?

Every tracked object gets a satellite catalog number, often called a NORAD ID. The ISS is 25544. Rocket bodies and debris receive numbers too, which is why the catalog is much larger than the number of working satellites.

The TLE format only has room for five digits, and in practice the usable range ended at 69999. CelesTrak reported that it ran out of five-digit catalog numbers with the addition of an object named Saramago on July 11, 2026. Newly cataloged objects receive numbers of 100000 and above, and CelesTrak does not provide TLEs for them.

The replacement is the Orbit Mean-Elements Message (OMM), a CCSDS standard that carries the same mean elements in XML, KVN, JSON or CSV with room for nine-digit catalog numbers. Space-Track also documents an interim Alpha-5 scheme that replaces the first digit with a letter inside a TLE, but CelesTrak does not emit Alpha-5. If a tracker you use still only reads TLE text, it will gradually stop showing newly launched objects.

How do ISS pass predictions work?

An ISS pass prediction is SGP4 run forward for your location. The tracker calculates when the station rises above your horizon, its highest point in the sky, and when it sets. A visible pass also needs lighting: according to NASA's Spot the Station, sightings occur within a few hours before or after sunrise or sunset, when sunlight reflects off the station while your sky is dark.

NASA describes the station as looking like a very bright star or airplane moving across the sky, but without flashing lights and much faster. Pass predictions usually include four values:

• Time: when the pass begins in your local time.
• Maximum height: elevation in degrees, from 0 at the horizon to 90 directly overhead.
• Appears: the compass direction where it first becomes visible.
• Disappears: where it leaves view.

The ISS orbits at 51.6 degrees inclination, at an average altitude of about 400 kilometers (250 miles). That inclination means it never passes directly overhead for locations farther north or south than 51.6 degrees latitude. Predictions more than a few days out can shift after a reboost, so check again on the day.

You can see the station's current position on the ISS tracker and its details on the ISS satellite page. Pass predictions for your location are part of GodsView Pro.

Why do Starlink satellites look like a train?

A Starlink train is a batch of newly launched satellites flying in single file along one orbit before they spread out. Space.com explains that the satellites are released into a low orbit and then use their thrusters to climb to an operational altitude of about 550 kilometers over the following days and weeks, separating as they go.

The train is easiest to see in the first days after a launch, while the satellites are still low, close together and reflecting sunlight near dawn or dusk. As they rise and spread, they become individually faint and blend into the wider constellation.

One practical tip follows from that: because the satellites are still maneuvering, early predictions for a fresh train drift faster than predictions for settled satellites, so refresh the data before you go outside to look. The Starlink tracker and the Starlink satellite page show the constellation; upcoming launches are listed on the space launch tracker.

How to track a satellite on GodsView

GodsView is a hosted web map, so there is nothing to install and no sign-in needed to view it.

1. Open the live satellite tracker, or switch to the 3D globe for an orbital view.
2. Turn on the satellite layer. Orbits come from CelesTrak and refresh about every hour.
3. Click a satellite to see its name and catalog details, then follow its ground track.
4. Use the ISS tracker for the station specifically.
5. For Starlink, ISS passes for your location, conjunction data from CelesTrak SOCRATES, launches from Launch Library 2, NOAA SWPC space weather and SatNOGS ground stations, use GodsView Pro.

Track the ISS live

What satellite tracking cannot tell you

A satellite map is useful, but its limits are worth knowing before you draw conclusions from it.

• It is not a live observation. Every position is predicted from the last published element set. If an object maneuvers, the map does not know until a new set appears.
• Names can be misleading. Debris and rocket bodies carry the name of their parent mission. A dot labeled with a well-known mission name may be a spent stage, not the payload.
• Accuracy decays. Error grows with element age and is worse for low, drag-affected orbits. Check the epoch before trusting a precise prediction.
• It does not show what a satellite is doing. Orbits reveal where an object is, not whether it is imaging, transmitting or switched off.
• New objects need OMM support. Since the 69999 rollover, objects with six-digit catalog numbers do not have TLEs from CelesTrak.

The data is public and useful, but it is a model, and models carry assumptions. If you use satellite data for research, the what is OSINT guide covers verification habits that apply here too. Satellite timing signals also underpin GPS, which is why GPS jamming and spoofing is a related topic for anyone following space infrastructure.

The short version

Satellite tracking is published orbit data plus a standard model. The 18th Space Defense Squadron maintains the catalog, Space-Track and CelesTrak distribute it, and SGP4 turns each element set into a position you can draw on a map. Check the epoch, expect kilometer-level error that grows with age, and remember that new objects now arrive in OMM formats rather than TLEs.

To see it in practice, open the live satellite tracker, follow the ISS, and compare its ground track with the next pass time NASA lists for your city. Then explore other layers, such as the live earthquake map, to see how different public datasets fit on one globe. You can read about other public data on the map in our guide to reading an earthquake map.

Open the 3D globe