7. From Satellites to Elevation
"A GNSS receiver does not tell you where you are. It tells you when you are. From that, and knowing where the satellites were or will be at a particular time, you can work out where you are." —Popular Geomatics Aphorism
The number on a screen is not one direct measurement beamed down from space. It's the end of a chain, one that starts with radio signals and passes through corrections, reference frames, height models, and mapping software before it ever becomes a coordinate anyone reads.
A Coordinate Appears
A rover reaches RTK FIX. A blue dot settles onto the map. Latitude, longitude, and an elevation stop wandering and hold still.
No satellite ever transmits a coordinate. Each one transmits a precisely timed radio signal and a description of its own orbit, and the receiver builds an actual coordinate out of that, before several more systems step in to decide what the result actually means.
Following one observation all the way from the sky to the map makes the whole chain visible.
The Satellites Only Provide Clues
A satellite broadcasts a signal carrying a timestamp. The receiver compares that timing pattern against what it actually received and estimates how long the signal spent traveling. Radio waves move at nearly the speed of light, so a tiny timing error becomes a surprisingly large distance error, which is exactly why a receiver never treats one satellite's signal as a finished answer. It combines several satellites at once and solves for its own position and its own clock error together.
The satellites in view never measure a rover's latitude, longitude, or elevation directly. What do they actually provide? Signals, and only signals, from which the receiver itself estimates distances and then builds a coordinate. The satellites hand over raw material. The receiver does the actual work.
The Trip Down Is Never Clean
Satellite clocks and orbit predictions are never perfect. The ionosphere and the lower atmosphere both delay the signal. Buildings, trees, vehicles, and water can all reflect it before it ever reaches the antenna. A standalone receiver models as much of this as it can on its own. A precision receiver goes further, leaning on correction information from a base station, a network, or another correction service, not simply to look steadier on screen, but to actually separate real geometry from the error a whole region happens to share.
Carrier Phase Is What Makes Centimeters Possible
A coded timing signal alone places a receiver within meters. Reaching centimeters depends on the carrier wave underneath that code, whose short wavelength lets a receiver detect extremely small phase changes. The hard part is knowing exactly how many complete wavelengths existed between satellite and antenna the moment tracking began, the same integer ambiguities introduced earlier in this series. Resolving those reliably is what a FIX status actually represents, a genuine milestone, and still not an answer to every question a coordinate needs answered.
Corrections Do More Than Add Precision
A correction source hands the rover more than precision. It also hands over a reference frame, since the rover's final solution is anchored to whatever coordinate and frame that source is using.
This is exactly why two rovers can both display FIX and still disagree. Both may have solved the actual measurement problem correctly, while sitting tied to different frames, realizations, epochs, or base coordinates entirely.
From Earth-Centered Numbers to Familiar Ones
Inside the positioning engine itself, a solution is naturally expressed in Earth-centered, Earth-fixed coordinates, three axes running through the planet's center and rotating along with it. Those X, Y, and Z values are excellent for computation and nothing like how anyone actually describes a location, so software converts them into geodetic latitude, longitude, and ellipsoid height on whichever ellipsoid the selected reference frame uses.
At this stage, the coordinate can already look beautifully precise, and its height is still, by default, an ellipsoid height rather than the familiar elevation printed on a benchmark or a topographic map.
A Coordinate Sometimes Needs a Transformation
A correction service can produce coordinates in one reference frame while a GIS project or a survey's control uses another entirely, and changing a label never safely bridges that gap on its own. A genuine coordinate transformation accounts for the real relationship between the two frames, which can involve translations, rotations, scale, and motion through time, depending on the frames and the accuracy actually required.
Switching a project's datum setting from one name to another does not automatically move a coordinate correctly. Software has to apply an appropriate transformation, using the correct realization and epoch, and simply relabeling a coordinate is not the same thing as transforming it.
Ellipsoid Height Becomes Usable Elevation
GNSS naturally supplies ellipsoid height, h. Most maps, benchmarks, and engineering projects want orthometric height, H, elevation tied to gravity and an actual vertical datum. A geoid model supplies the missing piece, N, the separation between the two, so that H = h − N. That conversion can happen inside the receiver, inside field software, or later inside a GIS package, and the final displayed number can look identical on screen no matter which path actually produced it, which is exactly why the geoid model and vertical datum belong in a coordinate's metadata rather than left assumed.
Latitude and Longitude Become a Map Coordinate
Latitude and longitude describe a location on a curved Earth. Most engineering drawings and GIS projects instead use a flat coordinate grid, State Plane or UTM among them, and a map projection is what converts curved coordinates into easting and northing, introducing its own predictable scale and direction effects along the way, since no curved surface flattens perfectly. Software may then convert units, apply a local scale factor, or distinguish grid distance from ground distance. Each of these is entirely reasonable when it's done deliberately. The trouble starts when a clean easting, northing, and elevation shows up on screen with no visibility into which of these operations actually happened.
The Whole Chain, Start to Finish
Laid out from beginning to end: satellite signals and orbit information reach a receiver's antenna, where clock, orbit, atmospheric, and local error get handled as well as possible. An RTK, PPP, or other precision method produces a solution, tied to a specific reference frame and observation epoch. A coordinate transformation gets applied if one is actually needed. The result becomes geodetic latitude, longitude, and ellipsoid height, then a geoid model converts that into orthometric elevation, and finally a map projection and a project's own units turn the whole thing into what actually appears in a GIS layer, a field map, or an engineering drawing.
No single link in that chain is mysterious on its own. The confusion comes entirely from seeing only the final number while every intermediate choice behind it stays hidden.
A familiar version of this chain, common across existing US projects, runs from GNSS observations with RTK corrections, through NAD83(2011) ellipsoid coordinates, through GEOID18's geoid separation, into NAVD88 orthometric elevation, and finally into State Plane or whatever projection the project actually uses. That specific sequence is not universal. A different country, correction service, agency, or project can easily use a different reference frame, geoid model, vertical datum, and projection instead. The habit worth building is identifying the actual chain in front of you, rather than assuming it matches the one just described.
Where the Chain Actually Tends to Break
Most field problems are almost never a genuine failure of the GNSS receiver itself. They happen at one specific link where a processing step gets misunderstood, skipped, or quietly applied twice: a base station entered with only an approximate coordinate, a correction service and a GIS project running on two different reference frames, a missing coordinate epoch, an app displaying ellipsoid height when orthometric elevation was expected, both the receiver and the app applying their own geoid correction at the same time, the wrong geoid grid selected for a location or datum, latitude and longitude assigned to the wrong map projection, feet and meters or two different definitions of the foot getting mixed together, an antenna height measured to the wrong reference point, or a pole that was neither fully extended nor level.
A receiver can sit happily in FIX through nearly every one of these mistakes. FIX describes the carrier-phase solution specifically. It was never built to certify the entire coordinate-processing chain sitting behind it.
Reading the Metadata Backward
When a coordinate looks wrong, working backward from the displayed value is usually faster than changing settings at random: which projection and units are actually being shown, is the height ellipsoidal or orthometric, which geoid model and vertical datum were used, was a coordinate transformation actually applied, which reference frame, realization, and epoch define the position, which correction source or base coordinate produced the solution, and what did the antenna itself actually observe. Each answer tends to point directly at the next place worth checking.
A coordinate is easy to treat as three plain numbers. At modern GNSS accuracy, it's better understood as the compact, final result of an entire processing history: signals measured, errors corrected, a reference frame and epoch giving the result meaning, a geoid model converting geometric height into usable elevation, and a projection placing the whole thing onto a flat map. Once that chain becomes visible, most apparent GNSS mysteries turn out to be ordinary configuration questions instead.
Everything in this chapter depends on reference frames, vertical datums, and geoid models staying where they are. In the United States, several of the most familiar pieces of that exact workflow are actively being modernized. Chapter 8 follows that change, and explains why future coordinates will make time an explicit, everyday part of positioning rather than something left implicit.