From WGS84 to NAD83 to ITRF: How Reference Frames Affect Your RTK Data

One antenna pillar in Maryland has three sets of coordinates, and all of them are correct. NAD83(2011) gives the pillar's position on 1 January 2010. ITRF2020 at the same 2010 epoch puts it about a metre away horizontally. In ITRF2020 today, it has moved another 26 cm with the North American plate. RTK then reproduces whichever frame its corrections carry, at centimetre level. This guide explains where WGS84 vs NAD83 differences come from, why epochs matter, and how to make RTK data land in the frame your project needs.

What WGS84, NAD83 and ITRF Actually Are

NGS describes a geodetic datum, or reference frame, as an abstract coordinate system with a reference surface that provides known locations to begin surveys and create maps (NGS datums overview, retrieved October 2026). In practice it fixes where the origin sits, how the axes point and what scale applies, and it comes with an ellipsoid for latitudes and heights. A realization is one published solution of that frame: a set of stations with coordinates and velocities, each carrying its own name or date.

ITRF realizes the International Terrestrial Reference System (ITRS), the conceptual system defined under the International Earth Rotation and Reference Systems Service (IERS). Recent solutions combine years of data from four space-geodetic techniques: VLBI, satellite laser ranging, GNSS and DORIS. The newest is ITRF2020, released in April 2022 and updated since without changing its datum; ITRF2014 is still offered in software and correction streams, ours included.

WGS84, the World Geodetic System 1984, is the frame of the US Department of Defense, and GPS broadcasts its satellite orbits in it. NGA has refined it several times, and each realization is named after the GPS week it took effect: G1762 from October 2013, aligned with ITRF2008; G2139 from January 2021, aligned with ITRF2014; and G2296 from 7 January 2024, aligned with ITRF2020, according to NGS's list of reference system codes.

NAD83, the North American Datum of 1983, is still the official geodetic datum of the United States, and Canada maintains its own realization, NAD83(CSRS). The current US realization is NAD83(2011) epoch 2010.00, computed from the NOAA CORS network and adopted in 2011 and 2012. NAD83 uses the GRS80 ellipsoid, which differs from the WGS84 ellipsoid by about 0.1 mm in the semi-minor axis, so the ellipsoid is not where the metre-level differences come from.

Why These Three Frames Aren't the Same Thing

Geocentric vs Plate-Fixed: The Core Difference Between WGS84 and NAD83

ITRF and WGS84 are geocentric: their origin is Earth's centre of mass, and no single plate carries them. Every plate moves within them, so the coordinates of a mark bolted to bedrock change a little every year. But NAD83 does a different job. It moves with the North American plate, so a mark in Kansas keeps nearly the same coordinates year after year, which is what deeds, plans and engineering records need.

NAD83 also has a different origin: NGS says it is misaligned to Earth's centre by about 2.2 m. The two effects add up. There is a fixed offset from origin and orientation, plus a difference that grows with tectonic plate motion, because positions move about 1 to 2 cm a year in many areas in ITRF2020, according to NGS.

WGS84 vs NAD83: How Far Apart Are They, Really?

So the frames fall into two camps. The global ones agree to within about 2 cm at the same epoch, so the ITRF2014 vs WGS84 gap is small too: ITRF2014 sits millimetres from ITRF2020, and NGA keeps WGS84 within 2 cm of ITRF. NAD83 sits 0.9 to 1.6 m away horizontally, and the gap grows by 1 to 2 cm a year. Picking ITRF2014 over ITRF2020 rarely shows up in a drone survey, whereas picking a global frame over NAD83 shows up every time the data meets NAD83 control.

Epochs: Why the Date on Your Coordinates Matters

In a global frame, a coordinate is only complete with its epoch, the moment in time it describes. Because the plates keep moving, the same physical mark has different ITRF coordinates in 2010 and in 2026 although nobody touched it.

Plate-fixed frames avoid most of this. They rotate with their plate, and their agencies publish coordinates for a reference epoch: 2010.00 for NAD83(2011), 2010.0 for NAD83(CSRS) in Canada and 2020.0 for GDA2020 in Australia. Those coordinates barely move, but every conversion to or from a global frame has to state the epoch of the global coordinates. A datum transformation between the two is a coordinate transformation in time as well as space. NGS's HTDP tool, for example, asks for a source frame and epoch and a target frame and epoch (HTDP user guide).

For RTK work, the practical rule is short. Record the frame, realization and epoch with every dataset, convert everything into one before you compare, and never compare a current-epoch ITRF position with ITRF coordinates published for 2010.

ITRF and the Multi-Constellation Reference Frames

ITRF: The Global Reference Behind Your GPS

The IERS publishes each ITRF through its ITRS Centre, and the International GNSS Service (IGS) extends it to a global network of more than 400 tracking stations; IGS orbits and clocks have followed ITRF2020 since November 2022. NGS anchors the US network the same way: since June 2025 it has published its CORS coordinates in ITRF2020 and derived the NAD83(2011) values from them. That shared anchor is what makes data from different systems comparable.

Multi-Constellation Frames: PZ-90, CGCS2000, GTRF, JGS and How They Tie Back to ITRF

Each satellite system defines its own terrestrial frame for its broadcast orbits, and each operator ties that frame to ITRF:

For an RTK user these differences stay in the background. In RTK the rover's position is the base station's coordinates plus a measured vector, as NGS's real-time positioning guidelines describe and as our guide to how RTK reaches centimetre accuracy explains. Your fix therefore lands in the frame the correction service assigns to its reference stations, whichever constellations you track.

Reference Frames in Real RTK Work

A Worked Example: One Coordinate, Three Frames

Take GODE, for example, a reference station on a pillar at NASA Goddard in Greenbelt, Maryland, whose antenna was first installed in 1993. NGS publishes its position in ITRF2020 with a velocity, and in NAD83(2011) at epoch 2010.00, in the station coordinate file (retrieved October 2026). Moving the ITRF2020 position along its velocity and converting it to ITRF2014 gives these answers:

The first two rows isolate the frame: at the same epoch, NAD83(2011) puts the pillar 0.93 m south and 0.41 m east of its ITRF2020 position, and 1.28 m higher in ellipsoid height. The second and third rows isolate time: in ITRF2020 the pillar has moved 25.7 cm west-northwest since 2010, while its horizontal NAD83(2011) velocity is under 2 mm a year. ITRF2014 for the same October day sits about 3 mm from the ITRF2020 position. A rover on this pillar with global corrections therefore reads about 1.2 m horizontally from the NAD83 datasheet, all of it from frame and epoch and none from accuracy.

What This Means for Your RTK Corrections Stream

So with RTKdata, the mountpoint decides the frame. Our caster at caster.rtkdata.com names the national-datum mountpoint SMART, while the regional hosts name it AUTO, and the global options follow the same pattern.

To reconcile a stream with a NAD83 project or a local GIS coordinate system:

1. Find the project's frame, realization and epoch on the control datasheets, in the GIS layer's metadata or from the client.
2. Pick the matching mountpoint. For a NAD83(2011) project in the US, that is SMART or AUTO.
3. Set the project's coordinate system to the datum the mountpoint delivers, so the software applies no second transformation.
4. Measure one known control point before collecting data.
5. Write the frame, realization and epoch into the deliverable's metadata.

Our coordinate systems guide lists the datum and epoch for every region. A steady offset against local control is usually a datum problem, not an accuracy problem.

When a WGS84 and NAD83 Mismatch Matters (and When You Can Ignore It)

Compare the possible offset with your tolerance. A metre-level frame difference is invisible next to a phone's position, which GPS.gov typically puts within 4.9 m under open sky (see RTK vs GPS accuracy). It also cancels out in a standalone job measured only against itself, such as stockpile volumes from a single flight.

It matters as soon as your data has to sit on top of something else: survey control, property lines, a CAD design surface, an earlier flight or a GIS layer from another source. In RTK for surveying, construction layout and repeat drone mapping, frame and epoch decide whether centimetre-level results line up. For important projects, NGS's guidelines expect teams to hold "the same datum, realized at the same control system monumentation" from design through construction.

NAD83's Replacement Is Coming: NATRF2022

NGS is replacing the three NAD83 frames with four plate-fixed frames, led by NATRF2022 for North America, and NAVD 88 heights with a new geopotential datum, NAPGD2022. This NSRS modernization is in its final stretch. NGS declared the definitions stable in May 2026 and released the supporting tools as a public beta on 15 September 2026, for testing only and not for operational work. Testing runs until at least March 2027, and NGS plans the approval vote and the official transition for 2027, without a fixed date. Until then NAD83(2011) and NAVD 88 remain the official US datums; Canada plans its own switch in early 2027.

The new frames remove NAD83's origin offset while keeping coordinates nearly steady on the plate. Each one is identical to ITRF2020 at epoch 2020.0 and then rotates with the stable part of its plate. An NGS map of the shift shows NAD83(2011) coordinates in the lower 48 changing by roughly 0.8 to 1.7 m horizontally. NGS also says coordinates will change by as much as four metres, with the largest shifts on the Pacific plate.

For RTK users, the preparation is the same as for any frame change: keep frame, realization and epoch in your metadata, keep ITRF2020 positions with their epochs where you can, and never mix new-frame coordinates with NAD83 control. ITRF2020 data with a known epoch will convert through NGS's published plate rotation and deformation model. Projects meant to create or update control in the current system must reach NGS by 13 January 2027.