NAD 83 is not one set of coordinates. It is a definition that has been realized several times, each realization being a new adjustment of the control network using better data and better technique. Coordinates change between realizations by centimetres to decimetres, which is invisible on a topographic map and fatal to a precise control survey.
On top of that, the ground moves. So each realization also carries an epoch — a date at which the published coordinates are valid — and comparing coordinates from different epochs on moving crust requires a velocity model, not just a readjustment.
Definition, realization, epoch
- Realization
- A specific set of published coordinates that gives users access to a datum. Named in parentheses after the datum, as in NAD 83 (2011). Changing the realization changes coordinates without changing the definition.
- Reference epoch
- The instant in time at which the published coordinates of a realization are considered valid, written as a decimal year such as 2010.00. Coordinates observed at another date must be propagated to the reference epoch before they are comparable.
The distinction repays attention. The definition of NAD 83 has been stable for decades: GRS 80, a specified orientation, a specified origin. The realizations have moved repeatedly. A surveyor who says my coordinates are NAD 83 has told you far less than they think.
The succession of realizations
| Realization | What it was | Typical shift from the previous |
|---|---|---|
| NAD 83 (1986) | The original simultaneous adjustment of the classical network | — |
| NAD 83 (HARN) / (HPGN) | Statewide high accuracy reference network readjustments, done state by state in the 1990s using GPS | Up to about a meter |
| NAD 83 (CORS96) | Readjustment tied to the continuously operating reference station network | Centimetres to decimetres |
| NAD 83 (NSRS2007) | National readjustment of passive marks constrained to the CORS | Centimetres to decimetres |
| NAD 83 (2011) | National adjustment at reference epoch 2010.00, the current widely used realization | Centimetres |
Parallel realizations exist for regions on other tectonic plates, because a frame fixed to North America cannot serve ground that is moving independently. NAD 83 (PA11) covers the Pacific plate area and NAD 83 (MA11) the Mariana plate area, both at epoch 2010.00.
Why epochs exist: the ground moves
The North American plate is drifting relative to the Earth as a whole at roughly two centimetres a year. Within North America there are additional motions: glacial isostatic adjustment lifting formerly ice-covered regions, subsidence over depleted aquifers and oil fields, and coseismic displacement that can move a region by meters in seconds.
NAD 83 handles plate-scale motion by being plate-fixed. It rotates with the North American plate, so a stable point in the plate interior keeps essentially the same NAD 83 coordinates over time even though its position in a global frame changes steadily. This is a design choice with a clear rationale: surveyors want a monument in Kansas to keep its coordinates, and the plate-fixed frame delivers that.
The design fails exactly where the plate-fixed assumption fails — near active margins. In coastal California, points move measurably in NAD 83 as well, because they are not riding rigidly with the plate interior. That is why western states care about epochs far more than eastern states do.
How coordinates move between epochs
Propagating a coordinate from the date it was observed to the reference epoch of the realization uses a velocity model. NGS distributes this capability in HTDP, Horizontal Time-Dependent Positioning, which combines plate rotation, a crustal velocity model and a catalogue of earthquake displacements.
position at epoch t₂ = position at epoch t₁ + velocity × (t₂ − t₁) + coseismic displacements between t₁ and t₂In a stable interior region a velocity of a few millimeters a year over a decade is a few centimetres — often below the noise of the work being done. In a tectonically active region a velocity of two to four centimetres a year over twenty years is a decimetre or more, which is not below anyone's noise.
What to do about it
- Write the realization and the epoch on every deliverable: NAD 83 (2011) epoch 2010.00, not just NAD 83.
- Check the realization of every published control coordinate you hold. Datasheets state it explicitly.
- Do not mix realizations in one adjustment. Transform first, then adjust.
- In tectonically active regions, treat the epoch as a first-class piece of metadata, not a footnote.
- When a network RTK service supplies coordinates, find out which realization and epoch its base coordinates are on. This is the most common way realizations get mixed without anyone noticing.
Where this is heading
The modernized frame that replaces NAD 83 is built around this problem rather than around it. It is defined by an explicit Euler pole rotation relative to the international terrestrial reference frame, with a published intraframe velocity model to describe motion that the plate rotation does not account for. Coordinates will be expressible either at a reference epoch or at the date of survey, with the transformation between them a defined and supported operation rather than a specialist correction.
The practical shift for a surveyor is that time becomes part of a coordinate rather than an afterthought. The habit worth building now is simply to record the observation date alongside every coordinate you produce.