Reference Systems, Frames, and Datums (Paper)

INTRODUCTION — Why This Topic Matters

Modern geodesy measures the Earth with centimeter to millimeter accuracy. This level of precision is needed for:

  • monitoring tectonic plate motions

  • studying sea level rise

  • tracking earthquakes

  • observing Earth rotation

  • precise satellite navigation (GPS, GLONASS, Galileo)

To compare global measurements consistently, all observations must be expressed in the same coordinate system.
To do that, we need:

  1. A Reference System → theoretical definition

  2. A Reference Frame → physical realization

  3. A Geodetic Datum → the set of fixed parameters that ties the system and frame together

The paper emphasizes that these three must remain conceptually separate, or else errors will propagate through global geodesy.

SECTION 1 — Key Definitions (Fundamental Concepts)

1. Reference System

A theoretical definition that states how positions are described.

It includes:

  • origin (e.g., Earth’s center)

  • orientation of axes (e.g., rotation axis)

  • scale (e.g., metre)

  • physical models (gravity field, Earth rotation model, etc.)

  • mathematical conventions

Example:
ITRS (International Terrestrial Reference System)
→ origin: Earth’s center of mass
→ Z-axis: Earth’s rotation axis
→ X-Y plane: equatorial plane
→ scale: metric (defined by speed of light)

Important: A reference system is purely conceptual.

2. Reference Frame

The practical realization of the reference system.

This consists of:

  • physical monuments (geodetic stations)

  • coordinates of those stations

  • velocities of stations

  • time series from GNSS, VLBI, SLR, DORIS measurements

Example:
ITRF (International Terrestrial Reference Frame)
→ actual coordinates of stations around the world
→ computed from real measurement data

3. Geodetic Datum

The connection between the system and the frame.

It fixes:

Origin

Where is (0,0,0)?
(e.g., Earth’s center)

Orientation

How do X, Y, Z axes point?

Scale

What is the length of 1 metre in the coordinate system?

The datum must be:

  • stable

  • independent from measurements used to define the frame

  • unchanging over time

This is essential so that real Earth movements (like plate tectonics) are not mistakenly interpreted as movements of the datum.

SECTION 2 — Fundamental Principles (Hierarchy)

The author states three rules that must NEVER be violated:

Rule 1

The reference system must not change because of how measurements are made.

Rule 2

The datum must be determined by independent observations (not by the same observations used to compute the frame).

This prevents measurement errors or network deformations from affecting the datum.

Rule 3

When computing a reference frame, algorithms must respect the datum and keep its parameters fixed.

Traditional 2D Geodetic Systems (Old approach)

Old triangulation networks fixed 4 parameters:

  1. latitude of origin

  2. longitude of origin

  3. orientation (azimuth between stations)

  4. scale (using prototype metre bars)

These parameters were fixed using astronomic observations, independent from the triangulation measurements.

→ This ensured the datum was consistent for centuries.

SECTION 3 — Modern 3D Reference Systems and Frames

Modern geodesy uses 3D Cartesian coordinates:

(x, y, z)

To fully define a datum in 3D, we need 7 parameters:

  • 3 translations (X₀, Y₀, Z₀): origin

  • 3 rotations: orientation

  • 1 scale: length standard

These parameters must be fixed — not computed from the reference frame.

FIXING THE 7 DATUM PARAMETERS IN MODERN GEODESY

1. Origin (X₀, Y₀, Z₀)

Given by the Earth’s center of mass (geocenter).
This is determined from gravity field coefficients (C₁₁, S₁₁, C₁₀ = 0).

By using satellite gravity models with these coefficients set to zero, satellite orbits naturally align to a geocentric origin.

Therefore:
The origin is automatically fixed and must never be re-estimated.

2. Orientation of Axes

The Z-axis could be defined using Earth’s principal inertia axis (via C₂₁, S₂₁), but this is not accurate enough.

Instead, the ITRS defines:

  • Z-axis = Earth’s rotation axis at epoch 1984.0

  • X-axis = intersection of Z-axis with Greenwich meridian

To maintain orientation over time, the system uses:

  • No-Net-Rotation (NNR) condition
    → the average horizontal tectonic motion is zero relative to the frame

But this is imperfect due to limited station coverage.

3. Scale

Defined by:

  • speed of light = fixed constant

  • distance measured via travel time of electromagnetic waves (VLBI, SLR, GNSS)

To realize scale correctly, we must model:

  • ionospheric delay

  • tropospheric delay

  • antenna phase center offsets

SECTION 4 — Problems in the Datum Realization

This is the core of the paper.

The author argues that ITRF does NOT properly maintain a stable datum.

Major Problem:

In practice, the datum is often derived from the same station coordinates that define the frame.

This is done using:

  • 7-parameter similarity transformations between different ITRF solutions.

BUT:

  • Stations move

  • Tectonic plates rotate

  • Networks expand

  • Local deformations occur (postglacial rebound, seismic strain)

When you compute transformation parameters between two epochs, the following station motions get incorrectly interpreted as:

  • translation → fake origin shift

  • rotation → fake orientation change

  • scale → fake enlargement of the Earth

This changes the datum itself, which violates the fundamental rules.

Detailed Examples of These Problems

A. Common translation of stations

Example: tectonic plates move northward as a block.

The least-squares solution thinks the whole network moved →
it inserts a shift in origin parameters.

Meaning:
The geocenter appears to move, even though it didn’t.

B. Common rotation of stations

Plate rotations show up as rotation of the coordinate axes.

This violates the no-net-rotation rule.

C. Expansion (scale change)

Because of postglacial rebound:

  • Arctic and Antarctic uplift
    → stations get farther apart

This appears in the solution as:

  • change in scale parameter

But the real scale of the Earth didn't change — only the crust moved.

SECTION 5 — Evidence from ITRF Transformations

The paper presents Table 1 with transformation parameters.

It shows:

  • Origin translations drift over time

  • Scale drifts

  • Orientation drifts

  • ITRF versions are inconsistent with each other

Meaning:

  • ITRF is not perfectly geocentric

  • ITRF orientation is not perfectly tied to Earth’s rotation axis

  • ITRF scale is affected by network expansion

  • Datum is not stable over decades

This is a fundamental problem because:

  • Long-term Earth studies need strict stability

  • Datum drift hides real geophysical signals

SECTION 6 — Recommendations

The author proposes strict rules to fix the ITRF.

✔ 1. Do NOT use station coordinates to fix the origin

Use gravity models only.

✔ 2. Do NOT use network similarity transformations

Keep datum independent of frame estimation.

✔ 3. Use raw observations (SLR, GNSS, VLBI)

Not pre-processed coordinates with built-in assumptions.

✔ 4. For station velocities, use a modern-day crustal model

Use APKIM (geodetic)
Not NUVEL (geologic average over millions of years)

✔ 5. Improve physical models

Atmosphere, hydrosphere, cryosphere, gravity field

✔ 6. Ensure long-term datum stability

This is essential for monitoring global change.