Foundations: Coordinate Systems, Plate Kinematics, and the Reading of Landscapes

Chapter 1: Foundations — Coordinate Systems, Plate Kinematics, and the Reading of Landscapes

The first question any regional geographer must answer is deceptively simple: where, exactly, is the thing I am looking at? On a planet that is simultaneously oblate, convecting, rotating, and re-sculpting its own surface at rates of centimeters per year, "where" is a moving target. The purpose of this chapter is to install the technical scaffolding that every subsequent continental analysis in this volume will lean on. We will move from the geodetic primitives that anchor a point on the sphere, to the kinematic machinery that drags continents across it, to a four-variable framework for reading the resulting landscape, and finally to a reproducible Python pipeline that turns raw GNSS time-series into plate-relative motion vectors. By the end, the reader should be able to defend a coordinate, a velocity, and a forcing argument with the same numerical rigor that an oceanographer brings to a CTD cast.

The narrative spine of the chapter is a research-vessel transit across the Mid-Atlantic Ridge between roughly 0° N and 10° N — a stretch of seafloor where the half-spreading rate is on the order of 13–16 mm yr⁻¹ on the African side and 23–27 mm yr⁻¹ on the South American side. The transit is useful not as scenery, but as a forcing function: it forces the reader to confront, within the span of a few days, the fact that the vessel's GPS receiver, the bathymetric swath mapped beneath its hull, and the basaltic pillow lavas being dredged from the rift floor are all referenced to coordinate systems that disagree by tens of meters unless explicitly reconciled.

1.1 The Shape of the Earth and the Geodetic Datum

The Earth is not a sphere. It is an oblate spheroid — flattened at the poles and bulged at the equator — and the figure that best approximates the global mean sea level surface (a particular equipotential of the combined gravitational and centrifugal field) is the reference ellipsoid. The currently dominant reference, WGS-84 (World Geodetic System 1984, last revised in its geopotential terms in 2020 with EGMI (Earth Gravitational Model 2020)), is defined by four fundamental constants: the semi-major axis $a = 6{,}378{,}137.0$ m, the flattening $f = 1/298.257223563$, the Earth's angular velocity $\omega = 7{,}292{,}115.0 \times 10^{-11}$ rad s⁻¹, and the Earth's gravitational constant $GM = 3{,}986{,}004.418 \times 10^8$ m³ s⁻². From these, the semi-minor axis is derived:

$$b = a(1 - f) \approx 6{,}356{,}752.3142 \text{ m}$$

The first eccentricity squared,

$$e^2 = \frac{a^2 - b^2}{a^2} \approx 0.00669437999014,$$

appears in nearly every geodetic formula. The radius of curvature in the meridian, $M$, and the radius of curvature in the prime vertical, $N$, govern how arc-length and azimuth accumulate on the ellipsoid:

$$M = \frac{a(1 - e^2)}{(1 - e^2 \sin^2 \phi)^{3/2}}, \qquad N = \frac{a}{\sqrt{1 - e^2 \sin^2 \phi}}$$

where $\phi$ is the geodetic latitude. For a position given in geodetic coordinates $(\phi, \lambda, h)$ — latitude, longitude, and ellipsoidal height — the corresponding Earth-Centered, Earth-Fixed (ECEF) Cartesian triplet is

$$ \begin{aligned} X &= (N + h) \cos\phi \cos\lambda, \\ Y &= (N + h) \cos\phi \sin\lambda, \\ Z &= \left[ N(1 - e^2) + h \right] \sin\phi. \end{aligned} $$

The reason this transformation matters operationally is that GNSS receivers — the modern workhorse of geodesy — internally solve for $X, Y, Z$ in an ECEF frame, then project back to $(\phi, \lambda, h)$ for human consumption. Whenever a user says "the receiver is at 8.5° N, 28.4° W," they are reading off the *back end* of a Jacobian. For any calculation involving inter-station baselines, satellite geometry, or tectonic velocity, working in ECEF is overwhelmingly more robust than working in geographic coordinates, because the metric is uniform and the partial derivatives of the forward and inverse transforms are well behaved.

The distinction between geodetic and geographic latitude is more than pedantry. Geodetic latitude is the angle between the equatorial plane and the normal to the ellipsoid at a point; geocentric (sometimes called geographic in the older sense) latitude is the angle between the equatorial plane and the radius vector from the Earth's center to the point. They differ by up to about 11.6′ at 45° latitude. Most mapping systems — and most laypeople — mean geodetic, but any spherical-Earth approximation (great-circle distance, initial bearing) that uses geocentric latitude in place of geodetic will accumulate errors that exceed the plate-tectonic signals we are trying to measure. The correction is

$$\tan \phi_{\text{geocentric}} = (1 - e^2) \tan \phi_{\text{geodetic}}.$$

A second, more subtle issue is the datum itself. WGS-84 is not the only ellipsoidal model in circulation. ITRF (International Terrestrial Reference Frame) realizations — currently ITRF2020 — are not strictly identical to WGS-84; they agree at the decimeter level in the most recent alignment (G1150 epoch), but historical WGS-84 realizations (WGS-72, WGS-84 original) can disagree with ITRF by up to a meter. For tectonic work, this is non-trivial. A velocity of 25 mm yr⁻¹ integrated over the operational lifetime of a GNSS station (often 20+ years) yields a cumulative displacement of half a meter — the same order as the datum offset. The correct practice, formalized by the IERS and adopted throughout this book, is to express all positions and velocities in a single, well-defined reference frame (we will use ITRF2020 with the IGS20 station coordinates for kinematic work) and to be explicit about the epoch at which positions are reported. An epoch is a specific instant at which the time-tagged position equals the listed value; positions at other epochs are obtained by integrating the velocity field.

1.2 Projections, Distortion, and the Cost of a Flat Map

Once a position exists on the ellipsoid, the next act of cartographic violence is to flatten it. No projection from the sphere (or spheroid) to the plane can preserve all four of distance, area, shape, and direction; choosing a projection is therefore choosing which distortions to tolerate. The classics, all of which will appear later in this book, are the Mercator (conformal — preserves local angles, distorts area catastrophically at high latitudes), the Lambert azimuthal equal-area (preserves area at the cost of shape), the equirectangular (a poor compromise used mostly for rasters and quick plotting), and the family of transverse Mercator projections that underpin UTM and MGRS grids.

The Universal Transverse Mercator (UTM) system divides the globe into 60 zones, each 6° of longitude wide, numbered from 1 at 180° W eastward to 60 at 180° E. Within each zone, a secant transverse Mercator projection is applied with two standard parallels (i.e., two lines of exact scale) located approximately 180 km east and west of the central meridian. The central scale factor is 0.9996, which means the central meridian itself is not scale 1.0 but rather slightly compressed — the result is that distortion across the zone is bounded to roughly 0.1% in length, a level that is acceptable for nearly all engineering, military, and medium-scale geoscience work.

The mathematics of a transverse Mercator projection is best written in terms of the isometric latitude and a series expansion. Given the geodetic latitude $\phi$ and longitude $\lambda$, with central meridian $\lambda_0$, the footpoint latitude $\phi_1$ is defined implicitly by

$$\int_0^{\phi_1} M(\phi)\, d\phi = k_0 \int_0^{\phi} M(\phi)\, d\phi,$$

where $k_0 = 0.9996$ is the central scale factor. The Karney (2011) series converges to sub-millimeter accuracy over the entire ellipsoid and is the formulation implemented in modern libraries such as PROJ's `proj=tmerc` pipeline. The easting and northing within a zone are then

$$ \begin{aligned} x &=