Traditional geodetic instruments such as theodolites rely on the gravity field for orienting their vertical axis along the local plumbline or vertical with the aid of a spirit level. After that, vertical angles (zenith angles or, alternatively, elevation angles) are obtained with respect to this local vertical, and horizontal angles in the plane of the local horizon, perpendicular to the vertical.

Levelling instruments again are used to obtain geopotential differences between points on the Earth's surface. These can then be expressed as "height" differences by conversion to metric units.

Table of contents |

2 The normal potential 3 Disturbing potential and geoid 4 Gravity anomalies |

The Earth's gravity field can be described by a potential as follows:

Note that both gravity and its potential contain a contribution from the centrifugal pseudo-force due to the Earth's rotation. We can write

The centrifugal force is given by

Here, *X*, *Y* and *Z* are geocentric co-ordinates.

Gravity is commonly measured in units of m s^{-2}, (metres per second squared). This also can be expressed as newtons per kilogram of attracted mass.

Potential is expressed as gravity times distance, m^{2} s^{-2}. Travelling one metre in the direction of a gravity vector of strength 1 m s^{-2} will change your potential by 1 m^{2} s^{-2}.

A more convenient unit is the GPU, or geopotential unit: it equals 10 m^{2} s^{-2}. This means that travelling one metre in the vertical direction, i.e., the direction of the 9.8 m s^{-2} ambient gravity, will *approximately* change your potential by 1 GPU. Which again means that the difference in geopotential, in GPU, of a point with that of sea level can be used as a rough measure of height "above sea level" in metres.

To a rough approximation, the Earth is a sphere, or to a much better approximation, an ellipsoid. We can similarly approximate the gravity field of the Earth by a spherically symmetric field:

It is more accurate to approximate the geopotential by a field that has *the Earth reference ellipsoid* as one of its equipotential surfaces, however. The most recent Earth reference ellipsoid is GRS80, or Geodetic Reference System 1980, which the Global Positioning system uses as its reference. Its geometric parameters are: semi-major axis m, and flattening .

A geopotential field is constructed, being the sum of a gravitational potential and the known centrifugal potential , that *has the GRS80 reference ellipsoid as one of its equipotential surfaces*. If we also require that the enclosed mass is equal to the known mass of the Earth (including atmosphere) GM = 3986005 × 10^{8} m^{3}s^{-2}, we obtain for the *potential at the reference ellipsoid:*

Due to the irregularity of the Earth's true gravity field, the equilibrium figure of sea water, or the geoid, will also be of irregular form. In some places, like west of Ireland, the geoid -- mathematical mean sea level -- sticks out as much as 100 m above the regular, rotationally symmetric reference ellisoid of GRS80; in other places, like close to Ceylon, it dives under the ellipsoid by nearly the same amount. The separation between these two surfaces is called the undulation of the geoid, symbol , and is closely related to the disturbing potential.

According to the famous Bruns formula, we have

In 1849, the mathematician George Gabriel Stokes published the following formula named after him:

The geoid, or mathematical mean sea surface, is defined not only on the seas, but also under land; it is the equilibrium water surface that would result, would sea water be allowed to move freely (e.g., through tunnels) under the land. Technically, an *equipotential surface* of the true geopotential, chosen to coincide (on average) with mean sea level.

As mean sea level is physically realized by tide gauge bench marks on the coasts of different countries and continents, a number of slightly incompatible "near-geoids" will result, with differences of several decimetres to over one metre between them; these are referred to a *vertical* or *height datums*.

The local direction of gravity or vertical or plumbline is on every point on Earth *perpendicular* to the geoid. On this is based a method, *astrogeodetic levelling*, for deriving the local figure of the geoid by measuring *deflections of the vertical* by astronomical means over an area.

In geophysics, these anomalies are often further reduced by removing from them the *attraction of the topography*, which for a flat, horizontal plate (Bouguer plate) of thickness *H* is given by

In case the terrain is not a flat plate (the usual case!) we use for *H* the local terrain height value but apply a further correction called the terrain correction (*TC*).