Defining Precision, Accuracy, Stability, and Drift
The quality of positioning within a reference frame is described in terms of precision, accuracy, stability, and drift:
Precision quantifies the ability to repeat the determination of position within a reference frame (internal precision) and can be measured using various statistical methods on samples of estimated positions. Although precision does not imply accuracy, high precision is a prerequisite for consistently high accuracy and is necessary to resolve changes in position over time. The precision of a reference frame itself (external precision) refers to the variation in the reference frame parameters (origin, orientation, and scale) that arise from statistical variation in the data used to define the frame.
Accuracy quantifies how close a position is to the truth. Strictly, it only applies to absolute physical quantities, such as distance between stations, but this report also uses it to mean accuracy of station position within a reference frame (internal accuracy). Precision contributes to accuracy, but accuracy also takes into account systematic biases arising from calibration errors or imperfect observation models. Accuracy can be assessed if there is a superior measurement technique that can be used as a standard, but since geodesy uses the highest-accuracy techniques, accuracy estimation is not straightforward for geodesy. Accuracy estimates for geodesy therefore typically involve an “error budget” analysis of systematic effects.
Stability refers to the predictability of the reference frame and station positions. The stability of the reference frame refers to the behavior (linearity and consistency) of its defining parameters, and the ability to predict accurately the future positions of the stations that are used to define the frame. That is, the ITRF parameter should not exhibit any discontinuity over the entire time span of the geodetic observations. Furthermore, the ITRF should remain internally consistent even as it is updated from time to time. The stability of a station refers to the ability to predict its future position within the reference frame. For example, local site stability typically implies that all stations at a specific site do not move relative to each other, and the site does not have non-linear motions relative to the ITRF. The deviation of measured station positions from their predicted positions provides information on geophysical processes that were not predicted. Stations of special geophysical interest (for example, for measuring topographic change in the Las Vegas Valley caused by groundwater effects) are obviously not well suited for defining the reference frame, but it is the stability of the frame that allows scientists to detect the interesting and important geophysical effects on the motions of these stations.
Drift refers to relative rotation, translation, and scale between different reference frames, which results in different velocities between stations given in each frame. Drift is a consequence of a lack of stability in one or both of the frames being compared, which in turn may result from systematic error in the measurement techniques, lack of precision in the measurements, or differences in the station motion models.