What is internal drift and external drift?

Question

For the Kriging model, I'm wondering what the difference is between internal and external drift? I understand the internal drift is part of the location (s) and the external drift is typically an additional variable that has a linear correlation with the location information.

Answer 1

External drift kriging (EDK) is a special case of Universal kriging. In the case of an external drift, the drift is not determined by the spatial position of the target point, at which the data should be interpolated. Making an educated guess, I think that what you mean with "internal" drift (simply as the opposite of "external" drift) is the drift functions, that determines the drift from the target point position.

The classical approach for a drift function is a regional linear drift (or more general: a regional polynomial drift [quadratic, cubic, etc.]) When a drift function is given, you don't have to provide the value for the drift "externally", where the name for EDK comes from.

TL;DR: For external drift kriging, you have to provide the drift value at each target point, where you want to interpolate your data. When you have a functional drift, you only have to provide that function and the drift value is determined by the position of the target point.

See for example: Webster R., Oliver M. A.: Geostatistics for Environmental Scientists

Answer 2

I think question is too vague. Kriging with external drift is a special case of Co-kriging, where drift is the co-variable (independent variable used for estimation). Some call it also Universal Kriging (see for example A minimal introduction to geostatistics with R or Kriging with External Drift). Different naming conventions, based on user background do not make thinks easier. I've never seen term "internal drift". Where have you found it?