[spatial statistics (use for geostatistics)]
In a spatial model, variation in the value of a variable that cannot be described by a mathematical function and is not spatially correlated: it includes measurement error and microscale variation (variation at a finer scale than that at which the data has been sampled). Random noise is one of the three main components—along with drift and spatially correlated variation—that contribute to the change in value of a variable over a surface. In a semivariogram, random noise is represented by the nugget. Random noise is sometimes called white noise.