MerQuR+Statistics Glossary

Statistics Glossary → Nugget, sill and range

Nugget, sill and rangeconcept

The three parameters that define a variogram model: the unexplained variability at very short distances (nugget), the plateau the variogram reaches (sill), and the distance at which spatial correlation dies out (range).

The semivariogram plots, against distance, half the average squared difference between pairs of points that distance apart. For a spatially correlated variable, nearby points are similar, so the variogram is low at short distances and rises with distance. The nugget is the value the variogram stays at, rather than dropping to zero, as distance approaches zero; it reflects measurement error and variation at scales smaller than the sampling interval. The sill is the plateau at which the variogram levels off and roughly corresponds to the total variance; the difference between sill and nugget is the partial sill (the spatially structured variance). The range is the distance at which the plateau is reached; points farther apart than this are uncorrelated.

Because the exponential and Gaussian models approach the plateau only asymptotically, the distance at which 95% of the sill is reached is used as the 'practical range'. The ratio of nugget to sill summarizes the strength of spatial dependence: if it is small, most of the variability is spatially structured and kriging gives good predictions; if it is close to 1, the spatial structure is weak. The parameters are estimated from the empirical variogram, so they are uncertain with little data and at large distances; fixing the nugget at zero forces kriging predictions to pass exactly through the data points.

An everyday example

Soil moisture is measured at 200 points in a field. Even two points 1 metre apart give slightly different values (the nugget: measurement error and small-scale differences). The differences grow with distance and stop growing beyond about 80 metres: that 80 metres is the range, and points farther apart than this tell you nothing about each other's moisture. The plateau the differences reach is the sill.

In MerQur

In Variogram + Kriging, the 'Fit nugget effect' checkbox; if unchecked, the nugget is fixed at 0. The sill and range are estimated together with the chosen theoretical variogram model.

Analyses where this concept appears

Variogram + Kriging (Spatial Interpolation)

Related concepts

Variogram (semivariogram)Kriging (simple, ordinary, universal, block)