Hello! In response to your question, it is my understanding that variograms may be used to evaluate polluted areas on both a "local" as well as a "regional" scale.

Perhaps the following research articles might be of some interest to you:

Spatial Analysis of Heavy Metals in Surface Soils based on Geostatistics (SUN et al., 2008)

http://144.206.159.178/ft/CONF/16423537/16423577.pdf

Inferences from Fluctuations in the Local Variogram about the Assumption of Stationarity in the Variance (CORSTANJE et al., 2008)

http://soils.ifas.ufl.edu/faculty/grunwald/home/PDFs/Geoderma_2008_localvariogram_Corstanje143_pp123.pdf

The Influence of Variogram Parameters on Optimal Sampling Schemes for Mapping by Kriging (Van GROENIGEN, 2000)

http://132.248.182.189/cursos/gest/Articulos/Geostatistics/The%20influence%20of%20variogram%20parameters%20on%20optimal%20sampling%20schemes%20for%20mapping%20by%20kriging.pdf

Variograms are used as part of the process to assess spatial variability, and thus can operate on any spatial scale. Variograms can be used to assess number of bacteria colonies on an agar plate right through to the number of animals of a particular species spread across the globe. It just depends on the scale you are working on. Obviously the more samples you have the better your assessment will be.

Look at that ppt at the end there is some theory on Kriging interpolation .. it uses the variogram

www.ce.utexas.edu/prof/maidment/giswr2007/visual/Surfaces.ppt

A variogram (or semi-variogram) is simply a technique for analysing the relationship between different sample points in space. It effectively asks how far two points must be seperated before they do not influence one another and how the strength of that influence differs up to that distance. The result can be used to then estimate values between points and map distributions (through krigging surfaces). As the previous comments have expressed these are scale independent, but do still require a considerable sample size, assumptions to be made about the type of model to be fitted to the data and the homogeneity, discontinuities and directional bias. If you are thinking of using it I would talk to a geostatistician first

Hi Clement,

The attached paper provides an example of how variograms can be used for local contamination of heavy metals. I'd echo the comments above regarding sample size. Good luck!

Joe