This links should help.

I would say variation below the scale of analysis (or sampling distance) is captured by the nugget and screening is just another phrase for gravity or distance decay. That is closer points have almost total influence at the expense of more distant points. Say in the situation of a compound that is highly volatile over distance (rapid decay over large distances)

Nugget

Natural phenomena can vary spatially over a range of scales. Variation at microscales smaller than the sampling distances will appear as part of the nugget effect. Before collecting data, it is important to gain an understanding of the scales of spatial variation in which you are interested.

Screen

The screening effect is the geostatistical term for the phenomenon of nearby observations tending to reduce the influence of more distant observations when using kriging (optimal linear prediction) for spatial interpolation (Journel and Huijbregts 1978, Chil`es and Delfiner 1999). This phenomenon is often invoked as a justification for ignoring more distant observations when using kriging.

http://help.arcgis.com/en/arcgisdesktop/10.0/help/index.html#//009z00000076000000.htm

http://arxiv.org/pdf/1203.1801.pdf

Well, these are two completely different phenomena. Nugget effect indicates the variability at zero lag distance (i.e., how much two samples situated next to each other can differ), thus indicating the residual error in any estimation. Screen effect is completely different and applies to cases when one sample can mask another, specifically when the "masked" one is directly behind the "masking" one with regards to the unknown. I personally worry more about the nugget effect. Hope this helps.

Many thanks Stuart E. Hamilton, Paul Alexandre for yours answers