Lognormal kriging: bias adjustment and kriging variances
Lognormality of spatial data occurs commonly enough for it to warrant continued study;, contemporary statistical and computational methodologies can shed new light on the old problem of block kriging for lognormal processes. There are a number of proposals available for block kriging, many of them discussed in an unpublished, 43-page, Centre de Morphologie Mathernatique "note" written by Georges Matheron in 1974. Loosely translated, the title of the note is, "The proportional effect and lognormality or: The return of the sea serpent". Our paper is meant to rein in the sea serpent, by comparing an optimal-prediction-based predictor with a permanence-approximation-based predictor, in the context of statistics for spatial lognormal data.