ABSTRACT
Present at the heart of any geostatistical approach and all spatial statistics, additivity is a rarely evoked concept in the literature. At most, it is mentioned as a prerequisite for calculations, without further considerations, stating in a rather vague way that a variable is additive if ‘its arithmetic average has a meaning’. In geotechnics, intrinsically nonadditive quantities have also been found, such as the rock quality designation or the fracture frequency (FF). Without presuming a linearity under a change of support, the directionality of these variables prevents using measurements taken in different directions in the same kriging neighborhood, even if they are corrected for their biases such as FF with the Terzaghi approach. The sampling theory is a sister discipline of geostatistics. Geostatistics aims to take advantage of the spatial correlations and the correlations between variables, and not make them disappear.
