Geostatistical sampling optimization of contaminated premises

This paper provides a methodological study of geostatistical approaches suitable to radiological evaluation in nuclear premises. By modelling the spatial continuity of activities, geostatistics provide sound methods to estimate and map surface activities, together with their uncertainty. The paper first recalls the main geostatistical principles, then investigate how it is possible to optimise the sampling strategy and avoid an exhaustive sampling, which usually leads to prohibitive costs. The kriging results of the sampling scenarios are compared to the exhaustive dataset in order to evaluate the sampling efficiency in terms of activity estimation and hot-spot detection.