Desegregation, interpolation and integration of spatial series

Abstract

So far most methods for conversion of areal data have referred to classical point interpolation procedures developed in spatial statistics for physical sciences or, have been conceived as "area-based areal interpolation methods" based on specific and somewhat unrealistic hypotheses. Addressing the desegregation problem as a part of a more comprehensive aggregation problem, the methodology that we present in this paper features in essence a logical and analytical continuity with all the previous studies in which the effects on statistical inference of areal aggregation have been considered. Within this context, a Bayesian solution is shown to be suitable for the porpoise and the advantages of adopting it are clearly demonstrated particularly for its capability of exploiting auxiliary information which can be easily accessed and manipulated thanks to the recent advances of Geographical Information Systems (G.I.S.).