Sull'analisi di serie spaziali non stazionarie. un metodo basato sul ricampionamento

Abstract

In this paper we consider the problem of estimating the unknown parameters of non-stationary spatial process when only a single replication of the process is available. The approach suggested here, to overcome the problem of estimating a number of parameters a lot larger then the number of observation, is based upon the general idea of resampling. A set of "moving subsample" are considered in the study area and the unknown parameters are estimated within them. An application of the method to some agricultural field trials data is also considered in the paper.