Statistical Inference Associated with Spatio, Stationary, Spatio-Temporal Processes-an Application to Kriging


  • Prof. T. Subba Rao – University of Manchester
  • FCUL (DEIO) – Campo Grande – Bloco C4 Piso 2 – Sala: 4.2.07 – 14:30h
  • Quarta-feira, 19 de Maio de 2010
 Abstract:X(s) ; sRd; d Z} find suitable spatial models for estimating trend over space, and use these to estimate the observation, say X(s), at an given location so:This later problem is often called ‘Kriging’ , named after a geologist from south Africa. He used these Kriging methods successful in estimating the amount of minerals. The estimation techniques involve defining a parametric form for the covariance function (and spectral density) which involves several parameters. One has to estimate these parameters before one uses them for Kriging. We describe various definitions used to high light assumptions. We describe the standard methods and also describe new methods of estimation which are quick to implement and give good results. We illustrate these methods with two real data sets

In many areas of physical sciences, such as Geophysics, geology, environmental sciences, in recent years statistical methods have played an extremely important role in analysis of real data sets. One of the important problems often considered is that given a set of observations ,say

In recent years attempts have been made to extend these to spatio temporal data sets. However, the methods proposed so far have not been fully investigated. Here we propose new methods based on frequency domain methods for estimation and also Kriging. These methods are now used to analyse real data sets.

We also propose methods for analysing nonlinear spatial processes.