On least squares estimation for long-memory lattice processes

JOURNAL OF MULTIVARIATE ANALYSIS

卷:100

Article

Beran, Jan¹  Schell, Dieter¹ 

[1] Univ Konstanz, Dept Math & Stat, D-7750 Constance, Germany

关键词: Long memory;    Fractional ARIMA process;    Lattice process;    Maximum likelihood estimation;    Anisotropy;   

DOI  :  10.1016/j.jmva.2009.04.007

来源: Elsevier

【 摘 要 】

A flexible class of anisotropic stationary lattice processes with long memory can be defined in terms of a two-way fractional ARIMA (FARIMA) representation. We consider parameter estimation based on minimizing an approximate residual sum of squares. The method can be applied to sampling areas that are not necessarily rectangular. A central limit theorem is derived under general conditions. The method is illustrated by an analysis of satellite data consisting of total column ozone amounts in Europe and the Atlantic respectively. (C) 2009 Elsevier Inc. All rights reserved.

【 授权许可】

Free