Fractionally Integrated Separable Spatial Autoregressive (FISSAR) Model and Some of Its Properties

Abstract

Spatial modelling has its applications in many fields. In time-series there exist a class of models known as long memory models where the autocorrelation function decays rather slowly. These types of time-series data are modelled as fractionally integrated ARMA processes. Spatial data may also exhibit a long memory structure and in order to model such a structure we introduce a new class of models called the fractionally integrated separable spatial autoregressive (FISSAR) model and discuss some of its properties. One way of estimating the parameters of the FISSAR model is also discussed in this article.