Abstract:

In this article, we address the numerical solution of a functional spatial regression model using Functional Data Analysis (FDA) and Generalized Least Squares (GLS) techniques. The proposed model considers the scalar response with spatial dependence in a continuous domain, incorporating this dependence with semivariogram models and interpolation methods such as ordinary kriging. The predictive variables form a multivariate functional random field modeled using cokriging interpolation methods. We apply generalized least squares to estimate the parameters and then implement it using climatic data from the Caldas region for validation.