A Cokriging Method for Spatial Functional Data with Applications in Oceanology

We propose a method based on a functional linear model which takes into account the spatial dependencies between sampled functions. The problem of estimating a function when spatial samples are available is turned to a standard cokriging problem for suitable choices of the regression function. This work is illustrated with environmental data in Antarctic where marine mammals operate as samplers. In the framework of second order stationarity, the application points out some di_culties when estimating the structure of spatial covariance between observations.