ABSTRACT
This chapter aims to teach the skills for exploring spatio-temporal data sets. Snapshots of spatial processes for a given time period can be plotted in numerous ways. A Hovmoeller plot is a two-dimensional space-time visualization in which space is collapsed onto one dimension and where the second dimension denotes time. An animation is typically constructed by plotting spatial data frame-by-frame, and then stringing them together in sequence. Spatio-temporal visualization in R generally proceeds using one of two methods: the trellis graph or the grammar of graphics. Visualization of data is certainly an important and necessary component of exploratory data analysis. Empirical orthogonal functions (EOFs) can reveal spatial structure in spatio-temporal data and can also be used for subsequent dimensionality reduction. The EOF decomposition is sometimes derived in a continuous-space context through a Karhunen–Loeve expansion, with eigenvalues and eigenfunctions obtained through a solution of a Fredholm integral equation.
