Spatial data in R

Introduction

R is a language for statistical computing. As such it has a strong capacity for storing, handling and manipulating numerical or categorical data. R has dedicated data structures and methods for specific kinds of data, e.g. time series data, but, as of yet, not for spatial data. Although one could argue that spatial data are just regular data with coordinates, life is much easier when they are structured. A large number of packages provide spatial statistical methods or interfaces to GIS, and many of them provide data structures and e.g. plotting methods for spatial data. There is however little coordination between them, and only very few provide functionality for more than one spatial data type.

A package called sp

Why do we need a package of spatial data classes?

• with support for a single package for spatial data, it is much easier to go from one spatial statistics package to another. Either the packages directly support reading and writing of data in these spatial classes, or they have to provide exchange to and from the package. This requires one-to-many links, which is easier to provide and maintain than many-to-many links.
• the package provides methods (functions) for plotting, printing and summarizing spatial objects, or combining them (e.g. overlay)
• packages with interfaces to GIS and geographic projection code support the new classes; some of this functionality is demonstrated here. Note that this functionality is now released on CRAN, in packages rgdal, maptools, and spgrass6.
• the new package provides Lattice plots, conditioning plots, and plot methods that combine points, lines, polygons and grids, scale bars, north arrows etc. (see the gallery)
• the package uses S4 classes, allowing for the validation of internal consistency