Information contained within remote sensing data sets is distributed not only within geographical space, but also within complex multi-dimensional spectral and feature spaces. Few visualization tools currently permit the exploration of these data spaces and the relationships between them. A novel graphical virtual reality (VR) toolkit has been developed which can be used to visualize multi-dimensional Earth observation data in a more effective way. This toolkit has been adapted for the purpose of examining the relationships between different techniques, used to extract quantitative sub-pixel information from images, such as fuzzy membership representation and spectral unmixing. The VR system permits the display, in an immersive graphical environment, of several different three-dimensional representations of the same data. This approach makes it possible to more effectively understand the structure and dispersion of information in different feature sub-spaces which can aid modelling and product extraction.