There is an increasing trend for the collection of multi-sensory quantitative data to support the mapping of geology and environment. In the United Kingdom and Ireland this trend has been led by the Tellus mapping programmes; large scale multidisciplinary surveys which have collected quantitative data by a combination of geophysical survey from the air and geochemical survey on the ground. Such datasets contain a huge amount of geological and environmental information. However, these datasets have tended to be analysed on a variable-by-variable basis rather than as an integrated representation of a single geoenvironmental system. Using the example of Northern Ireland, this paper presents a demonstration of the quality of information that can be extracted through an integrated approach using modern data analytics. Two tools are used: FastICA independent component analysis to unmix the full composition of Northern Ireland’s soil geochemistry into meaningful components, and the random forest machine learning algorithm to map these components in high-resolution according to their relationships with geophysical parameters. We find that when unmixed to eight independent components, each explaining different aspects of geological and surficial processes, the geochemical features of Northern Ireland can be interpreted concisely. High resolution mapping aids this interpretation, with the random forest approach providing more accurate maps than traditional IDW interpolation for all but one of the components. In addition, by recombining the high resolution maps of independent components into a ternary colour image, a highly detailed output is produced in which all the features of the region’s traditional geological map (and more) can be seen, all as a continuous and accurate fully quantitative representation of Northern Ireland’s geochemical composition.