Many natural processes produce “mixtures”, which consist of a combination of different parts that occur in varying abundances. Not surprisingly, mixtures are ubiquitous through the Earth Sciences in fields such as mineralogy, tracer analysis, sedimentology, remote sensing and even palaeomagnetism. Our interpretation of data sets consisting of samples from an undefined mixing system are often based on separating bulk signals into meaningful component parts. Specifically, many geoscience data sets are viewed as a collection of samples representing mixtures of a relatively small number of so-called “end-members”. In certain situations, however, the end-members of a given mixing system may not have been sampled and cannot be defined from theory. In such situations, how can we gain empirical information about end-members in order to develop a quantitative understanding of the mixture system we are studying?
Using examples from a variety of fields, I will show some of the history of “unmixing” and how recent developments have provided intuitive techniques for end-member identification and quantification. Such unmixing techniques take a general approach that can help in the characterisation and interpretation of a variety of data sets and have a wide spectrum of applications in the geosciences.