Muddled or mixed? Inferring paleoclimate from size distributions of deep-sea clastics

One of the outstanding problems of palaeoclimate reconstruction from physico-chemical properties of terrigenous deep-sea sediments stems from the fact that most basin fills are mixtures of sediment populations derived from different sources and transported to the site of deposition by different mechanisms. Conventional approaches to palaeoclimate reconstruction from deep-sea sediments do not distinguish between provenance and dispersal-related variations, and therefore often fail to recognise the true significance of variations in sediment properties. We formulate a set of requirements that each proposed palaeo-environmental indicator should fulfil, and focus on the intrinsic coupling between grain size and chemical composition. A critical review of past achievements in grain-size analysis serves to introduce a conceptual model of spatio-temporal grain-size variation based on dynamic populations. Each dynamic population results from a characteristic combination of production and transport mechanisms that corresponds to a distinct subpopulation in the data analysed. The mathematical-statistical representation of the conceptual model can be obtained by means of the end-member-modelling algorithm EMMA. Applications of the model to several ocean basins are discussed, as well as methods to examine the validity of grain-size-based palaeoclimate reconstructions. Palaeoclimate reconstructions of a high- and low-latitude basin illustrate the common degree of complexity of deep-sea grain-size records. © 2003 Elsevier B.V. All rights reserved.