A Multivariate Mixing Model for Identifying Sediment Source from Magnetic Measurements

Abstract

Abstract A sequential method for quantitative identification of sediment source components, based on magnetic measurements, has been developed and tested for sediments from the Rhode River, Maryland. Simulated mixing tests and multiple regression were employed to establish numerical relationships between source component proportions and the magnetic measurements of mixtures. On the basis of these multivariate mixing models, source components of three estuarine sediment cores were estimated by linear programming. The results strongly support the previous studies on this catchment which indicated a dramatic change in sediment source some 150 to 200 yr ago. Quantitative calculations are more useful and informative than purely qualitative descriptions.