Logratio Statistics and the Extraction of Environmental,Signals from Grain-Size Data

Key words: 
• logratio statistics /
• RQ-mode principal components analysis /
• grain-size data /
• sedimentary hydrodynamic environment

Abstract: Grain-size distributions of sediments contain important information on sedimentary source, dynamics and environment. The correct use of statistical techniques to extract as much as possible useful signals from grain-size data is of crucial significance. When applying multivariate analysis to grain-size data, however, the auther encountered the problem of "closure",which may mask true patterns significantly. The logratio statistics is an effective way to avoid this problem. This paper presents two case studies respectively in the western Baltic Sea and northern South China Sea. While the classical statistics failed in representing the natural clustering of sediment samples, the RQ-mode principal components analysis based on logratio transforms of grain-size data made a clear separation among sediment classes. The sedimentary regionalization based on this classification successfully reflects the spacial variation of sedimetary hydredynamic environments in these areas.