engleski

Example of Granulometric Analysis of Estuarine Sediments Using Conventional Graphical Methods and in Comparison with Random Models

Knowledge about any grain size distribution in riverine, estuarine or marine sediments is fundamental to study geomorphological processes, trace element transport, weathering processes, e.t.c. It is closely related to physical forces which are active during transport and deposition. There are many papers dealing with granulometric analysis of sediments, using different techniques for measuring. However, grain size data are mostly presented in graphical form. Statistical parameters are mostly calculated using methodology of Folk and Ward (1957). In this paper we present experimental results of granulometric measurements performed on sediments from Oere Estuary in Bothnian Bay, North Sweden. It is an example of clean, semiclosed, well mixed and Wind-dominated estuary. In addition to conventional granulometric parameters which we have calculated using graphical method, we propose a new methodology using Mathematical Morphology, for which we hope that will be useful to sedimentologists in the future. Mathematical Granulometric parameters are powerful shape descriptors used commonly in texture classification and shape modeling. The new methodology that is proposed adapts the classical mathematical granulometries to sediment sieving process. A random sample with proper grain descriptors and a random coverage could be used to model sediment heterogeneity and its granulometric size distribution would describe such a sediment model. Random sided shapes were tested that closely models the microscopic pictures of natural sediment grains. The combinations of different shaped primitives were tested and the ones that modeled the sediments gave the best agreement with the experimental grain size parameters. Results shows that particle morphology in addition to grain size should be considered when interpreting experimental granulometric measurements.

granulometric analysis; estuarine sediments; graphical methods; random modelling

nije evidentirano