(22) Measuring laboratory stratigraphy through image analysis

From Stratodynamics


Clara Orru1,2, Astrid Blom1, Victor Chavarrias3, Wim S. J. Uijttewaal1

1 Environmental Fluid Mechanics Section, Civil Engineering and Geosciences, Delft University of Technology, PO Box 5048, 2600 GA, Delft, The Netherlands.

2 Email: C.Orru@tudelft.nl

3 Escola Tecnica Superior d'Enginyers de Camins, Canals i Ports de Barcelona, Universitat Politecnica de Catalunya, C/ Jordi Girona, 31, 08034 Barcelona, Spain.


Grain-size selective processes in rivers lead to sorting of sediment in all directions, over various temporal and spatial scales. Sorting here indicates the spatial variation in the grain size distribution of the bed surface and/or substrate. Improvement of measurement techniques to define such spatial variation helps to provide new insights on grain-size selective processes. In this study a laboratory technique was developed to measure the spatial variation in the stratigraphy: image analysis combined with particle colouring and a sampling technique to remove thin layers of the deposit. Two experiments were conducted to evaluate the techniques. In both tests we used three well sorted grain size fractions with only slight overlap in grain size, within the range of coarse sand to fine gravel. The three grain size fractions were painted in three different colours. During the first test, patches composed of various mixtures of the three grain size fractions and various colour combinations were installed. Under submerged and unsubmerged conditions images were taken of the bed surface and the relative presence of each fraction was determined processing them with a colour segmentation algorithm (Figure 1), which provides the areal fraction of a specific colour (i.e. grain size). In the second experiment the above image analysis technique was combined with the sampling procedure. The technique was applied to measuring the stratigraphy of a prograding Gilbert delta (Figure 2). Areal images of the bed were taken at various elevations of the deposit. These elevations correspond to the top of horizontal layers of 1 cm thickness that were removed with a vacuum cleaner. The data for the stratigraphy resulting from the image analysis agree well with the ones based on the sieve analysis, which were necessary only to validate the method. The application of the technique to determine the stratigraphy of the Gilbert delta was therefore suitable to interpret the sorting processes which formed the deposit. In conclusion, the grain size distribution of the bed surface is measured with sufficient accuracy and combining the technique with an efficient and flexible sampling procedure the stratigraphy of the deposit is rapidly estimated. A large amount of data can be collected and processed quickly. Time consuming sieve analyses are avoided and it is possible to place back the sediment at its original elevation and continue the flume experiment. Only slight disturbances occur due to the removal of sediment layers.