Separating bathymetric data representing multiscale rhythmic bed forms: A geostatistical and spectral method compared

The superimposition of rhythmic bed forms of different spatial scales is a common and natural phenomenon on sandy seabeds. The dynamics of such seabeds may interfere with different offshore activities and are therefore of interest to both scientists and offshore developers. State-of-the-art echo sounding accuracy allows for the analysis of bed form dynamics on unprecedented spatial and temporal scales. However, the superimposition of bed forms complicates the automated determination of morphodynamic parameters of individual bed form components. In this research we present the extension and comparison of two well-known, automated signal-processing methods for the 1-D and 2-D separation of bathymetric data derived from multibeam echo soundings into different components that each represents a bed form of a particular length scale. One method uses geostatistical filtering, and the other uses a Fourier decomposition of the bathymetric data. The application of both methods in two case studies of the North Sea shows that both methods are successful and that results correspond well. For example, megaripples up to 0.83 m height could be separated from 1.49-2.28 m high sand waves, and regionally averaged lengths and heights of sand waves, as calculated in either method, differ only 0.42-8.2% between methods. The obtained sand wave migration rates differ 7-11% between methods. The resulting morphometric and morphodynamic bed form quantification contributes to studies of empirical behavior and morphodynamic model validation and is valuable in risk assessments of offshore human activities.