Abstract. This study is dedicated to the tidal dynamics in the Sylt-Rømø Bight with a focus on the non-linear processes. The FESOM-C model was used as the numerical tool, which works with triangular, rectangular or mixed grids and is equipped with a wetting/drying option. As the model's success at resolving currents largely depends on the quality of the bathymetric data, we have created a new bathymetric map for an area based on recent studies of Lister Deep, Lister Ley, Højer Deep and Rømø Deep. This new bathymetric product made it feasible to work with high-resolution grids (up to 2 m in the wetting/drying zone). As a result, we were able to study the tidal energy transformation and the role of higher harmonics in the domain in detail. For the first time, the tidal ellipses, maximum tidally induced velocities, energy fluxes and residual circulation maps were constructed and analysed for the entire bight. Additionally, tidal asymmetry maps were introduced and constructed. The full analysis was performed on two grids with different structures and showed a convergence of the results as well as fulfilment of the energy balance. A great deal of attention has been paid to the selection of open-boundary conditions, model validation against tide gauges and recent in situ current data. The tidal residual circulation and asymmetric tidal cycles largely define the circulation pattern, transport and accumulation of sediment, and the distribution of bedforms in the bight; therefore, the results presented in the article are necessary and useful benchmarks for further studies in the area, including baroclinic and sediment dynamics investigations.

No abstract

Pelagic chlorophyll-a concentrations are key for evaluation of the environmental status and productivity of marine systems, and data can be provided by in situ measurements, remote sensing and modelling. However, modelling chlorophyll-a is not trivial due to its nonlinear dynamics and complexity. In this study, chlorophyll-a concentrations for the Helgoland Roads time series were modeled using a number of measured water and environmental parameters. We chose three common machine learning algorithms from the literature: the support vector machine regressor, neural networks multi-layer perceptron regressor and random forest regressor. Results showed that the support vector machine regressor slightly outperformed other models. The evaluation with a test dataset and verification with an independent validation dataset for chlorophyll-a concentrations showed a good generalization capacity, evaluated by the root mean squared errors of less than 1 µg L−1. Feature selection and engineering are important and improved the models significantly, as measured in performance, improving the adjusted R2 by a minimum of 48%. We tested SARIMA in comparison and found that the univariate nature of SARIMA does not allow for better results than the machine learning models. Additionally, the computer processing time needed was much higher (prohibitive) for SARIMA.

Pelagic Chlorophyll-a concentrations are key for evaluation of the environmental status and productivity of marine systems. In this study, chlorophyll-a concentrations for the Helgoland Roads Time Series were modeled using a number of measured water and environmental parameters. We chose three common Machine Learning algorithms from the literature: Support Vector Machine Regressor, Neural Networks Multi-layer Perceptron Regressor and Random Forest Regressor. Results showed that Support Vector Machine Regressor slightly outperformed other models. The evaluation with a test dataset and verification with an independent validation dataset for chlorophyll-a concentrations showed a good generalization capacity, evaluated by the root mean squared errors of less than 1 µg L-1. Feature selection and engineering are important and improved the models significantly, as measured in performance, improving by a minimum of 48% the adjusted R2. We tested SARIMA in comparison and found that the univariate nature of SARIMA does not allow for better results than the Machine Learning models. Additionally, the computer processing time needed was much higher (prohibitive) for SARIMA.

No abstract