Sensitive parameters for hydrodynamic modelling of a multi-use case study

The sector of aquaculture cultivation is rapidly growing, since aquaculture is a promising food and protein supply for the increasing human world population. However, aquaculture is adding to the competition of marine space, especially in European Seas. One possible solution for relaxation of spatial competition is an approach of marine multi-use such as the combination of food and renewable energy production. Implementation of offshore aquaculture cultivation systems creates challenges for offshore engineers with respect to risks and huge costs. Large-scale hydrodynamic models, which are able to represent large and small-scale impacts, can serve as a tool to ease such challenges during the planning and assessment phase. One of these models is the calibrated and validated Dutch Continental Shelf Model (DCSM), developed by Deltares. This hydrodynamic large-scale and three-dimensional model covers the North-Western European Shelf and is based on the D-FLOW FM software. However, due to the complexity and thus computational costs of the DCSM, the application of the DCSM for various inputs and scenarios to understand multi-use in practice is limited. In this work, the DCSM is used for a case study in the North Sea. To minimise the computational amount, the considered area is cut-out from the DCSM. This nested model is based on initial conditions and inputs of the DCSM and the FINO3 platform. This platform was chosen as it is part of the UNITED project. The UNITED project is 4-year Horizon2020 EU project led by Deltares with 26 partners. The present study uses this nested model to investigate the sensitivities of input parameters. The variables water temperature, salinity and current velocities are selected, since these are the most important variables for mussel and seaweed cultivation, which are covered by this model. It is important to have information on the impact by changing the model input. Therefore, the parameters will be ranked according to their sensitivity. Since the used model still is a large and complex model, several sensitivity analysis techniques will be used. The Morris method will give a pre-liminary ranking of parameters. However, this method only changes one input at a time (one-at-a-time) and does not consider correlations between parameters. Therefore, it is planned that the method will be extended by copulas for the model input. Furthermore, it is planned to also give information about the variances of outputs. The Sobol’ variance-based method will be applied on the most influential parameters as previously detected, because the number of model runs is dependent on the number of parameters. The final results can later be used for model optimisation to allow efficient spatial planning of marine multi-use configurations.