The Danube river role in the Black Sea dynamics

Rivers play a fundamental role in shaping the physical and biogeochemical properties of coastal and marginal seas. However, the representation of riverine inputs in ocean models remains a major challenge due to the scarcity of in-situ observations, the limited resolution of regional models, and the complex multi-scale interactions occurring within estuarine systems. This PhD work addresses this gap by developing and validating a framework that couples high-fidelity estuarine modeling with mesoscale ocean simulations, using the Danube River–Black Sea system as a case study. The main objective is to improve the realism and cost-effectiveness of freshwater flux representation in regional ocean models and enhance the simulation of river–sea exchanges and their impact on coastal dynamics. The research introduces a modeling strategy integrating models of different complexity. A 3D, high-resolution finite element model was implemented over the Danube river–delta–sea continuum to resolve the thermo-hydrodynamic structure of the estuary. This model generates a Learning Dataset of pseudo-observations used to train and test a computationally efficient Emulator. The Emulator is a machine-learning-based, physically informed Estuary Box Model capable of representing estuarine exchanges in terms of discharge, salinity, heat fluxes, and salt wedge intrusion length. It closely reproduces the Learning Dataset, achieving performance comparable to the 3D model. Inland river temperature is identified as a key variable for accurately estimating temperature along the branches and at the river mouths, significantly influencing estuarine stratification. Later, the Estuary Emulator was coupled with a mesoscale regional ocean model for the Black Sea. Multi-annual experiments show that the coupled simulations improve the representation of the Danube plume, reproducing its structure, offshore extension, and seasonal variability. Including riverine temperature further enhances near-mouth thermal structure and stratification. Overall, this work demonstrates a physically consistent and computationally efficient approach for representing river–sea interactions in regional ocean models.