The Revival of Physical Modelling in Oceanology: Integration of Experimental and Computational Methods Based on Precision Laboratory Research

The article discusses the revival of interest in physical modelling in oceanology on a new technological basis. It is shown that modern computational methods (Large Eddy Simulation, Neural Differential Models) require detailed verification against reference data that only high-precision laboratory experiments can provide. Using the example of the Unique Scientific Facility ``Laboratory-Test Complex'' of the St. Petersburg Branch of the Shirshov Institute of Oceanology, the integration of physical modelling, digital twins, and machine learning methods is demonstrated. The results of experiments on convective vortex structures and the generation of vortex rings in a stratified environment are described. The prospects of using nonlocal hydrodynamics and neural network parameterizations for creating a new generation of ocean circulation models are substantiated.