Efficient data-driven flow modeling for accurate passive scalar advection in submesoscale domains

Abstract

Knowing the sea surface velocity field is essential for various applications, such as search and rescue operations and oil spill monitoring, where understanding the movement of objects or substances is critical. However, obtaining an accurate approximation of these advection processes is challenging, even with modern measuring equipment, such as high-frequency radar or advanced simulations based on oceanic flow models. Therefore this paper presents a data-driven framework to approximate sea surface velocity from spatially distributed observations, thus enabling efficient probability advection modeling across submesoscale domains. The system approximates transient flows by leveraging quasi-steady flow assumptions. To overcome the limitations of point measurements in capturing domain-wide circulation, the method employs a fusion of two simplified 2D flow models to approximate submesoscale dynamics, enabling complete velocity field reconstruction from scattered data. To ensure reliable flow dynamics, the approach iteratively adjusts boundary conditions in numerical simulations to align the simulated flow with observations. Experimental validation in Kvarner Bay using GPS-tracked drifters confirmed the system's ability to replace computationally intensive transient simulations by approximating flow fields based on model simplifications. The results demonstrate its efficiency across domains, making it a practical tool for real-world submesoscale applications requiring swift passive scalar advection.