Modeling Inertial Particle Dynamics in Large Estuaries: A Lagrangian Clustering Framework

This study presents a Lagrangian analysis of floating particle transport dynamics in the Pearl River Estuary region, addressing critical gaps in understanding long-term variability and density-dependent anisotropy. Based on a 40-year ensemble of 100 representative hydrodynamic scenarios simulated by a regional circulation model, a lagrangian transport model was used to generate trajectories for particles of three densities, together with neutrally buoyant water tracers. Zonal ($R_{uu}$) and meridional ($R_{vv}$) velocity autocorrelation functions were computed for each trajectory and classified using k-means clustering. This data-driven approach objectively partitions the trajectories into five centroids, further grouped into looping, transient and non-looping regimes. Spatially, looping regimes dominate estuarine channels and offshore recirculation corridors, whereas non-looping regimes prevail in semi-enclosed bays and sheltered coastlines. A systematic density-controlled transition is observed: decreasing particle density attenuates oscillatory memory and increases the probability of non-looping behavior due to enhanced windage and Stokes drift. In these systems, transport dynamics exhibit a strong anisotropy governed by the alignment of seasonal monsoons with the estuarine geometry, which controls the spatial distribution of zonal and meridional decorrelation scales. Nonlinear stochastic fitting further shows that looping regimes exhibit a consistent tidal period of approximately 12 hours across all densities, whereas transient regimes experience a collapse of coherent oscillations. Using the Pearl River Estuary as a prototype, this framework provides a physically interpretable and transferable tool for identifying transport regimes and guiding targeted monitoring strategies across highly dynamic, river-dominated coastal environments.