Machine learning for scalable and adaptive estuarine modeling: understanding dynamics and environmental patterns

Accurate modeling of estuarine systems is essential for understanding estuaries dynamics. However, it remains a major challenge due to site-specific hydrodynamics and the scarcity of in-situ observations. This study overcomes the complexity of diverse estuaries and limited local data through an unsupervised learning approach for classifying estuarine systems, with the aim of supporting more transferable and reusable modeling frameworks. Hydrodynamic and environmental data from multiple estuarine branches of four major European rivers—Po, Danube, Ebro, and Rhone—are analyzed using KMeans and KMedoids clustering algorithms. The analysis is performed with the goal of identify diverse estuarine behaviors shaped by key physical drivers such as salinity and river discharge. Both algorithms yield stable and interpretable groupings. These promising results suggest that unsupervised learning can serve as a valuable first step toward building scalable and generalizable models for estuarine monitoring.