Author(s): Mirian Jimenez; Laura Cagigal; Alba Ricondo; Beatriz Perez-Diaz; Pablo Alonso-Alguacil; Javier Sopelana; Cielo Fernandez; Sonia Castanedo; Fernando Mendez

Linked Author(s):

Keywords: Metamodel; Salinity prediction; Water level prediction

Abstract: Estuarine systems are particularly vulnerable to climate change, as rising sea levels and altered river discharge lead to inland saltwater intrusion. This intrusion disrupts biogeochemical cycles, ecosystems, and human activities such as agriculture and industry by increasing salinity levels. To address these challenges, we present a novel hybrid methodology that combines advanced numerical modeling with statistical techniques to predict salinity behavior in estuaries. The approach begins by identifying seven key parameters influencing estuarine salinity, including tidal range, river discharge, and storm surges. Synthetic scenarios are generated using Latin Hypercube Sampling (LHS), and representative cases are selected through the Maximum Dissimilarity Algorithm. These selected cases are modeled with the Delft3d numerical model, which is calibrated and validated using field data from an estuary in northern Spain. Principal Component Analysis (PCA) reduces data dimensionality, identifying dominant salinity modes, followed by interpolation through Radial Basis Functions (RBFs) to create an interpolation surface. Results demonstrate that this hybrid methodology effectively captures salinity concentration dynamics under varied conditions, providing rapid and reliable predictions. Additionally, a user-friendly viewer is implemented to enable managers to quickly access and interpret the results obtained.

DOI:

Year: 2026