Salt wedge intrusion events can severely affect and damage estuary ecosystems. Therefore predicting when they will occur is a considerable asset. A new study involving CMCC researchers uses machine learning and deep learning models to predict and forecast salt wedge intrusion and estuary salinity in the Po delta, Italy, yielding results that could help support smart management strategies worldwide.

Estuaries and their surrounding wetlands are home to unique plant and animal species, making them some of the most productive ecosystems in the world. However, estuaries are also delicate ecosystems that are increasingly threatened by salt wedge intrusion, which impacts their ecological balance and human-dependent activities.

In this context, accurately predicting estuary salinity is essential for water resource management, ecosystem preservation, and ensuring sustainable development along coastlines.

A recent CMCC study investigates the application of different machine learning (ML) and deep learning (DL) models to predict salinity levels within estuarine environments, using the Po River delta (the Mediterranean Sea’s second-largest freshwater inflow) as a case study by relying on data collected between 2016 and 2019 at the Po di Goro River mouth.

Leveraging different techniques, including Random Forest, Least-Squares Boosting, Artificial Neural Network and Long Short-Term Memory networks, the study attempts to enhance predictive accuracy in order to better understand the complex interplay of factors influencing estuarine salinity dynamics.

The map of the Po delta system (Google Earth image) along with its river mouths, namely, Po di Maistra, Po della Pila (which splits into Po di Tramontana, Po di Dritta, Po di Scirocco, Po di Bonifazi and Po di Bastimento), Po di Tolle, Po di Gnocca, and Po di Goro (yellow markers). The red path is the Po di Goro branch selected as the test case. The location of the Manufatto gauge station, which is close to the Goro mouth but in a secondary channel, is shown in the inset. Source: Saccotelli et al, 2024

The Po River estuary (Po di Goro), which is one of the main hotspots of salt wedge intrusion, was selected as the study area, with results highlighting an improvement in the machine learning performance, demonstrating the feasibility and effectiveness of ML-based approaches for estimating salinity levels due to salt wedge intrusion within estuaries.

The insights obtained from this study could significantly support smart management strategies, not only in the Po River estuary, but also in other locations, laying the groundwork for a broader discussion on the use of ML and DL based strategies for the monitoring of estuaries.

For more information:

Leonardo Saccotelli, Giorgia Verri, Alessandro De Lorenzis, Carla Cherubini, Rocco Caccioppoli, Giovanni Coppini, Rosalia Maglietta, Enhancing estuary salinity prediction: A Machine Learning and Deep Learning based approach, Applied Computing and Geosciences, Volume 23, 2024, 100173, ISSN 2590-1974, https://doi.org/10.1016/j.acags.2024.100173