Assessment of groundwater salinization dynamics in coastal aquifers using geophysical, hydrochemical, and machine learning approaches (Note: If you prefer a different focus within Geo-science, I can suggest alternatives.)

Project Abstract

In many coastal regions, freshwater availability is increasingly challenged by salinization driven by sea-level rise, intensive groundwater pumping, and climate-induced shifts in recharge, necessitating integrated approaches to characterize, predict, and manage groundwater quality. This study integrates geophysical imaging, hydrochemical analyses, and machine learning to elucidate the spatiotemporal dynamics of groundwater salinization in a representative coastal aquifer system. We deploy seismic refraction and electrical resistivity tomography (ERT) surveys to delineate freshwater-saltwater interfaces, identify preferential flow paths, and map lithological controls that govern salinity intrusion. Hydrochemical data, including major ions, stable isotopes, and groundwater age tracers, are used to characterize recharge sources, mixing processes, and groundwater residence times, enabling robust end-member mixing analyses and hydrogeochemical facies classification. High-resolution time-series measurements from monitoring wells capture salinity trends and reaction-transport processes under varying pumping regimes and seasonal recharge, providing a comprehensive picture of intrusion mechanics. To integrate these heterogeneous datasets, we develop a machine learning framework that combines supervised and unsupervised techniques. Feature engineering incorporates geophysical resistivity, porosity estimates, hydraulic parameters, well yield, land use, and climate variables. Supervised models predict salinity concentration and high-risk zones, trained on labeled observations across hydrological cycles, while unsupervised clustering reveals distinct salinity regimes and transition zones. We implement spatially explicit models, including Gaussian process regression and random forest, to quantify predictive uncertainty and identify nonlinear thresholds associated with aquifer resilience. The study also introduces a data assimilation component that updates model states with new measurements to improve real-time decision support for groundwater management. Results reveal a multi-layered salinization process driven by a combination of density-dependent saline water intrusion and advective-dispersive transport modulated by pumping intensity and recharge variability. Geophysical inversions indicate a persistent freshwater lens thickness reduction during drought periods, with salinity fronts advancing landward in zones of high pumping concentration. Hydrochemical signatures corroborate mixing between meteoric recharge and seawater end-members, with isotopic data suggesting recent recharge events contribute to transient dilution while older waters reflect longer residence times. The machine learning models achieve high predictive skill (R2 > 0.85 on validation sets) and effectively identify critical thresholds where small increases in pumping or sea-level rise precipitate disproportionate increases in salinity. Sensitivity analyses highlight the dominant role of hydraulic conductivity contrasts and boundary conditions at coastal margins, while scenario simulations demonstrate potential mitigation pathways, such as optimized pumping schedules, managed aquifer recharge, and land-use planning to reduce recharge contaminants. The integrated framework provides a scalable template for coastal aquifer assessment worldwide, offering actionable insights for water resource managers to anticipate salinization risks, design monitoring networks, and implement adaptive strategies under changing climate and groundwater usage. This study advances methodological integration in geo-science by coupling geophysical imaging, hydrochemical interpretation, and machine learning-driven decision support to enhance understanding and management of coastal groundwater salinization.

Project Overview

What This Project Is About

---

A straightforward study of how groundwater near coasts becomes salty over time. It combines simple tests of water chemistry, geophysical signals from the ground, and basic computer methods to find patterns and explain why salinization happens where and when it does.

---

---

The Problem It Addresses

---

Coastal groundwater can become too salty for drinking or farming, especially with sea level rise and human extraction. There is a gap in using combined data from field tests, ground measurements, and easy computer tools to predict salinization and guide water management.

---

---

Objectives of the Project

---

1. Describe current salinity levels in a coastal aquifer using simple tests.
2. Identify factors that contribute to salinization (tides, pumping, rainfall, geology).
3. Use basic machine learning to find patterns between environmental data and salinity changes.
4. Develop a practical guide for local water managers on monitoring and mitigation.

---

---

What You Will Do Step by Step

---

1. Review basic groundwater and salinity concepts in plain terms.
2. Collect water samples and measure salinity, temperature, and conductivity in the field.
3. Perform simple geophysical readings to infer soil and rock properties around the aquifer.
4. Analyze data with easy-to-use software to spot trends (no advanced coding required).
5. Test a few straightforward models to relate rainfall, pumping, and salinity.
6. Interpret results and discuss practical implications for local water supply.

---

---

Expected Outcome

Clear indications of what drives salinization in the chosen area, plus a simple decision aid for monitoring and basic actions to protect freshwater wells.