Abstract
Coastal ecosystems are increasingly vulnerable to climate-driven flooding, yet predicting subsurface water movement remains challenging due to soil heterogeneity. This study presents a hydrogeophysical framework that integrates time-lapse electrical resistivity imaging (ERI) with numerical flow modeling to characterize infiltration during an ecosystem-scale flooding experiment. By identifying local preferential flow paths and stratigraphic bottlenecks often missed by regional soil databases, the framework significantly improves the accuracy of soil saturation predictions, offering a robust tool for ecohydrological modeling in a changing climate.
Introduction
The emergence of "Ghost Forests" along coastal margins highlights the urgent need to understand how extreme flooding events impact terrestrial ecosystems. Traditional ecohydrological models often rely on point sensors or coarse national soil databases like SSURGO, which frequently fail to capture the complex "layer cake" of sand, silt, and clay that defines local subsurface flow. These "micro-heterogeneities" can create preferential flow paths—underground conduits that move water much faster than predicted—or restrictive layers that prolong saturation. This paper introduces a framework to bridge this gap using non-invasive geophysical imaging to "see" subsurface structures and track water movement in real-time.
Methods
The research was conducted at a site in Maryland as part of the Terrestrial Ecosystem Manipulation to Probe the Effects of Storm Treatments (TEMPEST) experiment. The framework employed a multi-modal approach to capture site-specific soil characteristics:
- Static Geophysics: Electrical Resistivity Imaging (ERI) and Ground Penetrating Radar (GPR) were used to map the initial site stratigraphy down to approximately 4 meters, identifying distinct soil horizons.
- Time-Lapse Monitoring: During a simulated 10-hour flood (300 $m^3$ of water), ERI was performed repeatedly. Because wetter soil is less resistive to electricity, this allowed researchers to "film" the water as it infiltrated the ground.
- Numerical Modeling: The Advanced Terrestrial Simulator (ATS) was used to couple surface and subsurface flow. Two models were compared: a "geophysics-informed" model using ERI/GPR data for layering, and a "database-informed" model using standard SSURGO data.
Results
The geophysical data revealed a critical silty clay layer between 1 and 2 meters deep that acted as a hydraulic bottleneck, restricting deep infiltration and maintaining high topsoil saturation longer than standard models predicted. Crucially, time-lapse ERI identified a preferential flow path in one profile where water bypassed this layer—a feature entirely absent in the national soil database.
The geophysics-informed model demonstrated a significant improvement in predictive accuracy, with an 11% to 48% reduction in Root Mean Square Error (RMSE) compared to actual field sensor data. Time-lapse ERI recorded a resistivity drop of up to 35% during the flood, which correlated strongly with in-situ moisture sensor readings ($R^2 > 0.53$).
Discussion and Conclusion
This study demonstrates that national soil databases, while useful for large-scale watershed studies, are insufficient for predicting local-scale responses to extreme events. The integration of geophysics allows for the detection of "hidden" features like clay discontinuities that dictate water movement.
While the current study utilized 2D imaging, future work aims to expand this framework into 3D volumes to better capture complex flow geometries. Additionally, incorporating solute transport modeling to track saltwater intrusion will be critical for understanding the formation of Ghost Forests. Ultimately, this hydrogeophysical approach provides a scalable method to improve the predictive power of ecosystem models facing increased coastal flooding.
Summary updated