2-8 Spatial estimation of the risk of groundwater impairment of subsurface infrastructure in coastal regions

Session: Coastal Hydrogeology in an Age of Rising Seas

Presenting Author:

Authors:

Abstract:

In many coastal regions, sea level rise and rapid changes in the intensity of rainfall events have caused a rise in coastal groundwater levels. This can affect subsurface infrastructure, especially in low-lying coastal areas. Previous work based on machine learning (ML) models has established that the risk of groundwater impairment is inversely proportional to the depth to the groundwater table (DTW). This study aims to establish the easiest and most efficient way of estimating DTW. We considered two methods for estimating the groundwater surface: topographical interpolations based on known groundwater elevations, which are referred to as topographical interpolations in ArcGIS, and traditional groundwater modeling using MODFLOW. Topographical interpolations rely on mean high high water (MHHW) and the height of ephemeral streams and ponds to constrain the position of the groundwater surface. The groundwater surface is then interpolated using a natural neighbor method. Creation of a MODFLOW model is more complex and requires knowledge of hydrogeological boundaries and appropriate hydraulic conductivities. To obtain hydraulic conductivity estimates, we used a United States Department of Agriculture (USDA) database, which includes representative hydraulic conductivities of the upper 2 m of the subsurface. This database is freely available for download. Occasional gaps in this database must be filled for it to be usable in MODFLOW. We compared three approaches to filling the gaps: Voronoi interpolation, inverse distance weighting (IDW) interpolation, and using an average of hydraulic conductivities for similar sediment types. These methods were tested in Beaufort County, SC. 

The results of the topographical interpolation and MODFLOW models closely matched each other. The results of MODFLOW simulations using the USDA database produce a satisfactory fit between simulated and observed water levels, showing r-squared values between 85% and 98% for different locations in Beaufort County, while using an average hydraulic conductivity resulted in an r-squared range of 62–93%. These methods offer a temporal and spatial characterization of risks, which could improve planning, infrastructural design, and adaptation strategies for coastal communities, as well as provide an efficient way of calibrating groundwater models.

Geological Society of America Abstracts with Program. Vol. 57, No. 6, 2025

doi: 10.1130/abs/2025AM-9906

© Copyright 2025 The Geological Society of America (GSA), all rights reserved.