Civil Engineering Faculty Publications

A Hybrid Physics-Guided and Machine-Learning Framework for Modeling Ground Subsidence in the Greater Houston Area

Document Type

Conference Proceeding

Publication Date

3-1-2026

Abstract

Ground subsidence, primarily driven by excessive groundwater extraction, poses a serious threat to heavily urbanized regions worldwide, including the Greater Houston area (GHA). This study introduces a hybrid framework that integrates physics-based modeling and machine learning to deliver robust, regional-scale forecasts of ground subsidence. Our approach uniquely combines multi-temporal InSAR and GPS data with geological and hydrological records, enabling the identification of both direct and indirect controls on subsidence. The resulting susceptibility map, validated by robust metrics, provides actionable insights for infrastructure planning and risk mitigation. Notably, the susceptibility map extends beyond current observations by incorporating multiple environmental and anthropogenic factors, allowing for the prediction of future risk zones even where subsidence has not yet been observed. This work advances the field by demonstrating how data-driven susceptibility mapping can reveal emerging hazards and guide proactive management strategies. Our findings highlight the importance of factors such as lithology, land use, and proximity to rivers, in addition to groundwater extraction, in controlling subsidence susceptibility in the GHA.

Comments

© 2026 American Society of Civil Engineers. All rights reserved, including rights for text and data mining and training of artificial technologies or similar technologies.

First Page

10

Publication Title

Geo Congress 2026 Site Characterization Rock Mechanics and Unsaturated Soils Selected Papers from Geo Congress 2026

Streaming Media

DOI

10.1061/9780784486740.002

Link to Full Text

Share

COinS