A physics-informed machine learning framework for high-resolution land subsidence prediction in the Greater Houston Area

Abstract Land subsidence induced by groundwater extraction poses significant threats to urban infrastructure and flood resilience in rapidly urbanizing coastal regions worldwide. This study presents a novel physics-informed machine learning framework that integrates Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) observations, consolidation theory principles, and convolutional neural networks (CNN) for comprehensive land subsidence prediction in the Greater Houston Area. Processing 155 Sentinel-1 SAR images spanning January 2020 to March 2023 generated a high-resolution deformation dataset of 10,652 persistent scatterer points validated through comprehensive integration with 54 continuously operating GPS reference stations, achieving RMSE of 2.1 mm/year and correlation coefficient of 0.96. The hybrid framework integrates twelve mapped conditioning factors representing geological, hydrogeological, topographic, and anthropogenic domains, supplemented by clay-thickness and growth-fault inputs, within a physics-guided CNN architecture, achieving an Area Under the Receiver Operating Characteristic (AUROC) of 0.9862, precision of 0.94, and recall of 0.92, which substantially exceed those of conventional machine learning approaches. Spatial analysis reveals that 18.66% of the study area (approximately 1,848 km 2 ) exhibits high to very high land subsidence susceptibility, with maximum observed rates reaching 24.3 mm/year in western Harris County near the Katy Prairie region. Unlike purely data-driven models that infer temporal dependence from empirical associations, the proposed framework embeds consolidation-derived temporal memory, allowing delayed aquitard response to influence subsidence susceptibility. Results indicate continued land subsidence despite significant regulatory interventions aimed at reducing groundwater extraction, with emerging subsidence hotspots identified in suburban areas experiencing rapid development. This framework provides critical insights for sustainable groundwater management, infrastructure planning, and flood risk mitigation in the Houston metropolitan region and offers a transferable methodology for similar coastal subsidence-prone areas worldwide.

Authors

Institutions

Publication Details

Journal
International Journal of Geo-Engineering
Published
2026-11-13
DOI
https://doi.org/10.1186/s40703-026-00298-0
Primary Topic
Synthetic Aperture Radar (SAR) Applications and Techniques
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

A physics-informed machine learning framework for high-resolution land subsidence prediction in the Greater Houston Area

Jinwoo An, Shinwoo Choi, Md Saffiquzzaman Chowdhury, Yong Je Kim
1 links
International Journal of Geo-Engineering
Synthetic Aperture Radar (SAR) Applications and Techniques
article

A physics-informed machine learning framework for high-resolution land subsidence prediction in the Greater Houston Area

Jinwoo An, Shinwoo Choi, Md Saffiquzzaman Chowdhury, Yong Je Kim
article en

Abstract

Abstract Land subsidence induced by groundwater extraction poses significant threats to urban infrastructure and flood resilience in rapidly urbanizing coastal regions worldwide. This study presents a novel physics-informed machine learning framework that integrates Persistent Scatterer Interferometric Synthetic Aperture Radar (PS-InSAR) observations, consolidation theory principles, and convolutional neural networks (CNN) for comprehensive land subsidence prediction in the Greater Houston Area. Processing 155 Sentinel-1 SAR images spanning January 2020 to March 2023 generated a high-resolution deformation dataset of 10,652 persistent scatterer points validated through comprehensive integration with 54 continuously operating GPS reference stations, achieving RMSE of 2.1 mm/year and correlation coefficient of 0.96. The hybrid framework integrates twelve mapped conditioning factors representing geological, hydrogeological, topographic, and anthropogenic domains, supplemented by clay-thickness and growth-fault inputs, within a physics-guided CNN architecture, achieving an Area Under the Receiver Operating Characteristic (AUROC) of 0.9862, precision of 0.94, and recall of 0.92, which substantially exceed those of conventional machine learning approaches. Spatial analysis reveals that 18.66% of the study area (approximately 1,848 km 2 ) exhibits high to very high land subsidence susceptibility, with maximum observed rates reaching 24.3 mm/year in western Harris County near the Katy Prairie region. Unlike purely data-driven models that infer temporal dependence from empirical associations, the proposed framework embeds consolidation-derived temporal memory, allowing delayed aquitard response to influence subsidence susceptibility. Results indicate continued land subsidence despite significant regulatory interventions aimed at reducing groundwater extraction, with emerging subsidence hotspots identified in suburban areas experiencing rapid development. This framework provides critical insights for sustainable groundwater management, infrastructure planning, and flood risk mitigation in the Houston metropolitan region and offers a transferable methodology for similar coastal subsidence-prone areas worldwide.

International Journal of Geo-EngineeringVol. 17(1)
The University of Texas at El Paso (US), The University of Texas Rio Grande Valley (US)
Sustainable cities and communities
Synthetic Aperture Radar (SAR) Applications and Techniques
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon