Hydrology, Cryosphere & Earth Surface Editors' Highlights

Machine Learning Predicts Subsidence from Groundwater Pumping

Machine learning and data on aquifer type, sediment thickness, and proxies for irrigation water use has been used to produce the most comprehensive map of land subsidence in the western U.S. to date.

Source: Water Resources Research


Excessive pumping of groundwater is leading to substantial land subsidence in many parts of the world. Monitoring land subsidence by remote sensing at sufficient resolution is costly, and GPS observations are usually too scattered to provide accurate maps. Smith and Majumdar [2020] use the machine learning method “random forests” to provide the first comprehensive map of land subsidence rates of the western United States. From these land subsidence rates, water loss from aquifers by excessive pumping in confined aquifers is also estimated. This shows that, apart from GRACE and water-balance based methods, there now exists a third route to estimating groundwater depletion in areas without piezometric data.

Citation: Smith, R. G., Majumdar, S. [2020]. Groundwater storage loss associated with land subsidence in Western United States mapped using machine learning. Water Resources Research, 56, e2019WR026621. https://doi.org/10.1029/2019WR026621

 —Marc F. P. Bierkens, Editor, Water Resources Research

Text © 2020. The authors. CC BY-NC-ND 3.0
Except where otherwise noted, images are subject to copyright. Any reuse without express permission from the copyright owner is prohibited.