Machine Learning Predicts Which Rivers, Streams, and Wetlands the Clean Water Act Regulates

RFF University Fellow Hannah Druckenmiller coauthored this study, which used a deep learning model to find that Supreme Court and White House rules offer dramatically different Clean Water Act regulation.

View Journal Article


Jan. 25, 2024


Simon Greenhill, Hannah Druckenmiller, Sherrie Wang, David A. Keiser, Manuela Girotto, Jason K. Moore, Nobuhiro Yamaguchi, Alberto Todeschini, and Joseph S. Shapiro


Journal Article in Science

Reading time

1 minute

Editor’s Summary

The Clean Water Act is a defining piece of environmental legislation in the US, but the waters that it protects from pollution have never been clearly defined. Greenhill et al. developed a machine learning model that uses geospatial data to predict which waters are covered by the Clean Water Act and trained and tested the model with jurisdictional determinations from the US Army Corps of Engineers. This work provides an estimate of the extent of protected waterways, as well as an understanding of the effects of Supreme Court and White House rules that have reinterpreted or changed the regulation. For a subset of sites with high predictive accuracy, their model can also act as decision support tool to expedite permitting. —Bianca Lopez


We assess which waters the Clean Water Act protects and how Supreme Court and White House rules change this regulation. We train a deep learning model using aerial imagery and geophysical data to predict 150,000 jurisdictional determinations from the Army Corps of Engineers, each deciding regulation for one water resource. Under a 2006 Supreme Court ruling, the Clean Water Act protects two-thirds of US streams and more than half of wetlands; under a 2020 White House rule, it protects less than half of streams and a fourth of wetlands, implying deregulation of 690,000 stream miles, 35 million wetland acres, and 30% of waters around drinking-water sources. Our framework can support permitting, policy design, and use of machine learning in regulatory implementation problems.


Simon Greenhill headshot

Simon Greenhill

University of California, Berkeley

Sherrie Wang headshot

Sherrie Wang

University of California, Berkeley

David Keiser headshot

David A. Keiser

University of Massachusetts, Amherst

Manuela Girotto headshot

Manuela Girotto

University of California, Berkeley

Jason K. Moore

US Department of Energy

Nobuhiro Yamaguchi headshot

Nobuhiro Yamaguchi

University of California, Berkeley

Alberto Todeschini

Alberto Todeschini

University of California, Berkeley

Joseph Shapiro headshot.jpg

Joseph S. Shapiro

University of California, Berkeley

Related Content