Machine learning predicts which rivers, streams, and wetlands the Clean Water Act regulates

Area of Science:

• Environmental Law
• Remote Sensing
• Machine Learning

Background:

• The Clean Water Act (CWA) regulates water resources in the United States.
• Supreme Court rulings and White House regulations have altered the scope of CWA protections.
• Assessing the impact of these regulatory changes on water resource jurisdiction is crucial.

Purpose of the Study:

• To determine the extent of water resources protected by the Clean Water Act under different regulatory regimes.
• To quantify the impact of the 2006 Supreme Court ruling and the 2020 White House rule on CWA jurisdiction.
• To develop a framework for assessing regulatory changes using machine learning.

Main Methods:

• A deep learning model was trained using aerial imagery and geophysical data.
• The model predicted 150,000 jurisdictional determinations from the Army Corps of Engineers.
• The study compared CWA protections under the 2006 Supreme Court ruling and the 2020 White House rule.

Main Results:

• The 2006 Supreme Court ruling protected two-thirds of US streams and over half of wetlands.
• The 2020 White House rule reduced protection to less than half of streams and a fourth of wetlands.
• This implies deregulation of 690,000 stream miles, 35 million wetland acres, and 30% of drinking water sources.

Conclusions:

• Regulatory changes significantly impact the scope of Clean Water Act protections.
• Machine learning offers a powerful tool for analyzing and implementing environmental regulations.
• The developed framework can aid in policy design and regulatory permitting for water resources.