America’s Hidden Methane Crisis: The Threat of Abandoned Oil Wells

Researchers are identifying hundreds of thousands of undocumented, orphaned oil and gas wells across the United States by training artificial intelligence to scan historical topographic maps. According to the U.S. Geological Survey (USGS) and the Berkeley Lab, these abandoned sites—estimated to number between 310,000 and 800,000 by the Interstate Oil and Gas Compact Commission—pose significant environmental risks, including methane leaks and groundwater contamination. By digitizing nearly 190,000 maps and using machine learning to detect standardized well symbols, scientists have already located 1,301 potential sites in four counties across California and Oklahoma.

How does AI find forgotten oil wells?

The detection process relies on teaching computer models to recognize specific visual patterns from historical records. Between 1947 and 1992, USGS quadrangle maps used a hollow black circle to mark drilling sites. Berkeley Lab researchers built a training dataset using roughly 100 California maps, teaching the model to ignore visual “noise” like cul-de-sacs or the letter “o.” Once the AI identifies a symbol, it cross-references the location against modern databases. Any symbol found more than 100 meters from a known, documented well is flagged as a potential orphan. Fabio Ciulla, a lead author of the study published in Environmental Science & Technology, notes that this approach extracts information from historical data at a scale previously impossible for human analysts.

Why are these abandoned wells an environmental hazard?

Unplugged wells act as conduits for toxic substances to enter the ecosystem. According to the Berkeley Lab, these boreholes can leak benzene and hydrogen sulfide into the air and discharge brine or chemicals into local groundwater. The most significant climate risk is methane, a greenhouse gas roughly 28 times more potent than carbon dioxide over a century. Because these wells lack financially solvent operators, they remain “lost” and unmonitored. Unlike modern sites that require strict closure protocols, early drilling regulations were inconsistent, often leaving these wells open or sealed with makeshift, ineffective plugs.

How do ground teams verify the AI results?

Verification involves a combination of high-tech sensors and traditional field work to confirm the AI’s predictions. Researchers use magnetometers—devices that detect the magnetic field disturbances caused by buried metal well casings—to survey the ground in grid or spiral patterns. In testing, the Berkeley Lab team found that verified wells were located an average of just 10 meters from the AI-predicted coordinates. To scale this effort, the Consortium Advancing Technology for Assessment of Lost Oil and Gas Wells (CATALOG) is developing a multi-layered toolkit. This includes drones equipped with methane sensors that calculate gas concentrations based on wind speed and direction, offering a more efficient way to scan large areas like the 1.5 million-acre Osage Nation.

What is the future of well remediation?

The goal is to turn invisible environmental hazards into manageable, recorded sites. Once a well is located, the standard remediation process involves filling the borehole with cement to permanently block leaks. The current effort, led by a group of five national laboratories, focuses on making the detection toolkit affordable for state and local agencies. New regulations now mandate the measurement of methane emissions both before and after the plugging process to quantify the reduction in greenhouse gas impact. As the AI models improve, researchers expect the number of identified, pluggable wells to grow, creating a clearer path for national environmental restoration.

Explore more: Have questions about how environmental technology is changing the energy landscape? Share your thoughts in the comments below or subscribe to our weekly newsletter for the latest updates on sustainable infrastructure.