A breakthrough in artificial intelligence has allowed researchers to identify nearly 1,300 previously hidden oil wells in Texas, according to a report from Lawrence Berkeley National Laboratory. The technique, which uses computer vision and machine learning, has the potential to greatly aid the state’s oil and gas regulators in finding and addressing long-abandoned wells.
Texas plans to spend $334 million by 2027 to plug these uncapped wells, which can lead to environmental harm and costly cleanup efforts. However, the exact number of such wells is unknown, with estimates ranging from 1.2 million nationwide.
Researchers used a combination of historical topographic maps, satellite imagery, and field surveys to confirm their findings, successfully flagging over 40 previously undocumented wells. The accuracy of these identifications averages about 10 meters, making this technique “stunningly accurate.”
The development aims to empower regulators and local leaders with more-informed decisions regarding wells that pose health, safety, and environmental risks. As the oil industry shifts towards carbon dioxide injection, finding uncapped wells becomes increasingly important.
With millions of dollars already allocated for well plugging in 2026 and 2027, authorities are seeking additional funds to address the growing number of leaking wells. This technology has the potential to significantly reduce costs associated with cleanup efforts and mitigate environmental damage caused by these wells.
The project is expected to expand nationwide, using federal funding from the U.S. Department of Energy. Experts warn that ignoring these hidden wells can lead to groundwater pollution and other hazards, making this breakthrough timely and significant.
Source: https://www.houstonchronicle.com/news/investigations/article/orphan-old-oil-wells-undocumented-ai-texas-report-19954499.php