Photo credit: www.sciencedaily.com
Addressing America’s Orphaned Oil and Gas Wells
Across the United States, remnants of nearly 170 years of commercial drilling exist in the form of hundreds of thousands of abandoned oil and gas wells, often referred to as undocumented orphaned wells (UOWs). These wells are not recorded in official documentation and lack known operators, rendering them hazardous due to potential leaks of oil and harmful chemicals.
Improperly sealed wells pose a threat to nearby water sources, as they can leak contaminants such as benzene and hydrogen sulfide into the air. The environmental implications extend to climate change as well, with these wells being sources of methane emissions—an impactful greenhouse gas that is approximately 28 times more potent than carbon dioxide over a century.
In their quest to locate these orphaned wells and assess methane emissions, researchers are harnessing advanced technologies, including drones, laser imaging, and an array of sensors. Given the vastness of the contiguous United States, researchers are employing a combination of modern artificial intelligence (AI) and historical topographic maps to identify potential locations where undocumented wells may be found.
Fabio Ciulla, a postdoctoral fellow at the Lawrence Berkeley National Laboratory, highlights the importance of integrating both contemporary technology and historical data. “AI can enhance our understanding of the past by extracting information from historical data at a scale that was previously unattainable,” he noted, emphasizing the utility of AI in conjunction with older data sources.
Since 2011, the U.S. Geological Survey has made available a collection of 190,000 scans of topographic maps, dating from 1884 to 2006. These maps are geotagged, allowing precise spatial reference for each image’s data.
Ciulla meticulously compiled quadrangle maps, which use consistent symbols for oil and gas wells, specifically hollow black circles, from the mid-20th century. “Identifying these circles is relatively straightforward,” Ciulla stated, but extracting this information manually from thousands of maps is not feasible. This is where AI plays a crucial role.
To enable the AI system to effectively identify the relevant symbols amidst diverse visual data, the Berkeley Lab research team created a digital training set by manually marking wells on around 100 maps from California. Once the AI was trained to recognize the correct symbols, it became capable of scanning other USGS maps with similar symbols. The geographic referencing allowed the software to cross-check coordinates for undocumented wells against known locations.
In their analysis, the researchers focused on attributes of well symbols that were situated over 100 meters away from any known wells to mitigate potential mapping errors. Additionally, they developed a tool that allows researchers to quickly verify findings made by the algorithm, ensuring accuracy in symbol interpretation.
The result of their AI-driven analysis across four counties, noted for their historical oil production—Los Angeles and Kern in California, and Osage and Oklahoma in Oklahoma—yielded 1,301 potential undocumented wells. To date, the team has verified 29 of these wells using satellite data and an additional 15 through field surveys, with ongoing investigations needed for further validation.
From the Map to the Field
Initial verification of suspected undocumented wells relies on remote assessments. Researchers leverage satellite imagery and historical aerial photography to identify possible indicators like oil derricks and pump jacks. In cases where surface structures are not visible, researchers engage in ground surveys to physically locate the wells.
At designated locations, researchers search for surface indicators and, if absent, employ magnetometers to detect underground metal casing from the wells, which disrupts the magnetic field. Upon concluding a survey, researchers log their findings, documenting coordinates and checking for methane leaks if a well is confirmed.
The Berkeley Lab team observed that the average actual location of verified UOWs was within 10 meters of the predicted locations from the AI and mapping data. They assert that this AI methodology represents a groundbreaking step for identifying UOWs at a county scale, with the potential for broader application across different regions.
The AI initiative is part of a larger collaborative effort known as the Consortium Advancing Technology for Assessment of Lost Oil & Gas Wells (CATALOG), which includes contributions from multiple national laboratories and aims to address this complex issue comprehensively. The Interstate Oil and Gas Compact Commission estimates that there may be anywhere from 310,000 to 800,000 undocumented orphaned wells across the United States.
Challenges regarding regulation and well abandonment practices contribute to the current situation, as older drilling regulations often permitted wells to remain open or inadequately plugged. Properly identifying these wells is crucial for ensuring they are effectively sealed, which involves filling them with cement to prevent toxic emissions from escaping into the environment.
Through CATALOG, the focus is on enhancing methods for locating, assessing, and prioritizing wells for mitigation. The initiative aims to develop cost-effective technological solutions that can be broadly applied, ensuring all regions have access to effective monitoring and plugging strategies.
With a unique testing ground in the Osage Nation, the project benefits from local partnerships that provide insights into equipment usability and data accuracy. Craig Walker, director of Osage Nation Natural Resources, describes the collaboration as invaluable, noting that AI applications have substantially improved the discovery and management of undocumented wells in the area.
Sebastien Biraud, a leading investigator within CATALOG, is looking at affordable sensor technology to measure methane efficiently. He advocates for a practical approach to assess leaks, emphasizing the need for quick evaluations of methane emissions in various scenarios, including prior to and after the sealing of wells.
From the Field to the Sky
In addition to ground surveys, CATALOG researchers aim to enhance their detection capabilities through drone technology. Equipped with various sensors, drones can cover significant areas that would otherwise be logistically challenging to navigate on foot.
Current drone operations involve multiple types of sensors to improve detection accuracy. For instance, one drone carries a methane sensor that analyzes air quality as it flies, while another employs hyperspectral imaging to detect methane plumes.
The consortium is also exploring aerial methodologies such as LIDAR for ground imaging, thermal cameras for hidden leaks, and even smartphone applications leveraging built-in sensors. Ciulla points out the value of this multifaceted approach, likening it to layering different data streams to create a comprehensive understanding of the situation.
As CATALOG’s endeavors continue, the commitment to addressing the environmental impacts of undocumented orphaned wells becomes increasingly critical. Biraud concludes, “It is essential to find effective solutions that minimize our emissions while collaborating with local stakeholders to achieve meaningful environmental impacts.”
The AI mapping initiative utilized resources from the Department of Energy’s National Energy Research Scientific Computing Center, with funding from the U.S. Department of Energy’s Office of Fossil Energy and Carbon Management, specifically targeting the urgent issue of undocumented orphaned wells.
Source
www.sciencedaily.com