In the third quarter of 2024, plug-in EVs made up 25.3 percent of light-duty vehicle sales in Colorado, according to a report from the Northeast States for Coordinated Air Use Management, a Boston-based clean air consortium. 

“Coloradans are purchasing electric vehicles at a higher rate than any other state because they are affordable, quiet and a great ride,” Gov. Jared Polis (D) said in a statement.   

California came in second place at 24.3 percent, followed by Washington State at 23.5 percent. Next in line was the District of Columbia at 19.4 percent, Nevada at 16.3 percent, Oregon at 15.7 percent and New Jersey at 14.4 percent.  

These top seven places have all adopted criteria for zero-emissions vehicles initiated by California, which can set stricter-than-federal emissions standards under the Clean Air Act. Other states can then follow suit.  

But the ability of California — and its fellow tailpipe rule adopters — to maintain such standards remains uncertain, as the incoming Trump administration has repeatedly vowed to revoke such permissions.  

California Gov. Gavin Newsom (D) has launched a wide-ranging campaign to “Trump-proof” his state from possible interventions, including from those related to clean vehicle policies.  

For example, Newsom recently pledged to revive the Golden State’s EV rebate program if the incoming administration decides to scrap the existing $7,500 federal tax credit. 

California’s rebate program was phased out in November 2023, after investing $1.49 billion in funding 594,000 vehicles.  

Colorado, meanwhile, still offers a $5,000 state tax credit to those who purchase or lease a new EV by the end of 2024. That credit will then decrease to $3,500 in 2025.  

Citing the clean air consortium’s data, Polis noted that the demand for electric cars continues to rise — a trend he attributed to “competitive state and federal rebates” that help people save money on these purchases.  

“EVs are helping us reach our bold climate goals and protect our clean air,” he added. 

Welcome to The Hill’s Sustainability newsletter, I’m Sharon Udasin — every week we follow the latest moves in the growing battle over sustainability in the U.S. and around the world.

Researchers have been harnessing artificial intelligence (AI) tools to relocate “lost” oil and gas wells, in hopes of permanently shutting down these dangerous artifacts. 

Hidden hazards: There are likely hundreds of thousands of oil and gas wells across the country that have no formal records or owner documentation, according to the scientists.  

• These so-called “orphan” wells are the relics of nearly 170 years of commercial drilling.
• But if they weren’t properly plugged after operations ceased, the wells could leak oil or chemicals into nearby water resources.
• They could also emit toxic benzene, hydrogen sulfide or the greenhouse gas methane into the air.  

Merging the old and new: To predict where some of these undocumented wells might be, the researchers paired modern AI tools with historical topographic maps.  

They published their findings Wednesday, in Environmental Science & Technology. 

Sifting through scans: The scientists were able to access the topographic maps through the U.S. Geological Survey, which has uploaded 190,000 scans of such resources, created between 1884 and 2006. 

• The maps were geotagged, meaning that each pixel corresponded to coordinates that could be easily accessed.
• Those issued between 1947 and 1992 also used a consistent, hollow black circle, to flag oil and gas wells. 

Educating AI: The researchers, from the Department of Energy’s Lawrence Berkeley Laboratory, were able to train AI to identify these circles.  

• AI needed to be able to do so amid a plethora of visual information and on maps with different terrains, colors and conditions.
• The scientists ended up using a digital tool to manually mark wells on nearly 100 maps in California, as part of the training regimen.  

Even AI has a learning curve: “This problem is equivalent to finding a needle in a haystack,” senior author Charuleka Varadharajan, a Berkeley Lab scientist, said in a statement. 

“We are trying to find a few unknown wells that are scattered in the midst of many more documented wells,” Varadharajan added.   

Screening and verification: But once AI was trained to find the circles and ignore false positives, such as cul-de-sacs, the algorithm could be applied to any of the maps, the scientists found.  

When they set out to verify their findings — first via remote images and then out in the field — the researchers saw that the wells were located an average of 10 meters from AI’s predicted locations.  

The past and future: Lead author Fabio Ciulla, a postdoctoral fellow at the Berkeley Lab, stressed that although AI is a “rapidly evolving technology, it should not be exclusively associated with modern data sources.”  

“The more we go into the future, the more you can also use the past,” he added.