PINEWOOD, MINN. -- The rusty, 1950s-era drilling rig bored noisily in the sandy soil, stopping at a depth of 27 feet. As workers lifted out the spiral bit, a whiff of petroleum drifted up the hole. "That's some Bemidji crude oil," said Jared Trost, a hydrologist for the U.S. Geological Survey in charge of the rig.

Thousands of gallons of crude oil gurgle underground at this spot off a gravel road 12 miles northwest of Bemidji, Minn. It isn't a new shale oil discovery.

The crude oil is the stubborn remnant of a massive 1979 pipeline rupture. Over the past three decades, this site has become a science center unlike any other in the world. Using data collected mostly from bore holes, scientists have produced a gusher of research discoveries, including some that have influenced U.S. pollution cleanup policy.

"We are trying to take advantage of a bad situation and learn what we can from it," said Mindy Erickson, a USGS hydrologist based in Mounds View who manages the research site.

The site is slightly larger than a football field, has no buildings and consists mainly of hundreds of wells drilled for sampling and other research. The pipeline that ruptured, spilling nearly 450,000 gallons of crude oil in 1979, runs through the middle of it.

Much of the spilled oil was cleaned up in the weeks after the accident, but techniques used 35 years ago left about 100,000 gallons in the ground.

"Today, you wouldn't walk away with that much oil in the ground — that just wouldn't happen," said Scott Lounsbury, senior environmental manager for Enbridge Energy, the Calgary-based pipeline owner.

Enbridge didn't actually walk away. It helped launch a research project that's still going 31 years later.

"It is a unique site," said Bruce Bauman, research program coordinator for the American Petroleum Institute, a Washington-based trade group that supports spill research. "It is certainly the most studied site in the world."

Americans consume 18 million barrels of oil per day. It moves mostly by pipelines, which have 350 accidents a year, spilling more than 50,000 barrels of hazardous liquids. Cleaning up oil spills is expensive, and even as techniques improve, it's rarely possible to recover every drop that seeps underground.

Learning from nature

Pollution control experts once doubted that subsurface microbes could break down oil because of the lack of oxygen. But scientists working at the Bemidji site published scientific studies in the 1990s that showed otherwise.

"Those results got nationwide attention," said Barbara Bekins, a USGS research hydrologist based in Menlo Park, Calif., who coordinates research at the Bemidji site.

As a result, "natural attenuation" was recognized as a cost-effective way to remove underground pollution, especially from urban gas station tanks where digging up blocks of property isn't an option.

Scientists on the Bemidji Crude Oil Research Project also have made discoveries about the spread of pollution "plumes" underground, pioneered ways to measure natural breakdown of oil and learned about how petroleum-eating bacteria work.

Research at the site has shown that bugs rapidly eat toxic, water-soluble compounds such as toluene and benzene. The spread of such compounds in the site's water table has halted about 500 feet of the spill. "Toluene is 100 percent gone," Bekins added.

Much of the remaining oil floats atop the water table, about 2-feet thick. But it hasn't moved far, suggesting that even a major oil spill doesn't necessarily pollute an entire aquifer. That's been a worry about the massive Ogallala aquifer in Nebraska, where antipipeline activists oppose building TransCanada's proposed Keystone XL crude oil pipeline.

"The position that … one pipeline release to a sandy aquifer is going to destroy the aquifer forever, the science doesn't back that up," said Paul Meneghini, a senior environmental manager for Enbridge.

Scientists also found that bugs are not as fond of eating Bemidji's thick oil, so it could take years of natural attenuation before the last of it is gone.

Pumping out the Bemidji oil is not considered an option. In 1999, or 20 years after the spill, Enbridge made a second try at the request of state regulators. Another 30,000 gallons of oil was suctioned out through wells over four years.

Unfortunately, water came out with the oil, and had to be injected back into the ground. That spread subsurface pollution in what researchers call the "remediation plume." The good news is that bacteria are eating that plume too, according to a recent USGS study.

It's one of about 270 research papers about the Bemidji site that have appeared in journals, from such prestigious ones as Nature and Science, to the lesser-known, aptly named "Environmental Forensics."

Visit with a drill rig

Scientists rarely need to visit the site. Each year, a midsummer sampling event draws about 35 to 40 researchers. Most of the scientific work is done elsewhere, at institutions ranging from such East Coast universities as Rutgers in New Jersey and Syracuse in New York, to the University of British Columbia and the Juárez Autonomous University of Tabasco in Mexico.

Enbridge has kicked in about $600,000 in recent years to keep the research going. USGS puts in $180,000 a year, and jointly administers the site with Enbridge, Beltrami County, which owns the land, and the Minnesota Pollution Control Agency, which regulates the cleanup.

For new phases of research, Trost and two other USGS employees spent a week at the site in May. They sank two wells with the rusty drilling rig, and installed electronic moisture and vapor sensors.

Trost, who started working with the USGS as a student in 2006, is among the new generation of scientists now taking over from early spill researchers who have retired. He graduated from Augsburg College in Minneapolis and went on to earn a master's degree in water resource science from the University of Minnesota in 2010.

Trost said one of his interests is a simple method to capture and measure gases released by microbes as they degrade oil. Measuring the gas at the surface reveals how fast oil is degrading below ground.

The technology is being used by energy giant Chevron and others to monitor underground oil degradation, API's Bauman said. It also has turned out to be surprisingly useful in monitoring the success of ethanol spill cleanups. "A lot of unique sampling methods have been developed here and are starting to get noticed," Trost said.

'It blew oil like a geyser'

The rupture happened Aug. 20, 1979, on a 34-inch diameter pipeline carrying light crude oil from Canada. It is one of five Enbridge lines on a corridor across northern Minnesota carrying oil to a terminal and other pipelines in Superior, Wis.

On the morning of the rupture, the telephone rang at the home of Willis Mattison, who was then the regional director of the state Pollution Control Agency. He got to the site after daybreak. Cleanup crews already were at work.

"Everything was black and oily," recalled Mattison, who retired in 2001.

The 450,000-gallon release is the seventh largest crude oil pipeline rupture in Minnesota. "It blew oil like a geyser into the air," Mattison said. Wind splattered the oil over an area the size of a football field. Oil flowed into a wetland pond and percolated underground.

"The first thing they wanted to do was set it on fire," Mattison said during his first visit to the site in more than 30 years. "We didn't allow that. We told them to get down and start skimming and pumping."

The most visible sign of the accident today is a large area of bare, eroded soil where little grows. Scientists are still studying why, after nearly 35 years, the oil-tainted soil still repels water, keeping moisture from plant roots. One theory: Oil-eating bacteria in topsoil don't like Minnesota winters.

Nearby, a wetland that in 1979 was layered in oil now looks like any other wilderness pond. The thick oil had been vacuumed away after the rupture. Microbes that thrive in wetlands have broken down much of the rest of the pollution, researchers say.

Recent studies have looked at the genetics of Bemidji bacteria. Nicole Fahrenfeld, assistant professor of civil and environmental engineering at Rutgers University, has worked with other scientists to get DNA profiles of "known hydrocarbon degraders."

The research revealed where certain bugs are degrading the most oil. Other researchers are still studying the oil-loving bacteria, and looking for new ones.

"Everybody would like to find the new bacteria with a new ability," Fahrenfeld said.

David Shaffer • 612-673-7090 • @ShafferStrib