A flight crew spent 18 days flying a grid over the New Mexico Permian and logged 529 super emitters, and 18 of those sites were still venting on the day the survey ended

The aircraft climbs to survey altitude just after dawn and turns onto the first line.

Below, nothing looks wrong.

Pump jacks nod, tank batteries sit on their graded pads, and gas moves toward the pipelines heading for the coast, all of it perfectly ordinary from three thousand feet.

Aerial campaigns are expensive and episodic, but continuous satellite coverage keeps improving and aerial systems are now being validated down to 13 pounds an hour.

The camera does not see any of that.

It reads heat signatures in the infrared, and what it maps is a pattern that ground crews walking the same pads almost never encounter.

Why a few broken valves swamp an entire basin

Methane escapes from oil and gas equipment on a hard physical curve.

Most valves, tanks and pipe joints leak at rates too small to measure from the air, and there are a great many of them.

Then there are the others.

A relief valve sticks open, a compressor seal degrades, a tank hatch stays unlatched, and that one site starts releasing at hundreds of times the background rate.

The proportion is the whole point.

Fix a handful of those and you erase a disproportionate share of the damage, which is also why an inventory built from engineering estimates and spot checks misses the total by so much.

What a stuck valve looks like from three thousand feet

On the screen it is a plume.

A pale smear leaving a pad and drifting downwind, sharp at the source and dissolving into nothing a few hundred yards out.

The source is usually mundane.

Not a blowout, not a rupture, just a seal that stopped sealing on some afternoon nobody recorded, on a pad that a truck visits once a month if that.

Nothing on the ground announces it.

The equipment keeps running, the site keeps producing, and the only outward sign is a faint shimmer that the human eye does not register and that no gauge on site is watching for.

So it continues, sometimes for years on end.

The survey on the record

The campaign ran from 30 April to 17 May 2024, roughly 13 flight days inside an 18 day window.

It covered about 4,600 square miles of the New Mexico Permian, taking in 276,000 wells, 1,100 compressor stations, 175 gas processing plants and around 16,800 miles of pipeline.

The instrument had near full sensitivity above 220 pounds an hour.

That figure is not arbitrary. It is the federal threshold that defines a super emitter, and everything above it was logged.

The tally came to 1,380 plumes from 529 distinct sources, most of them leaking somewhere between 110 and 1,100 pounds an hour.

Roughly 300 of those sources turned up on more than one survey day, and 18 of them were seen every single year going back to 2019.

Comparison against satellite data over the same weeks put this group at about half of everything the region emitted.

Where the bridge fuel argument runs into trouble

Gas has been sold for two decades as the sensible step between coal and renewables.

That case rests entirely on one assumption, which is that the methane arriving at the burner tip is most of the methane that left the ground.

The persistence data settles that question.

Methane traps heat at more than 80 times the strength of carbon dioxide over a twenty year horizon, so a leak rate that reads as trivial in percentage terms is not trivial in effect.

How trivial depends on the route.

One analysis puts the break even against coal at around 1.4 percent for domestic gas, and as low as 0.2 percent once the fuel is liquefied in Texas and shipped to Asia.

Which matters most where a pipeline system is being built to move more Permian gas toward the Gulf Coast.

What a list of addresses can change

The encouraging part is that methane is a product.

Gas that escapes through a stuck valve is gas that never reaches a customer, so the business case for fixing the worst sites is unusually clean.

Detection at scale is the hard part, because super emitters do not announce themselves on a schedule.

Aerial campaigns are expensive and episodic, but continuous satellite coverage keeps improving and aerial systems are now being validated down to 13 pounds an hour.

Half of what this survey flagged had been unknown to anybody.

Not to the operator, not to the state, not to the inventory that is supposed to add all of this up at the end of the year.

Once it was handed to operators and state agencies, it got fixed, which turns a list of addresses into a list of repairs rather than another argument about estimates.

That is the useful outcome.

The 18 that never stopped had been venting through five years of inspections, permits and quarterly reports, and it took an airplane to notice.

The published paper has the persistence data. The survey summary has the mitigation results.