How would you track the different stages of oil production in the Permian Basin?
The Permian’s spectacular growth has upended the global oil market. And the oil market is hungry for every bit of information about it. Yet significant knowledge gaps exist.
Keeping tabs isn’t easy across 75,000 square miles covering West Texas and New Mexico with 493 active rigs, according to the latest count.
Another challenge is delays in public records. As outlined in Part One of this blog series, state regulators require oil companies to report milestones, but don’t have strict timelines.
Now take a second and imagine what you could observe from space.
Converting about one acre of scrubland into an oil-producing well follows a similar pattern. Each step involves certain pieces of equipment and changes to the physical landscape. Every site must be cleared of brush and prepared for the dozens upon dozens of trucks and heavy machinery arriving around-the-clock for months on end.
Not convinced? Look for yourself. Open Google Earth and find Midland, Texas — the heart of the Permian. The whole area surrounding Midland is a tidy patchwork of dots connected by rows and columns.
The closer you look those features become recognizable as well sites and dirt roads, like the image below. With quality satellite imagery, it’s not difficult to locate a drilling rig or type of truck parked somewhere.
You can clearly see what is on the ground. You can also make an educated guess about the stage of development (Check out these videos for a refresher on rig setup, drilling and fracking + well completion).
Yet there are limits. First, this is a static picture or snapshot. You will need to revisit the same site on a regular basis, which costs money.
Second, the area of coverage is small. You would have to repeat the same exercise hundreds of times to cover the Permian Basin alone. This isn’t a feasible method to gain an understanding of broad trends.
Like a regular camera, however, you can widen the field of view. Details on the ground are lost, but the tradeoff is acceptable if you can still follow what’s going on.
As an example, we picked a spot (5km x 5 km) in the Permian Basin where we knew a pair of wells was developed and collected satellite imagery from there between January 2017 and August 2018.
You can see the land begin untouched but then roads, well pads and water storage tanks appear. The well pads remain square-shaped, but their interiors change because of the coming and going of different equipment and activity.
We selected about a dozen days to illustrate the changes (Video 1). It may take a minute for your eyes to adjust to the lower resolution.
Video 1 Source: Copernicus Sentinel data 2017/18, processed by European Space Agency
The video above is essentially what a naked eye would observe. But it can be supplemented with infrared (i.e. invisible) images because the Sentinel-2 satellite carries a multispectral instrument.
One type of infrared image emphasizes vegetation (Video 2), while the other senses water (Video 3).
Video 2 Source: Copernicus Sentinel data 2017/18, processed by European Space Agency
Video 3
Source: Copernicus Sentinel data 2017/18, processed by European Space Agency
Collectively, this provides a panorama allowing someone with the right training the ability to draw some conclusions…if the weather cooperates.
One thing you’ll notice in all three videos are days when clouds obscure the view. An optical camera simply doesn’t work in those conditions.
What does work is synthetic aperture radar (SAR), which sends microwave pulses to earth and records the signal. Utilizing SAR ensures reliability of coverage in any weather, day or night.
The “belt-and-suspenders” approach makes a lot of sense here. Include lots of tools and data sources. And then develop machine-learning algorithms to tie it all together.
“The most important thing is understanding the physics,” said Corey Miller, senior imaging scientist at Ursa. “What is the sensor measuring and how does it show up in the data.”
Miller worked on remote sensing projects at the US Naval Research Laboratory prior to joining Ursa.
“I’ve worked in multiple domains – acoustic, optics, and now radar – but the problem is the same,” he said. “You’re trying to extract information from the signals themselves.”
