SAN FRANCISCO—The sky has filled with eyes, and NASA is starting to notice.
Over the past several years, venture-funded start-ups have hurled hundreds of inexpensive satellites into orbit. For-profit companies have used smartphone technology to make compact satellites that look down at Earth and monitor its every oceanic gurgle, erupting volcano, or forest conflagration. Hundreds of these satellites might gaze down at the same time; they are organized in what are called (rather poetically) “constellations.”
NASA has now taken heed of these new arrangements. Earlier this year, it asked 36 scientists to figure out whether imagery and data from three satellite companies could be put to serious scientific use. On Thursday, the San Francisco–based start-up Planet announced that it is one of the three companies participating in the pilot program.
Among NASA’s goals: Decide whether data from the three satellite companies can be used to create a dashboard of what are called “essential climate variables.” These core clues to planetary health—which include figures tracking the size of leaves, the health of Arctic permafrost, and the extent of groundwater reservoirs—could function as a kind of early-warning system for environmental upheaval.
This program also reflects a potential shift for NASA. The space agency is already preparing to send a human crew to orbit in a commercial spacecraft later this year. It might soon rely on for-profit companies when collecting scientific data, too.
The announcement is something of a coup for Planet, which operates what it describes as the largest private constellation of Earth-observing satellites ever assembled. Planet’s leaders have long described their company as a boon not just for the financial and defense industries—the usual customers for this kind of data—but for scientists and humanitarians. Now they have the NASA deal to prove it.
But it could raise more difficult questions for researchers. Science is conducted largely as a public good, and researchers can vet one another’s work by checking it against publicly available data. If that basic data is no longer publicly available, it could mean that major earth-science research relies on proprietary data.
I recently visited Planet’s headquarters in San Francisco. Housed in an unassuming former warehouse in the city’s SoMa neighborhood, its office melds the feel of the National Air and Space Museum in Washington, D.C., with the unfinished, exposed-brick-and-metal aesthetic that the writer Kyle Chayka has called “AirSpace.” One floor is an open-plan office with rows of sleek monitors and small meeting rooms. The next floor up is a shining clean room where dozens of satellites are manufactured every year.
Planet is now the dominant satellite-imagery start-up in the Bay Area. In 2015, it bought RapidEye, giving it five military-grade satellites. Two years later, it acquired its rival small-satellite manufacturer Terra Bella in a deal with Alphabet. Today roughly 120 Planet satellites float in orbit—most of them are about the size of a shoebox—allowing the company to photograph every spot on Earth at least once a day.
Planet can also boast association with a large research community. More than 100 peer-reviewed papers cite Planet data. Researchers have used Planet imagery to monitor Arctic lakes, track ships, and tally the biomass of forests.
Jamon Van Den Hoek, a professor at Oregon State University and one of the 36 scientists currently adjudicating the NASA pilot project, told me that Earth imagery now functions as a crucial data source for scientists. The pixels are the data, he told me: “These aren’t just for reference. These aren’t just for a base map. These are the data you analyze.” A certain shade of pixel might say whether a forest is thriving or clear-cut, or it might suggest that a once open meadow has been swallowed by a city.
As part of the NASA pilot program, Van Den Hoek has used Planet data to help search for and study Earth’s “missing million” people—refugees and internally displaced people who live in informal camps and settlements in the least mapped parts of the planet. The project would constitute the first systematic analysis of how these settlements change over time, he said.
Not all of this is new: Scientists have been using pixels as data for the past half century. Since 1972, satellites in the U.S. government’s Landsat program have systematically photographed every speck of land on Earth, every 16 days, without fail. Landsat, now one of the largest and most powerful tranches of earth-science data, is an invaluable scientific resource. In the 1980s, it revealed the extent and severity of Amazon deforestation; now it captures the massive changes to Earth’s surface wrought by climate change. One of the most widely cited satellite data sets, a global survey of forest loss created by the University of Maryland professor Chris Hansen, is powered by Landsat data.
But the program’s future is more uncertain—and its fate is tied to Planet’s. The next Landsat satellite, dubbed Landsat 9, is due to launch late next year. But NASA and the U.S. Geological Survey have considered using a new approach with its successor, Landsat 10. They could replace it with two satellites instead, mimicking an EU program. Or they could try replacing Landsat with a swarm of satellites, creating a publicly owned version of Planet’s constellation.
Some lawmakers have even proposed that the government rely entirely on private-sector data for Landsat 10. Van Den Hoek said that seemed unlikely, at least for now. “People who hold the purse strings may want that to happen, but no one at NASA wants that to happen,” he said. Planet, too, supports the Landsat program and doesn’t want to see it change significantly, a spokesperson told me. The company’s satellites revisit the same speck of land more often than Landsat’s do, and its cameras have a higher resolution. But its craft are unable to capture as many types of light.
Planet tries to make as much of its data available to as many researchers as possible, and some universities now have blanket licenses to much of its imagery. The company remains a commercial enterprise (albeit one that has not yet turned a profit), but Joe Mascaro, an ecologist who now directs academic partnerships for Planet, told me that the replicability of research is a “core principle we would do our best to meet.” And if Planet explores “future, larger contracts with NASA,” says Trevor Hammond, Planet’s vice president of communications, it “would go in with its eyes open” about the tension between open science and closed data.
Van Den Hoek emphasized that Landsat and Planet are good at different tasks. Landsat could capture widespread shifts to the land: urbanization, deforestation, the loss of polar ice. Planet excels at more fine-grained tasks. “You can ask questions that you could never ask before,” he said. “Huge portions of sub-Saharan Africa rely on small-scale agriculture for daily subsistence. You can’t measure that with Landsat data.”
NASA is also working with Spire Global, a Bay Area start-up that collects high-quality weather data, and Maxar, a more established player that owns the WorldView spy satellites. Peter Platzer, the chief executive of Spire, told me in an email that NASA plans to spend $100 million on small-satellite projects over the next few years.