Sustainable agriculture has an unlikely ally: satellites

The race to remove CO₂ from our atmosphere on. In an effort to reduce carbon on a significant scale, people are looking to the ground. The top meter of the world’s soil holds more than three times the carbon currently in our atmosphere, and if we manage our land better, it could absorb even more.

This is good news for farmers. Companies and individuals desperate to offset their emissions by purchasing carbon credits are willing to pay farmers to use sustainable farming practices and sequester carbon in their fields. The problem? The process of verifying that a field is absorbing the extra carbon isn’t easy: physical samples must be regularly collected throughout the land and sent to a lab for processing.

Sign in perennial, a Boulder, Colorado-based startup that says it has the answer. While studying at Brown University, Chief Innovation Officer David Schurman met CEO Jack Roswell and President Alexei “Alex” Zhuk, passionate engineers from family farms in Michigan and Ukraine, respectively. When they got to Brown, they were surprised to find that “agriculture as a whole had basically been forgotten” by technologists, Zhuk says. Today, their ambition is to create “the infrastructure that underpins the full vertical of the soil carbon market,” Roswell says. “No technology is going to solve a problem unless it solves the problem at scale and in a cost-effective way,” Roswell says. “We actively monitor every area of ​​carbon removal and net emissions in the U.S. and beyond.”

Jim Kellner, professor at Brown University and Perennial’s chief scientific officer, explains that the company’s technology relies on multispectral satellite imagery. This means measuring the light reflected from Earth in narrow bands across a wide range of the electromagnetic spectrum, capturing information invisible to the human eye. Kellner says that analyzing the spectrum of reflected light can accurately identify carbon in soil, even using satellite imagery with a spatial resolution of only 10 meters. By comparing the amount of light reflected at different wavelengths, “you can learn to identify materials, even without a picture,” he says.

Satellite images are fed into a machine learning algorithm, along with environmental data about the area in question, such as elevation and climate, to produce a measurement of soil carbon. To accurately train the algorithm, the team collected thousands of soil samples by digging holes in fields across the US to calibrate their models for different climates and crop types. By training their model on these representative physical measurements, the team enabled the algorithm to remotely quantify carbon in the ground. The company sees this as an important step towards unlocking the market for soil carbon. “If you’re solving the problem of quantifying carbon, but it depends on sending someone into a field with a stake or a shovel, you’re not going to go global,” says Zhuk.

This is all very well, but do farmers really want to switch to sustainable farming practices and change the way they grow food? Zhuk believes that the answer is positive. In the context of severe soil erosion around the world and rising farm chemical prices, he hopes Perennial will give farmers the financial incentive they need to abandon environmentally damaging practices and restore their land. “Our approach provides standard measurements anywhere in the world—a farmer in Ethiopia who puts a ton of carbon into the soil will be recognized and paid the same as a farmer in Iowa, overcoming borders and conflicting verification standards,” he says.

Right now, the company is working on training its algorithms in new countries and continents, as well as new land types like pastures and pastures in addition to fields. Zhuka’s goal? “To move agriculture from an industry that feeds us to an industry that contributes greatly to offsetting our emissions and climate change.”