Editor’s note: Today's article comes from David Schurman, Co-founder & Chief Product Officer, Perennial. David shares how Perennial has been able to increase the scope and precision of its soil carbon level maps with its migration to Google Earth Engine—enabling Perennial to double revenues while helping customers fight climate change through regenerative agriculture.
Four hundred parts per million.
Climate scientists had been warning us for years: Breaching that threshold of atmospheric carbon meant the planet was on the path toward overheating. And I remember exactly where I was when it happened.
As an intern at the National Oceanic and Atmospheric Administration, I was analyzing air samples in an atmosphere lab in Colorado when the news came through from the Mauna Loa Observatory, a sister institution in Hawaii where the 400 ppm carbon level was first measured. It was then that I knew I wanted to help find a way to heal the planet.
But the more research I did, the more frustrated I became. Human beings generate 50 gigatons of carbon emissions every year. I was looking for a climate remediation solution that was both fast and scalable. But the geoengineering techniques (we called them “planet hacks”) I was studying—from light-reflecting aerosols to carbon-sucking machines—either came with dangerous side effects, like drying up the monsoons on which billions rely for water, or would take too long to start making a difference.
A solution to climate change? You’re standing on it.
By this time I was working at NASA’s Jet Propulsion Laboratory, where we were analyzing Martian rocks and soil for signs that life might have existed there in the distant past. Peering at rocks and dirt on a planet 140 million miles away led me to the idea for how to preserve life right here on Earth: soil. A solution was literally right under my feet.
Elements of digital soil mapping.
Industrialized agriculture has released tens of billions of metric tons of carbon from the soil, leaving unfarmable dirt closer to what you might see at a construction site in its wake. But because soil is scalable, readily available, and behaves predictably, it has the potential to repair the damage 100 years of industrialized farming—and other types of carbon-belching industrialization—has done to our planet.
Regenerative—or “sustainable”—agriculture has the power to turn dirt back into soil. I co-founded Perennial with longtime colleague and friend Jack Roswell to do just that.
Locate, measure, monitor, report, and verify
For regenerative agricultural practices to take hold, you have to be able to find the most promising available arable land, measure carbon and greenhouse gas levels, monitor the soil over time, and report and verify potential benefits to the climate.
Perennial uses Google Earth Engine to create maps of soil health, changes to carbon content, and other important soil ecosystem metrics. Earth Engine’s scalability, flexibility, reliability, and cost-effectiveness are enabling us to provide this service anywhere on the planet.
Perennial has integrated over 350,000 soil samples into our digital soil mapping models. But the process is about more than simply mapping measurements we take. We combine hundreds of data points from dozens of other sources—30 years’ worth—using our own algorithms to develop biophysical indicators from satellite data that predict and describe soil conditions. Through this rigorous approach, we can perform audit-ready measurements of soil carbon and ecosystem health with a fraction of the soil samples and farm information that would have otherwise been required.
Soil sampling in the field to accurately capture measurements.
In this way, Perennial’s datasets are enabling food, fiber, and agriculture companies to mitigate climate change by incentivizing farmers to adopt regenerative agriculture practices. And by measuring, monitoring, reporting, and verifying the climate impact of regenerative agriculture, we’re enabling those companies to provide such incentives with proof of their impact.
Ultimately, by building a foundational model for soils anywhere on the globe, our algorithms—the product of AI- and ML-based modeling—hold the key to making the Earth itself a giant, orb-shaped sink for atmospheric carbon.
Displaying carbon use efficiency.
But that kind of AI modeling requires vast computing resources. That’s why we decided to migrate all of Perennial’s remote sensing data, orchestration pipelines, and data derived via AI modeling to Google Cloud Platform, and all the data points on lab-analyzed soil samples, which we use to train our AI models, to BigQuery.
Fostering climate resilience with Google Cloud
The decision to go with Google Cloud was an easy one. Having worked with all three major cloud platforms, I knew that Google Cloud was the only one whose individual components feel organically integrated. That tight integration makes them easier to use, while their reliability means our engineers can focus on product development, not troubleshooting. Its scalability meant we’d be able to map data anywhere on Earth. And Google Cloud fit seamlessly into our existing workflows, which meant we’d be able to get up and running quickly.
Google is also the founder of much of the cloud-based geospatial ecosystem (who hasn’t used Google Maps?) and the team at Google Earth Engine shares our belief that artificial intelligence can be the technology that underlies climate resilience. So we really wanted to become part of the Earth Engine community.
Displaying microbial soil temperature index.
Working with Google partner Woolpert Digital Innovations, we built a proof of concept. We then moved hundreds of thousands of rows of training data and all our orchestration pipelines to Google Cloud, as well as tens of thousands of lines of code and hundreds of thousands of data points to BigQuery, in just three months. By anyone’s measurement, the migration has been a success.
Less carbon, more revenue
Our AI model-iteration speed is 10 times faster than before, the volume of data we crunch in iteration is 10 times larger, and the area of the globe covered by our data is five times bigger—all while lowering our engineering costs.
But the numbers that count are what those faster speeds, larger volumes, and bigger areas translate into: the ability to verify the removal of hundreds of thousands of metric tons of atmospheric carbon. And because our data cover more turf, we can serve more clients across a larger swath of the globe. As a result, we doubled our revenue between 2023 and 2024, and we’re on track to multiply 2024’s revenue by five in 2025.
Regenerative agriculture is not only changing how the global food system works, but how agricultural ecosystems are managed. By returning carbon to the soil rather than releasing more of it into the atmosphere, sustainable farming practices can help keep the planet from overheating so that Earth will still look like Earth—and not Mars—100 and even 1,000 years from now. And Google Earth Engine, Google AI, and the entire Google Cloud Platform are enabling Perennial to help get farmers paid for putting those sustainable farming practices into action.
