Lars Dyrud is the Chief Executive Officer of EarthOptics, and he wants to make soil healthier.
Soil carbon sequestration is an essential piece of the climate-change mitigation puzzle. Nearly 45 percent of global soil is used for agricultural purposes, and soil can store about 2.3 times the amount of carbon compared to the carbon in atmospheric CO2 and 3.5 times more than in living plants.
According to Global X, their Charting Disruption 2022 report found that by 2050, cumulative investment across clean technologies is expected to reach between $94 and $131 T, depending on the decarbonization scenario.
EarthOptics’ new Soil Carbon Project labeling initiative is designed to help growers and the food industry quantitatively demonstrate to consumers that the climate-smart products they purchased contributed to the world’s carbon neutrality goals.
“Consumers will know that every time they buy food in the grocery store with the carbon weight attached, they will know that the labeled amount of carbon was added to the soil the year that food was grown, where it was grown,” added Dyrud.
The critical value is creating a long-term incentive structure connecting consumer choices to grower and farming activities that result in more carbon in the soil each year,” adds Dyrud.
Using AI and ML
To make this work, EarthOptics is using machine learning (ML) and teaching their artificial intelligence (AI) to read every attribute in the soil from Ground Penetrating Radar and Electromagnetic Induction scanning of the attributes in the soil.
Dyrud says their approach is focused on the Carbon. “Soil Carbon is only a small percentage of soil mass, and small changes are hard to measure. A one-ton change in soil carbon (standard carbon credit quantity) over an acre of soil that weighs 2,000 tons is difficult to measure.”
“The attributes surrounding compaction and low carbon are highly inter-related, so as EarthOptics rolls along the fields scanning and identifying compaction, knowing the conductivity of Carbon and taking soil samples for validation with machine learning, we are teaching the AI to read the attributes in the soil at an accuracy rate,” added Dyrud.
Dyrud says sample and pixel rate are currently measuring at an accuracy of less than one sample per acre, with the standard rate at one sample per five acres. “The current soil sampling method is labor-intensive fieldwork with a level of inaccuracy because of the competency in sampling, and it hasn’t changed in 80 years,” said Dyrud.
“Traditionally, you’d take a sample, box it up, send it to the lab, and the lab runs a test. But by using AI and ML, the accuracy is simplified to driving a pattern over the field, viewed from GPR data, with scanned information uploaded to the AI and then published to the cloud and available for the farmer wants to view.”
Dyrud says the AI gets better with each field uploaded to the cloud and ML helps untangle the complexity by interpreting sensor data from the field to only isolate the influencing factor of interest as long as there are enough soil samples to train the ML model.
Climate-smart packaging awareness
“Consumer packaging companies who partner with EarthOptics will be able to accurately measure and display on qualifying climate-smart product packaging the amount of carbon removed from the atmosphere and stored in the soil where the agricultural ingredients were harvested,” said Dyrud.
Dyrud says the soil labeling project is designed to educate consumers and people in the food and food production industry on their role in removing carbon from the atmosphere.
“For example, all plants are built from atmospheric carbon. Less than half a dollar more for a box of “carbon cornflakes” could generate hundreds of dollars per acre for food producers and a significant incentive to drive both producer practice and innovations in how to get more carbon in the soil,” said Dyrud. “All of this can be driven by consumer choice.”
EarthOptics has raised $10.3 M, led by Bayer’s venture arm to scale adoption of its AI/ML soil carbon monitoring tools across farmlands.