By Kayla Voigt
The first thing fourth-generation farmer Ori Ben Ner looks at each morning? The weather.
While his grandfather and great-grandfather depended on their gut instincts or consulted a farmer’s almanac, Ben Ner uses a series of detailed sensor outputs and predictive analytics that give up-to-the-minute readings on weather, soil health, and moisture.
With some areas of the country averaging only 1.2″ of rain per year, Israeli date farmers like Ben Ner carefully ration their water usage. But even in places that see plenty of rainfall, water matters to the bottom line: Farmers spend $2.2 million on irrigation each year and stand to lose up to 20 percent of their yields from weather events like early-season hail or midsummer drought.
In the U.S. alone, 80-90 percent of all water usage goes toward agriculture. “California exports most of its water inside the almond,” says Ben Ner. “The best way to save water is to irrigate precisely, based on the exact needs of the specific crop you’re growing.”
“California exports most of its water inside the almond. The best way to save water is to irrigate precisely, based on the exact needs of the specific crop you’re growing.”
–Ben Nur, date farmer, Israel
This kind of water usage isn’t sustainable. As climate change threatens agricultural systems around the world, artificial intelligence (AI)-driven technology hopes to help farmers increase yields while decreasing the amount of water they use—adding up to a smarter, more sustainable operation.
A More Sustainable Irrigation Regime
When he’s not tending his farm, Ben Ner runs SupPlant, an AI-fueled irrigation system designed to help farmers optimize their water usage and conserve more water overall.
SupPlant places five sensors around the plant: on the tree, in deep soil, in shallow soil, and on the trunk, leaf, and fruit. These sensors measure stress on every aspect of the plant, from soil nutrition to water usage and plant genetics. The sensors then digest all of that information every 10 minutes, pairing new data with historic weather and soil data from similar crops and climates. This delivers a set of recommendations on a plant-by-plant basis that feeds into other pieces of agricultural technology like irrigation systems, delivering the exact amount of water a plant needs each day.
Take, for example, the date. Israel exported 138,400 tons in 2019 as its primary cash crop. But the fruit demands more water than almost any other tree on the planet, with each unit consuming approximately 20-25 million liters of water each year. “In most cases, the limiting factor isn’t land or labor but water,” says Ben Ner. “You need precise watering to make sure the plant is getting what it needs. Last year, we cut one grower’s water usage by 40 percent with a more precision-based irrigation regime through SupPlant, which allowed him to plant 50 percent more trees and yield 32 percent more dates.”
“Last year, we cut one grower’s water usage by 40 percent with a more precision-based irrigation regime through SupPlant, which allowed him to plant 50 percent more trees and yield 32 percent more dates.”
–Nur
Sustainable, high-yield agriculture is exactly why researchers like Dr. Steven Mirsky from the USDA Agricultural Research Service spend hours pouring over data from popular American-grown crops like soybeans, corn, and alfalfa. “Farmers don’t want to know what’s going on down the street. They want to know how to manage that specific field,” says Dr. Mirsky. “AI increases their ability to achieve that site-specific management, to quantify spatial patterns and dynamics, and improve models for the future.”
Dr. Mirsky is working to create a similar plant recognition system with cameras that train AI models to understand indicators of stress—water, soil, or genetics—that feeds back into a larger recommendation model for farmers to use in their practice, leading to higher yields with less water.
These cameras visualize what stress looks like on an individual plant level. Plants naturally contract each night as they accumulate water and open each day while evaporating that water into the air, as part of their lifecycle. “We’re asking, ‘When does that leaf curl?'” says Dr. Mirsky. “We already know the indicators of water stress in plants like these, and now we can put a time stamp on that and train models to trigger different actions based on what they’re seeing in the plant itself.”
Historically, scientists gathered agricultural data from observing one or two farms over a long period of time in strictly controlled environments. But with AI, Dr. Mirksy’s research can take into account the true variability in farming styles, practices, and weather patterns across the country. “We’re developing a lot of the cyber infrastructure and data acquisition systems and protocols to work across large coordinated networks of scientists,” he says. “We have a national network of 250 farms, and now we can start using AI and machine learning to find the relationships across multiple factors, understand those interactions, and make recommendations from it.”
Precision-Based Agriculture at Scale
Sensor- or camera-based technology has one thing in common: It generates massive amounts of data. “What AI allows us to do is extrapolate a lot of information to see not just individual plants but how they’re performing over massive landscapes,” says Dr. Mirsky. “AI is transforming our ability to understand how to manage farms more efficiently.”
“What AI allows us to do is extrapolate a lot of information to see not just individual plants but how they’re performing over massive landscapes. AI is transforming our ability to understand how to manage farms more efficiently.”
–Dr. Steve Mirsky, USDA Agricultural Research Service
What makes AI so powerful is the connection to other pieces of technology in a farmer’s operations, like their irrigation system. “AI automates that process so that the correct amount of irrigation is triggered based on the readings from the sensors, and that can change by the hour,” says Ben Ner. “We’re using technology to replace what used to be the basic intuitions of the farmer.”
Rather than worry over specific readings or anxiously look at the sky and hope for rain, these models allow farmers to focus on quality and yield instead of day-to-day watering schedules. AI enables a win-win scenario for farmers by boosting output through tailored irrigation systems that use less water.
“It’s all about giving farmers the most information we can about different management practices, and using AI to inform that so we can make better recommendations,” says Dr. Mirsky.
Climate Change Is a Daily Experience
While farmers can’t control the weather, they can control how they respond to it. That’s because, for them, climate change is a daily, lived experience. “The daily practices of farmers have some correlation to the weather,” says Ben Ner. “It’s never this big phenomenon of climate change. It’s the minor changes in the weather that affect farmers on a daily basis.”
Sensor- and camera-readings can determine real-time decisions, but it’s historical weather data that gives farmers a better sense of planning and timing for whatever nature throws their way. “The U.S. collects massive amounts of weather data,” says Dr. Mirsky. “If I’m trying to understand processes like nutrient cycling or crop productivity, we need that real-time weather data. Now growers can say, ‘Okay, I want to do X, what’s the probability that will hurt my yields, based on historical weather dynamics?'”
No two years are exactly the same, and no two fields are exactly the same. But the more data collected—soil moisture, nutrition level, plant stress, growth, and weather—the more prepared farmers will be to respond to everyday weather changes and optimize their operations.
“We want to keep that at minimum stress and maximum growth,” says Ben Ner. “When you look at the big picture while maintaining that level of precision, you will see greater yields. We’re consistently able to achieve a 10 percent increase, and sometimes in that first year we see up to 45 percent increase just by being more precise, which is dramatic.”
More data—alongside AI and machine learning to make sense of it—makes it easier for farmers not only to stay in business as the world changes around them but to implement more sustainable operations in the face of climate change. “With AI we’re only limited by our creativity and how we think of the applications, and it’s totally transformative,” says Dr. Mirsky. “I’ve never been more hopeful and inspired by what’s possible.”