Andrew Nelson’s family has farmed in the hills of eastern Washington State for five generations. Rows of wheat, peas, lentils and barley spread across their 7,500 acre farm. Yet each spring these hills have provided constant stress when attempting to forecast frost – a task that eludes typical weather forecasting services.
“Sometimes the difference between the bottom of my hill and the top of my hill can be 10 to 20 degrees different,” he said. “We can very regularly have frosts that the weather forecast will never predict.”
Pack big data into a small package: Ground sensors feed data to artificial intelligence algorithms that combine with data from the National Weather Service to predict how sensitive certain areas of its fields are to frost.
“There are so many risks in farming, and so many unknowns and uncertainties, that if I can do something a little less risky…it really helps me a lot,” Nelson said, who this year added the algorithms that prevented him from fertilizing too soon. — which can damage plants.
These futuristic farms go by many names – smart or digital farming, or precision or data-driven farming – but a growing number are using data and algorithms to inform everything from when to plant seeds to where to water, which techniques release the most carbon into the air.
According to a 2021 survey, 87% of U.S. agricultural businesses currently use AI, and the agricultural robotics market is expected to reach $30 billion by 2026.
Nelson, who is also a software engineer, combines a program to remove clouds from satellite images via AI with another program that collects data from drone images and ground sensors to create a model capable of identify weeds. This allowed him to target the herbicide only to the areas that need it and spend 38% less.
Nelson’s farm has been a proving ground for Microsoft agricultural research projects since 2017, allowing Microsoft to prove its technology can work in the real world, said Ranveer Chandra, general manager of Microsoft Research for Industry. This week, his team released an open-source algorithm suite called Project FarmVibes.
Chandra said the goal is for researchers around the world to use it to develop data-driven technology that farmers won’t need a software degree for.
“We should be able to tell what (a certain section of a farm) looks like right now, what it looked like in the past, and what it will look like in the future…not just above the surface, but also below the surface,” he said.
And the stakes are high.
“Given the current food security problem, given the climate change and water issues, we have no choice but to start adopting these technologies very quickly,” he said. . “Working with Andrew, we have shown that (this technology) is mature, but the ecosystem needs to evolve.”
There remain challenges to the mass adoption of this technology, including the digital and economic divide faced by small-scale farmers, especially in developing countries. There are more than half a billion farms smaller than 2.5 acres worldwide, Chandra said, and less than 13% use some form of digital technology.
Even in the United States, farmers have to recoup the cost of new technologies within a year, Nelson said.
Pandora’s box
Until recently, smart farming was limited to large companies or researchers, those with strong internet connections and those who can afford to experiment, said Maanak Gupta, cybersecurity expert and Tennessee professor. Tech University. Yet, as technology becomes cheaper and connectivity becomes more accessible, “adoption in smart agriculture certainly opens Pandora’s box of cyberattacks.”
The area of greatest concern is data privacy (who can see it) and collection (who can keep it and where) – especially in an industry where soil composition or seed yields are considered proprietary information.
“Suppose that before you spray water, your machine measures that the soil (moisture) is too low. So it will automatically trigger on a ground water sprinkler,” he said. “But what if it’s fake data? What if your soil already had enough of it and you don’t need any more water? It’s going to completely ruin their entire estate.
“When someone can disrupt the entire economy of a nation dependent on agriculture…attacks in a well-coordinated manner can truly qualify as agroterrorism,” he added.
It also has huge implications for the entire food supply chain, which is already shaped by insights from algorithms and big data.
“We have to be careful about which data falls into whose hands. Food security is critical infrastructure for the country,” Chandra said, adding that Project FarmVibes does not currently collect or store data.
According to Gupta and Chandra, the key to finding solutions is giving farmers control over the data collected on their farms.
“(Data-driven insights) will help farmers be more profitable. It also helps consumers to be healthier. And it makes the whole supply chain much, much more efficient (reducing) food waste…helping ensure food safety,” Chandra said. “These are all things that data-driven agriculture could illuminate.”
#Farmers #harvest #data #increase #yields