NC State Uni and Lenovo adapting to climate change with artificial intelligence

Researchers are processing large amounts of geospatial data using deep learning algorithms to help farmers make real-time decisions about water and energy usage.
Written by Aimee Chanthadavong, Contributor

Researchers at the North Carolina State University Center for Geospatial Analytics (CGA) are using artificial intelligence (AI) and machine learning (ML) to help farmers better adapt their crops to changing climates.

Speaking to ZDNet, CGA associate director Ranga Raju Vatsavai said his team of researchers has been working in partnership with Lenovo for the last two years to develop AI and ML solutions to help farmers preemptively identify ways to best optimise water and energy -- and ultimately address the threats to food insecurity.

"Our area of research is to extract actionable knowledge from the datasets. Food, energy, and water are a good application because the population is going to reach 10 billion by 2050. Right now, we are utilising 70% of fresh water for agriculture," he said.

"The challenge is by 2050, we're going to need almost twice the food production we have today. What that means is that we need 60% more food, 50% more energy, and there will also be a 10% increase in water usage.

"Of course, we can grow more crops, but we can't grow land. The question is how can we use technology efficiently to increase crop productivity."

See also: The Internet of Wild Things: Technology and the battle against biodiversity loss and climate change (TechRepublic cover story) | Download the free PDF version (TechRepublic)

Lenovo workstations, such as the ThinkStation P920 equipped with high end GPUs together with Lenovo's LiCo AI framework, along with the ThinkStation P330 Tiny are being used by CGA to support deep learning algorithms that process large amounts of geospatial data collected from sensors in the field.

Vatsavai said the data sets are allowing CGA to provide real-time updates to farmers about how they can improve water and energy conservation. 

"What we're trying to do is predict ... when [farmers] can schedule their irrigation, so it only happens when they need it. This is compared to standard irrigation system that you usually program ahead of time, and it will spray water whether it's needed or not," he said.

Similarly, the same methods have been applied to the way fertiliser is used, which when unused, can flow into nearby water sources and contaminate it, Vatsavai said.

"Right now with fertilisers, you program your tractor and it moves across the field with it. It does not change, it does not recognise whether there's soil or whether the crop is there, or whether the location needs it or not," he said. 

"Last year alone more than $1 billion of fertiliser was wasted in the mid-west region in the United States … and this is where our research is focused on, to understand these challenges by using appropriate sensors, data, and machine learning."

Vatsavai added the same application can be applied to help farmers identify whether diseases are spreading across their crop or if weed is growing, which ultimately will allow them to address the problem earlier, rather than wait for the problem to escalate. 

11 September, 2019 8.40am AEST: Correction the university involved in the partnership was North Carolina State University, not the University of North Carolina. 

Related Coverage

Optimizing IoT for blueberry farms: A learning experience for Intel

ZDNet's Stephanie Condon tells Karen Roby how Intel improved its Connected Logistics Platform after conducting a pilot project with blueberry farmers.

SwarmFarm scales up with AU$250,000 QLD government grant

The Queensland agtech startup plans to open a new on-farm office facility and hire 15 additional staff.

Farmbot delivering remote watering solutions to Aussie farmers

It uses algorithms to detect abnormal behaviour in water levels to deliver near real-time reports to farmers.

Farmbot delivering remote watering solutions to Aussie farmers

Grants of up to AU$30,000 to cover two-thirds of the cost of IoT equipment for suitable farmers.

Editorial standards