From Navigation to Irrigation - AI Powered Water Management

Professor Matthew Wilson on the left and researcher Xander Cai on the right, holding the prototype sensor, which will be mounted to a pole above fence height and linked to a computer system which runs the AI technology. Amy Riach

A brand-new venture for GPS signals: moonlighting for a water management system on New Zealand farms.

Using artificial intelligence and satellite technology, a new research project designed to measure soil moisture may help to guide Kiwi farmers making informed decisions about when and where to irrigate.

Researchers at the University of Canterbury (UC), are developing a system that aims to provide real-time information about moisture levels across entire paddocks, helping farmers save on water use, and respond to increasingly variable weather conditions.

The project, led by Professor Matthew Wilson and PhD student Xander Cai, of UC's Geospatial Research Institute, is focused on addressing one of agriculture's ongoing challenges: accurately measuring how much water is available in the soil.

“Water is one of the most critical and constrained resources in agriculture,” Wilson told the Guardian.

“Right now, farmers are often making decisions based on limited or incomplete data. This project is about giving them the information they need to use water more efficiently and sustainably.”

The research team is developing the ANZ Soil Moisture Data Assimilation System (ANZ-SMDAS), which combines data from ground-based sensors, satellite signals, and advanced artificial intelligence models, to monitor soil moisture.

Unlike traditional soil monitoring methods, which rely on sensors installed at fixed points in the ground, the new system is designed to provide a more comprehensive picture of conditions across an entire field.

“Relying on ground sensors alone introduces significant limitations,” Wilson said.

“They can certainly be helpful, but they have a fixed location, so their coverage is limited, they need installation, and require disturbing of the soil to do that.”

“It’s a bit of a leap to say that one measurement at that precise location of the sensor is representative of a whole field,” he explained.

The innovative, AI driven system, uses existing signals from global navigation satellites such as GPS in a new way.

“Those signals are used for navigating, of course, that's what everyone is familiar with: when you're doing your satellite navigation in the car, and we are using those same signals but in a completely different way,” Wilson said.

The technology analyses satellite signals that bounce off the land surface. Changes in the reflected signals reveal how much moisture is present in the soil.

“What we're most interested in is the change in that signal that occurs as a result of soil moisture,” he said.

“We are able to compare the direct signals and the affected signals, and the differences between them change depending on the moisture of the soil, all based on signals that already exist.”

Professor Wilson described the approach as using “signals of opportunity”, repurposing existing satellite transmissions rather than relying on expensive dedicated satellites or radar systems.

The physical sensor itself is expected to be relatively simple, consisting of a GPS antenna mounted on a pole, connected to a solar-powered computer and cellular communication system.

Artificial intelligence running on the device will process data before transmitting it for further analysis.

“We're pulling together data from lots of different sources, including other satellite information, that would allow us to predict moisture beyond that physical location of the sensor,” Professor Wilson said.

“The sensor itself is measuring about 100 by 100 metres, and the data assimilation will allow that to be expanded far beyond that location.”

Researchers believe the system could benefit a wide range of agricultural sectors, including dairy, grazing and cropping operations.

“Better soil moisture data means better decisions,” PhD student Cai explained.

“For dairy farmers, it can help optimize pasture growth. For cropping systems, it can improve irrigation timing and reduce water waste. Ultimately, it supports both productivity and sustainability.”

Wilson told the Guardian, irrigation management is likely to be the primary use of the technology, but the information could also help guide decisions about fertiliser application and overall farm management.

“There is the ability to have quite precise irrigation in terms of the hardware for irrigating, but there isn't always the data to feed into that,” he said.

“The idea is to provide that information to those systems.”

The project comes as farmers in New Zealand and Australia face increasing pressure on water resources due to climate variability and more frequent drought conditions.

Wilson and Cai are collaborating with the University of Newcastle, and Monash University in Australia, as well as the Soil Cooperative Research Centre.

Prototype sensors are expected to be deployed on farms in New Zealand and Australia later this year, with field trials planned around September and October.

“We're looking to deploy prototypes on-farm in New Zealand and Australia this year,” Wilson said.

The three-year project is currently in its early stages, with researchers expecting initial outputs from the system within the next two to two-and-a-half years.

“We're not just developing a new technology, we're building a system that will be publicly accessible and can be used day-to-day by farmers,” said Wilson proudly.

“The goal is to deliver practical tools that make a real difference on the ground.”