Using military-grade satellites in the war on weeds
Getting a bird’s-eye view of cropping paddocks gives growers and their advisors a wealth of information to assist decision-making. The recent proliferation of satellites and developments in sensor capability have thrown the door open to significant advances in precision weed management.
Tim Neale, digital agronomy specialist and DataFarming managing director, says the commercial availability of super-high-resolution imagery from the latest generation of military-grade satellites, plus algorithms to distinguish between crop and weed plants, can turn any modern sprayer into a spot sprayer without any up-front capital expense.
“The images we can currently access from these satellites clearly show plants within a 30 cm by 30 cm area on the ground,” he says. “Using algorithms to detect and identify weeds in fallow or within a crop, we can generate a map that the grower can load into their spray equipment and use the boom’s nozzle or section control to spray only the areas of the paddock with weeds present.”
Tim is leading a new 4-year GRDC and SmartSat CRC funded project to harness the potential of remote sensing technologies to assist growers and agronomists. On the weeds front, the team is tackling one of the hardest problems first – detecting annual ryegrass in cereal crops. Other priorities are all fallow weeds, brassicas in winter crops, and hard-to-kill weeds like fleabane, feathertop Rhodes grass and barnyard grass in summer row crops.
“The visible spectrum for humans is three different bands of light – red, green and blue. But the latest generation satellites can see up to 350 bands of light,” he says. “It is now possible to discriminate between different plant species to separate target weeds from the crop.”
“We are collecting data for green-on-green detection of ryegrass in wheat this winter and hope to have algorithms ready to field test in 2025.”
The system is well-advanced for remote green-on-brown weed detection in fallow, having been field-tested last summer. Tim says this service will likely be available on a limited, early-release basis to growers for the coming summer fallow season at less than $10 per hectare for green-on-brown spray coverage maps. Everything is done remotely, with the spray map file emailed to the grower ready to upload or pushed directly to the machine through connections with programs like the John Deere Operations Centre.
“To use the service, growers will just need to provide their paddock boundaries and have the ability to upload a shape file to their spray gear with nozzle and or section control capability,” he says.
The spray map includes a buffer of a couple of metres around the detected weeds. Typically, herbicide is applied to around 15 to 20 per cent of the paddock, vastly reducing the volume of product used compared to blanket spraying. Decoupling the weed detection from the spraying also means that the operator knows how much product to put in the tank ahead of time.
“One of the major advantages of satellite weed detection is that growers do not need to invest in more machinery to increase their spot-spraying capability,” says Tim. “There are no cameras or sensors to maintain on-farm and no need for an operator to collect the imagery.”
With a rapidly increasing number of these high-grade satellites orbiting the Earth, imagery is available soon after capture, and in the near future, revisit times over the same location will be daily or even multiple times a day, reducing the potential impact of cloud coverage.
Tim says the available satellite images provide scalability with 5000 to 10,000 ha captured in a single pass, which reduces the cost per hectare to capture and generate the spray maps compared to using drones to collect the imagery.
The team has also demonstrated the algorithms’ capability to distinguish between a row crop like sorghum and a hard-to-kill weed like barnyard grass. Using the map, the grower could use a shielded sprayer or inter-row cultivator to target the weeds within the cropping season.
The project will also investigate the opportunity to use satellite images to monitor for changes in weed populations over time and identify small incursions of emerging or herbicide-resistant weeds. Giving growers and their advisors early warning of changes will help with effective herbicide choices for the field while targeting small areas with suitable chemistry to avoid the resistant gene spreading.
Weed detection and mapping is a powerful tool for growers to use as they implement the WeedSmart Big 6 tactics to stop weeds setting seed, and minimising the risk of herbicide resistance.
“With growers and agronomists managing increasingly large production areas, it is very easy to miss early warning signs,” says Tim. “There is also great scope to use this technology for monitoring and managing crop nutrition, pests and diseases.”
GRDC codes: DFL2312-001RTX, DFL2304-001RTX, DFL2304-002RTX
Grain Automate is a Grains Research and Development Corporation (GRDC) initiative on behalf of Australian grain growers aimed at accelerating the adoption of machine automation, autonomy and digital technologies in the Australian grains industry.
More resources
Satellite imagery to detect and monitor weeds
Can drones provide early warning of herbicide resistance?