Next-generation analytics to improve profit from integrated weed management

What would weed management look like if we could predict where and when weeds will emerge? Utilising this information in the planning phase to employ every tool available (tailored pre-emergent mixes and application, site-specific seeding rates, competitive varieties in weedy zones, and strategic spray-topping) to achieve improved, longer-lasting control.  With the result being less reliance on herbicides due to minimal weed seed set and the weed seedbank in decline?

Today’s weed control tools, such as camera-guided sprayers and site-specific applications, help reduce chemical use and improve timing. However, they still focus on herbicides and treat the weed outbreaks rather than the underlying cause. A new national project led by the Australian Herbicide Resistance Initiative (AHRI) at The University of Western Australia with investment from the Grains Research and Development Corporation (GRDC), aims to flip the script from “spray what you see” to “predict what will grow.” By combining spatial imagery, farm records, and weed biology, this project will forecast the location, size and spread of likely weed patches before they appear. Pre-armed with this knowledge, farmers and consultants can apply a well-thought-out, pre-emptive, integrated weed management program, allocating valuable herbicide chemistry where it delivers impact and maximising crop competition, delivering greater efficiencies across the farming system.

AHRI has developed a new weed-prediction model that uses satellite imagery (Sentinel-2 and Planet) to monitor crop and weed signals throughout the season and across extended crop rotations (approximately five years) and flags vegetation anomalies associated with high-density weed patches. Partnering with Farmanco in Western Australia, we are validating the model using growers’ fields to record weed density, species mix, and the underlying soil weed seedbank.

From this research, AHRI hopes to predict weed dynamics in growers’ fields, rather than simply reacting to them, with the aim of turning predicted weed maps into management plans.

Responsible for delivery of this project:  Dr Mike Ashworth, Dr John Duncan