- New research is investigating ways to better use remote sensing technology as a tool to detect weeds in complex ecological landscapes
- The project brings together researchers, governments and weed management experts
- The research will look at airborne and satellite technology combined with machine learning
New research aims to improve the way remote sensing technologies can be used to detect weeds in complex ecological landscapes.
The project, led by Charles Sturt University through the Graham Centre for Agricultural Innovation, aims to better use the technologies to rapidly detect weeds in mixed landscapes, such as native grasslands and woodlands, coastal dunes, alpine areas, and other diverse natural systems.
The aim is to help land managers to effectively target control to prevent weed spread, and reduce herbicide use.
Lecturer in livestock production management at the Charles Sturt School of Agricultural, Environmental and Veterinary Sciences, Dr Jane Kelly said remote sensing technology is already used in a number of agricultural and environmental contexts but challenges remain.
“Complex natural systems make it difficult to distinguish weeds from other vegetation, especially during early stages of weed growth,” Dr Kelly said.
“Remote detection techniques can also be expensive, technically-challenging and inaccessible for many land managers.”
Dr Kelly said the two-year project will test the limitations of available remote sensing technologies, including airborne and satellite platforms and high resolution RGB (Red, Green, Blue), hyperspectral and multispectral sensors, in combination with machine learning, to detect weeds.
“The project will use three nationally-significant ‘model’ weed systems, including hawkweeds, African Lovegrass and Bitou Bush to investigate the limitations of this technology,” Dr Kelly said.
“They’ve been selected on the basis of the differing growth habit, colour and texture of the weeds, and for the variety of invaded landscapes and weed management objectives.”
Study sites will be located within Kosciuszko National Park, the Monaro and the North, Central and South Coasts of NSW.
Dr Kelly said the project also aims to provide robust guidelines for using remote sensing in weed detection across a variety of landscapes.
“The intent is to build an online portal and a community of practice to share tools and knowledge between researchers, governments, natural resource management groups and a range of end-users,” Dr Kelly said.
This project is supported Charles Sturt through funding from the Australian Government’s Established Pest Animals and Weeds Management Pipeline Program - Advancing Pest Animal and Weed Control Solutions.
It brings together drone, machine learning and remote sensing researchers Associate Professor Lihong Zheng and Dr Remy Dehaan from Charles Sturt, Associate Professor Felipe Gonzalez from the Queensland University of Technology, and weed management experts from the NSW Government.
Other partners include South East, North Coast and Murray Local Land Services, Mid Coast, Eurobodalla and Bega Valley Shire Councils, the Illawarra District Weeds Authority, and XAg Australia.