Science & Technology (Commonwealth Union) – The application of artificial intelligence (AI) continues to broaden new horizons in various areas of science.
The satellite imagery which is added together with AI, presently has the ability to precisely mark two invasive weed species in Australia, bringing about a promising new tool for the handling of these persistent plants.
Researchers from Charles Darwin University (CDU) and Charles Sturt University (CSU) investigated the use of SkySat satellite images alongside AI algorithms to detect and map African lovegrass (Eragrostis curvula) and bitou bush (Chrysanthemoides monilifera ssp. rotundata).
African lovegrass is an extremely invasive perennial grass that contributes heavily to the $4 billion annually spent on direct control of agricultural and environmental weeds.
Bitou bush, recognized by the Australian Government as a Weed of National Significance, is a highly aggressive shrub that invades coastal dune ecosystems. It forms dense thickets that suppress native vegetation and can drastically diminish coastal biodiversity.
Detecting these species—particularly African lovegrass—is both costly and challenging because infestations often span large areas and occur within mixed landscapes.
Researchers analyzed SkySat satellite images from sites across New South Wales using two machine learning algorithms. One model identified African lovegrass with 89.9% accuracy and bitou bush with 86.1% accuracy.
Glen Shennan, a CDU Spatial Analyst and co-author with expertise in African lovegrass, said this technique could become an essential tool for rapid and cost-effective monitoring of invasive plants in Australia.
Mr Shennan indicated that sampling on the ground is very labor-intensive. Using drones and satellites can greatly reduce costs and allows repeated surveys to track where these species are spreading.
He further indicated that we can mark vulnerable parts that is needed to block these species from spreading to, and we can direct management funds and mitigation funds to where it is most required.
Mr Shennan stated that there is an urgent demand for fast, affordable ways to identify these species, especially African lovegrass.
He pointed out that African lovegrass grows aggressively and is highly opportunistic, allowing it to outcompete native grass species.
Mr Shennan indicated that it is not palatable or nutritious and sheep and cattle try and avoid it as much as they can.
“There’s a lot of work going into managing it, but it is herbicide resistant and the only thing that will kill it, it adapts to very quickly.
“It’s very fast growing, and grows whenever the weather is right, especially in droughty summers. It likes disturbed ground so if you have a fire come through, it’s the first thing that will come back.”
Mr Shennan stated that the high level of accuracy was especially important for detecting African lovegrass, as the plant is hard to recognise in its early stages.
He pointed out that it closely resembles poa tussock when it is young and even skilled botanists struggle to tell them apart, which is why satellite imagery and certain drones are so useful.
Mr Shennan indicated that these technologies can detect colour differences invisible to the human eye, and they hope this will allow them to track its growth patterns.
The ability of enhanced analysis of changes in color, shape and other details with enhanced capabilities is what sets modern research apart with modern technology such as AI.
The project received funding from the Australian Government’s Department of Agriculture, Water and the Environment. It was co-written by CDU remote sensing lecturer Dr Richard Crabbe and CSU senior lecturer in livestock production management Dr Jane Kelly.
Next steps for the research include growing the dataset, working more closely with government agencies, and improving the models so they can consistently tell apart visually similar species.
The paper, Investigating the Potential for the Detection of African Lovegrass and Bitou Bush Using SkySat Earth Observation Satellites, appeared in the journal Weed Research.
