Scientists at James Cook University have developed artificial intelligence systems that can forecast drought conditions.
Subscribe now for unlimited access.
or signup to continue reading
A trial conducted by Dr Bithin Datta and a team of students has looked at five years' worth of data for the Ross River catchment near Townsville, using AI tools to examine patterns signalling an impending drought.
"I was really surprised at some of the results we got; it was quite successful," Dr Datta told AAP.
Their AI system was able to accurately predict some of the key indicators of drought, such as soil moisture, salinity, groundwater levels and dam levels, between three and six months ahead of time.
Dr Datta said the technology's potential uses could be a game-changer for farmers and for urban water management, with enormous economic benefits.
"If you knew a drought was coming you could take remedial measures ... manage water supply, look at different cropping patterns," he said.
The prediction system, known as an artificial neural network, can identify relationships between variables that humans can't pick out and figure out which information isn't relevant.
In a decision-making process that mimics the human brain, it can use the patterns it's identified to make predictions.
Dr Datta said his team want to develop their work so it can be used more widely.
"I really want to take it forward and I really think it works," he said.
Droughts followed by extreme flooding have long been a problem across northern Queensland's 100 sq km Ross River catchment, where sugarcane is the main commodity, followed by beef cattle.
In the 1930s, drought conditions were so severe the river stopped flowing entirely and water had to be transported to Townsville by train.
Two metres of rainfall in the catchment in 14 days in the summer of 2019 caused widespread damage in Townsville and surrounding areas.
Australian Associated Press
