Add Category
Brad Wooldridge's family have been farming sheep in Western Australia since the 1950s.
One of the biggest challenges a farmer like Mr Wooldridge can have is deciding how many animals to put on his paddocks.
"We were always told to increase stocking rates to increase profits," he said, but he knows from experience that there's a delicate balance to strike.
"You must increase stocking rates in good seasons and reduce it in bad seasons."
Otherwise, the grass won't be able to keep up with the rate it's being munched, and the pastureland ends up "overgrazed" and damaged.
Overgrazing isn't just bad for the land. It can mean less wool, of lower quality, slower lamb growth, more lamb deaths, and fewer lambs the following year. And it can also mean less pasture for every millimetre of rain.
And to make matters worse, an overgrazed pasture will set less seed.
"So, the next year, regardless of how good the season is, you'll be compromised before you even begin. It just snowballs," Mr Wooldridge said.
Satellite solutions to replace guesswork
In the past, Mr Wooldridge has tended to overestimate how many animals he could feed.
But recently he's been tapping into high-tech solutions to take some of the guesswork out of stocking rates.
Mr Wooldridge has been making use of satellite data that measures reflected light from paddocks. This tells him how green they are and how much "biomass" they contain. In other words, how many animals his paddocks can feed.
"It's just so good," Mr Wooldridge said.
He's also looked at historical satellite data on biomass and matched it with the seasons to help him predict three months out how much feed he'll have.
Now he can better match his stocking rate to the land's carrying capacity. And by avoiding overgrazing he gets much more pasture for every millimetre of rain than he ever did before.
The secret to this particular development in "digital agriculture" is calibration of the satellite data to local conditions — something Mr Wooldridge continues to work on with a Queensland-based company called Cibolabs.
To do this he takes sample ground measurements of paddock biomass — either by cutting dry grass and weighing it, or by using a handheld tool that measures growing pasture — and sends the data to Cibolabs.
The organisation then uses the data to train algorithms that turn the satellite data into reliable "biomass" maps of farmlands.
Mr Tickle said numerous ongoing measurements from many farmers can feed back into the system continuously so it learns to be more accurate.
This machine learning is at the heart of apps like Google Maps that use a continuous feed of data from the smartphones of millions of drivers to predict how long it will take to drive from A to B.
Mr Tickle said the computer models that use farmer inputs can predict within 10 per cent accuracy the amount of biomass in every 10 square metres on the ground, and are updated every five days as new satellite data comes in.
Thankfully, given problems with internet connectivity around the country, the amount of data that needs to be uploaded and downloaded for such a system is quite small, said Mr Tickle.
"We might process a terabyte of data using cloud computing, but the farmer only has to download a few megabytes of maps or numbers for a spreadsheet."
Replacing a year's worth of work
The biomass maps can help farmers plan their animal movements — something that is particularly useful on large farms.
Mr Tickle points to some properties in Australia that are millions of hectares in size.
Traditionally, getting a sense of how much grass is left can be a tough call, requiring a yearly plane survey or a week's drive around the property.
But, said Mr Tickle, he can now use calibrated satellite data to very quickly tell property owners what the situation is.
"It would have taken a government research agency 12 months to do that in the past."
Richard Heath, executive director of the Australian Farm Institute welcomed such developments. He said relatively few companies were turning satellite imagery into genuinely useful products that made a difference to farmers' bottom lines.
"Data is only useful if it is transformed into management information," he said.
"Artificial intelligence and machine learning capabilities are speeding up the ability of turning data into useful insights."
Dr Heath said it would also be useful to combine satellite data on available feed with data from GPS collars on animals that track where they are actually feeding.
The animals won't necessarily choose to feed where there is most grass, he said —for example, they may prefer certain areas because the grass tastes better. Knowing this can also help farmers better manage their farms.
James Rowe of the Cooperative Research Centre (CRC) for Sheep Industry Innovation said properly calibrated satellite data was "potentially very useful" because the data is very cheap but in itself is only a "snapshot".
Apart from the forecasting methods like Brad Wooldridge uses, Professor Rowe points to tools like ASKBILL — developed by the Sheep CRC and now licenced by the University of New England.
ASKBILL uses a "full system approach" that combines information on pastures, soils and stock with weather data.
"We can then provide the farmer with predictions about how much pasture they will have every day for the next six months," added Lu Hogan of the University of New England.
Much of the satellite data being used by farmers like Mr Wooldridge have been made available via Geoscience Australia's Digital Earth Australia program.
The open-source software it runs on has been crunching decades of land and water data collected by ESA and NASA satellites and turning it into a form than can easily be accessed.
The data has been used by scientists and governments to manage disasters, natural resources and urban planning, but is now slowly working its way into the hands of everyday people on the ground.
Cibolabs was one of three companies recently awarded Federal government grants for developing on-farm uses for satellite data.
And the Geoscience Australia technology has also gone global, being used by countries around the world, with the latest development — due to be launched in November at the intergovernmental GEO Ministerial Summit — called "Digital Earth Africa".
The first step will be to provide maps of water from space.
"Governments and NGOS [in Africa] will be able to use it to target aid and drought relief activities," said Stuart Minchin from Geoscience Australia.
"Individual villagers will be able to use it to understand which water bodies within walking distance may have water for their family."