“We have a traditional image of a farmer picking up some soil and running it through his hand – invariably a male farmer – and that’s not really how things operate any more, and it’s not how things will operate in the future,” Fielding said.
“It is more likely that the soil will be analysed through x-ray fluorescence from the sky via satellite or through a UAV (unmanned aerial vehicle). The data will become more and more specific as we do that. It is impossible to go around and test every square metre by running the soil through your hand, but it is imminently possible to do that from the air through sensors. And it is imminently possible to run an automated system across a paddock.”
Fielding described the amount of data generated by this sort of personalised and localised activity as inconceivable.
“When you have got all this data, it becomes impossible for people to make all of these decisions. So increasingly we will see algorithms (computer software) making the decisions. We will see that the sheer weight of data makes it impossible for people to manage this themselves.”
He illustrated a possible use with an example based on hyperspectral imaging. Hyperspectral imaging aims to obtain the light spectrum for each pixel in an image. While the human eye sees visible light in three bands – red, green, and blue – spectral imaging divides the spectrum into many more bands.
“Research has shown that this hyperspectral imaging data can be used to identify a diseased plant within four hours of it contracting a disease. Something that usually takes a week for the human eye to spot this specific disease can now be done using this data in a way that humans couldn’t possibly ever do. Within just a few hours it can identify that the disease is spreading.”
Fielding said the algorithms used to make these decisions would become the intellectual property of the businesses that developed them, effectively causing a shift towards a business model with decision-making-as-a-service at its core.
Another trend identified by Fielding was increasing use of automation, although this would not all be plain sailing.
“Some of the most challenging problems in automation (in primary industries) deal with the fact that this is not a factory environment we’re talking about, this is the real world.”
“What you get out of automating in the primary industries is similar to what you would get in most industries. You can reduce waste by using the best practice time after time. A robot can pick a kiwifruit in exactly the same way without damaging the calyx hundreds of thousands of times in a row without making a mistake. It improves the quality for the same reason – everything can be handled in a uniform way. You can use sensors to detect parts of the animal that you want to keep and cut out, and you can do it in a completely consistent way.”
Big data the future of farming?
Massey University Professor of Precision Agriculture, Dr Ian Yule, said the first thing to recognise was that it was not just business as usual.
“It is not just putting your PC farm data on to a cloud – that is not what big data is. We might have historically collected data on finances and physical performance of the farm, it is collecting that but along with that, collecting a lot of other data that is in the environment.”
“So big data might also be monitoring equipment, for example. One example would be John Deere tractors, which have a modem that can send out data to track the machine performance; there is a whole raft of stuff, some of it useful to the farmer, some of it is useful to the service agent and some of it is useful to the manufacturer.”
But data collection is just the first step. That information then needs to be collated, analysed, and interrogated.
“But this requires that the data analytics are much better than they are at the moment,” Yule said.
“Analytics looks at trends in data and is a fairly large science in itself. Medical records for humans would be an obvious example”.
Yule said a potential outcome could be having measurement devices on cows all the time, using analytics to notify the farmer of any change in behaviour.
“We might collect a cloud full of data but we then only want specific pieces of management information from it.”
