Big data, big wins for ag

Almost everything can be turned into digital information in a world awash with sensors - movement of vehicles, livestock, eyes across a page; temperature of soils and cows in oestrus; plant growth, plant species, genetics, and smartphone usage patterns.

When those data points are strategically layered on each other, they become 'big data', a resource to be interrogated for hidden patterns and trends not available to our experience of the world through our six senses.

“Big data is a capability, not a thing,” wrote Steve Sonka, emeritus professor of agricultural strategy at the University of Illinois.

Professor Sonka argued that big data is moving “from hype to agricultural tool” in the Australian Farm Institute’s latest Farm Policy Journal, which examines the value of big data from various perspectives.

For Professor Sonka, big data “is the capability to extract information and insights where previously it was economically, if not technically, not possible to do so.”

That’s similar to the perspective of another author in the journal, John McLean Bennett, who looks at big data through the prism of US Sectary of State Donald Rumsfeld’s infamous 2002 comment on “unknown unknowns”, in reference to Iraq.

Rumsfeld’s point - that the facts we don’t know exist are sometimes the most important - are in Dr McLean Bennett’s view the facts that big data should give us access to.

“Should”, because what tends to happen instead is “information overload leading to decision-making paralysis”.

All the journal’s five papers agree that the ability to usefully analyse data is central to making big data practical.

In their paper, James Rowe of the Sheep CRC and Rob Banks of the Animal Genetics and Breeding Unit (AGBU) illustrated what happens when a production sector finds a way to convert data into action.

Since 1990, genetic gain in the Merino sheep sector has amounted to about 62 cents per ewe per year - about a third of its potential, according to a 2009 study.

The terminal meat sheep breeds were tracking on the same path until the turn of the century, when breeders began doing cross-flock comparisons and acting on the results.

Since 2000, genetic gain in the terminal breeds has produced cumulative returns of 202 cents per ewe per year.

When data is used to track results across sectors, matters become simultaneously more complicated and potentially more rewarding.

The group of four European researchers who contributed the journal’s European perspective on big data are interested in whether data can save the small family farm.

They see hope through collaboration in “regional clusters” where information and communication technology (ICT) is used to reduce the transaction costs on small “food webs”.

“The internet plays an important role in these clusters by matching local demand with supply and managing the last-mile logistics,” the researchers wrote.

But having hopefully suggested that “the trend to big data may not be detrimental to the position of small family farms”, they look at a more imminent reality and wonder whether they have been too optimistic.

The trend might equally be for big data to support closely integrated supply chains, where the farmer locks into a proprietary system and accepts stability over freedom. That type of system is already used by Australia’s supermarket duopoly.

For the New Zealand dairy researchers who contributed to the journal, big data has an important role in a rural economy where farms are growing larger, labour scarcer, and management more complex.

But they also saw problems of dependency. If one component in an integrated system fails, the whole system can fail.

There is also the looming shadow of trust. Who owns the data, and who can get their hands on it?

Professor Sonka wrote that data is flowing both ways: from the farm to inform the consumer about the products they are buying, and from the consumer to the farm to inform producers about tastes and buying patterns.

“As farmers and managers become increasingly familiar with the notion that data has value, many have questions regarding which entities benefit economically from use of data from farm operations,” he said.

In the United States, work has begun on clarifying the legalities of data ownership, how data is collected, and how it is used in the future.

If, as Professor Sonka predicts, big data and ICT become as essential to farmers as their smartphones have become in less than a decade, then Australia will have to set to work asking the same questions, and soon.

Farming today is awash in technology, and a lot of that technology collects data, a lot of it automatically - but proportionally little of it is used.

Of the Australian grain growers with the capability to create yield maps, only 30 per cent actually use the capability, and even fewer use it to its full capability.

That frustrates Tim Neale of consultancy Precision Agriculture.

Mr Neale has seen huge shifts in productivity from paying attention to what the data is saying, and acting on it.

He cites the case of the Victorian grain grower who lost $52,000 in a 2010 canola crop because of waterlogging over a quarter of the 140-hectare block.

Brought in to look at the problem, Mr Neale built a topographic map of the paddock from elevation data automatically collected by the grower’s on-board GPS equipment, and referenced it against yield and satellite data, also readily available.

VR nitrogen maps (Tim Neale/Precision Agriculture)

From that, he was able to determine the optimal site for a $5000 drain through part of the paddock, and suggested the grower change the direction of his tillage runs to help water flow off the block.

The following year only 4 per cent of the paddock was waterlogged, and the wheat crop it grew made $42,000 - a cumulative $94,000 gain from acting on data the grower already had in his tractor cab.

Yield stability map (Tim Neale/Precision Agriculture)

The case study is part of a wider theme for Mr Neale. He sees a great deal of attention paid to new crop varieties or livestock genetics that promise an incremental gain in productivity, when back at home in farmers’ paddocks and herds there can be a five-fold difference between the best and worst soils or animals.

Data gives producers the tools to address that fivefold difference, often at modest cost, Mr Neale said, yet those opportunities are sailing past producers daily.

Part of the problem is that equipment manufacturers and others in the data business have entered into an arms race of proprietary systems that don’t talk to each other, at least without considerable effort.

Mr Neale is working with the CRC for Spatial Information on a project called Intelligence Bottler, a kind of digital Tower of Babel that translates digital languages in all directions.

If successful - and Mr Neale said results are already promising - the promise of big data comes a whole lot closer because data can be pulled from all sources, irrespective of paint colour, and put to work on a united platform.

Machinery manufacturers would still sell machinery, and the data loggers carried by the machinery, but for companies like Precision Agriculture, the ability to automatically parse different forms of data could mean the difference between dealing with hundreds of customers and 10,000 customers.

In fact, Mr Neale can’t see anyone losing from making data more universal.

AFI Farm Policy Journal, “From little data big data grow” is available online.