What’s the deal with big data?
Not long ago, Nick Guetterman’s brother asked him, “What’s the deal with data?” The combines on his family’s farm in Johnson County, Kan., have been equipped with yield monitors since 1997, but there’s still a lot to learn.
The term “big data” alone brings forth questions of how to manage the all-encompassing set of data farmers collect — from yield data, soil maps and weather history to disease and weed pressure, different seeding rates, and timing of fertilizer applications. “Managing all that data and figuring out what to do with it once you’ve got it is slow and time-consuming,” Guetterman says.
That’s one of the likely reasons farmers using precision agriculture have stuck with guidance systems like autosteer and swath control, which have a quicker return on investment. “Row shutoffs have been great, especially by preventing overlapping and skipping on point rows. There aren’t weeds growing where you’ve skipped; the corn isn’t overlapped; and you aren’t wasting seed,” Guetterman says. “It’s much easier to put your finger on what you’re saving than other technology.”
Making sense of it all
On the other hand, getting a return on investment from a yield monitor isn’t always easy. Making sense of big data, as Matt Danner puts it, “starts with a lot of pockets of small data.” Yield monitors have to be calibrated and verified, and each piece of small data must be correct, or the entire data set becomes skewed.
“If I punched in the wrong variety from the day I put the bag in the planter, it’s going to be incorrect throughout the year,” says Danner, who farms in Carroll County in western Iowa. “All these tiny pieces have to be 100% right from the get-go, or you’ve proliferated an inaccuracy into the cloud throughout the entire system.”
How a prescription performs doesn’t mean much without a control variable to compare it to. “If you’re going to test a prescription and you have no control strip to compare it with, you have no way to benchmark it,” says Danner, adding that only one variable should be tested at a time. “It’s one thing to increase the seeding rate with the same variety, but with two varieties and different rates, you’ve put four variables in the air.”
Making a variable-rate prescription in the first place takes time — several years of yield data, soil maps and personal knowledge of the farm, especially considering the variability in the last few years, notes Brooks Hurst, who farms in Atchison County in northwest Missouri. “If you take one year of data and try to make a variable-rate fertilizer or planting map, you’d be making a mistake, and would keep making self-fulfilling prophecies when it may have been an anomaly of a year,” Hurst says.
What’s the value?
This isn’t to say there isn’t value in data. The bigger the data set a grower has access to, the greater the potential to improve on-farm decisions. Take that data set and expand it across the county on the aggregate level, and farmers can benchmark themselves against others in similar conditions.
“My data isn’t as valuable to me on my farm alone. It’s a lot more valuable if I can look at all of Atchison County. Then the entire county is a test plot for me,” Hurst says. “I can compare farming practices and inputs against everyone that participated and gain some valuable knowledge for my farm. But on the same token, I don’t want the raw data to get in the public domain.”
He’s referring to the geospatial reference identifying where each piece of data originates. If shared on the aggregate level, he says it should identify how a percentage of the entire database performed with a certain practice, rather than identifying a specific farm or acre. “We have to be aware of the fact that the value of this data is the geospatial reference,” he says. “But it’s also what eliminates the privacy you think you have.”
Service providers can massage value out of individual farm data by taking the time to apply different layers to a geospatial point. “I am the IT guy, the law division, the secretary and everything else. But there are some things we’re going to have to start partnering on,” Hurst says. “I spend a lot of time working on equipment and managing field operations, so it’s hard to find time to massage a lot of raw data.”
A fair trade for data
It’s because of this specific geospatial reference that sharing data takes a certain level of trust, bringing up the question of property rights: “Who owns the data?” The concern is when the data falls into someone else’s hands, it might be used for something the farmer doesn’t approve of, including using it for precision marketing, pricing equipment or seed specific to the farmer, or disclosing it to third parties who might do the same.
While producers can negotiate the terms of the agreement to prevent these things, sharing data with a service provider is also beneficial to the service provider. So Danner says monetary compensation is necessary. “After I’ve shared the data, I no longer control it. So it was truly a transaction,” he says. “You bought something from me.”
Guetterman notes the jury is still out on a number of data ownership questions. “Is it our data, or is it theirs? Should they be paying us for this data instead of us paying them? Those are all questions to be asked.”
This article published in the November, 2014 edition of WALLACES FARMER.
All rights reserved. Copyright Farm Progress Cos. 2014.
Precision Farming Management