Mike Cerny has been mapping yields on his Wisconsin farm since 1994. Fourteen years later, he is just beginning to get a picture of where site-specific management might pay off.
“Precision farming only came in about 12 or 14 years ago,” says Joe Lauer, a University of Wisconsin agronomist. “So we're just now getting to the point where we have enough data that we can start to predict performance.” And that's what site-specific management is all about, he says. “We have to move from describing what happened in a field to predicting what's going to happen. If you can't predict performance, there's no point in doing precision farming.”
Cerny raises corn, seed soybeans and seed wheat on 1,200 acres near Sharon in southeast Wisconsin. He farms a mix of soils: “some good, black prairie dirt; some gently rolling hills; some clay soils.” His land's productivity ranges from very high to medium-low, he says.
He bought his first yield monitor in 1987, and in 1994 invested in a GPS satellite receiver and GIS mapping software. “I'm one of the idiots who got pretty bloody” in the early days of precision-agriculture technology, he says. “Back then, it was a huge investment — well over $20,000.”
IN THE LATE 1990s, Cerny added geo-referenced grid soil samples to his database. Today, in addition to crop yields, he can pull up and map several GIS layers of data about his operation, including soil nutrient maps, elevation maps and real-time planting maps.
Over the years, though, he's asked himself if his ever-growing data sets were really doing him much good. “It's hard to put it all together.”
He's not the only one scratching his head, Lauer says. “I see growers with all their yield maps thinking, ‘What should I do with them?’” Like Cerny, they are “wrestling with how to identify management zones and where to move inputs.”
Recently, Lauer and Cerny examined 12 years of Cerny's crop yield data — six years of corn and six years of soybeans — collected over six rotations. They divided 300 acres of cropland into 50-meter geo-referenced data cells (164 ft. × 164 ft.) and calculated the five-year average corn and bean yields for each cell. They also looked at each cell's year-to-year yield variation from the average.
“We wanted to find out if, by doing this, we could predict what the yield would be in the sixth year in each cell,” Lauer says. “Bottom line, we found that we could predict the next year's performance.”
In corn, for example, the six-year analysis showed a consistent, 26-bu. difference between high-yielding and low-yielding cells. That was a bit of a surprise, Cerny says, because the cropland was “all one soil type, with little elevation variance and fairly uniform fertility. That's not where you'd expect to see much difference, and yet we did.” So these stable-, high- or low-yielding zones could be candidates for special management, Lauer says.
Cerny is using the data to vary corn plant populations within fields. “We do know that plant density influences yield dramatically,” Lauer says. “And we know there are different optimum populations in different yielding environments.”
Working with 120-ft. GIS zones, Cerny bumps up corn plant populations to 35,000 in reliably high-yielding areas, and drops them back to 29,000 in consistently below-average zones. He's not cutting his input costs, but hoping to boost his seed returns. Still, he's not sure if he's putting more money on the bottom line. “Seeding rates to me are a real guessing game.”
When it comes to drainage decisions, though, Cerny's years of yield mapping have definitely paid off, he says. He has pinpointed areas where water problems caused variable yields. “I always knew they needed drain tile, but I didn't know how bad,” he says. “You forget how fast a bad spot destroys the whole field average.”
Rather than installing uniform drainage systems in those fields, Cerny used his yield and elevation maps to “put tile where it was needed. We probably installed more tile than we would have without the yield maps. But we put it in better locations.” He even tiled some rented land in exchange for a seven-year lease with a reimbursement provision for early cancellation.
THE RESULTS WERE plain to see on subsequent yield maps: In formerly wet areas, yields jumped from fewer than 100 bu./acre to 220+ bu./acre.
Cerny is experimenting with variable-rate fertilizer applications. He uses his geo-referenced soil samples to manage soil pH. If you've got pH problems, variable-rate lime application offers a clear payback, says Gregg Carlson, agronomist and precision agriculture expert at South Dakota State University.
Cerny is also using his database to generate prescription fertilizer maps for phosphorus (P) and potassium (K). On one 60-acre field with a sharp range of high and low soil test values, his P (0-46-0) rates for a two-year broadcast application last fall ranged from zero in one area of the field to a high of 336 lbs./acre in another zone. Likewise, K (0-0-60) rates ranged from 60 to 541 lbs./acre.
Higher fertilizer prices are sparking interest in variable-rate P and K applications, says DeLon Clarksean, a precision-agriculture specialist for Farmers Co-op Association in Canby, MN. Clarksean advises about 25 southern Minnesota growers who are using variable-rate technology (VRT). He says VRT helped some farmers cut their fertilizer bills by up to $70/acre last year in parts of fields.
Cerny is not certain that VRT lowers overall input costs. His total fertilizer bill is about the same as it would be with uniform application, he says, “although I might put on four times as much fertilizer in one area as another.” Whether VRT pays off on the revenue side is still a question for him. “I've asked that, and never really got a clear answer.”
Why, for instance, are some zones consistently high-yielding despite low fertility levels, he wonders. “You expect it to answer all the questions you have, but what it does is raise more questions,” he says.
Still, Cerny adds, boosting returns is not his only fertility-management goal. “Where it's reallya benefit is to the environment. We're putting fertilizer where we need it,” reducing over-application and cutting the risk of nutrient loss, he says.
“There's a whole list of inputs you could move around the field,” says Lauer. Lots of different approaches to variable-rate management are being tried, but there's little solid research to back them up, he says. “What's the probability that any given area is going to respond in any given year? We don't know that yet.”
But one thing is certain, Lauer says. Before you can make reasonable predictions about the outcome of site-specific management practices, “what you need, fundamentally, are many years of yield maps. I like to say, a farmer's children will be the real beneficiaries. If you can hand them 20 or 30 years of data, they'll be able to make some pretty good predictions.”
Continue on Page 2
Mike Cerny, Sharon, WI, has been using GPS precision agriculture tools for nearly 15 years. Here are a few of his tips for managing data.
Invest in computer power. Buy plenty of memory and high-end graphics capability.
Download and back up. Any time you geo-reference fieldwork, download your data that same night. Back up your data on a CD and put it in your bank safety deposit box. “It's a valuable asset.”
Collect good data. Properly calibrate your yield monitor, and be aware of weather and field conditions that can distort the readings. “If you're basing your decisions off bad data, you're sunk.”
Make in-season observations. Yield maps show you where yield variations occurred within fields, but they don't tell you why. To interpret your data, you need to observe things like weed and insect pressure, disease and standing water.
Focus on key performance data. “I tried working with a software program that had 75 data layers. That's too many.”
Make a commitment. Be prepared to spend a lot of time digging into your data. “If I had $5 for every hour I've spent in front of my computer, I'd be retired by now.”
