By Harold Watters
I love data. To hear the ag engineers talk about big data you would think they invented the concept. But agronomists and crop advisors have been working with big databases for a long time. And this is the time of year when data rushes in.
I think the reason ag engineers like the big data concept is that they don’t understand how to read a dozen yield reports, two to three state variety trial reports and a string of farmer trials to come up with a recommendation for a hybrid or a variety.
When yield trial data is coming in, I get daily updates from several companies. I have my favorites that I picked out earlier in the year to watch, or maybe even last year. So far I hear one of the hot ones I considered last year is a bit of a dog this year — but I’ll keep checking as more data comes in to make sure.
Where do I start?
I go to company yield trials because they have treated me well in the past. But, with corn and soybean prices being lower, I will look around to make sure I have the best genetics. Peter Thomison at Ohio State University says that hybrid selection is our biggest decision in making money for next year. In scanning the past several years of Ohio State corn performance trials, I see a range of 60 bushels of corn per acre from the highest yielding to the lowest yield. Meaning that if I pick the top hybrid (or something similar) versus the lowest I am likely to pick up a 25-plus% improvement in yield — under the same conditions. I look for trials close to me. I also look for sites that have soils similar to mine, or maybe have a similar rainfall pattern.
For soybeans I do the same thing. I dig, I hunt and I check my other data information to find a good fit for my location and my clientele. While I trust my seedsman to provide good information, I still want to check — that’s “trust but verify”.
What about technology?
It’s about yield. Today in Ohio, with the exception of just a few acres, we can manage our weeds. So the herbicide technology may not be that important — look at yield first, then look at the herbicide resistance package. In corn, consider insect protection. What do I need? What’s my past experience? Am I in a continuous corn program? And if so, do I use a refuge as the label says we must. If I am running the ragged edge on keeping insects at bay, do I risk buying no insect protection? Probably not, I may buy above ground protection but skip below ground if I rotate. Again look at the data, where are the yields coming from? Today, any technology is likely to have good yielding genetics.
What about soil test data?
I map soil test data and check that I didn’t screw up on sampling. Then I check past history — as far back as I have — and do not use just this one year. Soil test data that we get today should become part of a big library of data. And looking to that past history may tell more than the information from this fall’s sampling.
How can history help?
Let’s suppose that based on past history and crop removal, expected potash levels are off in part of the field. It was dry at my client’s place this year, and typically when we have dry conditions potash levels will be down. What about dry weather at harvest and for an extended period before we pulled samples? If we don’t have weather knocking K out of our corn stalks, then we will also likely see lower potash levels than expected. If our history from our library of past soil samples is right, then we shouldn’t make any rash decisions on upping the potash rates. Again look at all the data, past included.
What about yield monitor data?
I am hearing some producers say they are off a mile, others say they are within 12 pounds in the bin. In any case, double check that a calibration was done this year. Consider moisture, yield level and speed. It is imperative to do a good job of calibration, and then double check throughout the season. Otherwise, the data may not be usable. What is done up front is critical to gathering that good information.
Data management is critical. I like to use paper first, get a good view by eye, then integrate that into making a decision. The human brain is a pretty good calculator. The process to manage big data may help to make a good decision, but unless you put the right information in you won’t get the right answer back out.
Watters CCA/CPAg, [email protected], office phone 937-599-4227