Precision Agriculture: It’s Coming

The planter. It has a 60-foot wingspan and 36 individual row units spaced at 20 inches apart. Each row unit is responsible for planting a seed about 1–2 inches deep in the soil and 7 inches apart. It’s a big machine with its tanks of seed and fertilizer sitting (often covered in dust) on top. And it’s complex — perhaps one of the most data-centric pieces of equipment big ag has to offer.

60 foot planter

I want to talk about the planter not only because it’s a truly neat machine and capable of some pretty interesting things (especially the more advanced versions), but mostly because it’s a good case example of what precision agriculture means today. It epitomizes how big ag can use data to make its actions much more accurate and refined. It’s also a good reflection of how “precision agriculture” isn’t really new (although many folks talk about it like it’s some big new idea) — and how big ag has been pursuing this concept for years.

Let’s start with the seeding plan. The decision of what to plant, where to plant, how deep to plant, etc. is all baked into what is called a “seeding prescription”. These prescriptions are developed by agronomists and are customized and tailored to each field. If you want to think about a highly complex data problem, here you have it. The agronomist takes in 40+ variables into account for every single field. This includes: weather conditions, soil type, soil nutrients, soil moisture, elevation, previous yield, previous crop, etc. Fed into a model, the agronomist creates a map of the field that indicates what variety should be planted, at what population rate, for any given portion of the field.

When it comes time to do the actual planting, each row unit (see photo below) takes in this seeding plan and calibrates it accordingly. This means that it’s automatically adjusting its spacing and down force given the instructions that have been preset. Originally, I assumed it was pretty uniform for each field (implying a precision of hundreds of acres). However, as I quickly learned, it’s much more precise than that.

Individual row units that release the seed into the soil. For seed corn, specific rows take in male seeds vs. female seeds.

On the monitor in the cab, there’s a complex array of numbers being displayed. You see terms like “singulation” (accuracy of planting location), “good ride” (accuracy of planting depth), “seeding” (population of seeds per acre) — all accompanied by some set of numbers that change rapidly as the tractor drives through the field. It’s phenomenal how the machine tracks all this data and makes sure that it aligns with the seeding plan. And each acre is different. For example, when planting on a hill, our population numbers dropped to about 31–32K seeds / acre because there’s generally less nutrients and worser growing conditions. But as we drove down the hill, the numbers on the monitors increased to closer to 35–36K seeds / acre as the seeding plan had taken into account the elevation delta and planned accordingly.

So to say that agriculture “isn’t precise” seems a bit off-base. Planting is an impressively data-driven and precise process, especially given how it’s able to be so specific across such a large number of acres. This means 18K different combinations of recommendations. I wouldn’t call this “imprecise”.

Now perhaps where people get the notion that agriculture is “less precise” is that not all growing processes can offer this type of precision. In terms of the food chain, the order from most precise process to least precise process goes something like:

  • Planting
  • Crop Protection Application
  • Side-Dressing (Fertilizer Application)
  • Irrigation
  • Harvesting
  • Spraying
  • Tillage

While planting is the most advanced, the others aren’t all that far behind. In fact, many of these processes have equipment that can offer the precision at the acre level — it’s just a matter of farmers investing in these more advanced modifications.

Personally, I was surprised at how data-driven some of these procedures were. However, I imagine there is still room for improvement. For example, seed prescriptions are made in advance of the season and weather predictions can change at any given time. I would assume that there could be a way to automatically adjust and update seed prescriptions based on new, incoming data that gets taken in during the start of the growing season. Although, Tom Farms does have a secondary or “later” soybean prescription that takes into account a few additional datapoints, it seems there may be opportunity to do something ever more agile in the future. Again, it may only provide incremental improvements, but with more data comes more accuracy and it looks like we have the tools to implement this degree of precision.

My planting experience gives me optimism about the future of data in agriculture. We’re already on a good data-driven path and I’m confident that farming enterprises welcome the innovative use of existing data or new data. Planting has set the standard for precision and I hope to see some of these other processes follow in line.

The planter with some nicely planted rows of commercial corn