Big data is a phrase that has integrated this world of technology across industries. It’s about capturing relevant data from a huge number of sources, and translating it into something that people can use. Big data provides actionable insights to solve problems at scale and at speed. In this world of ag, we have billions of dollars of venture capital funding pouring into agriculture through technology builds. Big data has been at the center of that.
There are several ways big data can be advantageous to agriculture. It depends on your goals. What do you want to accomplish with the data? Obviously, big data is enabled by computing power. We have much more capacity because of server farms and cloud computing. These let us collect more and crunch through more data.
In ag, the topic of big data is relatively recent. Before yield monitors, we made many decisions in ag based on what I would call small data, which was a lot of replicated trials. A replication in a trial might be 25-feet-long, replicated three times, and becomes an observation. So now with yield monitors and all the other devices, we’re able to collect data at a high resolution. In a hundred-acre field, we would divide that field into 4,000 unique observations that are geo-referenced, tied with a lat, long, yield value and hundreds of layers of data underneath.
There’s a ton of data being collected today in ag from many different sources. Much of it is public. It seems like there are newer companies trying to take advantage of public data and the complexity of sourcing it and putting it together into some usable format. Public data is not drilled down to a level where I think it’s all that helpful. If you’re a company selling an analytics package to a grain trading company, you don’t need it refined. With that type of data, you’re trying to understand global yield trends and how it will move the supply chain. So a lot of the public sources aren’t as valuable to a grower as they are to other stakeholders.
MYTHS OF BIG DATA
Let’s talk about some of the myths out there on big data. We often hear the words “weather modeling,” but what we’re talking about is predicting the future. It might be future weather or future performance.
All models are based on assumptions. It’s about understanding specific geographies within fields, and how they’re similar to geographies in other fields. It’s almost like the more data you get, the more it lets you break it apart into more meaningful insights. The power of what we have in ag is that you have different growing environments every year. As a company, we get to observe different growing environments within the same year. So, Nebraska or Minnesota can have a dramatically different growing environment than Indiana and Ohio. For example, you could see how a hybrid or variety performs in the same year in dramatically different growing environments because you’re seeing it across these big geographies. It’s highly dependent on believing in the idea that agronomy is local, that agronomy and geography have a really close relationship with each other. It relates to the idea of big data, and aggregating it across multiple different agronomic environments. So how do we give it enough credibility that people can make decisions?
Over the years, the ability to aggregate data geographically has been a big deal. The ultimate power of all this is at a subfield level because that’s where you drive change. Every farmer who works with us wants to see beyond their own operation. They want to see agronomic practices, trends and rates.
When a farmer starts working with us, they usually want to see the biggest data set possible, meaning they want to see data from a large geography. However, we believe that local data is king in agriculture. The bigger, richer data set from a local perspective is more powerful because there are more things that are relevant and stay the same. We almost went through a decade where it seemed like the whole seed industry on corn was going to fixed-year numbers. The only way you could drive yield was to drive population. No matter what size of database, we saw a trend. We were marching up 400 or 500 seeds per acre for a decade because that’s what it took in order to drive yields.
Now, we’ve gone through almost a decade where it seems like there’s more flex in numbers. We’re producing much higher yields at lower populations. However, when we were going through that match up in population, growers started looking at row width. We had this phenomenon where everybody was chasing 20-inch corn or even narrower corn. The plants were on top of each other, and needed to be more spaced out. In the data, 20-inch corn was a South Dakota and southern Minnesota phenomenon. That’s where we were seeing the most 20-inch corn. We had people outside of that area that wanted to drill down. They wanted to see data outside of their area because they were trying to make a decision about switching to a narrower row of corn. This was a way to space out the plants as they continue to drive the population.
Since we’re capturing data off the planter, as it goes across the field, we’ve been able to calculate planting speed. One of the very early signs was we had a report that showed the faster they planted the corn, the better and higher the yield. That’s an example where faster planting speed was correlated to higher yields, but when you actually interviewed the grower and talked to the grower about what happened in that field, parts of the field worked up rough. And so, they slowed down because they were trying to maintain seed-soil contact. As they went into those areas that worked up rough, they slowed the tractor down and slowed the planter down. In the part of the field that worked up great, they planted at normal speed or higher speed and, sure enough, that was where the higher yields were. The rougher areas worked up rough, so the real correlation was to field conditions of planting, but it showed up as planting speeds. So, it was an example where you can have correlation, but it doesn’t necessarily mean causation.
One of the many myths people believe about big data is that if you haven’t got involved already, you’re probably too late.
Many growers are sitting on data and no one has helped them put it to use. There are growers who quit caring about yield data because they haven’t been able to use it. One of our successes is that we grab that historic yield data and try to use it to capture the variability within fields. You can go from zero to big data really quick in farming. You can start at any time and begin creating value right away.