Farming with big data
A teacher, farmer and software engineer discuss the future of managing streams of digital data on farms.
November 30, 2020 By John Dietz
“Easy-button” farming with big data may seem like science fiction, but it’s not far away.
Many farms and ranches already use technology that creates an abundance of digital data on multiple platforms. However, a lack of program-to-program and inter-company compatibility make it complicated to combine this data in a usable way.
What do an Alberta college, a farmer from Moosomin, Sask., and a software engineer have in common? They’re all determined to be ahead of the curve when it comes to smart agriculture. Whether designing, testing or using the latest big data ag solutions, they are committed to making operational decisions with the assistance of smart ag technology.
Olds College Smart Farm
The Smart Farm was started by Olds College in Olds, Alta., in 2010. Its mandate is to use leading-edge technology to demonstrate how to grow crops and manage livestock better. Giving students the opportunity to work with experimental agtech, it offers a post-diploma certificate in agriculture technology integration and a diploma in precision agriculture “techgronomy.”
“We can demonstrate almost any kind of leading-edge technology you can think of right now,” says Alex Melnitchouck, chief technology officer of the Smart Farm. “One of our goals is to bring in the best technologies from around the world, implement them and use them for applied research.”
The farm’s newest machine technology is the DOT Power Platform. Researchers and students are evaluating the autonomous equipment platform for its technical, economic and environmental footprint using multiple technological metrics.
“DOT is fully integrated into the Smart Farm operations. One field where we have many layers of data was seeded by DOT and sprayed by DOT in 2020,” Melnitchouck says.
The Smart Farm recently began working with hyperspectral and thermal imagery. Compared to the traditional red-green-blue or near-infrared imagery, the hyperspectral analysis includes between 12 and 500 spectral bands that allow for a much more detailed analysis of crop conditions. Thermal imaging allows for instant and accurate stress detection in field crops.
The Smart Farm collects these growing season images with a drone, and also accesses images from the Teledyne DESIS sensor on the International Space Station.
“We’re trying to figure out applications of spectral analysis for applied farming – like detecting nutrient deficiency or water content. Otherwise, it will be a custom-based tool and way too expensive,” Melnitchouck says. “Application to farming is a bit of a grey area. Eventually, you want to bring benefits to real farming on the Prairies.”
Even perennial problems, like establishing an appropriate fertilizer rate mid-season to achieve high yields, are addressed by researchers working with big data. The Smart Farm has several sensors measuring real-time air and in-ground temperature and moisture levels. In 2020, a highly experimental sensor from Teralytic was added to the Smart Farm’s toolbox. It is supposed to provide real-time data on nitrogen (N), phosphorus (P) and potassium (K) in the soil.
“This year, we started with one field and collected a high-density soil grid by sampling every acre,” Melnitchouck explains. “The data includes over 20 different soil characteristics: organic matter, pH, N/P/K, calcium, magnesium, micronutrients and base saturation. We also did soil electro-conductivity mapping and collected several hyperspectral images. Combining all those layers, we can address many questions.”
However, software compatibility remains an issue. “To do pretty much anything on the Smart Farm, I still have to use more than 10 programs and they’re completely different. I hear about integrating yield information with weather data, but how? The two types of information are used for completely different purposes.”
The Smart Farm isn’t equipped to develop new data-integrating software, but it tries to be objective and explain concepts for integrating data streams. “In everything, we try to make sure that anything we do could be replicated anywhere on any farm,” Melnitchouck adds, noting a program starting this fall is a combination of high-tech and practical applications.
“We want [students] to be empowered by knowledge and expertise to choose what exactly works for their farm. We’re trying to explain the most complicated things in agriculture in very simple words or terms.”
One smart farmer
Around 2010, Kristjan Hebert returned to his family’s 4,500-acre farm at Moosomin, Sask., ready to farm after a short career in accounting. Now, he’s a managing partner of Hebert Grain Ventures, a 22,000-acre operation with malt barley, hard red spring wheat, canola, yellow peas and fall rye, and co-founder of Maverick Ag, an agricultural financial business risk management and consulting company in Saskatoon that uses diverse data in their operations.
“Big data might get you headed in the right direction, but until you start collecting that data locally and on your own farm, it’s really hard to use it to affect a decision – and decisions are what make us money,” Hebert says.
There are four types of data for farming purposes: financial, machinery, agronomic and market-related. Each type has specific variables, and quite often, the variables don’t talk to each other and the types don’t connect at all. To make his decision-making easier, Hebert wants an approach that synchronizes data between platforms.
“My yield data from my combine automatically pulls into [the combine’s platform]. For that to automatically pull over to my accounting platform, it has to have an API (application program interface). It doesn’t go automatically from place to place, so you have to manually do it. When you try, you end up double-typing and having a lot of overlap,” he says.
Hebert’s ideal set-up would involve three screens from which he can run his whole operation. “One screen has your live agronomy-type data, like soil moisture, growth stage, heat units, the yield trend and it recommends decisions for today. The middle screen has all your people and equipment data. You quickly look there and see you’ve got two sprayers free and two operators free. If you check the ‘Yes’ box, work orders go out automatically to the sprayers and the operators. Your third screen is financial. It automatically pulls in the costs to run the sprayers and the costs for the fertilizer. It also increases the expected revenue on your pro forma based on the results of that action. To me, that’s how it flows across. That’s how this customer wants it to work.”
Engineering the future of big data
Eric Smith is the engineering manager of precision agriculture at the Winnipeg headquarters of JCA Electronics. As an original equipment manufacturer supplier to implement and machinery makers, JCA Electronics is often contracted to develop application systems.
“Typical projects are in the autonomy area, and we do everything from developing control software to ISOBUS-compatible systems, to mobile apps and cloud systems,” Smith says. “We manage data from sensors on equipment all the way up to the cloud.
“Everybody wants to use data and a ton of data is available, but the challenge is to get the data off the machine and from different machine manufacturers. Most of it talks to other stuff from the same company and nobody else. Quite often you can’t get the data off your machine and into the cloud [but] that’s starting to change. These companies are starting to talk together and working on technologies that allow cloud systems to talk to one another and allow farmers to get their data into the cloud.”
Smith says it is to everyone’s advantage to push for compatible communication standards like ISOBUS, which is a standard from the 1990s for farm electronics.
“It will help everybody and allow farmers to choose the best product instead of sticking with a single colour,” he says.
Getting the streams of big data to merge into the format that Hebert would like to see on three screens isn’t far-fetched anymore, Smith says. Some of the technology Smith has worked on goes even further beyond Hebert’s dream.
“A couple of years ago we investigated augmented reality and came up with an interesting application using Microsoft HoloLens so you could visualize your application data,” he says.
HoloLens 2 was introduced commercially in February 2019. It is a ‘mixed reality’ operating system using the Windows 10 platform and a head-mounted display, or smart glasses.
“The HoloLens is like a big set of glasses that you put on your head. You still can see reality behind it, but things can be drawn and added to what you see. For this proof of concept, we created an artificial field and created a tractor to drive a path on that field back and forth,” Smith says. “Then we generated data as if you were driving a path on the field and overlaid that data on top of the field. Then we took it a couple of steps further.”
The HoloLens empowers the use of voice commands to talk to and control the data simulation. When the API is further developed, in theory a person could operate several DOTs sitting at one desk, knowing exactly what each is doing. The machine app could be augmented with real-time agronomy and weather data.
Smith believes that with accurate geotags, various streams of data can be collected and visualized for operators to have a greater understanding of what’s happening on a farm.
“I think AR (augmented reality) for farming will be here soon. It’s just a matter of time.”