Big Data

Purdue Researchers Will Help Farmers Make Big Decisions

By Keith Robinson - Published May 18,​ 2015

Tractor CabHuge increases in the volume of data about seeds, soil, fertilizer and yields, and the ability to send the information to and from computers in tractor cabs add new dimensions to precision agriculture. Farmers can equip their tractors with systems and devices that monitor and track all sorts of data, such as seed spacing and fertilizer dispersal. Yield data from sensors in combines help refine plans for the next season’s planting. (Photo © Andrew Sacks/AgStock Images/Corbis)

Big data.

Big potential.

Big challenges.

Agriculture becomes more high-tech every year, and researchers at Purdue University are zeroing in on how the gazillion bits of digital information being created every day—"big data" as they are often called—can help farmers become more efficient and even more productive. That need will only continue to become greater as a world population grows from 7 billion people today to a projected 9 billion by 2050.

That's a lot of mouths to feed. There's also the reality that farmers not only will have to produce much more food than they do now, but they will have to do it on perhaps less land around the world while also protecting the Earth's natural resources as much as possible to sustain agriculture for future generations.

"Although farmers are more efficient now than they have ever been throughout history, there still are challenges ahead of us, and we are using technology in our work to find solutions," said Karen Plaut, the College of Agriculture's senior associate dean for research and faculty affairs.

Maximize Yields, Minimize Impact

The benefits of agronomic data generated in the field is clear to Katy Martin Rainey, an assistant professor of agronomy.

"It's all about maximizing yields and minimizing environmental impact," she says.

Rainey sees the potential in using drones to obtain data that can help in the assessment of soybean crops' health and be another tool to predict yields. The unmanned aerial vehicles, or UAVs, hover above the plants and take digital maps that can be layered with data from other sources for deeper, more defined analysis.

Integrating data from multiple sources can be difficult because of varying data formats and incompatible software and hardware.

"That's really the big challenge," Rainey says. "But when you collect a lot of data and integrate it, you can do a lot with it." The information, for example, can help researchers develop better seeds and delve deeper into the intricacies of how crops grow. And special cameras record infrared waves that can tell the plant canopy's temperature; a warm canopy might indicate a pest problem or that a specific part of the field might need irrigating.

Purdue Agriculture's plant sciences initiative, part of the Purdue Moves program to develop more research and educational opportunities across the university, has a "big data" element. The university is building an Automated Plant Phenotyping Facility at the Agronomy Center for Research and Education to develop automated systems that aid researchers by collecting billions of field measurements showing differences in plant characteristics, such as canopy development, leaf area and photosynthetic ability. The facility will open in 2016.

Boosting Production Abroad

soils mapMapping farm soils in Central America will help farmers reduce fertilizer application and decide which fertilizers would be best for their crops. This soil map of the Namasigue Watershed in Honduras shows percent soil organic matter mapped at four different scales, ranging from the landscape scale to the farm field level. Colors of the boundaries outlined on the 1:50,000 scale map correspond to the bordered maps to the right. (Map courtesy Purdue University, USGS, ArcGIS)

Purdue's work in collecting and building agronomic data extends abroad to help farmers in other countries produce more food. Phillip Owens, an associate professor of agronomy, is heading a three-year project to map the farm soils of the entire Central American countries of El Salvador, Honduras and Nicaragua—areas comprising 143,400 square miles, about four times the size of Indiana.

The research, conducted in collaboration with Catholic Relief Services and regional and national partners, is funded by a $5 million grant from the Howard G. Buffett Foundation. The objective is to assess soil fertility down to the farm level, indeed even to sections of individual fields, with the goal of reducing application of fertilizers and helping farmers decide which fertilizers would be best for their crops. That, Owens says, would help farmers weigh production costs against possible yields—inputs versus outputs.

Using historical information in the form of soil characteristics records, geology, digitized maps of land elevations and locals' understanding of soil functions, he and postdoctoral researcher Jenette Ashtekar of Cleveland and research soil scientist Minerva Dorantes of Chicago are developing practical field-scale maps of soil properties. Based on the data they analyze, they will produce a soil suitability index that will help farmers know not only which crops would grow best on their land but also where there is greatest potential for soil erosion, where water on their farm is most and least available and how water is likely to drain.

Possibly the greatest contribution of this project stems from training local people to use and continually update their countries' maps.

"We're putting big data into the hands of small farmers," Owens says. "We're setting a platform for better economic growth and to continue to improve their countries."

But convincing farmers that they should change their ways based on new, high-tech research could be difficult because their livelihoods are engrained through generations of experience.

"People don't like to experiment with their own livelihoods," Owens says. "In the end, it will always take an expert to analyze the new information and make recommendations to them." (See "Purdue Ag Economist: Give 'Big Insights.'")

For the Greater Good

Information in the form of data from individual research projects already conducted also could be very helpful to the agriculture industry, policymakers and researchers if it were readily available. Agronomy professor Sylvie Brouder was an organizer of a meeting in the Washington, D.C., area in 2013 on the topic of developing a system to provide open access to agricultural research data.

Brouder says data from her research on crop nitrogen responses, coupled with research conducted by others, could help modelers project how nitrogen applications would affect the environment of an entire watershed without conducting a full series of their own field experiments.

"The theory is that my data is more useful in aggregate with your data rather than as stand-alone," Brouder says. "Traditionally, data in agricultural research have no purpose beyond the current research. But it really does have value beyond that one experiment."

Tool for Farm Management

Dennis Buckmaster, a professor of agricultural and biological engineering, says farmers also would benefit from data that could help them know how to best market their crops, such as by assessing the demand during and after harvest, and determining what percentage of the crop they might sell at harvest and store on the farm for selling later, and whether to ship it by truck or rail.

Buckmaster, who co-coordinates Purdue's Agricultural Systems Management program, last year helped to create the Open Ag Data Alliance, an industry organization including Purdue's Open Ag Technology Group. Part of the alliance's mission is to develop protocols to make data systems more compatible.

"Data only becomes 'big' when you gather it and layer it into something useful," he says. "Otherwise, it's just pockets of information.

"To be more useful, data needs to move among systems and be exchanged from one software package to another. But farmers should control how this happens. Controlled systems which aggregate and share information with trusted parties are being developed quickly."



Purdue Ag Economist: Give 'Big Insights'

Tractor CabScott Downey.

Associate professor of agricultural economics Scott Downey has a tip for farmers:

"Focus on big insights, not big data."

Downey wrote a blog item ("What's all the fuss about big data?") in which he says Purdue University research shows that farmers overwhelmingly trust local supplier representatives more than they do manufacturers or specific brands.

Downey, an associate director of Purdue's Center for Food and Agricultural Business, also says some local suppliers have seen that their business is changing from selling tangible products to providing an intangible, information-based service that brings specific insights to the farmers who trust them with the data.

He says local companies will have an advantage if they get good at analyzing data aggregated from hundreds of square miles of farmland—not just from one individual farm.

"Local dealers and retailers should be talking about how they can cooperate with farmers to find the best farming practices for the farmers in their area—using local soil types, water availability and genetics from a wider geography than any one farmer could have on their own.

"The local supplier who brings this type of value will justify the trust their farm customers have placed in them."

Whether at the manufacturer level or the local level, Downey says "big data will be driving agriculture, and there's a race to control access to it."

"The winner in that race will be those organizations that figure out that big insights are way more important than the big data."