Guide to big data analytics tools, trends and best practices
A comprehensive collection of articles, videos and more, hand-picked by our editors
The Flint River Partnership is a government-funded nonprofit group in Cairo, Georgia, that's working to develop and promote the use of more efficient farm irrigation processes in a 27-county region of southwest Georgia. The region produces large quantities of peanuts, cotton and corn, and the partnership is looking to reap the benefits of new weather analytics capabilities based on big data tools and information from sensors and other sources.
The partnership -- which is jointly led by the Flint River Soil and Water Conservation District, the U.S. Department of Agriculture's Natural Resources Conservation Service and The Nature Conservancy -- teamed up in late 2013 with IBM and the University of Georgia on a big data project aimed at creating more accurate weather and soil-condition forecasts to help farmers in the Lower Flint River Basin optimize their irrigation plans.
The project involves the use of IBM's supercomputer-driven Deep Thunder analytics service to crunch data collected from thousands of in-field weather stations as well as satellite data and information from commercial weather data networks. The system generates highly localized forecasts that predict conditions at 10-minute intervals over a 72-hour period and are updated every 12 hours. Plans call for adding sensor data on soil temperatures and moisture levels and information collected by GPS-enabled farm equipment to the forecasting model this year.
At this point, the project is in trial mode, said David Reckford, director of the Flint River Partnership. He said the model will be tested on the ground during the course of the 2014 growing season to see how well it works "per farmer, per field" -- and whether it's more accurate than existing forecasting systems, including a cloud-based application that uses an analytical model developed by the USDA to support the development of irrigation schedules.
Big data tools could change forecasting game
The forecasting task is daunting: The goal is to pinpoint expected conditions in 1.5-square-kilometer areas across 550,000 acres of irrigated farmland. But Reckford said that if the big data system works properly, it could "put weather forecasting on steroids" and pay big economic and environmental dividends. "The implications of knowing the weather three days out in agriculture is a game-changer, I would argue," he said. "This really is a new way of seeing technology as a way to change how we do what we do."
David ReckfordDirector, Flint River Partnership
For example, farmers could use the sensor-driven forecasts to set variable-rate irrigation systems to water heavily in some spots, lightly in others and not at all elsewhere. In addition to reducing water consumption and irrigation costs, that potentially could help increase crop yields by avoiding overwatering or letting fields go dry because of faulty forecasts, Reckford said. It also could help keep farming viable in the region well into the future: "If our farmers here can use resources in the most efficient way, you have a more sustainable production system over the long term," he said.
Ironically, another possible benefit of the big data analytics project that Reckford envisions is a reduction in the number of soil-condition sensors that need to be installed in fields to collect data for use in forecasting. He said the sensor units can cost $3,500 to $5,000 to install and operate for a single growing season, including data transmission equipment and services. Currently, sensors typically are deployed at a rate of one per hundred acres, according to Reckford. One aim of the pilot project, he said, is to see if the forecasts produced by Deep Thunder can free up farmers to "get away from putting sensors in every field."
Wider reach for weather analytics effort?
Reckford said the project eventually could also have implications beyond southwest Georgia if makers of irrigation systems adopt similar analytics technologies and farmers elsewhere buy into the concept. Cost issues need to be addressed, though. For now, the big data initiative is primarily being funded through grants received by the Flint River Partnership. At some point, that would have to change to a cost-per-user model -- and one of the objectives of the trial run is to assess costs and tangible financial benefits to point the way toward a pricing structure for the weather analytics service. "Our goal is to show what can be done," Reckford said. "We help get things off the ground. Ultimately, this is going to have to be commercialized."
Another issue the partnership will examine as it tests the service is how best to deliver the forecast details to farmers in a way that's easy to access and understandable. For now, farmers have to log into a Web portal, wait for the forecast information to load and then view graphs, data tables or animations depicting expected weather conditions. "There's definitely a cool factor in being able to see all that information," Reckford said. But some busy farmers think it's burdensome to have to log into a system, he added. "They might find it easier to get a text."
Convincing farmers to trust the forecasts is another test point for the project, Reckford said, noting that many have "deep-seated frustration" over inaccurate weather forecasting. "If you're about to plant a crop and the weather forecast isn't right, you can be looking at significant issues or unnecessary delays," he said. "There are economic ramifications to this."
Craig Stedman asks:
What's your top piece of advice for organizations looking to take advantage of sensor data as part of big data analytics initiatives?
0 ResponsesJoin the Discussion