Optimizing Variable Rate Seeding for New York State Farmers

It’s all about the algorithm. There’s no shortage of opportunities to collect data on today’s farms. Topography measurements, soil types, soil nutrients and crop yields can all be mapped on a foot by foot basis. And the tools that will allow a farmer to customize production practices are becoming fairly widespread. The challenge? How do you turn the data into actionable information, and profit, for a farmer.

This article was originally published in NYFVI’s 2016 annual report. Additional work has been funded by both NYFVI and the New York Corn & Soybean Growers Association.

The project, led by Savanna Crossman at the New York Corn and Soybean Growers Association, has been gathering data from farms across the State to create a New York specific algorithm for variable rate seeding. The model, drawing on all the data collected through the research, identifies optimum hybrid/variety placement, and plant population in corn and soybeans given varying soil and climate conditions.

Corn and soybean data have been gathered on over 2500 acres across three growing seasons on fourteen farms. The dataset includes six major data types: seeding rate, type of hybrid, topographical information, NRCS soil survey maps, Veris soil sampling data and grid soil sampling data. Each data type consists of many variables which are analyzed individually and as interacting networks.

To better understand which variable has the greatest effect on yield, the project is using many different statistical approaches, including random forest regression. It showed that while the drivers of yield may vary given the field, crop or year, the key variable tends to remain the same in each field over time.  It is important to see stable relationships over time in order to write a planting prescription that the team and the farmer will have confidence in.  

In 2016, five farms began testing the prescription maps on over 500 acres. Baseline data were also collected on an additional 1600 acres to add to the dataset. The team also started to test hybrids to understand which hybrid will perform best in each of the varying soil types.

As the work progresses, a farmer will be able to link his farm’s soil data to the project model and generate a prescription map, providing customized guidance on planting depth and seeding rate for his fields. The ability to manage fields at a sub-field level using the algorithm will increase productivity, profitability and help farmers be good stewards of the land.

Results from the 2016 validation tests are still being compiled, but the algorithm has predicted to increase profit by up to $60/acre depending on the field and hybrid.

The project team includes five farmers and agribusiness professionals, as well as a PhD student from Cornell University.  The project also receives financial support from DuPont Pioneer, New York State Aid to Localities Appropriations , and the Soybean Check-Off program.

“I think this project will increase adoption of precision techniques and equipment/software by bridging the gap between equipment capabilities and agronomic knowledge.”

Rodman Lott
R. Lott & Sons, LLC
Seneca Falls, New York

Leave a Reply

Your email address will not be published.

This site uses Akismet to reduce spam. Learn how your comment data is processed.