This article appears in the January 2021 issue of Potato Grower.
During certain stretches of the growing season, potato growers must monitor the nitrogen status of their crop on a regular basis in order to fertilize in the most efficient and sustainable way possible. This is often done by collecting petioles—the part of the plant that connects leaflets to stems—from numerous plants in each field, and mailing them in for a fast-turnaround lab analysis for nitrate. Within a few days, growers receive results indicating whether more nitrogen fertilizer is needed.
The system works, but it could be better, says Yi Wang, assistant professor and extension sustainable vegetable production specialist in the University of Wisconsin–Madison Department of Horticulture.
“There are downsides to this common approach,” says Wang, whose research and outreach efforts focus on the needs of Wisconsin’s potato and vegetable growers. “Collecting the petioles is time-consuming and labor-intensive, and sometimes the results can be misleading because a lot of factors can affect petiole nitrate numbers, such as weather conditions or the time of day of sample collection. Plus, the results don’t catch spatial variation [of nitrogen needs] within the field.”
To address these issues, Wang is leading an effort to develop a set of high-tech tools that will give potato growers a potentially easier, faster and more comprehensive way to assess the true nitrogen needs of their crops. The project, funded by the USDA National Institute of Food and Agriculture, involves collecting and processing data from a hyperspectral camera—mounted to a UAV or low-flying airplane—flown over potato research plots grown at different nitrogen levels, and then developing computer-assisted models to link the imagery with in-season plant nitrogen status and end-of-season yield, quality and economic return.
“The ultimate goal of the project is to assist potato growers with their nitrogen management using a platform that blankets the entire field in a timely manner, unlike the traditional petiole nitrate testing,” says Wang. “My collaborators and I hope to develop an online program that will translate the hyperspectral images into information about when to apply fertilizer and how much to apply, so that maximum profitability can be achieved for the growers with minimum environmental impacts.”
Hyperspectral cameras are powerful pieces of equipment, able to capture images that detect hundreds or thousands of spectral bands of sunlight reflected from the crop canopy, says Trevor Crosby, a graduate student working on the project in Wang’s laboratory.
“Factors that cause variation in canopy health—such as nutrient status, water status or disease pressures—are all related to the spectral reflectance and therefore can be visualized in the hyperspectral images,” he says. “We use image processing to extract the most useful information for our research project.”
And there’s certainly a pile of data to process. One flight over a 70-by-150-meter research field can collect dozens of images, each with hundreds of spectral bands. It takes a lot of time to crunch the resultant data, so the research team is looking to expedite the image processing.
The challenges of this complex project led Wang to bring on two key collaborators. Phil Townsend, professor in UW–Madison Department of Forest and Wildlife Ecology, is a national leader in utilizing remote sensing technologies. Paul Mitchell, professor and extension specialist in the Department of Agricultural and Applied Economics, will help with the economic analysis that informs the computer model’s nitrogen application recommendations.
“Dr. Townsend’s group has created a program that makes image processing really efficient,” says Wang. “We are very excited about plugging into that.”
Crosby is taking the lead on collecting ground measurements for the project, gathering a wide array of data from the field research plots at different potato growth stages, including leaf area index, leaf and vine total nitrogen content, and environmental factors such as soil moisture and temperature, solar radiation and wind speed. At harvest, he will measure total tuber yield and size profile.
Crosby will then go on to develop advanced models to link the hyperspectral imagery with the ground measurements. There are two steps of modeling. First, Crosby will use the in-season imagery to predict real-time crop nitrogen status. This work will be co-advised by Wang and Townsend. Second, with guidance from Mitchell, Crosby will use the modeled in-season crop nitrogen status, together with the environmental factors data, to predict end-of-season tuber productivity and economic return.
“With all the issues in [Wisconsin] around nitrates in groundwater, we need to find ways to make better use of our fertility inputs, and we are hopeful that Yi’s new project can help direct those efforts,” says Andy Diercks, a fourth-generation potato grower at Coloma Farms, LLC. “The potential is significant. Yi’s new project represents an opportunity to really leap forward [in nitrogen management].”
Wang shares the findings of her research efforts widely with the state’s potato and vegetable growers via the UW Vegetable Crop Update e-newsletter, grower meetings, farm visits, field days and her YouTube channel. She has developed good working relationships with farmers around the state, and many eagerly await her research findings and applications.
“Hyperspectral imaging has the potential to show the plant’s response to deficiencies in inputs before the human eye can see that response,” says Diercks. “If we can gain a few days in responding to nutrient stress, the impact to the health of the plants would be quite significant, not to mention the possibility of using fewer inputs to remedy the situation—which would be a serious win-win.”
Nicole Miller is the news manager for the University of Wisconsin–Madison’s College of Agricultural and Life Sciences.