Drones match farm planning effectiveness of more expensive tech, study finds

UNIVERSITY PARK, Pa. — Environmental scientists and water resource managers need precise, high-resolution maps to reveal areas that farmers should avoid when planting crops, to limit polluting waters with phosphorus from fertilizer or manure. Making those maps has depended on an expensive, sometimes unavailable technology, but a team led by Penn State researchers has developed a cheaper approach that can be just as effective.

The researchers’ novel system, detailed in a paper available online ahead of publication in the June issue of Computers and Electronics in Agriculture, uses drones and photogrammetry, a technology that develops reliable 3D spatial information by analyzing overlapping 2D photographs. With this system, the team can map hydrologically sensitive areas — locations where water tends to collect or flow, creating high runoff risk — and phosphorus critical source areas, where phosphorus is likely to wash into streams and pollute them. They found that the drone-photogrammetry approach was cheaper, more accessible and nearly as accurate as conventional mapping.

The team tested the accuracy and resolution of maps created with the new method against maps made using a technology called LiDAR, which stands for light detection and ranging. It is a remote-sensing technology deployed from aircraft or satellites that uses laser pulses to measure distances to the Earth, creating precise, high-resolution maps. LiDAR is accurate but expensive and not always accessible, according to study co-author and team leader Patrick Drohan, professor of pedology in Penn State’s College of Agricultural Sciences.

“Our new technique uses a small drone to take hundreds and hundreds of photographs, essentially duplicating what a LiDAR model does,” Drohan said, explaining that the LiDAR model is only as accurate as the most recent data obtained from an overflight, but the new approach can obtain new data as needed. “We employed a technique called ‘Structure from Motion’ photogrammetry — stitching photographs together that are taken from many different angles and slightly different positioning to create a three-dimensional surface of a landscape.”

The drone-based approach could allow water resource managers worried about sediment and nutrient pollution to analyze agricultural landscapes even if a LiDAR overflight has not occurred in recent years, or if landscape alterations have happened since the last LiDAR flight, Drohan pointed out.

“It doesn't take very long to fly over a typical size farm in Pennsylvania, so this is a way that we can more rapidly update areas that might be being targeted for best-management practice implementation, or a property that is eligible for financial assistance to install some type of runoff attenuation feature, such as a riparian buffer,” he said. “Most phosphorus losses originate from a small proportion of watershed areas, following the established 80:20 rule where approximately 80% of phosphorus losses originate from 20% of watershed area.”

The team studied four farm sites in eastern Pennsylvania and created elevation models from drone imagery and compared them to existing LiDAR data from 2017. They checked accuracy using 400 to 1,000 ground control points per site. Then the researchers used both datasets to map hydrologically sensitive areas and phosphorus critical source areas.

They found that drone maps matched LiDAR very closely. In elevation accuracy, the correlation was 0.999 — almost perfect. In the mapping of hydrologically sensitive areas and phosphorus critical source areas, maps from drones and LiDAR were nearly identical, with differences less than 1.53%.

“The drone method gives almost the same answers as LiDAR, meaning that drones plus structure from motion photogrammetry are a viable, cheaper alternative to LiDAR,” Drohan said. “Our approach can be used for farm planning, reducing nutrient runoff and environmental protection. Instead of paying for expensive LiDAR scans, farmers and researchers can use drones and photogrammetry to map runoff and pollution-risk areas with nearly the same accuracy, making precision agriculture more accessible.”

The research was conducted within the U.S. Department of Agriculture’s experimental watershed in Northumberland County, a sub-catchment of the Mahantango Creek Watershed that ultimately drains to the Chesapeake Bay.

Study first author Jhony Armando Benavides-Bolaños, a researcher and professor at the Universidad del Valle in Cali, Colombia, earned his doctoral degree in soil science and international agriculture and development at Penn State. He was advised by Drohan in the Department of Ecosystem Science and Management.

Contributing to the research were Daniel Guarín, Carboneers, Utrecht, Netherlands; Dimitrios Bolkas, associate professor of surveying engineering at Penn State Wilkes Barre; and Alejandro Pérez Y Soto-Domínguez, Universidad Nacional de Colombia.

This work was funded by the Pennsylvania Department of Agriculture, and the Broadening Extension Through Student Training program, the Harrar Scholarship, and the International Agriculture and Development Competitive Grant Program Award.