Dr. Antonio Ray Asebedo, Assistant Professor of Precision Agriculture at Kansas State University (KSU), has been studying the combination of drones and multispectral cameras in an effort to improve recommendations for nitrogen management.
“Right now I see them as having the potential to be a very valuable tool for use in agronomy for making nitrogen recommendations. Variable rate nitrogen recommendations cost farmers a lot of money. With an appropriate algorithm or a brain, if you will, we’re trying to give that drone an agronomic PhD so that it can help you decide the yield for your farm and also what kind of fertility program you need to run.”
“With an appropriate algorithm (or a brain, if you will), we’re trying to give that drone an agronomic PhD.
— Dr. Asebedo
He tested the MicaSense series RedEdge and found that “the selection of the five bands: blue, green, red, red edge, and near infrared are ideal for nitrogen recommendation algorithms.”
Most of KSU’s existing algorithms were created for active optical sensors, and Dr. Asebedo not only wanted to see if those algorithms would work for RedEdge, but also to see if he could make optimized nitrogen recommendation algorithms specifically for the camera.
“These algorithms will not only help determine the rate, but also the optimal time. The timing of nitrogen applications is key. If you really want to boost yields and improve nitrogen use efficiency, timing is probably just as important, if not more important than rate. This system, using drones and crop monitoring, lends itself to an algorithm that can take advantage of that temporal resolution and help determine the right time to apply.”
He studied five locations in Kansas, applying different levels of treatment to separate plots in each location. Multiple times throughout the growing season, he collected soil samples, scanned the plots with a ground-based active optical sensor, and captured drone-based aerial multispectral imagery with RedEdge.
He used the multispectral images to produce multiple analytical map layers of the area, including NDVI. He then compared the NDVI with another NDVI map derived from the ground-based active optical sensor and made recommendations using one of KSU’s standard algorithms. When the crops matured, he harvested with a plot combine, estimated grain yield, and evaluated grain moisture as well as nitrogen concentration.
“If you really want to boost yields and improve nitrogen use efficiency, timing is probably just as important, if not more important than rate.
— Dr. Asebedo
He found that collecting multiple times at key growth stages provides for better assessment, reducing application rates without sacrificing yield. He also discovered that data captured with RedEdge could indeed be used in algorithms originally designed for active optical sensors. It turns out that RedEdge and the active optical sensor produced similar NDVI which were normalized across time.
“A good way to think about variable rate nitrogen is to take a big field and break it down into many small fields. You can fly over the field with your multispectral camera and you can then determine how to break the field up. We’re not just talking about symmetrical polygons. We let the crop dictate how to break up our map based on its interaction with the soil; imagery lends itself better to this.”
“This is site-specific agriculture: you can address local soil variations, crop interactions, and the impacts of weather too. By addressing specific areas in a field by their individual needs, we can make more optimized nitrogen recommendations in comparison to blanket nitrogen applications that are typically applied.”
Combining RedEdge data with current KSU algorithms significantly increased agronomic efficiency. The RedEdge/KSU intensive management treatment applied the least amount of nitrogen and better evaluated the nitrogen needs of the winter wheat in comparison to single-application blanket treatments applied in the early spring.
To take full advantage of RedEdge data, Dr. Asebedo is working on custom algorithms as he continues his research in the 2016–17 season. The primary goal is to achieve agronomic efficiencies between 60–70% without yield reduction.
“The end goal for these algorithms and systems is to produce more yield on less nitrogen. Which means that the farmer is going to make more profit per acre, produce a higher quality grain, and more of it. Which means that we help feed a world of 9 billion people by 2050. And because we’re using less nitrogen and applying it when the crop is most efficient at taking it up, we’re reducing the environmental impact too, which plays into making a sustainable agricultural system. So it winds up being a win-win for everybody.”
