This research project was conducted by Ms. Katia Rizzardi at the University of Georgia during her TAPAC internship experience. The project formed the foundation for Ms. Rizzardi’s M.S. thesis which was awarded by the Università degli Studi di Padova. Ms. Rizzardi’s major professor was Prof. Francesco Morari. Also on her advisory committee were Dr. George Vellidis, Dr. Eric Prostko from the University of Georgia and Dr. Ivan Sartorato from the Università degli Studi di Padova. Download Katia’s M.S. thesis here.
Anyone who has seen weeds growing in fields knows that they are not evenly distributed. Over the past few years, a few studies have been done in Georgia and other states showing that significant reductions in herbicide use can be achieved by spot spraying weeds rather than treating them with conventional broadcast spraying. We know that with the reduction in herbicide use, there is also a reduction in herbicide cost. What we don’t know is whether this reduction in herbicide cost can offset the cost of automating spot spraying. Spot spraying is also known as variable rate application (VRA).
To answer these questions specifically for peanuts, we began a small study at the University of Georgia’s Ponder Farm near Tifton during the 2006 growing season. Our objective was to evaluate the technology while also accounting for the costs and benefits of using VRA. We established four treatments: conventional weed control, manual weed control, 100% VRA, and a mix of conventional and VRA.
For VRA we selected two different technologies. The first was the commercially available WeedSeeker® hooded sprayer (NTech Industries, Inc., Ukiah, CA). The WeedSeeker® uses advanced optics and computer circuitry to sense the presence of a weed. When a weed enters the WeedSeeker® sensor’s 12″-wide field of view, it signals a spray nozzle to deliver a precise amount of herbicide. The WeedSeeker® was used to control weeds in the alleyways between the peanut beds throughout the season and between the rows of peanuts within the bed early in the growing season (Figure 1). Once the peanuts began covering the bare soil of the bed, we controlled weeds emerging above the peanut canopy with a height-selective sprayer.
The height-selective sprayer was a prototype developed in-house and consequently went through several revisions during the growing season. To trigger spraying, we used an infrared sensor. The sensor consisted of an emitter and a receiver. The emitter sent an infrared beam over a distance of up to 40″ to the receiver. If the beam was interrupted by a weed, hand, or other obstacle, the receiver transmitted a signal indicating an interruption. We captured this signal with a data logger and used the data logger to control solenoids which applied a short burst of herbicide. For field use, the height-selective sensor was installed on the same tool bar as the WeedSeeker® sensors. The height of infrared sensor above the canopy was adjustable to allow for plant growth (Figure 1).
Our 100% VRA treatment relied on the height-selective sprayer to control weeds emerging from within the peanut rows. Because of technical difficulties, we were not able to deploy our height-selective sensor until about a month after weeds emerged above the peanut canopy. Once deployed, the height-selective sensor performed well. But, despite good performance from the sensor, we were not able to control weeds because they had become so large that we were not able to effectively apply herbicide to them from our tool bar. Good weed control was achieved between the rows and in the alleyways.
Our mix of conventional and VRA used a banded application of post-emergent herbicides over the peanut rows during the first month after planting. Weed control between the rows and the alleyways was done with the WeedSeeker®. This treatment resulted in very good weed control and did not require the use of the height-selective sensor later in the season.
To evaluate the four treatments, we collected data on the effectiveness of the WeedSeeker® and the height-selective sensor in recognizing and killing weeds (Figure 2) as well as kept a careful record of operating costs and the cost of herbicides for each treatment. At harvest, yields were collected and measured from each treatment. Statistical comparisons were made between conventional and the mix of conventional and VRA treatments. Because of the technical difficulties, a comparison of 100% VRA will have to wait for a future project. The complete results can be found in Ms. Rizzardi’s M.S. thesis (download thesis).