Precision Farming Solutions for Weeds in Soybean Farms

The weed control measures used in soybean production need to be overhauled. The development of resistance in weeds has made current technology and management increasingly expensive and comes accompanied with heavy yield losses. Precision management solutions can reduce chemical use in several ways.

Current Weed Management in Soybeans

Soybean is important for the global economy as it is used as a source for oil and protein (for people and livestock) and as biodiesel.

Soybean is the crop with the largest monoculture in the world and with 90% of it grown in Argentina, Brazil, the United States, China, and India.

Current Technology

Weeds are the number one problem in soybean cultivation in all these countries and are estimated to result in a 37% reduction in yields on average.

The current measures are clearly not working for soybean farmers.

The main reason for this is the development of resistance in weeds. Worldwide, regardless of the weeds species and the soybean cultivars used, farmers use only a few broad-spectrum herbicides and similar agricultural technology in the monoculture.

In the USA more than 93% of soybean crops use GR (glyphosate-resistant) technology, and in Brazil and Argentina, it is 80% and 99%, respectively. Roundup or glyphosate is used as the herbicide to control weeds growing in these GMO crop fields.

GR Technology Creates Resistance

Since glyphosate is a broad-spectrum herbicide also used for many other crops and weeds, it is extensively and frequently used over large areas. This has led to the development of chemical resistance in the weeds. So, farmers spend an increasing amount of herbicides and are yet not able to control the negative impact of weeds on their soybean yield.

Of the top ten weeds found in soybean fields in the USA, half have already developed resistance to herbicides, and they are the

• Common waterhemp
• Giant ragweed
• Lamb quarters
• Horseweed/ marestail
• Common sunflower

The other five weeds are Velvet leaf, Giant foxtail, Woolly cupgrass, Burcucumber, and Annual morning glory.

Precision Farming Methods can Reduce Reliance on Chemicals

Research to find new herbicides, which can control the resistant weeds, would take years to develop and these new treatments will also ultimately result in resistance. Hence, it is necessary to reduce the reliance on chemicals completely or as much as possible.

One of the solutions is the wider use of precision farming methods based on site-specific control.

Drew Lyon, an extension weed specialist from the Washington State University, says, “The new site-specific weed management technology is currently popular in high-value crops.”

Figure 1: Weed treatment can be restricted to patches, grids in patches, or individual plants, Christensen et al. 2009. (Source: Site‐specific weed control technologies)

Instead of treating the whole soybean farm, it is possible to treat only patches/subfields, grids in patches, or target individual weed plants to control them, as shown in Figure 1. The amount of herbicide or treatment is also proportional to the weed density using variable rate application.

To use precision farming methods, we need

1. Weed scouting systems
2. Weed management models
3. Precision weed control implementation

There are numerous methods and approaches used. Recommendations depend on the scale at which weed implementation will be carried out in soybean farms.

Weed Scouting

Figure 2: Using cameras of different light wavelengths for weed identification, Martin Weiss et al. 2008. (Source: Precision farming for weed management: techniques)

Instead of walking in your fields, use precise and rapid technology. Essentially, cameras using various wavelengths of light such as visible (red, green, and blue), infra-red, or near-infrared are used to take a picture or see the conditions on the field. A combination of the wavelengths is also sometimes used to scout, identify, and locate weeds.

The camera differentiates weeds from the soil and the soybean, based on color, shape, and size of weeds and crops, as shown in Figure 2.

• In patch and grid level treatments, an image of the field is taken by remote sensing by drones that are flown over the fields by farmers or by drone operators. Satellite imagery uses the same technology.
• In spot treatment where individual weed plants are treated, sensors attached to the boom feed the images directly to the computer to help in decision making.

Weed Management Models

During scouting by drones, many photos are taken, that are “stitched” using stitching software to get a composite map of the field, which is then analyzed.

In spot application, the pictures of the ground are analyzed in real-time.

Figure 3: The images a–e, show maps derived from remote sensing on (a) soil classification, (b) weed distribution map, (c) weed control recommendations, (d) application map, and (e) yield map (as a result of soil quality and weed control), Weiss et al. 2008. (Source: Precision farming for weed management: techniques)

A combination of computers, artificial  intelligence (or decision-making algorithms), and machine learning analyze the images to identify the weeds. Machine learning uses technology that allows facial recognition. When thousands of images of a certain weed at various growth stages are shown, the software learns to identify a weed.

This technology can also count weeds in a given area, for example per square meter to find out the intensity of infestation. Then using threshold levels based on the crop and weed type, the intensity is calculated creating treatment models. For example, Figure 3 b shows a model with four zones,

• Grids with no weeds
• Low-density grids with 10-25 weeds/m²
• Medium-density grids with 25-60 weeds/m²
• High-density grids where there are more than 60 weeds/m²

Specific weed maps can be created for each weed or a group (grasses or broad-leaved species) as in Figure 3c.

As it is possible to detect different species of weeds from drone images as well as sensors on application booms, species-specific herbicides can be used, reducing reliance on any single chemical.

Precision Weed Control Implementation

Farmers can use insights on weed identity, density, and location, to control weeds, with either chemical or mechanical treatments.

In either case, farmers can opt to use conventional technology, and still cut costs by reducing herbicide use, since areas with no weeds get no treatment at all. This can reduce areas to be sprayed by around 60% or even more, by targeting only the weed patches.

However, there is advanced variable rate application technology already on the market.

Chemical Method

The smaller the treated area, the greater is the reduction in herbicide use.

“Several weed sensing systems and precision implements have been developed over the last two decades,” says Dan Crummett. “The lack of knowledge about the economic and environmental potential for increasing the resolution of weed control,” remains an important barrier in increasing the resolution of weed control, he explains.

Weed maps depicting the occurrence of one or more types of weeds are used to develop prescriptive maps, using economical thresholds. The treatment is adjusted based on intensity, so not only is a treatment restricted where necessary, but even the amount of herbicides can be varied, resulting in huge savings in chemical use.

Depending on the scale of application there are many technologies available on the market.