Weed detection in soybean crops & Variable-rate application

About the project

An agronomist opted for drone turnkey solutions to scout weeds in post-emergent soybean. Engaging a drone operator made the process of getting remote-sensed aerial imagery analyzed by Agremo software solutions convenient and effortless.

Manual scouting for weeds in a 95.43-acre farm cannot provide accurate estimates of the percentage of damage nor the exact location of infestations. Precision farming techniques that rely on drone imagery analysis are the simplest option for an agronomist to obtain objective and historical data on weeds.

Weeds are endemic and regenerate from seedbanks in the soil, so maps of an infestation can help in current control and future risk assessment.

The drone operator, who had previously used Agremo web app, chose to reuse this software service, given its

  • Simplicity: Anyone can use Agremo analyses and navigate the simple workflow provided by the company website
  • Accuracy: Agremo’s analytics are more accurate and reliable in comparison to others

After taking high-resolution drone images and stitching them to get a map, the drone operator uploaded it to the Agremo web app to use Weed Detection analysis.

The report showed that only 39% of the farm was weed-infested, while the analyzed shapefile, demarcated fine, potential, and weed stress zones. Our new tool helped the agronomist create a prescriptive variable spraying map based on the analyzed shapefile. The farmer uploaded this map to his John Deere tractor to spray affected areas correctly.

By reducing treatment to 39% of the total farm area, the farmer saved $1422.86 in second herbicide spray costs. Moreover, the overapplication of chemicals that lead to weed resistance and environmental damage was considerably diminished.

Customer requirements and challenges

Danielle Brown (name changed), an agronomist, was working with a farmer who was growing soybeans. The crop ranks second in importance in the USA, after corn, in terms of acreage. The agronomist hoped to optimize weed control through informed decisions and provide farmers with time to implement effective corrective measures. So, she contacted William Blake (name changed), a drone operator to assist her and to discuss methods of data collection.

Weeds, the number one scourge in soybean cultivation, can reduce yield by an average of 37%. Though the weed species that affect a field can be different, the control and management practices are similar. Repeated use of the same herbicides has built weed resistance and has produced super-weeds by affecting non-target wild plants.

Hence, Danielle had to control weeds with the minimum use of herbicides. At the same time, the reduction of spraying can boost farmers’ ROI.

To control weeds before they affected the current yield goals, Danielle wanted to use an early warning system that could:

Client needs

  • Identify weed infestations
  • Calculate the exact extent of the field infested with weeds
  • Locate the weed-infested areas

Existing process / Field condition

The traditional scouting method (walking) was not a feasible option, especially for the large monoculture farms that Danielle advised. The method was inefficient, inaccurate, subjective, and not suited for data management or resource use. To be more specific:

  • Inefficient: Entire large farms are difficult to monitor manually and reports would be incomplete.
  • Inaccurate: Visual assessment of weed impact on plant growth and weeds zones is subject to human error.
  • Not standardized: Manual scouting is subjective and will differ from person to person. Furthermore, it is especially challenging when it includes working with third-party contractors.
  • Data management: It is difficult to rerun or update analysis, or get a historic perspective of the field conditions.
  • Wasteful utilization of resources: Applying insurance sprays was a waste of resources and money. It was also unlikely to optimize yield.

William had experience with drone imagery analysis for weed stress and normally advised a whole crop monitoring for best results.

The farmer Danielle had already applied the pre-emergence spray for the entire field of 95.43 acres of soybeans, following traditional practices. However, weeds can interfere with soybean establishment by competing for nutrients, water, and light, up to 35 days after crop emergence, and thereby impact yield.

How Agremo approached the challenge

William recommended Agremo software’s weed detection analysis as the best solution for the post-emergence soybean weed scouting. Agremo had an analysis program designed for various phases of crop growth, which were conducted in collaboration with professional agronomists.

The agronomist agreed to use the post-emergence imagery analysis by Agremo when William explained his reasons for choosing Agremo software solutions:

Agremo software analyses combine technology and human intelligence and go beyond NDVI index-based analysis of chlorophyll levels of plants. We use several sources, such as spectral data, and the form and size of plants. Agremo analysis algorithms were developed over the years using real customer data from thousands of fields of over 100 species. Hence, green weeds cannot be mistaken for crops.

Our unique technology relies on Artificial Intelligence, Machine Learning, and Computer Vision to identify, classify, and quantify complex spatial patterns within image data, giving accurate analysis results that others cannot match.

Design and Plan

Collect data

Analyze data

Deliver and Apply

The process and the solution

As he had prior experience with Agremo, William offered to get the remote-sensed aerial imagery and analyze them. First, he registered a new account for the farmer, so they could conduct follow-ups if necessary. Then, he completed the familiar three-step process:

The first step - Get farm photos

With his drones, William took several pictures of the farm, making sure they had an overlap.

The second step - Prepare a map

Using his own account at DroneDeploy, he stitched the photos to get a seamless map of the entire soybean field.

The third step - Upload 2D map

The prepared map was then uploaded onto the Agremo web app. William chose weed analysis and entered crop details and its growth stage.

The fourth step - Results

He received the PDF report and an analyzed map as a shapefile, with three weed zones demarcated by color: fine or no stress zones (green), potential stress zones (yellow), weed stress zones (red). Only 39% (37.22 acres) of the 95,43 acres showed potential and weed stress and needed to be treated. The remaining 58.21 acres were fine

Try Agremo Analytics

Free demo - no credit card required

What were the decisions informed by Agremo analysis?

William immediately shared the shapefile and document with Danielle, so that she could advise the farmer on the proactive measures he had to take.

Danielle, in turn, shared the Agremo weed analysis report with the farmer and showed him the exact places on his farm that were infected. She advised him to spray only the affected 37.22 acres to cut costs and prevent environmental damage.

With the help of the Agremo tool, the agronomist made a variable spraying map.

The farmer uploaded the prescriptive map to his John Deere tractor, which could read it. During the spraying process, the farmer could easily find the problem spots and apply the correct amounts. Thus, Danielle helped the farmer:

  • Optimize weed control
  • Reduce the use of herbicides and wear-and-tear of his equipment
  • Spray only weed-affected areas
  • Cut the costs of herbicides
  • Minimize over-application of herbicides, which in turn reduced development of herbicide resistance in weeds, runoff, and leaching into water sources
  • Less impact on non-target species – both plants and animals

Return on Investment

Danielle and William helped the farmer improve his profits and still maintain optimum yields, by integrating drones into his day-to-day workflows. The powerful data they provided him made once-difficult tasks simple.

During the first pre-emergence herbicide treatment, before performing Agremo weed analysis, the farmer spent $22 per acre as he sprayed the entire field (see Table 1).

The weed analysis report showed the farmer could restrict his second treatment to only 39% of his farm. The prescriptive map further helped calculate the exact amount of herbicides he needed. The agronomist recommended two sprays to treat the broadleaved and grass weeds separately, which cost the farmer $17 per acre and $14 per acre, respectively.

The drone scouting-turnkey services cost $4 per acre. Yet, the farmer saved 14.91 $/acre by not spraying 61% of his farm, which had no weeds.

Relying on traditional farming approaches and spraying the whole farm would have cost him $61.00 per acre. Informed decisions made using precision farming methods reduced his costs to $46.09 per acre. That amounted to a saving of $1,422.86 for the entire farm.

Data has to be quick, comprehensive, and highly accurate to be useful in agricultural operations. Growers can then perform fewer unnecessary measures and add more efficient practices, and the overall ROI will be quickly visible.

Agremo services have been designed for use by agronomists, drone operators, or directly by farmers.

Analyze your field

Free demo - no credit card required