Case Study: Wild Oat control efficiency using UAV imagery

Case Study: Wild Oat control efficiency using UAV imagery


wildoatsAccording to government sources, wild oats are considered to be one of the most prominent and costly grassy weed that affects fields across the Prairies.  By impacting yields, lowering grades, and more factors, it ends up costing farmers as much as $500 million per year. (,wild-oats.html)

By competing for nutrients, moisture, and sunlight, wild oats can quickly choke out intended crops such as canola, barley, wheat, and flax if left unchecked.  While good seeding practices can help to control this weed’s impact on yield, most producers choose to use herbicide applications at various times of the year.  However, these additional applications can significantly increase costs by as much as $12-$16/ac.

In an effort to minimize spray application costs for a particular field (seeded in Canola), a producer approached AgSky Technologies Inc. and requested UAV imagery of his field for the purpose of identifying areas most affected by wild oats.  AgSky arrived on-field and began mission preparations according to a previously-prepared flight plan.  It was determined that multi-spectral imagery would be used to capture as much information as possible and improve the data quality.

Flight Specifications and Conditions

Using our eBee Ag UAV from senseFly, flights were conducted on the subject field with weather at 26°C, overcast cloud cover, and 4-6m/s average wind speed from South direction.

Two cameras were used as image sensor payloads, one with the RGB True-Color Canon IXUS/ELPH, and the second with a NIR modified Canon S110.  Flight 1 captured 257 16.0mp RGB images at approximately 4.1 cm/px, covering 237 acres with about 70% image overlap.  Flight 2 captured 366 12.0mp NIR images at about 80% overlap, totaling 242 acres at 4.2 cm/px.

Several additional images were captured by hand from the ground to help assessment and comparison before leaving the field.

Raw image maps

NIROrthoRGBOrthoBack at the office, several initial maps were generated using Postflight Terra 3D v3.4.46 for photogrammetry and orthomosaicing.  The maps were cut down to about 179 acres of the immediate field area only, and transferred into GeoTIFF format for use in QGIS and other tools.

Initial review of the RGB orthomosaic showed green vegetation (likely wild oats) visible when examined at very close detail (approximately 4 cm/px), but was difficult to determine the affected areas when viewed at a greater distance.  This would be unsuitable for what the farmer wanted as manually noting the “green” areas of the True-color images would be very time consuming, however it was still an excellent confirmation layer.  The NIR orthomosaic would hold much more promise for easily identifying high-density vegetation areas once a NDVI (Normalized Differential Vegetation Index) was processed.

Both the RGB and NIR orthomosaics show effects of variable lighting resulting from broken cloud cover during the UAV capture flights. While this is much more apparent in the NIR imagery (light-dark striping across horizontal), the NDVI processing is able to compensate for the variable light conditions and significantly reduce impact on the final report.

NDVI Processing and Zoning

The next step in the field analysis was to create a region for the core crop area that would potentially be sprayed or otherwise treated.  Using tools built into Postflight Terra 3D (new since version 3.4, still relatively undocumented and beta), detailed NDVI “reflectance” maps were generated along with zoned SHP files.  This process could have easily been completed in other farm management software packages such as Ag Leader SMS Advanced, FarmWorks Office, Ag Data Mapping Solution, or manually via GIS software such as ArcMap, QGIS, etc.  Postflight’s tools have improved significantly and greatly streamlines the process of converting UAV imagery into Ag-specific reports and VR (Variable Rate) controller data.

Screenshot of Postflight Terra 3D v3.4.46

Screenshot of Postflight Terra 3D v3.4.46

From Postflight, a NDVI map was generated showing the index areas that had a high-density of vegetation (suspected to be wild oats, given the preliminary field scouting), and then transferred into a series of SHP polygons with rate data for the areas most affected (NDVI index of 0.09 or higher, for this particular field).  From this SHP data, a Raven or similar type of sprayer/floater VR controller should be able to apply product to the designated areas only, thereby reducing costs and saving time compared to applying to the entire field.


Image Comparison

To validate the findings of the NDVI, the typical process would be to visually inspect the field at particular areas, such as high-index and low-index areas to confirm the presence (or absence) of wild oats and other vegetation.  In this case study, it was decided to use photographs to visually record the field condition and then compare to the generated map(s).  A summary infographic is shown below.


From the time on-site and the captured photographs, it was clear that the high-index areas were made up primarily of wild oats.  Based on the known seeding date of the field crop and the plant/leaf stage, very little index density could be attributed to canola.  A colleague more experienced in agronomy reviewed the NDVI report and then visually scouted the field, confirming the proliferation of wild oats across the field.

Cost / Benefit Conclusions

With current chemical and application costs, it was projected that treating the entire field (136 acres) would cost between $2000-$2500 CAD.  By using the NDVI data and report analysis prepared by AgSky Technologies, the producer could potentially reduce the area for application down to 52 acres, saving as much as $1500 in application costs.  By accounting for the UAV flight and data report cost (approximately $600), the producer would still be able to save $900 net.  These savings would be further increased on larger field areas, as well as other beneficial uses of the captured data to increase crop yields.

As AgSky continues to review uses and applications of UAV imagery for producers, we will post additional case studies demonstrating real-life examples and the benefits that are available.  Without a doubt, drone technology presents a significant value to the agriculture industry.  For those who are ready to leverage this technology on their farm, an overview of our services is available here.

1 Comment

  1. Very insightful article. I am also a SenseFly eBee user. If you don’t mind, would you elaborate on – “Postflight’s tools have improved significantly and greatly streamlines the process of converting UAV imagery into Ag-specific reports and VR (Variable Rate) controller data.”

    Which specific tools were used in PostFlight to generate the NDVI, the Ag-specific report, and the Variable-Rate application controller data?

    I truly appreciate any help/insight you could provide. I’m hoping to apply a similar workflow to cotton/soybean/wheat farming in the Southeast.

    Madison Dixon