Top Crop Manager

Features Precision Ag
Drone mapping targets kochia patches

Mapping provides the basis of site specific weed control.

March 5, 2024  By Bruce Barker

If ever there was a weed that called out for a targeted approach for control, kochia is the one. Prolific, with multiple flushes throughout the growing season, stress-tolerant, genetically diverse and resistant to four different herbicide Groups, including the mainstays of control – glyphosate, dicamba and fluroxypyr. Yet kochia is not very crop competitive, and so is often found in marginal areas of a field that are saline or around barriers in a field like power lines and wellheads, and even in hybrid canola production fields in strips where the male plants are mowed down.

“Kochia is a big concern right now. We wanted to look into whether we could manage it with variable-rate or site-specific strategies,” says Lewis Baarda, field-tested manager with Farming Smarter at Lethbridge, Alta. “We thought if we could narrow down the scope and manage kochia in 10 or 20 or 30-acre areas instead of the entire field, maybe it would be a little easier for growers to deploy the tools on the toolbelt to tackle it.” 

A weed mapping and kochia management research project was set up with 75 per cent of funding coming from the Canadian Agricultural Partnership, and the remainder self-funded by Farming Smarter. FMC Canada also provided support and AJM Seeds provided field imagery. The project ran from 2019 through 2023. The objectives were to investigate tools for mapping kochia growth zones in a field and to evaluate site-specific kochia management using different control methods.


Farmer co-operators with three fields in southern Alberta at Burdett, Medicine Hat and Scandia were selected for the project. Soil samples, electrical conductivity, weed counts and yield maps were collected for each field. Drone imagery was used to map kochia growth in the fall.

Two processing techniques imagery types were used in the project. One used NDVI, which is a common index that uses the Red and NIR (near-infrared) portions of the spectrum to identify greenness.

The second was excess greenness (ExG). Excess greenness is a similar index but is computed differently. It uses band arithmetic to determine greenness in a way that doesn’t require an NIR band. You could use RGB bands to compute this. The most basic form of imagery gathering would be RGB (red, blue, green). 

“Like TVs that use those three colours to produce all the colours that you see. So, ExG allows one to reproduce a facsimile of the more common NDVI using lower quality (fewer bands) drone imagery,” says Baarda. “We found ExG and NDVI to be equally effective for the purposes of our study.” 

These approaches were used in the fall after the crop had matured and turned brown but kochia was still growing and green. 

“At that time of year, this was an opportunity for us to fly drones over the fields and identify where kochia is,” says Baarda.

 The technology worked well with crops that matured relatively early, but was less effective on later maturing faba bean.

Baarda says the imagery technology was relatively simple to use without the need to zoom in to distinguish between crop and weed or supervise the technology because it is well proven already. He says the only parameters required were georeferenced data and colour imagery collected by the drone.

“Our idea was to stay away from more expensive and high-tech drones to develop a tool that would be accessible to farmers in terms of price as well as expertise needed,” says Baarda.

The data layers collected were overlaid and integrated using geographical information systems (GIS) and spatial analysis using the ESRI ArcGIS program. Kochia management zones were identified and used for site-specific kochia management. Ground truthing with weed counts verified the mapping.

“In our case, the drone produced an NDVI map for us that we received from DroneDeploy software. This map can be directly input into whatever software a farmer would use to send a map to their sprayer. In the case of ExG, which we computed for the sake of the project, those calculations were done in ArcGIS, then exported to the farm software.

“Across the board, I would say we were 80 to 90 per cent accurate in identifying kochia,” says Baarda.

Site-specific control
The next step was to utilize the maps in a site-specific approach. One way was to use a site-specific herbicide application. Mapping was used to apply Authority herbicide (Group 14) pre-plant to only the areas of the field with kochia. This resulted in a savings of 70 to 75 per cent in herbicide and application costs. The results found similar control of kochia using a site-specific application as if the entire field had been treated, producing a significant cost savings.

 Another example would be to take the same approach to chemfallow, targeting additional effective modes of action to areas of the field with kochia. This approach is even more timely, as the drone can be flown during the chemfallow season to target kochia patches in-season.

Eventually, the Holy Grail would be the use of drone spraying to target mapped weed patches. Currently, there are regulatory hurdles since the Pesticide Management Regulatory Agency has deemed that each herbicide must be assessed for drone spraying as an application method. 


Stories continue below