Anez Consulting has devised an innovative method of estimating crop yield and analysing the health and vigor of plants—using a drone-sourced surface model to visualise biomass. Waypoint spoke with the firm’s precision ag specialist to learn more.
Over the past year, agronomy service provider Anez Consulting of Little Falls, Minnesota, has begun employing what appears to be a unique method of analysing crop health and estimating yields: when methods such as NDVI analysis struggle to illuminate in-field variability, its team uses Pix4Dmapper software to transform eBee Ag drone imagery into a digital surface model (DSM), which is used to visualise the biomass variability of full-canopy crops.
The company’s precision agriculture specialist, Michael Dunn, explains: “I’m originally from Northern California where LIDAR is often used on manned aircraft to remotely measure the height of trees. When I learned of the 3D mapping potential of the Pix4D software, I realised we could do the same thing on corn and other crops utilising photogrammetry instead of LIDAR.”
By taking the difference between the drone-sourced DSM at full canopy and the topography of the field, Dunn isolates the height of the crop and maps this out across the entire field. “This is a new way for us to look at the condition, health, and vigor of corn (and other crops), across different soil types and management zones to delineate crop health and yield potential,” he says. “It is not a method we totally rely upon and use frequently, but it can be a great tool when other methods fail to accurately highlight in-field differences.”
By taking the difference between the DSM at full canopy and the topography of the field, Dunn isolates the height of the crop and maps this out across the entire field.
According to Dunn, mapping out relative biomass can enhance the estimation of crop yields, validate management zones and more. “Relative biomass should parallel yield potential in most crops; where relative biomass is greater, yield potential is also higher,” he asserts. “Therefore, using a relative biomass map and extrapolating ground-truthed yield estimates over that map would likely provide a better estimate than conventional methods, which consist of trying to find the ‘average’ field condition and doing a random sample. NDVI would also be a good layer to do this kind of extrapolation, except for instances of shadows, glare, and white-balance issues that may skew values.”
“This approach is also more accurate than conventional imagery at identifying planter skips, drown-outs, lodging, and other events that affect crop height and density; especially where the colour of the crop is not dramatically impacted,” he adds.
Relative biomass mapping – example 1
One area where the management zones and biomass map conflict is in the lower-left corner of the field (red in the biomass map and purple in the zone map immediately above). While the zone map shows this area displaying high yield potential, the relative biomass map indicates there is very low yield potential. “This is due to early spring drown-outs,” Dunn says. “Had this area been drained, it would have had some of the best corn because the soil’s organic matter and water availability there are optimal for plant growth and development.”
Relative biomass mapping – example 2
“Many details are illuminated in the relative biomass map that are much less obvious in the orthomosaic, mainly due to excessive sun glare,” notes Dunn. “The west end of the field (1) is planted with a different corn hybrid than the rest of the field, and at point 2 corn residue was wind-rowed and left near the surface where it inhibited plant-available nitrogen from intercepting roots. This resulted in shorter corn where the residue was heaviest. At this point, the corn had recovered much of its color, but its height was still affected.”
Relative biomass mapping – example 3
Proving by sampling
In order to verify the validity of the relative biomass mapping method, Dunn and his team sampled corn from the same field based on the relative biomass map above (example 3).
Corn was sampled from within the same row, with the same soil type. In the biomass map’s red zone, five consecutive plants were harvested at the first node above the ground. These were then weighed. This process was then repeated in the map’s green and blue zones.
“The plants in the red and green zones were beginning to tassel, which indicates there were stress factors accelerating maturity. These areas will likely yield less than the blue zone (but will have better dry-down times for harvest). Plants in the blue zone were tall, lush and looked very healthy. However, if weather conditions began to favour the development of fungal disease, the blue zone would likely be the most afflicted by such pathogens, due to lower air circulation and less light penetration within the canopy),” Dunn says.
Dunn asserts that while there is still much work to be done to refine the accuracy of this approach, the underlying theory is simple and solid.
“Where plant population is high, plants will naturally grow taller as they compete for sunlight, which should result in higher biomass. Where plant population is lower, even on good soils, plants will grow shorter, resulting in lower biomass. These are the underlying principles that allow us to infer relative biomass using plant height alone,” he says, adding that the eBee Ag farming drone used to collect the data is key to making such work possible. “Without the drone we couldn’t achieve these results, as no satellite or manned aircraft could provide such detail.”
Watch Michael Dunn’s agriculture drone application video: