EC cannot help make VR strategies more consistent
Key Result: While soil electrical conductivity (EC) measurement is a strong indicator of soil clay and moisture content, it cannot help to make variable rate (VR) fertilizer programs more consistent. Producers using VR should be prepared to use a specific strategy for each field each year.
Project title, Principal investigators: “Understanding soil variability for effective zone management in precision agriculture –
an evaluation of sensor based soil mapping tools,” Ken Coles, Farming Smarter
Funding: Alberta Canola ABC
Satellite images, historical yield maps, terrain analysis and representative soil samples are often used in various combinations to characterize different zones. Farming Smarter initiated this study to see if soil electrical conductivity (EC) measurements could improve the accuracy and effectiveness of these zones and the crop response predictions and prescriptions used on these zones.
The study compared two EC sensors: EM38-MK2 (EM38) and Veris MSP3 (Veris). Soil EC maps from both sensors were found to be strong and consistent indicators of the presence of clay and soil moisture. However, the study revealed that mapped EC data could not be used for a direct estimation of the spatial distribution of macronutrients in the soil.
The project then tested five different zone delineation methods in each of the 10 fields studied. They were:
Surface Geography: Zones were created using a subjective assessment of visual spatial differences in terrain, moisture, salinity, etc.
Grid Soil Sampling: Soil sample nitrogen measurements were spatially interpolated using the kriging method. Resulting values were divided equally into three zones.
Historic Yield: All available yield maps were normalized, then pooled to create an average normalized yield map. Resulting values were divided into three zones equally.
EC: A single EM38 deep EC map was put through a cluster analysis procedure to objectively determine zone boundaries and number of zones.
Composite: A single representative EM38 deep EC layer and a single representative yield layer were pooled and put through a cluster analysis procedure to objectively determine zone boundaries and number of zones.
All five methods had some level of success at identifying regions that yielded differently from one another. However, the study could not consistently identify an effective variable nitrogen management strategy for these zones.
Among the zone delineation methods, there was reasonable success identifying regions that yielded differently from one another, as the study did so in about 50 per cent of instances. These results varied across delineation methods, with the grid soil sampling method being notably less effective than the others. However, the project was unable to consistently identify unique yield responses to nitrogen, indicating that grain yields in the zones identified did not respond differently to nitrogen.
Yield correlated with EC data in roughly 20 per cent of instances, on average, but this fluctuated significantly by year. Correlation exceeded 30 per cent of instances for 2010 and 2013 yield data but was as low as zero per cent for 2012. This shows how variable yield patterns can be from year to year. In fact, yield maps from various years only had strong correlations to one another in
10 per cent of instances project-wide. Elevation correlated strongly with yield data in 26 per cent of instances. This places significant limits on the capability of soil sensor or elevation data to predict grain yield in a given field in a given year.
The study found that universal strategies for zone delineation were largely ineffective, which begs for close scrutiny of VR strategies. It is unlikely that the strategies tested would help a producer reduce nitrogen inputs and associated costs.
In conclusion, this study questions the viability of a formulaic approach to zone delineation and management. Producers should be prepared to develop a specific variable-rate strategy for each field each year.