Forecasting potato tuber yield using a soil electromagnetic induction method

Timely forecasting of crop yield is vital for precision agriculture management practices. This study used on-the-go proximal soil sensing using electromagnetic induction (EMI) readings of apparent ground electrical conductivity (ECa) to map ECa and forecast potato tuber yield in four fields across Atlantic Canada. The ECa data, measured in the horizontal co-planar (HCP) configuration mode of the DualEM-2 instrument, were segmented to the top 0.30-m thickness of the soil layer using a standard method to compare mapping/prediction accuracy. Results showed that ECa correlated well (R² = 0.81–0.90) with a 1:5 soil-to-water ratio solution's electrical conductivity (EC1:5). The actual tuber yield, which moderately varied (CV = 18.9–27.5%) across the fields and significantly correlated with ECa, explained more than 55% of the yield variability (R² = 0.57–0.66). The forecasted tuber yield calculated from cubic regression models of the actual tuber yield versus ECa was non-significantly different from the actual tuber yield (RMSE = 12.2–18.3%; R² = 0.57–0.66). Interpolated maps of the predicted and the actual yields, and their correlation analyses, showed similar trends of variations within the study fields (r = 0.69–0.80). The higher values of cation exchange capacity, calcium, phosphate, potash, organic matter and soil moisture content in the New Brunswick soils compared to the Prince Edward Island soils resulted in an overestimation of the predicted tuber yield than the actual yield at the lower ECa values, and an underestimation of the predicted tuber yield at higher ECa values for New Brunswick. The results revealed that the province-based calibrations produced more accurate predictions when compared with the single calibration by combining all of the data from New Brunswick and Prince Edward Island. The non-destructive prediction of potato tuber yield can enable the development of precision agricultural techniques and management practices for yield forecasting, in addition to making informed decisions for enhanced potato productivity. Highlights: Potato cropping is highly important to the economy of Atlantic Canada. The DualEM-2 sensor was used to forecast and map tuber yield to advance crop management. The tuber yields significantly correlated with ground conductivity that explained '55% of variability in yield. The DualEM-2 sensor can accurately predict potato yield under agricultural conditions similar to Atlantic Canada.