Evaluation and estimation of boreal podzol soil nutrient concentrations using electromagnetic induction sensors

Monitoring the spatial variability of soil properties by intrusive methods can be complicated and time-consuming; nevertheless, digital mapping of apparent electrical conductivity (EC_a) can assist in the investigation of shallow podzol soils. We hypothesized that EC_a measured as a soil proxy using electromagnetic induction (EMI) sensors can be used to estimate spatiotemporal variability of soil nutrients. This study evaluated the relationship between EC_a and selected soil nutrients (ammonium – NH₄⁺, nitrate – NO₃⁻ and orthophosphate – PO₄³⁻) and developed regression models to predict those nutrients. Multi-coil (MC) and multi-frequency (MF) EMI sensors were selected to predict nutrients within two land uses (grassland – GL and agricultural land – AL). The study revealed that MC-EMI responded statistically significantly (p-value < 0.05) relative to MF-EMI to correlate with soil NH₄⁺, NO₃⁻ and PO₄³⁻. Although highly significant correlations (p < 0.001) were observed between EC_a values and nutrients, simple linear regression (SLR) models suggested nutrients did not effectively explain the EC_a variations (low coefficient of determination and high root mean square error). Multiple linear regression (MLR) models were more effective than SLR in representing EC_a with the inclusion of saturation percentage and bulk density with each nutrient. MC EMI-based MLR models provided better nutrient predictions than the MF EMI sensor, possibly due to the sensor’s sampling depth differences and sensitivity to moisture availability in the soil. The study also revealed that if textural variability and organic matter content remain temporally stable, soil water content acts as the main driving factor for both EC_a and nutrient variability. While these results suggest the potential use of EMI sensors to rapidly assess the spatial and temporal variability of podzol soil nutrients; further research on different agronomic treatments and their effect on EC_a and soil nutrient relation is required to improve nutrient prediction accuracy.