Quasi-3D mapping of soil moisture in agricultural fields using electrical conductivity sensing

Abstract Knowledge of real time spatial distribution of soil moisture has great potential to improve yield and profit in agricultural systems. Recent advances in non-invasive electromagnetic induction (EMI) techniques have created an opportunity to determine soil moisture content with high-resolution and minimal soil intrusion. So far, EMI has mainly been used for homogenous soil conditions, which are not common in agriculture and results are mainly validated by excavated pits or calibration models using soil samples on a transect. This study converts apparent electrical conductivity data recorded with a Dualem-1Hs EM-metre for two surveys of variable moisture conditions (dry and wet season) with 2475 and 2174 data points over 5.4 ha, in a field with a contrasting vertical soil profile into spatiotemporal management zones. A least square inversion algorithm was used to determine electrical conductivities for individual soil layers of 0–0.5 m, 0.5–0.8 m and 0.8–1.6 m. Soil samples from the depth of 0.5 m and 0.8 m were used for soil moisture calibrations. A laboratory experiment under controlled conditions developed electric conductivity vs volumetric water content relations with power law functions for required soil depth slices with R2 values between 0.98 and 0.99. Subsequently, EMI data were converted to volumetric water contents for each layer and predictions were spatially displayed. Median change between the measured apparent conductivity and inverted values range from 6 to 17 mS m-1 resulting in 3–7% difference in volumetric water prediction. These EMI based soil moisture predictions were compared with neutron moisture metre measurements, with Pearson R values of 0.74 and 0.95 for the wet and dry season surveys, respectively. The method is robust and offers a comparatively fast method to estimate the soil moisture status in fields and subsequently make informed management decisions.