EM38 for volumetric soil water content estimation in the root-zone of deep vertosol soils

Abstract

Electromagnetic induction sensors, such as EM38, are used widely for monitoring and mapping soil attributes via the apparent electrical conductivity (EC a) of the soil. The sensor response is the depth-integrated combination of the depth-response function of the EM38 and ‘local’ electrical conductivity (EC az) at depth. In deep, Vertosol soils, assuming the instrument depth-response function is not perturbed by the soil and where volumetric moisture content at depth ( θ v( z)) dominates EC az, EM38 should be capable of predicting average moisture content without recourse to mathematically complicated, and unstable profile inversion processes. Firstly a multi-height EM38 experiment was conducted over deep Vertosol soils to confirm the veracity of the EM38 depth-response function and test the concomitant hypothesis of the EM38 response being an integrated (i.e. additive) combination of depth-response function and θ v( z). Secondly, depth profiles of moisture content were used to calibrate the EM38 to infer average θ v( z) within the ‘root-zone’ of crop plants—here taken to be surface—0.8 m and surface—1.2 m. EM38 calibration was performed using soil samples acquired from both extracted cores and excavated pits. Mathematical summation of measured θ v( z) from sectioned cores and the known depth-response function of the EM38 was found to explain 99% and 97% of the variance in measured EC a for horizontal and vertical dipole configurations at multiple sensor heights above the ground. Average θ v from surface to 0.8 m ( θ ¯ 0.8 ) and surface to 1.2 m ( θ ¯ 1.2 ) explained only 37% and 46% of the variance in on-ground EC a for vertical dipole configuration measurements compared to 55% and 56% of the variance for horizontal dipole configuration. In a separate validation experiment, the shape of the vertical moisture profile proved highly influential in determining the ability of the calibration equations to infer underlying average moisture content, especially where the depth profile shapes differed between sensor calibration and subsequent field validation (for example following rainfall or irrigation).