Characterization of field‐scale dryland salinity with depth by quasi‐3d inversion of DUALEM‐1 data

Abstract

AbstractTo understand the limitations of saline soil and determine best management practices, simple methods need to be developed to determine the salinity distribution in a soil profile and map this variation across the landscape. Using a field study in southwestern Australia, we describe a method to map this distribution in three dimensions using a DUALEM‐1 instrument and the EM4Soil inversion software. We identified suitable parameters to invert the apparent electrical conductivity (ECa – mS/m) data acquired with a DUALEM‐1, by comparing the estimates of true electrical conductivity (σ – mS/m) derived from electromagnetic conductivity images (EMCI) to values of soil electrical conductivity of a soil‐paste extract (ECe) which exhibited large ranges at 0–0.25 (32.4 dS/m), 0.25–0.50 (18.6 dS/m) and 0.50–0.75 m (17.6 dS/m). We developed EMCI using EM4Soil and the quasi‐3d (q‐3d), cumulative function (CF) forward modelling and S2 inversion algorithm with a damping factor (λ) of 0.07. Using a cross‐validation approach, where we removed one in 15 of the calibration locations and predicted ECe, the prediction was shown to have high accuracy (RMSE = 2.24 dS/m), small bias (ME = −0.03 dS/m) and large Lin's concordance (0.94). The results were similar to those from linear regression models between ECa and ECe for each depth of interest but were slightly less accurate (2.26 dS/m). We conclude that the q‐3d inversion was more efficient and allowed for estimates of ECe to be made at any depth. The method can be applied elsewhere to map soil salinity in three dimensions.