LAUSR

Abstract The introduction of irrigated agriculture through a network of canals has seen enormous growth in agricultural output in northeast Thailand. However, the canals are leaking causing recharge water to interact with an underlying sequence of geological units rich in salts, which has caused secondary soil salinization. Herein, we measure apparent electrical conductivity (ECa) from EM instruments (e.g. Geonics EM38, EM34 and DUALEM-421) along a section of the Kham-rean Canal of the Muang Pea project in Thailand. We inverted the ECa to generate electromagnetic conductivity images (EMCI) using EM4Soil and evaluate these 2-dimensional models of estimates of the electrical conductivity (σ) distribution with depth, against soil properties, including, measured electrical conductivity of saturated soil paste extract (ECe – dS/m). Using a linear regression (LR) calibration we determine the suitability of each instrument to predict ECe from σ. We found a moderate coefficient of determination between σ and ECe by inverting ECa using either a EM38 on the ground (R2 = 0.66) or a height of 0.5 m (0.65). This was also the case inverting DUALEM-1 (0.63) and DUALEM-2 (0.64) ECa. A strong R2 was achieved by inverting DUALEM-421 (0.72) ECa, with generally weak R2 achieved with DUALEM-4 (0.34) and EM34 (0.51). Using a leave-one-out cross validation, good agreement in ECe predictions were achieved by inverting DUALEM-421 (Lin’s = 0.83), followed by moderate predictions when EM38 on the ground or a height of 0.5 m (0.79) and EM34 (0.66) were used. Our interpretation of predicted ECe suggests the approach can identify non-saline and highly saline areas. Our interpretation of the predicted ECe shows non-saline and highly saline areas, where the former represents a likely recharge area. In the centre of the study, the recharge area was characterised by large silt content (>70%), moderately sodic (6