Evaluation of calibration methods for interpreting soil salinity from electromagnetic induction measurements

The EM-38 electromagnetic induction sensor of Geonics Ltd. (Canada) is a most useful instrument for rapid field identification and mapping of soil salinity. Interpretation of instrument measurements in terms of meaningful parameters of soil salinity is difficult, however, due largely to the non-uniform response distribution with depth. Various models have been proposed that allow the conversion of measurements made on this instrument to the electrical conductivity of the bulk soil (ECa, as measured with the four-electrode probe) or EC of the saturation extract (ECe). Seven of these models were evaluated in this study by comparing predicted with measured values. Four allow the estimation of ECa at 0.30-m depth intervals down to 0.90 m, two allow the estimation of a single valued ECe, weighted for instrument response with depth, and one estimates ECe for 0.30-m intervals down to 0.90 m, The performance of the models varied greatly, and likely reasons for poor prediction are discussed. For the model that produced the most accurate estimate of ECa, the estimated values were, based on 95% confidence limits, within 0.8 dS m(-1) of the values predicted using the regression equations. For the models that predict weighted ECe, the corresponding value was typically about 2.2 dS m(-1). While salinity measurements made with the EM-38 are not highly accurate, the strength of the technique is that measurements of reasonable accuracy can be made very rapidly. Categories of soil salinity for large areas can be readily established, which represents useful information for salinity management and certain research applications.