Soil moisture estimation and within-field zone delineation using electromagnetic induction

Three empirical salinity models (designated as Models R1, R2, and MF) were examined for real time soil moisture estimation and one-time calibration of electromagnetic induction sensors using apparent soil electrical conductivity. Field data were collected at two sites representing a range of soil types in Central Texas: high shrinking-swelling Vertisols in Temple (Site A) and light textured soils at Texas A&M University Research Farm near College Station (Site B). Model R1 was not found applicable for real time soil moisture determination at either site, even though the calibration parameters were adjusted on site-specific bases. Poor model performance was attributed to differences in amount and type of clay and inherent salinity levels between the sites of model development and validation and the sites used in this study. Model R2 developed for Site A yielded an R2 of 0.72, predicted soil moisture within ±0.02 g g−1, and was deemed applicable for real time soil moisture estimation and one-time calibration of electromagnetic induction sensors. Model R2 developed for Site B gave an R2 of 0.80 and predicted soil moisture within ±0.03 g g−1, but could not be considered applicable for real time soil moisture estimation. Model MF predicted soil moisture within ±0.03 g g−1 at Site A, and was considered appropriate for real time soil moisture estimation and one-time calibration of electromagnetic induction sensors using apparent soil electrical conductivity. At Site B, Model MF could not be evaluated since saturation data were not available. However, based on the performance at Site A, Model MF was anticipated applicable at Site B as well. At Site A, a strong relationship between apparent soil conductivity and soil profile depth to the restricting layer yielding an R 2 value of 0.45 on profile color variation basis and 0.68 on soil profile with 50% clay content-basis, was found. Within-field zone delineation capturing variation in potential water supplying capacity across the field could be possible at Site A by estimating soil profile depth apparent soil conductivity, and moisture content predicted with Models R2 and/or MF. However, at Site B, because no restricting layer was found, zone delineation could be accomplished using regression models (R2 = 0.85) relating soil moisture, apparent soil conductivity, and clay content, developed under the auxiliary approach.