Predicting saturated hydraulic conductivity in a sandy grassland using proximally sensed apparent electrical conductivity

• The large spatial variability of soil Ks, ECa and physical properties were found.
• We found a good correlation of Ks and soil properties to the DUALEM-21S-ECa data.
• The relation was established and validated to predict Ks, using ECa, as a proxy data.
• Detailed map of Ks was produced with satisfactory accuracy for hydrological modeling.

Finding a correspondence between soil hydraulic properties, such as saturated hydraulic conductivity (Ks) and apparent electrical conductivity (ECa) as an easily measurable parameter, may be a way forward to estimate the spatial distribution of hydraulic properties at the field scale. In this study, the spatial distributions of Ks, of soil ECa measured by a DUALEM-21S sensor and of soil physical properties were investigated in a sandy grassland. To predict field scale Ks, the statistical relationship between co-located soil Ks, and EMI-ECa was evaluated. Results demonstrated the large spatial variability of all studied properties with Ks being the most variable one (CV = 86.21%) followed by ECa (CV ≥ 53.77%). A significant negative correlation was found between ln-transformed Ks and ECa (r = 0.83; P ≤ 0.01) at two depths of exploration (0–50 and 0–100 cm). This site-specific relation between ln Ks and ECa was used to predict saturated hydraulic conductivity over 0–50 cm depth for the whole field. The empirical relation was validated using an independent dataset of measured Ks. The statistical results demonstrate the robustness of this empirical relation with mean estimation error MEE = 0.46 (cm h− 1), root-mean-square estimation errors RMSEE = 0.74 (cm h− 1), coefficient of determination r2 = 0.67 and coefficient of model efficiency Ce = 0.64. The relationship was then used to produce a detailed map of Ks for the whole field. The result will allow model predictions of spatially distributed water content in view of irrigation management.