Predicting plot-scale water infiltration using the correlation between soil apparent electrical resistivity and various soil properties

The identification of runoff source areas is essential for Integrated Water and Resources Management (IWRM). Although direct methods for the determination of steady-state water infiltration in soils (Inf) do exist, these are tedious and time-consuming. Geophysical techniques offer an alternative, however, geophysical data are often misinterpreted, especially in terms of the inter-relationships between soil apparent electrical resistivity (Rho) and Inf and several other soil physical or chemical properties. This paper evaluates the magnitude of the extend Rho measurements might allow prediction of Inf. This study was conducted in the Kwazulu-Natal province of South Africa where surface runoff arising from the steep slopes has a large impact in land degradation. Measurements of Rho with an RM-15 resistance meter were taken within a 10 × 30 m plot showing similar sandy-loam Acrisols but different proportions of soil surface coverage by plants (from 0–5% to 75–100%), depth to the clayey Bw horizon (D2B), top-soil (0–0.1 m) water content (θ) and bulk density (BD). There was a low correlation between Rho and Inf obtained under controlled conditions of rainfall (30 mm h−1during 45 min) at fifteen 1 m2 micro-plots (r2 = 0.30). However, the correlation with the normalized Rho (Rhon) as if D2B, θ, and BD were constant over the study plot and equal to their average value, was much higher (r2 = 0.66), pointing out the need to consider the complex and multiple correlations between soil properties and Rho in an attempt to map the spatial variations of Inf. Finally, the use of Rhon as a co-kriging co-variate appeared to significantly improve the short range spatial prediction of water infiltration in soils and thus IWRM implementation.

► We evaluate the magnitude of the extend soil apparent electrical resistivity can predict water infiltration in soils (Inf).
► We examine a 10 × 30 m plot showing similar sandy-loam Acrisols but different proportions of soil surface coverage by plants.
► There was no direct correlation between Rho and Inf because of multiple correlations.
► The use of Rho as a co-kriging co-variate appeared to significantly improve Inf prediction.