The dielectric behaviour of clay soils and its application to time domain reflectometry

Abstract

The dielectric behaviour of 3 soils, a sandy loam (Red Chromosol), a highly structured non-swelling clay (Red Ferrosol), and a self-mulching swelling clay (Black Vertosol), was investigated using a waveguide and network analyser technique in the frequency range 3.0 GHz to 4.5 GHz. Curves relating the real part of the relative permittivity to water content are presented and compared with the general Topp curve. The Chromosol generally followed the Topp curve, but the Ferrosol and Vertosol both had curves below the Topp curve. The Ferrosol showed a maximum horizontal offset of 0.05 m³/m³ from the Topp curve in the mid soil-water content range of 0.2–0.3 m³/m³ offset from the Topp curve of 0.10 m³/m³, with a maximum of 0.12 m³/m³ occurring at a soil water content of 0.4 m³/m³.

Similar dielectric curves were obtained for the Chromosol and Vertosol using time domain reflectometry (TDR). With this method, the Chromosol showed very close agreement with the Topp curve, but the Vertosol again gave a curve below the Topp curve, similar to the one obtained using the waveguide and network analyser, but with a smaller maximum horizontal offset of 0.08 m³/m³. The difference between the waveguide and TDR Vertosol curves was mainly attributed to low bulk densities in the waveguide where packing was difficult. Some was also attributed to the wider spectrum of frequencies used by TDR. Use of the Topp curve for TDR measurements in the Vertosol would underestimate its water content by at least 0.06 m³/m³.

These results are in good agreement with others obtained from similar soils. Deviations from the Topp curve are attributed to bound water associated with the clay particles and this depends on clay mineralogy and clay content. The presented calibration curves improve the accuracy of TDR measurements in these types of clay soils.

A field comparison between water contents measured by TDR and gravimetric sampling in a similar Black Vertosol is presented. This calibration showed that soil water contents can be severely overestimated by using TDR with long probes and cables. This unexpected and opposite result is discussed in terms of attenuated high frequencies in the 15-m-long connecting cable used, errors in depth of probe placement, and changes in bulk density and DC conductivity.