Assessing Field and Laboratory Calibration Protocols for the Diviner 2000 Probe in a Range of Soils with Different Textures

Frequency domain reflectometry (FDR) downhole sensors have been increasingly used for soil moisture field monitoring because they allow measurement, even continuously, along a soil profile. Moreover, they can also be installed with minimal soil disturbance around the access tube. The objectives of the paper were to assess the field and laboratory calibration protocols for a FDR capacitance probe (Diviner 2000) for a range of soils characterized by different particle size distributions and shrink/swell potential and to propose a practical and effective protocol on the basis of undisturbed soil samples, accounting for soil shrinkage/swelling processes characterizing swelling clay soils. The experiments showed that on coarse-textured soils, field calibration under wet, moist, and dry conditions allows estimations of the volumetric soil water content, with root-mean-square error (RMSE) values always lower than $0.058{\mathrm{cm}}^3·{\mathrm{cm}}^{-3}$. On the contrary, the problems occurring in the field on finer-textured soils, which are characterized by a clay content ranging between 36.7 and 45.1% and moderate to high shrink/swell potential, did not permit identification of suitable calibration equations and then accurate estimations of the soil water content. For such soils, in fact, a great dispersion of the experimental data and consequently high error values associated with the site-specific calibration equations, i.e., up to $0.121{\mathrm{cm}}^3·{\mathrm{cm}}^{-3}$ for the soil characterized by the highest clay percentage, were observed. The laboratory experiments were carried out by using undisturbed soil monoliths which, compared with sieved soils, have the advantage of accounting for the natural soil structure surrounding the access tube and monitoring the soil shrinkage processes occurring in clay soils during sensor calibration experiments. The Diviner 2000 calibration equations obtained in the laboratory were characterized by error values generally lower than those obtained in the field and always smaller than $0.053{\mathrm{cm}}^3·{\mathrm{cm}}^{-3}$. Finally, in the range of a soil water content between approximately 10% and the maximum observed, the scaled frequency measured by the sensor was almost constant at a decreasing soil water content. This circumstance can be ascribed to the normal phase of the shrinkage process determining the compensative effects between the reduction of the volumetric soil water content and the increasing soil bulk density. The maximum variations of scaled frequency were observed in the range of the soil water content, for which the resulting soil bulk density was approximately constant. The knowledge of the soil shrinkage characteristic curve therefore assumes a key role when calibrating FDR sensors on shrinking/swelling clay soils.