Correcting the moisture–permittivity conversion functions for selected soil-specific properties in the 20 MHz–3 GHz frequency range for accurate soil moisture measurement

The application of an appropriate dielectric permittivity—volumetric water content calibration function is important for accurate measurement of soil water content with the use of dielectric sensors. The paper explores the performance of a new class of models incorporating terms dependent on the maximum achievable permittivity value for a given soil, as well as loss tangent and bulk electrical conductivity of a given soil sample. The models were trained on two data sets, together consisting of 687 dielectric spectra of samples of six soils and of various water content, bulk electrical conductivity, temperature, and bulk density. The test set consisted of 158 spectra of four new soils. All dielectric spectra were measured with the use of a coaxial transmission-line cell system in the frequency range from 20 MHz to 3 GHz. Water content of all samples was determined using the standard oven-drying method. The results indicated that the inclusion of the maximum permittivity term improved the accuracy of the volumetric water content determination with respect to the simple linear model, especially at frequencies higher than about 500 MHz, while the inclusion of both loss tangent and bulk electrical conductivity correction terms enhanced the performance of the models at low frequencies. The further development of the presented models could lead to the development of fast and simple single-point soil-specific calibration procedure, as well as to the improvement of the measurement accuracy of the low-frequency impedance sensors. • Accurate soil moisture measurement requires application of an appropriate conversion function. • Proposed corrections for selected soil properties improve the RMSE of the models. • Incorporating maximum achievable permittivity improves the correction function.