An improved model and field calibration technique for measuring liquid water content in unfrozen and frozen soils with dielectric probes

Dielectric soil moisture probes can be used to obtain frequently logged field observations of volumetric water content, but because this is an indirect method, it is challenging to ensure that you have a suitable calibration relationship and parameters. Challenges are associated with (a) limitations in the probes' ability to accurately measure the soil bulk dielectric constant and (b) limitations in obtaining reliable direct observations to calibrate the probe against, due to the spatial heterogeneity in the field. Furthermore, in soils that freeze, we do not have a robust approach to account for the effect of ice on the instrument. In this study, we propose a calibration relationship for all dielectric probes that is physically based, parsimonious, accounts for errors in the bulk dielectric constant, and quantifies the uncertainty in the liquid water content caused by the presence of ice. We show that our relationship has a better performance and more realistic parameter values than existing approaches. To calibrate a dielectric probe in the field, we show that it is necessary to account for spatial heterogeneity in samples taken for calibration. There is value in taking simultaneous measurements of the bulk dielectric constant (using the same dielectric probe as is being calibrated) and gravimetric samples for water content at a number of points in space. Taking too few samples is likely to be misinformative-we show that it is better to not calibrate the probe at all than to calibrate it with too few data points.

Automated summary

Dielectric soil moisture probes provide a frequently logged field observation of volumetric water content, but calibration is challenging due to limitations in accurately measuring the soil bulk dielectric constant and obtaining reliable direct observations for calibration. This study proposes a physically based calibration relationship that considers errors in the bulk dielectric constant and quantifies the uncertainty caused by the presence of ice. The proposed relationship outperforms existing approaches and highlights the importance of accounting for spatial heterogeneity in calibration samples.