An Ice Correction Model for Dielectric Sensor to Improve Accuracy of Soil Water Potential Measurement in Frozen Soils

Soil water potential is a key variable that determines water flow and indicates water availability and content. Although diverse techniques and methods have been developed for water potential measurement in unfrozen soils, most commercial techniques such as dielectric soil water potential (DSWP) sensors are not applicable for frozen soils due to potential errors of the dielectric measurement caused by ice formation within the sensor probes. Rather than directly recording sensor readings, these commercial DSWP sensors have to indirectly estimate soil water potential under equilibrium conditions by temperature measurement and the Clapeyron equation. Here we develop a mathematical model to correct the ice induced measurement error of the DSWP sensor in frozen soils. The DSWP sensor tested in this research was self-developed with a perforated cylinder coaxial (PCC) probe filled with a porous medium (gypsum). The central electrode was a temperature sensor (Pt100) replacing the usual steel pin. Sensor performance was demonstrated in a non-saline, loamy soil, by installing one sensor (DSWP-1) in a wetted soil sample prior to soil freezing, with two additional sensors (DSWP2 and DSWP-3) installed later in frozen soil, prior to stepwise thawing. The results show that ice formed in the porous gypsum of the DSWP-1 sensor during soil freezing, as would be expected in the field with liquid water initially present to enter the probe. In comparison, no ice formed in the porous gypsums of the DSWP-2 and DSWP-3 sensors during the stepwise thawing process, allowing directly measurement of soil water potential in frozen soils and acting as a reference to develop an ice correction model based on a dielectric mixing model for the ice-formed DSWP-1 sensor to improve accuracy of soil water potential measurement. With the ice correction model, the RMSE between the readings of the DSWP-1 sensor and the reference values determined by the DSWP3 sensor was decreased from 1.065 MPa to 0.491 MPa. In general, the ice correction model developed here can be used to improve accuracy of soil water potential measurement during soil freeze-thaw process using the DSWP sensor or other dielectric sensors and may extend application of the DSWP sensor to frozen soils under nonequilibrium conditions in the field.