A stochastic linear model to evaluate the dielectric constant-electrical conductivity relation of a sandy soil using TDR records

Despite the importance of evaluating soil pore water electrical conductivity from soil bulk electrical conductivity in ecological and hydrological applications, it still has not been worked out very well. This study was carried out on a sandy soil under laboratory conditions. We used a Dynamic Linear Model (DLM) and Kaman filter (Kf) to estimate soil pore electrical conductivity. Time series of soil dielectric constant and bulk electrical conductivity were measured by time domain reflectometry (TDR) at different depths in a soil column to modify the deterministic Hilhorst (2000) model into a stochastic model. We analyzed the linear relationship between dielectric constant and bulk electrical conductivity in order to estimate precisely pore water electrical conductivity. As a result, the observed and measured of dielectric constant were reasonably agree well. However, although the linear relation between soil dielectric constant and bulk electrical conductivity which was realized in homogenous soil, the offset of this linear relation varies for each depth. © 2018 IEEE.Postprint (published version