Evaluation of a Direct-Coupled TDR for Determination of Soil Water Content and Bulk Electrical Conductivity.

Signal degradation in coaxial cables and interconnects is a long-standing problem in the practical deployment and calibration of time domain reflectometry (TDR) for soil water content monitoring. Acclima, Inc. has recently commercialized a low-cost TDR sensor (TDR-315) with all electronics required for waveform acquisition embedded in the probe head thereby avoiding signal degradation. Firmware acquires pertinent waveform features and calculates apparent permittivity (K_a), bulk electrical conductivity (σ_a), and water content values that are transmitted to the data logger via SDI-12 communications. We calibrated ten TDR-315 sensors and conventional TDR for K_a and σ_a measurements using acquired waveforms in air, deionized water, and electrolytic solutions of CaCl₂. Additionally, soil water content calibrations were carried out for TDR-315 probes and conventional TDR using a Pullman clay loam soil. Lastly, a column displacement experiment using Pullman clay loam and electrolytic solutions of CaCl₂ (0.25 and 7.3 dS m^-1) was carried out to evaluate the sensitivity of K_a to σ_a (0.65 to 2.8 dS m^-1) for both measurement technologies. Permittivity calibrations of the TDR-315 could be accomplished with conventional methods of waveform interpretation used for TDR. Conventional calibration of σ_a using long time amplitudes yielded a linear response for σ_a ≤ 0.3 S m^-1 above which the response was nonlinear; but successfully accounted for by the firmware conductivity calibration. The fitted water content calibrations of the Pullman clay loam using the acquired waveforms and firmware-reported K_a of the TDR-315 were nearly indistinguishable from conventional TDR calibrations. Response of the two measurement technologies in a lossy soil during changing solution conductivities demonstrated that, in contrast to conventional TDR, travel time measured using acquired TDR-315 waveforms was insensitive to σ_a up to 2.8 dS m^-1. Firmware-calculated K_a was satisfactory at σ_a < 2 dS m^-1compared with estimates evaluated using full waveforms and could be improved at greater conductivities by sampling at finer time resolutions.