Location: Soil and Water Management Research
Title: A reevaluation of time domain reflectomery propagation time determination in soils Authors
Submitted to: Vadose Zone Journal
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: October 2, 2013
Publication Date: December 19, 2013
Citation: Schwartz, R.C., Casanova, J.J., Bell, J.M., Evett, S.R. 2013. A reevaluation of time domain reflectomery propagation time determination in soils. Vadose Zone Journal. 13(1)1-13, doi:10.2136/vzj2013.07.0135. Interpretive Summary: Time domain reflectometry (TDR) is a standard method used to measure soil water content. This method requires the evaluation of the velocity of signals in waveforms recorded by instruments. Current computer programs used to interpret waveforms can yield large water content errors. User input of program parameters is often necessary. This causes problems when a large number of water contents are evaluated over a long period in time. Our objectives were to develop a new computer program that requires no user input, and to reduce water content errors. The new adaptive computer program permitted the consistent evaluation of water contents, and required no user intervention. Using the new method to analyze difficult waveforms, water content errors were 13% of that obtained using a standard technique.
Technical Abstract: Time domain reflectometry (TDR) is an established method for the determination of apparent dielectric permittivity and water content in soils. Using current waveform interpretation procedures, signal attenuation and variation in dielectric media properties along the transmission line can significantly increase sampling error in estimating the time, t2, at which the pulse arrives at the end of the probe. Additionally, manual adjustment of waveform analysis parameters is frequently required in current software to accommodate changes in media properties when processing large time series of TDR measurements. An algorithm entitled "adaptive waveform interpretation with Gaussian filtering" (AWIGF) that circumvents difficulties with current methods was developed and evaluated. The algorithm filters signal noise using Gaussian kernels with an adaptively estimated standard deviation based on the maximum gradient of the reflection, at the termination of the probe. Two fitted parameters are required to scale the smoothing level for a given step pulse generator. Additionally, the maximum second derivative is used to evaluate t2. AWIGF determined t2 was compared with TACQ, a standard waveform interpretation algorithm. The strategies of AWIGF permitted the determination of t2 without parameter adjustment when the loss characteristics of the media changed, such as with an increase in soil water content and bulk electrical conductivity. Using the new method, the sampling error of t2 was less than 0.06 ns over a wide range of media properties and less than or equal to that obtained with TACQ. In strongly attenuated waveforms, the water content sampling error determined with AWIGF was 0.005 m3 m**3 compared with 0.038 m3 m**3 obtained using TACQ