Comparison of travel time analysis and inverse modeling for soil water content determination with time domain reflectometry

[1] Travel time analysis algorithms are used to extract dielectric permittivity (and thereby water content) from time domain reflectometry (TDR) waveforms. A limitation of travel time analysis is the need to define subjective analysis parameters for reproducible measurements. The aim of this paper is to compare three methods to remove subjectivity from the analysis of TDR waveforms: (1) travel time analysis with optimized analysis parameters, (2) fitting a simulated waveform based on a model with frequency-dependent permittivity, and (3) fitting a waveform based on a model with an apparent dielectric permittivity. We performed an outflow experiment on a sandy soil sample and simultaneously collected soil water content measurements (determined from outflow) and duplicate TDR measurements. After optimizing the travel time analysis parameters by minimizing the difference between duplicate measurements, we found little difference in accuracy and reproducibility between the three methods. However, we showed that small deviations from the optimal travel time analysis parameters have a significant influence on both the accuracy and the reproducibility of travel time analysis. A comparison of the two inverse modeling approaches showed that the inclusion of frequency-dependent permittivity did not greatly improve the fit between measured and modeled waveforms and that a frequency-independent (apparent) permittivity adequately describes the TDR measurements made in this sandy soil. We conclude that the removal of subjectivity resulted in similar accuracy and reproducibility for all three methods. Nevertheless, we consider inverse modeling an interesting alternative to travel time analysis because it is a fully automated analysis with high accuracy and reproducibility.