Premium
Complex Permittivity Model for Time Domain Reflectometry Soil Water Content Sensing: II. Calibration
Author(s) -
Schwartz R.C.,
Evett S.R.,
Bell J.M.
Publication year - 2009
Publication title -
soil science society of america journal
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 0.836
H-Index - 168
eISSN - 1435-0661
pISSN - 0361-5995
DOI - 10.2136/sssaj2008.0195
Subject(s) - reflectometry , soil water , permittivity , saturation (graph theory) , dielectric , soil science , water content , power law , materials science , calibration , empirical modelling , mineralogy , conductivity , environmental science , analytical chemistry (journal) , time domain , geology , chemistry , mathematics , geotechnical engineering , statistics , optoelectronics , combinatorics , chromatography , computer science , computer vision , programming language
Despite numerous applications of time domain reflectometry (TDR), serious difficulties in estimating accurate soil water contents under field conditions remain, especially in fine‐textured soils. A complex dielectric mixing model was calibrated for fine‐textured soils (24–45% clay) and its accuracy was evaluated and compared with empirical calibrations. The Ap and Bt horizons of two soils were packed into columns and adjusted to volumetric water contents (θ) ranging from air dry to near saturation. Travel time and bulk electrical conductivity (σ 0 ) were measured using TDR at temperatures ( T ) of 8, 22, and 40°C and using three coaxial cables to obtain a range of input spectrum bandwidths (ω S ). Apparent permittivities ( K a ) were predicted using the complex permittivity model with measured θ, T , σ 0 , ω S , and soil bulk density, and fitted to measured K a by optimizing specific surface area ( A s ), the power‐law exponent ( a ), and an empirical polarization loss factor. Measured K a was best approximated using the power‐law dielectric mixing model with a semiempirical effective frequency estimate and a = 0.68. Predicted A s increased with increasing clay content, cation exchange capacity, and measured specific surface areas. The two‐parameter power‐law calibration removed temperature bias in θ estimates and reduced the RMSE in θ estimates by an average of 0.006 m 3 m −3 compared with an empirical calibration. Empirical models predicted field θ with oscillations of up to 0.022 m 3 m −3 in phase with soil temperatures resulting from permittivity temperature dependencies. In contrast, the calibrated dielectric mixing model removed or dampened in‐phase θ fluctuations to <0.004 m 3 m −3 , which permitted the detection of more subtle changes (<0.02 m 3 m −3 ) in θ.