Quantifying and correcting for clay content effects on soil water measurement by reflectometers

Authors

Jasreman Singh, University of Nebraska-LincolnFollow
Tsz Him Lo, University of Nebraska-LincolnFollow
Daran Rudnick, University of Nebraska-LincolnFollow
Suat Irmak, University of Nebraska-LincolnFollow
Humberto Blanco-Canqui, University of Nebraska - LincolnFollow

Date of this Version

2-20-2019

Citation

Published in Agricultural Water Management 216 (2019), pp 390–399. doi:10.1016/j.agwat.2019.02.024

Comments

Copyright © 2019 Elsevier B.V. Used by permission.

Abstract

The presence of clay particles increases the specific surface area of a soil and can affect the calibration of electromagnetic soil water sensors including reflectometers. To quantify and correct for this effect in two relatively new reflectometers, three TDR315 and three CS655 sensors were installed in each of five soils with clay content ranging from 5 to 49%. As the soils were dried in a temperature controlled room, sensor reported soil volumetric water content (θ_v) according to the factory calibration was compared against reference θ_v determined by weighing the soils. Sensor reported θ_v was similar to reference θ_v in the sand soil (root mean square difference (RMSD)<0.02 m³m⁻³), but the discrepancy was larger for the clayey soils. An increase in clay content tended to cause TDR315 to underestimate low θ_v and tended to cause CS655 to overestimate θ_v in a concave down pattern. At the levels evaluated in this experiment, differences in clay content produced a larger effect than differences in temperature (24 versus 35 °C) and salinity (0 versus 3.09 g/L CaCl₂) on factory calibration accuracy of both sensors. Soil specific empirical calibrations developed using a square root mixing model fitted the experimental data very closely (R²>0.93) for both sensors. By estimating calibration coefficients based on clay content alone to recalculate sensor θ_v from sensor reported apparent relative permittivity, RMSD from reference θ_v was reduced by approximately 36% for both sensors as compared with using the factory calibration. Applying the same procedure to independent literature data tended to improve θ_v accuracy of both sensors increasingly as clay content increased above 20%. The findings suggest that a simple, user-friendly correction for clay content effects may provide initial practical improvement over the factory calibration of a reflectometer in clayey soils.