Submitted to: International Symposium on Soil Water Measurement Using Capacitance Impedance and Time Domain Transmission
Publication Type: Proceedings
Publication Acceptance Date: 5/8/2007
Publication Date: 10/24/2007
Citation: Rowland, R.A., Guber, A.K., Pachepsky, Y.A., Gish, T.J. 2007. Comparison of plot scale average gravimetric soil water contents with data from calibrated multisensor capacitance probes. Proceedings of The Second International Symposium On Soil Water Measurement Using Capacitance, Impedance and Time Domain Transmission. Beltsville, Maryland, USDA, October 28 - November 02, 2007. Paper 1.6:1-9. Interpretive Summary: No Interpretative Summary.
Technical Abstract: Multisensor capacitance probes (MCPs) provide unparalleled spatial and temporal resolution to soil water content measurements. They are utilized in many applications where soil water availability needs monitoring. The objective of this work was to assess errors in plot scale soil volumetric water contents estimated from single MCPs. Four 1 m2 plots of sandy loam soil at the field B of the USDA-ARS OPE3 experimental watershed in Beltsville were instrumented with MCPs with sensors set at 10 cm increments from 10 to 90 or 100 cm. The plot scale water contents were computed by averaging water content measured in undisturbed samples taken around an MCP probe within each plot at 10 cm increment down to 100 cm. Soil texture and bulk density were also measured at 10-cm increments down to 100 cm.. The measured scaled frequencies were converted to volumetric water contents using three laboratory calibration curves obtained by Paltineanu and Starr (1997) using data on Californian, Australian, and Maryland soils. The MCP volumetric water content values were generally greater than plot averaged. For further analysis, errors were defined as differences between plot-averaged and MCP measured values of water contents. Application of the Maryland calibration (Australian calibration was nearly identical to the Maryland calibration) gave the errors about 2 vol. % smaller compared to the California calibration. The observed distribution of errors with the Maryland calibration was close to normal with average and standard deviation about 2 vol. % and 3 vol. % respectively. Application of regression trees to data from depths of 70 cm and shallower showed that the errors were affected by soil clay content and bulk density. Applying the calibration correction based on clay content and bulk density made the difference between MCP lack-of-fit and intrinsic variability in measurements non significant (P<0.01). Depending on the monitoring purposes, field calibration may be desirable to eliminate the bias in the MCP representation of plot averaged values.