Skip to main content
ARS Home » Research » Publications at this Location » Publication #220876

Title: How Well Do Data from Multisensor Capacitance Probes Represent Plot-Scale-Average Soil Water Contents?

Author
item Guber, Andrey
item Rowland, Randy
item Pachepsky, Yakov
item Gish, Timothy

Submitted to: International Agrophysics
Publication Type: Peer Reviewed Journal
Publication Acceptance Date: 12/11/2009
Publication Date: 1/4/2010
Citation: Guber, A.K., Rowland, R.A., Pachepsky, Y.A., Gish, T.J. 2010. How Well Do Data from Multisensor Capacitance Probes Represent Plot-Scale-Average Soil Water Contents? International Agrophysics. 24:43-49.

Interpretive Summary: Multisensor capacitance probes (MCP) are a relatively new addition to the family of devices for determining soil water content. The MCP devised has shown great promises in irrigation scheduling, evaluating water needs of plants, estimating soil hydraulic properties, estimating groundwater recharge and infiltration losses, and other soil water-related fields. The individual sensors are located at different depths in the same access tube. It is often beneficial to know how representative measurements of soil water are in the middle of a experimental plot for the average water content across the plot. Such relations may be of importance, for example, when the crop status across the plot is estimated and has to be related to soil moisture status, or when and the MCP is used to track water infiltration. The differences between MCP and plot-averaged water contents may arise because of systematic error in manufacturer MCP calibration and random variability of soil properties. The objective of this work was to assess the systematic and the random components of the difference between MCP and plot-averaged water content across 1x2m plots with heterogeneous coarse-textured soil planted to corn. The soil around four MCP was sampled at a distance of 50 cm from the MCP on three dates with distinctly different water contents. Both systematic and random differences between MCP and plot-averages were encountered. The manufacturer calibration led to the overestimation at low water content and probably to the underestimation at high water contents. The removal of this bias had only a marginal effect on the overall difference. The dependence of the differences on the sand and silt content in soil was demonstrated. Correcting MCP measurements for the sand and silt texture resulted in a 20% decrease in root-mean-square difference between these measurements and plot-average water content. Site specific MCP calibration appears to be an important task before using MCP in soil water monitoring.

Technical Abstract: Multisensor capacitance probes have shown great promises in irrigation scheduling, evaluating water needs of plants, estimating soil hydraulic properties, estimating groundwater recharge and infiltration losses, and other soil water-related fields. It is often beneficial to know how representative are measurements of soil water in the middle of an experimental plot for the average water content across the plot. The differences between MCP and plot-averaged water contents may arise because of systematic error in manufacturer MCP calibration and random variability of soil properties. The objective of this work was to assess the systematic and the random components of the difference between MCP and plot-averaged water content across 1 m2 plot of the heterogeneous coarse-textured soil under corn. The soil around four MCP was sampled in triplicate at the distance of 50 cm from the MCP on three dates with distinctly different water contents. Both systematic and random differences between MCP and plot-average were encountered. The manufacturer calibration led to the overestimation of small water content and probably to the underestimation of high water contents. The removal of this bias had only a effect on the overall difference that was apparently random. The dependence of the apparently random differences on the soil textural composition was demonstrated with regression trees. Correcting MCP measurements for the texture resulted in 20% decrease in root-mean-square difference to the value of 0.032 cm3 cm-3. Site-specific MCP calibration and evaluation of the random upscaling error can be a useful before using MCPs in soil water monitoring.