An Efficient Calibration Technique for Heat Dissipation Matric Water Potential Sensors

Abstract

Heat dissipation sensors are used to measure matric potential in soils. A van Genuchten equation can be used to fit the relationship between measured heat dissipation and matric potential. Calibration is required for each probe because of intrinsic variability in the properties of the porous material. However, calibration is time-consuming (months), requiring numerous measurements taken by a pressure plate apparatus over their operational range. Here the feasibility of minimizing the number of measurement points required to reliably characterize the calibration curve is explored. A two parameter (m = 1−1/n) and a three parameter (m is a free parameter) van Genuchten type model is used for the calibration. We explore how reducing the number of calibration measurement points impacts the resulting calibration curve, that is, what is the information content that each measurement provides, and how significantly is the calibration performance degraded by removing measurement points. We also consider how measurement errors during the calibration process, which are understood to be non-uniform over the range of matric potential values, result in uncertainty in the calibration curve, and explore how this uncertainty can be minimized. Different measurement locations (i.e., matric potential values) are found to contain different information content. A two-parameter-model calibration using four calibration points is recommended; and the specific location of these four points is essential to maximize the accuracy and the efficiency of the calibration. As a rule of thumb, the four points should be uniformly distributed on a log-scale over the pressure range from 20 to 1000 kPa.