This study evaluated the performance of three soil water content sensors and a soil water potential sensor. The evaluation was performed using &thetas; v data collected in the laboratory and in fields near Greeley, CO. Soil water content/potential values measured by the sensors were compared with corresponding values derived from gravimetric samples, ranging from the approximate permanent wilting point (PWP) to field capacity (FC) volumetric water contents. Calibration equations of sensor-measured &thetas;v were developed based on the laboratory and field data, and were compared with the factory-recommended calibrations. In addition, laboratory tests using an additional salt (calcium chloride dihydrate) concentration with varying soil water content were carried out to determine the effects of the soil bulk electrical conductivity on CS616, TDT, and 5TE sensor readings.
Under laboratory and field conditions, the factory-based calibrations of &thetas;v did not consistently achieve the required accuracy for any sensor. The MBE of the factory calibrations of &thetas;v for the CS616 sensors ranged from 0.032 to 0.337 m3 m −3 in the three soils. The factory calibration for the TDT sensors produced MBE values in the range of 0.007 to 0.061 m3 m−3 in the three soils. The MBE of the factory calibrations of &thetas;v for the 5TE sensors ranged from 0.004 to 0.024 m 3 m−3 in the three soils. The factory calibration for the Watermark sensors produced MBE values in the range of 0.082 to 0.200 m3 m−3 in the three soils.
Additional salt (calcium chloride dihydrate) concentrations in the laboratory caused the CS616 to give an error reading. Also, the higher concentrations increased the MBE of the factory calibration of the TDT sensor by 0.026 m 3 m−3 in the sandy clay loam (Site A), and 0.066 m3 m−3 in the clay loam (Site C), and increased the MBE of the 5TE sensors in the soils from Sites A and C by 0.172 m3 m−3 and 0.162 m3 m −3, respectively.
Field tests indicated that using the calibration equation developed in the laboratory to correct the data obtained by CS616, TDT, 5TE and Watermark sensors in the field at Site A were not consistently accurate in every treatment. However, they were more accurate than the factory calibration equations. Applying the laboratory-derived calibration equation developed for the CS625 sensors at Site B (loamy sand) was accurate at the 30- and 61-cm depths. However, using this equation resulted in an overestimation of &thetas; v by 0.032 m3 m−3 at the 91-cm depth. Using the laboratory equations developed for the Watermark sensors at Site B accurately measured &thetas;v at the 61- and 91-cm depths (RMSE = 0.014 and 0.024 m3 m−3, respectively).
Results from field tests at Sites A and B indicated that a linear calibration of the TDT and 5TE sensors (and a logarithmic calibration for the Watermark sensors) could reduce the errors of the factory calibration of &thetas; v to 0.020±0.035 m3 m−3 (2±3.5%). These tests also confirmed that each individual sensor needed a unique calibration equation for every soil type and location in the field. Furthermore, the calibrated van Genuchten (1980) equation was not significantly more accurate than the calibrated logarithmic equation.
Analysis of the &thetas;v graphs from the field data indicated that the CS616, 5TE and Watermark sensor readings were influenced by diurnal fluctuations in soil temperature, while the TDT was not influenced. Therefore, the TDT sensor was overall the most robust of the four sensors that were evaluated. Additionally, it is recommended that the soil temperature be considered in the calibration process of the CS616, 5TE, and Watermark sensors through either a correction equation or taking readings from the sensors during times that the soil temperature is similar (for example, every day at noon). (Abstract shortened by UMI.)
