Accuracy calibration and evaluation of capacitance-based soil moisture sensors for a variety of soil properties

Abstract

Accurate measurement of soil moisture (θ) is key to hydrology and agriculture research. Soil moisture sensor technology is the predominant method for measuring θ, and such measurements are used as a standard for evaluating results from remote sensing and data assimilation. Therefore, improving the θ measurement accuracy of soil moisture sensors is of great significance. This study used the capacitance-based soil moisture sensor (5TM, Decagon Devices, Inc.) as an example to illustrate the necessity of calibration. The 5TM soil moisture sensor calculates θ by measuring the dielectric constant (ε) of the soil medium, and ε is affected by soil properties (texture, salinity, soil organic matter, etc.). Consequently, a common calibration model (CCM) was developed to calibrate θ from the 5TM sensor by incorporating the soil properties using soil samples collected from 17 sites in 13 provinces in China at 4 different depths. First, the θ change experiments were conducted in the laboratory for each soil sample. Second, a linear calibration model (LCM) was applied to calibrate the 5TM measured soil moisture (θ5TM) based on “true” soil moisture (θtrue) assessed through the gravimetric method. The results indicated (1) high correlation coefficient (R=0.95) was found for θ5TM and θtrue, but with a high root mean square error (RMSE) of 0.051 m3m−3, and a more significant underestimation with increasing θ; (2) LCM calibration results (θLCM) showed a higher R (0.99) and a lower RMSE (0.017 m3m−3). Finally, the CCM was established through relating the LCM coefficients (aLCMandbLCM) and soil properties based on multiple regression, with RMSE of 0.126 m3m−3 and 0.023 m3m−3 for aCCMandbCCM respectively. The CCM calibrated result (θCCM) showed an RMSE of 0.02 m3m−3 and R of 0.98. CCM can almost replace LCM in terms of similar accuracy. In this study, a CCM for soil moisture sensors is proposed, which provides a new approach for soil moisture sensor calibration.