Evaluation and Calibration of Soil Moisture Sensors in Undisturbed Soils
Gary Feng, Ruixiu Sui
Abstract
<b>Highlights</b> <p /> Evaluated and calibrated soil moisture sensors TDR315, CS655, and GS1 in both field and laboratory. The three sensors were tested in six predominant undisturbed soil types across Mississippi state. Sensors with built-in factory calibration overestimated soil moisture, performed better in sandy loam than clayey soils. It is imperative to calibrate sensors in situ or in undisturbed soils. <b>Abstract</b>. There is increasing interest and demand for growers and researchers to use soil moisture sensors for water management and research in the state of Mississippi. Previous studies demonstrated great challenges in the selection and deployment of sensors for accurate monitoring of soil moisture due to the high variability of soil types and properties. In this study, we evaluated and calibrated three soil moisture sensors, TDR315 (Acclima), CS655 (Campbell Scientific), and GS1 (Meter Group), on-site in fields and in undisturbed soil samples in the laboratory. We selected six predominant soil types from across Mississippi, including four clayey soils (Sharkey clay loam, Brooksville silty clay, Tunica clay, and Dowling clay) and two loamy soils (Commerce sandy loam and Leeper sandy loam) for the tests. Results indicated that these three types of sensors with built-in factory calibration generally overestimated the volumetric water content (θ<sub>v</sub>) of the soils in both field and laboratory. The θ<sub>v</sub> readings of these sensors were closer to the θ<sub>v</sub> determined by gravimetric method in the sandy loam than in the clayey soils. The three types of sensors basically met their factory root mean squared error (RMSE) criteria in loamy soils, but they varied in their performance<b>. </b>The RMSE of TDR315 ranged from 0.03 to 0.17, while CS655 and GS1 had greater RMSE, ranging from 0.03 to 0.29 cm<sup>3</sup> cm<sup>-3</sup>. Specific calibration of TDR315 and CS655 was conducted for each of the six soils using combinations of on-site field data and laboratory undisturbed soil data. Field validation on clay and sandy loam soils suggested that the accuracy of the soil water content measurements by CS655 and TDR315 was acceptable (R<sup>2</sup> > 0.6) with proper and undisturbed soil-specific calibration.