Evaluation of low-cost capacitive moisture sensors in three types of soils in the Cerrado, Brazil

Abstract

Soil moisture sensors enabled new automation and computerization solutions in agriculture. However, its use still requires prior calibrations to attest its accuracy to estimate soil moisture. Thus, in this study a prototype for soil moisture monitoring was developed using the capacitive sensor SKU:SEN0193 integrated with the Arduino microcontroller. The prototype was calibrated based on the relationship between volumetric moisture and sensor output voltage in deformed samples of Red-yellow Latosol (RYL), Regolitic Neosol (RN), and Red Latosol (RL) of the Cerrado region, Federal District, Brazil. The calibration of sensors resulted in a second-degree polynomial model with R² between 0.93 and 0.96 for RYL, 0.89 and 0.92 for RN, and 0.86 and 0.88 for RL. Root mean square errors (RMSE) of volumetric moisture were 0.08, 0.12, and 0.15 cm³.cm^-3 for RYL, RN, and RL, respectively. Sensors at operating voltage of 5.00 V showed less variability of readings, with coefficient of variation (CV) between 0.21 and 1.54%, while sensors at 3.30 V showed CVs between 3.27 and 17.84%. It is recommended to calibrate the SKU:SEN0193 for each soil type in order to obtain reliable estimates of water content for RYL, RN, and RL.