A cost-effective capacitive sensor system, SEN0193, was ruggedised, calibrated, and evaluated in both laboratory and field settings to assess sensor-to-sensor variability resulting from sensor placement in the soil. Experimental data underwent regression analysis to develop a model for predicting soil moisture levels using the output voltage of the capacitive sensor, as recorded at the Analog-to-Digital Converter (ADC) of the microcontroller. The accuracy of the developed low-cost sensor system was demonstrated through the evaluation of Mean Absolute Error, Root Mean Square Error, and Relative Absolute Error, yielding values of 1.56%, 0.36, and 0.65, respectively. A comparison was conducted between the field-calibrated soil sensing system and a commercial SM150T sensor to measure Volumetric Water Content (VMC) in a sugarcane field. The Spearman rank correlation coefficient for Volumetric Water Content (VMC) prediction using the new low-cost sensor and the commercial SM150T sensor exceeded 0.98, indicating a strong and positive correlation between the two sensors' readings. Throughout the entire field testing period, the low-cost capacitive soil moisture sensor system exhibited consistent and reliable performance, with no practical issues reported. After soil-specific calibration, the low-cost capacitive sensors in the group demonstrated performance on par with commercially available sensors. This finding suggests that these sensors can be efficiently employed for irrigation management, with minimal impact on irrigation efficiency.
Raheja et al. (Wed,) studied this question.