Automated Low-Cost Soil Moisture Sensors: Trade-Off Between Cost and Accuracy

Abstract

Automated soil moisture systems are commonly used in precision agriculture and environmental monitoring. Using low-cost sensors, the spatial extension can be maximized, but the accuracy might be reduced. In this paper, we address the trade-off between cost and accuracy comparing low-cost and commercial soil moisture sensors. The analysis is based on the capacitive sensor SKU:SEN0193 under lab and field conditions. The laboratory tests aimed at evaluating the response speed, best supply voltage, temperature dependency, calibration, and applicability for controlled infiltration column tests (one meter high). Laboratory tests indicated that the sensor is temperature and voltage-sensitive. The use of 5.5 V as supply voltage for the sensors drastically reduced the correlation between output and degree of soil saturation, thus we suggest the use of 3.3V. Soil temperature had a negligible impact on the sensor output: 0.27% of soil saturation degree per degree Celsius. For field implementation, a low-cost monitoring station was built using Arduino as a microcontroller and tested during three months. The sensors could represent daily and seasonal oscillation in soil moisture resulting from solar heating and precipitation. In addition to individual calibration, two simplified calibration techniques are proposed: universal calibration, based on all 63 sensors, and a single-point calibration using the sensor's response in dry soil. The monitored wetting front was compared to the one estimated by Hydrus model and had a high correlation. The low-cost sensor performance was later compared to commercial sensors based on five variables: (1) cost, (2) accuracy, (3) qualified labor demand, (4) sample volume, and (5) life expectancy. Commercial sensors promote soil moisture measurements with high accuracy at a high acquisition cost. On the other hand, low-cost sensors, such as the SKU:SEN0193, provide data with medium accuracy at a very low acquisition cost, enabling spatial monitoring through multiple-point measurements. Thus, the use of the SKU:SEN0193 sensor is suggested in projects with budget limitations with short duration where there is a medium requirement accuracy or when the spatial variability of soil water content is considerable. Despite the physical fragility of the hardware used (sensors and monitoring station) and the lower accuracy when compared to other commercial sensors, this work demonstrates through the case study of the SKU:SEN0193 sensor the possibility of using low-cost technologies for monitoring environmental variables.