Versatile simplistic correction of T-higrow sensors for improved soil moisture measurement accuracy

Abstract

Abstract The use of soil moisture sensors for irrigation can help reduce water and energy consumption and risks of groundwater contamination, which are essential aspects for pursuing sustainable development goals. However, increased adoption of this technology is limited by calibration requirements, technical complexities, and sensor costs. In this work, a simplified method for reducing the measurement error of a recently released low-cost soil sensor (T-Higrow) is presented. The method only requires measurements of a dry sample from the target soil, which are inputted into a simple correction formula to reduce the measurement error at higher moisture levels. The requirements of the proposed method are simple enough for most labs or extension services. This method was compared to the commonly used linear, polynomial, and logarithmic regression models based on repeated bench-scale experiments within 0%–35% moisture range in silt and sandy loam soils and in silica sand. Uncorrected sensor readings correlated well with soil moisture (r: 0.94–0.98), but with significant overestimation (25%–60% error). The simplified correction method showed comparable error reduction to regression models across all soil types. All methods reduced error down to 2%–10% (0.02–0.1 cm3 cm−3) and maintained high correlations (r > 0.94), except for logarithmic regression which reduced correlation by around 3%. Variability amongst sensor measurements was generally low (Standard Deviation: 0.01–0.03) particularly at moisture ranges below 20%, this was also the case for sensor-to-sensor variability (Standard Deviation: 0.01–0.03). Sensor evaluation and calibration works are needed to increase the accessibility to this technology for improved water and energy conservation.