Playing in the sandbox: An experimental set-up for comparison of soil moisture profile sensors

To enable an efficient and economical use of limited water resources, sensing techniques to determine root zone soil moisture are gaining importance. Because of their easy handling and ability to provide simultaneous measurements in different depths, so-called soil moisture profile sensors (SMPS) exhibit high potential for climate-smart agriculture. However, determining soil moisture with reasonable accuracy is a complex task. Especially clay content and soil temperature influence the soil dielectric permittivity and might thus affect the electromagnetic soil moisture measurement of the SMPS.

To date, an accurate and easy-to-use method for the evaluation of long SMPS is not available. To this end, we designed a laboratory and a field experiment to better discriminate between changes in soil dielectric permittivity and sensor variability due to environmental effects. The tested SMPS are the SoilVUE10 (50 cm) from Campbell Scientific, the Drill&Drop (60 cm) from Sentek, as well as the SMT500 (50 cm), which is an early prototype from TRUEBNER. The following questions were addressed: (1) How high is the measurement variability of the vertical measurement sections of an SMPS? (2) How strong is the sensor response influenced by changes in temperature? (3) What is the SMPS accuracy compared to reference TDR measurements and how high is the sensor-to-sensor variability? We addressed questions 1 and 2 by placing the SMPS into a container filled with well-characterized fine to medium sized sand (type F36, Quarzwerke Frechen). The sand was water saturated and the temperature of the container was stepwise increased from 5 to 40 °C using a water cooling/heating. Question 3 was addressed by setting up a 2 x 2 x 1.5 m sandbox, also filled with F36 sand at a field site. The sandbox is sealed watertight to the sides and to the bottom and provided with a drainage layer of 20 cm gravel. The water level inside the sandbox can be controlled by pumping water in or out using piezometer tubes, which are permeable in the drainage layer. The SMPS were installed into the sandbox and the measurements were compared against reference measurements made using CS610 TDR probes with TDR100 (Campbell Scientific) and against SMT100 (TRUEBNER) TDT measurements.

Preliminary results using factory calibrations indicate that all tested SMPS have their shortcomings regarding the accuracy of soil moisture estimation. The D&D probe shows a high agreement between the measurement depths and a fair temperature stability, but the soil moisture content was underestimated compared to the reference measurements. In comparison, the SoilVUE10 displayed larger variability between different measurement depths, as well as between different sensors. In addition, the soil moisture was overestimated at high soil moisture content and the accuracy declined strongly above a soil temperature of 25°C. The SMT500, albeit a prototype, performed well at low soil moisture but strongly overestimated the soil water content under saturated conditions. Our experimental setup has generally proven useful for the characterization of SMPS. It clearly showed that the accuracy of the soil moisture estimates obtained with the SMPS is quite variable, especially at high soil moisture content.