Soil & Water Res., 2020, 15(2):75-84 | DOI: 10.17221/227/2018-SWR

The laboratory calibration of a soil moisture capacitance probe in sandy soilsOriginal Paper

Marina Campora¹, Anna Palla^*,2, Ilaria Gnecco², Rossella Bovolenta², Roberto Passalacqua²

¹ Genoa, Italy

² Department of Civil, Chemical and Environmental Engineering, University of Genoa, Genoa, Italy

Determining and mitigating landslide risk is a technical-scientific objective, particularly for the protection and proper territorial management and planning. The slope stability depends on the pore pressure distribution, which is influenced by the saturation front propagation through the unsaturated zone, whose monitoring is useful to understand any possible instabilities. Such monitoring may be undertaken by sensors based on the measurement of the relative dielectric permittivity. Reliable relationships between the measurement and the soil moisture are necessary. The main objective of this study is to assess a laboratory calibration protocol for a specific capacitance sensor (Drill & Drop, Sentek Sensor Technologies). Two monogranular sands have been selected for the calibration purpose. The laboratory tests were performed under three relative density values (D_R equal to 40%, 60% and 80%) for seven volumetric water content values (θ_v ranging from 0.00% to 36.26%). Based on the experimental measurements, the soil-specific calibration curves were determined at an assigned relative density value; in particular, a simple power law is adopted to describe the probe's reading as a function of the volumetric water content. The results point out that the relative density values slightly affect the tests, thus, the soil-specific calibration curves are derived based on a simple regression analysis fitting the whole set of the laboratory tests validated for each sand. The calculated coefficient of determination (R² = 0.96/0.99) and root mean square error (RMSE = 1.4%/2.8%) values confirm the goodness of fit. In order to propose more general fitting curves, suitable for both the investigated sands, multiple linear regressions are performed by considering θ_v and the mean grain size, D₅₀ as independent variables; again, the R² and RMSE values equal to 0.97 and 2.41%, respectively, confirm the suitability of the calibration curve. Finally, the laboratory calibration curves are compared with the manufacturer-supplied curves, thus, enhancing the need for the soil-specific calibration.

Keywords: particle size distribution; regression analysis; relative density; soil water monitoring; volumetric water content

Published: June 30, 2020  Show citation