Calibração de sensores FDR (Frequency Domain Reflectometry) para estimativa da umidade do solo

Abstract

A work was carried out at the Center for Agricultural and Environmental Sciences of the Federal University of Bahia, Cruz das Almas campus (12o40´S, 39o30´W), BA, aiming at the calibration of two FDR (Frequency Domain Reflectometry) sensors for estimating soil moisture. A weighing system with a capacity of 59 kg was developed and tested for use in gravimetry as a standard calibration method for FDR sensors to estimate soil moisture. The calibration model found was of the linear type (r2 = 0.999) and the repeatability, hysteresis and linearity errors were of the order of 0.119, 0.068 and 0.009% of the end of scale (EF), respectively. The limits of the accuracy error, which represents the overall performance of the system, were ±0.083% of FE. The calibration of the FDR probes was divided into two steps: laboratory and field calibration. The laboratory work consisted of removing monoliths from six depths (10, 30, 50, 70, 90, and 110 cm) and after saturation, their volumetric moisture (θ) was measured, through mass variation and the corresponding values of the D parameter (a dimensionless unit that expresses the variation in the frequency of the electromagnetic wave emitted by the sensors) of the two probes, obtaining ordered pairs and then equations with θ = f(D) were adjusted. In the field calibration, a trench was opened and ten access tubes for the probes were inserted, five on each side of the trench distributed equidistantly. The readings with the probes were carried out at the same depths as in the laboratory work and two undisturbed samples were taken, for gravimetric analysis, around the tube at the same depths as the probes were read. Ordered pairs ( D and ) were obtained, then models were fitted with θ = f(D). The results showed that the calibration of the probes in the laboratory was more adequate, as it presented better coefficients of determination at all depths and that the exponential model in the laboratory calibration presented a better performance than the linear one.