Design and Performance Analysis of Split Ring Resonator-Based Sensor for Soil Moisture Content Characterization.

This paper addresses the need for compact, low-cost soil moisture sensors operating at low microwave frequencies that can provide accurate, texture-aware (i.e., sensitive to different soil particle-size distributions such as sand and loam) characterization of soil permittivity and moisture content. Conventional techniques and many existing microwave resonator sensors are constrained by limited penetration depth, relatively large or complex structures, and calibration procedures that do not robustly account for different soil textures and moisture ranges. A dual-port microstrip square split ring resonator (SRR) sensor on Rogers RO3010 (Rmit University, Melbourne, Australia) is designed for operation at 1.3 GHz and analyzed using full-wave 3D electromagnetic simulations. The structure employs a T-shaped feedline and a shunt quarter-wavelength matching section to achieve strong field confinement in the sensing region and effective impedance matching. Soil is modeled as sandy and loamy superstrates over practical agricultural moisture ranges, with their complex permittivities drawn from reference datasets. Empirical calibration models are then developed, including polynomial curve fitting between resonance frequency shift and real permittivity, machine-learning-based calibration using resonance frequency and transmission loss features, and multiple linear regression linking moisture content to both real and imaginary permittivity components. The sensor exhibits a resonance frequency shift of about 115 MHz over 0-30% moisture for sand and 0-40% for loam, with a maximum sensitivity of 3.4%. Calibration models achieve mean absolute error below 1.22%, root mean square error under 1.58%, and coefficients of determination R2 > 0.98 for both soil textures. These results demonstrate that a compact 1.3 GHz square SRR sensor with data-driven calibration, i.e., empirical models learned from simulated and measured S-parameters, enables sensitive, reproducible, and texture-aware soil moisture estimation suitable for agricultural and environmental monitoring.