While Ridge regression demonstrated strong predictive performance in estimating soil resistivity from moisture, suction, and temperature data, incorporating pore-structure-informed features can substantially enhance this modeling framework. Future work will integrate empirical data from the 3D-printed soil pore volume to validate and refine model inputs, particularly bulk density and porosity proxies derived from compaction and root penetration studies. Using segmented CT images and printed soil analogs, hydraulic conductivity and resistivity tests will be conducted to establish direct pore-volume–resistivity correlations, offering an additional layer of interpretability for machine learning algorithms.
This direction opens a pathway toward multiscale soil characterization, where machine learning predictions based on bulk field data are grounded in physical models of pore connectivity and plant-root interaction. Moreover, the fabrication techniques demonstrated in related studies on composites and thermally modeled extrusion printing offer scalable and bioinspired strategies to simulate natural soil behavior (Akib et al., 2024; Rahman et al., 2023; Ufodike & Nzebuka, 2022). Ultimately, coupling machine learning with additive manufacturing and experimental root growth analytics holds transformative potential for applications of Enabling Rural Innovation (ERI) in sustainable agriculture, green infrastructure, and smart geotechnical design.
Authors:
Soil Resistivity at the Geo-Congress 2026, held March 9–12, 2026, in Salt Lake City, Utah, featured technical sessions, workshops, and GSPs (Geotechnical Special Publications). GSP 379, published by ASCE Library, covers topics such as soil properties, modeling, and computational geomechanics. The conference highlights geotechnical innovation, including AI in engineering and seismic design.
