Geophysical Methods for Geotechnical Parameters′ Estimation in Shallow Subsurface Investigations: A Review

Geotechnical investigations have traditionally relied on invasive methods, such as boreholes and laboratory tests. However, these techniques can be time-consuming, expensive and spatially limited, often resulting in discontinuous subsurface characterisation that fails to capture the full extent of soil heterogeneity. Geophysical methods are becoming increasingly integrated into conventional geotechnical investigations to overcome these spatial and economic constraints. This review examines how seismic, electrical resistivity and electromagnetic surveys provide rapid, nondestructive and spatially continuous subsurface data that can be used to infer geotechnical information across a range of shallow infrastructure and environmental applications, typically within the upper 30 m of the subsurface, at site-specific scale. Methods such as multichannel analysis of surface waves (MASW) estimate shear-wave velocity profiles, which can be related to soil stiffness and small-strain shear modulus through appropriate constitutive assumptions and site-specific calibration. Electrical resistivity tomography (ERT) is well suited to mapping moisture content distributions and detecting seepage anomalies in embankments and slopes. A key finding of this review is that the reliability of geophysical–geotechnical correlations is strongly parameter-dependent: Relationships grounded in physical theory, such as that between shear-wave velocity and small-strain shear modulus, are more robust and transferable than empirical correlations such as those between resistivity and SPT-N values, which tend to be highly site-specific. In all cases, reliability depends on integration with complementary geophysical methods and calibration against direct geotechnical measurements, since soil heterogeneity and environmental conditions can make interpretation ambiguous. Advanced data fusion approaches, including belief function theory and geostatistical interpolation, show practical potential for combining heterogeneous datasets and managing spatial uncertainty. Emerging techniques such as machine learning and Full waveform inversion (FWI) offer further possibilities for quantitative parameter estimation, though their routine application remains constrained by computational demands and the current lack of standardised field validation datasets. This review identifies current knowledge gaps and outlines directions for developing more reliable, quantitative frameworks for geophysical–geotechnical integration in infrastructure assessment and environmental applications.