Normalized, Not Absolute: Transferable Prediction of SPT-N from Trend-Based Resistivity Descriptors for Resource-Efficient and Sustainable Site Investigation

Site investigation is costly, slow, and locally destructive, with boreholes spaced by budget rather than by ground variability. Portable electrical resistivity offers a rapid, low-disturbance alternative, but correlations built on absolute resistivity transfer poorly between locations. This study asks whether the limitation lies in the measurement or in its representation. In four boreholes in tropical sandy soils of the Khorat Plateau, Thailand (601–1729 m apart), Wenner-array resistivity was measured at 0.5 m depth increments alongside standard penetration tests (SPT) and index testing, yielding 63 paired observations. Each smoothed log-resistivity profile was described by three within-borehole quantities: the relative electrical state (x), its squared deviation (x2), and the transition intensity (g). SPT-N was predicted by linear regression under leave-one-borehole-out cross-validation. Absolute resistivity predicted SPT-N poorly (R2 = 0.133) and added nothing to depth alone (0.597 versus 0.617); the descriptors raised cross-borehole performance to R2 = 0.733 (RMSE 13.5 versus 16.2 blows), kept an advantage without smoothing (0.677), and laboratory indices transferred worst (−0.262). With only four boreholes, the gain is not yet statistically definitive (bootstrap 95% CI −0.04 to +0.26), so the study is presented as a proof of concept: the transferable information in a resistivity profile appears to lie in its shape rather than its magnitude, supporting borehole targeting and more resource-efficient site investigation.