From 2-D seismic refraction to 3-D subsurface characterization: Unpacking the role of univariate spatial interpolation techniques with borehole validation

Abstract

This study evaluates the accuracy of three spatial interpolation methods IDW-Anisotropic, IDW-Isotropic, and Closest Point in generating 3-D geological models from 2-D seismic refraction data, validated against borehole lithology and SPT N-values. Conducted at a site underlain by granitic formations and residual soils, the analysis aimed to determine the most reliable method for representing subsurface heterogeneity. Statistical assessment revealed that IDW-Anisotropic provided the best performance, with a mean velocity of 1649.45 m/s and standard deviation of 1429.76 m/s. Its model exhibited a right-skewed distribution (skewness = 1.23653) and leptokurtic nature (kurtosis = 1.02792), effectively capturing sharp velocity transitions associated with stratigraphic boundaries. In comparison, IDW-Isotropic showed lower variability (SD = 471.72 m/s), negative skewness (−0.58528), and high kurtosis (2.80095), resulting in over-smoothed subsurface representations. The Closest Point method yielded the highest mean velocity (3041.85 m/s) but exaggerated high-velocity zones, leading to inconsistent layer delineation. Validation with borehole data confirmed that IDW-Anisotropic most accurately aligned with lithological changes, particularly at depths where seismic velocities exceeded 2400 m/s, indicating the transition to unweathered granite. The findings show that directional interpolation significantly enhances the reliability of geophysical modelling in complex tropical weathering environments, providing critical support for geotechnical and subsurface decision-making.