Smoothed Adaptive Weighting Average (SAWA) interpolation for geotechnical site characterisation using limited data

Accurate site characterisation is essential in geotechnical engineering, but in practice, it is often constrained by the limited availability of field data such as cone penetration tests (CPTs). When combined with the complex three-dimensional variability of geological materials, this limitation makes the reconstruction of reliable ground models a persistent challenge. Widely used interpolation methods, such as Kriging and Inverse Distance Weighting (IDW), involve trade-offs between accuracy and efficiency. Adaptive IDW (AIDW) incorporates data-density-based weighting but remains limited by reliance on globally defined data. To overcome these challenges, this study introduces the Smoothed Adaptive Weighting Average (SAWA) method, a robust spatial interpolation approach for efficient and accurate ground modelling from sparse data. SAWA employs an adaptively adjusted interpolation radius at each location, determined by local data density. This strategy enhances resolution in data-rich regions while ensuring reliable prediction in sparsely sampled areas. In doing so, SAWA directly addresses the growing demand for data-density-aware site characterisation, driven by increasingly large geotechnical datasets and emerging digital twin applications. Case studies on both synthetic and real-world data confirm its strong balance between accuracy and computational efficiency, establishing SAWA as a promising tool for scalable 3D ground modelling in data-limited practice.