Evaluation of Various Spatial Interpolation Techniques for Generating Synthetic CPT Data Profile

Abstract

In order to incorporate the influence of collected in-situ data, the spatial correlation between the data and the foundation needs to be explored. Statistical information of the soil property can be estimated from available field data obtained from testing at discrete locations across the site. In this research, several well-established spatial interpolation methods like ordinary kriging (OK), simple kriging (SK), inverse distance weight (IDW), spline, natural neighbor (NaN), and universal kriging (UK) were incorporated to evaluate the best method for generating synthetic cone penetration test (CPT) data. To remove the spikes, continuous five points averaging was done to generate the smoothed tip resistance. For the analysis, the spatial interpolation was performed in each foot (depth wise). Six CPT cases were investigated in this study. According to the results, four out of six cases, if the first priority is given to bias factor followed by coefficient of variation (COV) and root mean square error (RMSE), the best three spatial interpolation techniques are IDW, OK, and SK sequentially, based on their performance. For the other two cases, in one case, the best three spatial interpolation techniques are OK, IDW, and SK, sequentially, and the other case shows SK, IDW, and OK sequentially are the best three spatial interpolation techniques.