Similarity quantification of soil parametric data and sites using confidence ellipses

This paper presents a confidence ellipse-based method to evaluate the similarity of soil parametric data using the database from the site investigation reports. Then, the obtained similarity assessment results of parametric data are used to further estimate the site similarity via two proposed strategies, namely the mean and weighted mean approaches. The former referred to the average of parametric data similarity degrees, while the latter was the weighted average, and the weight was calculated using the coefficient of variation (COV) of each parameter. For illustration, the liquidity index (LI) dataset was firstly used to explore the performance of the presented method in the evaluation of parametric data similarity. Subsequently, the site similarity was assessed and the effects of numbers and weights of selected parameters for study were systematically studied. Lastly, the transformation models about the relationships between Cc and x as well as between Cc and e0 were constructed to illustrate the application of the similarity analysis in reduction of transformation uncertainty. Results show that the greatest site similarity degree is at about 0.76 in this study, and the maximum decrease of transformation uncertainty can reach up to 18% and 25.5% as union parametric data similarity degree increases. Moreover, the site similarity degree represents the whole similarity between two different sites, and the presented union parameter similarity degree maintains a good agreement with transformation uncertainty. (c) 2021 China University of Geosciences (Beijing) and Peking University. Production and hosting by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/ licenses/by-nc-nd/4.0/).

Automated summary

This paper presents a confidence ellipse-based method to evaluate the similarity of soil parametric data, and further estimates the site similarity through two proposed strategies. The results show the effectiveness of this method in evaluating parametric data similarity and reducing transformation uncertainty.