WHAT DOES THE SOIL PARAMETER ESTIMATED FROM A TRANSFORMATION MODEL REALLY MEAN?

Jianye Ching

《Journal of GeoEngineering》 13卷3期 (2018/09) Pp. 105-113

https://doi.org/10.6310/jog.201809_13(3).2

摘要

In practice, a transformation model is frequently used to estimate the design soil parameter at a design site. A possible concern is that a transformation model is typically constructed by non-site-specific data points. The purpose of this paper is to address the question: Can these non-site-specific data points be used to derive any meaningful site-specific estimate for the design site? Our analysis results show that the design soil parameter estimate obtained from the transformation model is meaningful, in the sense that the resulting 95% confidence interval (CI) is close to genuine, as long as the design site and training sites belong to the same population. In this study, the term ＂same population＂ may not only be defined based on geographical locations, but it can also indicate similar soil deposition histories as well. A more controversial scenario is that the design site and training sites do not belong to the same population. In this case, the resulting 95% CI is not genuine. In the case that the design site population is a subset of the training site population (e.g., the Finland population is a subset of the general population), the resulting 95% CI is conservative.

關鍵字

Transformation model ； design soil parameter ； confidence interval ； uncertainty