3D Geostatistical Integration Based on Site-specific Correlations with Borehole data and Geophysical data

Abstract

The decision of 3D strata information is necessary in various constructions site. In order to determine the 3D strata information in the site, borehole datasets obtained by borehole tests are mainly used. Borehole data provides relatively clearer geotechnical information compared to other geotechnical investigation methods. However, it has a limitation in the insufficient acquisition of continuous geotechnical information in entire site. In contrast, geophysical tomography covers a larger area and is relatively continuous data due to be investigated by 2D cross section. However, geophysical tomography indicates the geophysical characteristics, which should be transformed into geotechnical information by empirical method. Therefore, it is possible to obtain more reliable strata information by combining borehole data and the geophysical measurements using geo-statistical methods. The borehole datasets are generally regarded as true value. However, even borehole datasets include outlier because of various uncertainties: the inherent soil variability, the measurement uncertainties, and the transformation uncertainties. Therefore, the borehole datasets have to be optimized to obtain reliable information. In addition, the borehole data and geophysical data have site-specific correlations which it makes difference in each site. Therefore, the site-specific variability must be considered to estimate reliable information. In this study, the 3D geo-statistical integration method, which combines borehole data and geophysical data, considers site-specific geotechnical variability was proposed. First, the borehole datasets are optimized to obtain reliable information by using outlier analysis based on cross validation. Then, the site-specific geo-layer criteria are determined to transform geophysical data into 3D strata. Finally, 3D strata information considering site-specific correlations in target area is determined by using 3D geo-statistics. In addition, the proposed method was applied to validate applicability on 3 testing sites in Korea.